‘7‘5‘ 3? my a 0 Statistical Study of Selected Trace Elements with Reference to Geology and Genesis of the Carlin Gold Deposit, Nevada GEOLOGIC ‘w Statistical Study of Selected Trace Elements with Reference to Geology and I Genesis of the Carlin Gold Deposit, Nevada By MICHAEL HARRIS and ARTHUR S. RADTKE GEOLOGICAL SURVEY PROFESSIONAL PAPER 960 UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON : 1976 mm? S" _ . UNITED STATES DEPARTMENT OF THE INTERIOR THOMAS S. KLEPPE, Secretary GEOLOGICAL SURVEY V. E. McKelvey, Director Library of Congress Cataloging in Publication Data Radtke, Arthur S. 1936— Statistical study of selected trace elements with reference to geology and genesis of the Carlin gold deposit, Nevada. (Geological Survey Professional Paper 960) Bibliography: p. 20—21. Supt. of Docs. no.: I 19.161960. 1. Trace elements~Statistical methods. 2. Geochemistry—Nevada—Carlin. 3. Gold ores—Nevada—Carlin. 1. Harris, Michael, 1949— joint author. II. Title: Statistical study of selected trace elements... III. Series: United States Geological Survey Professional Paper 960. QE516.T85R3 553'.4l'0973l6 » 76—608006 For sale by the Superintendent of Documents, US. Government Printing Office Washington, DC. 20402 Stock Number 024-001—02771—2 — ‘, CONTENTS Page Abstract ______________________________________________________________________________ 1 Introduction __________________________________________________________________________ 2 Acknowledgments ________________________________________________________________ 2 Geologic setting ______________________________________________________________________ 3 Genesis of the ores ____________________________________________________________________ 3 Geology of the ore bodies ______________________________________________________________ 7 Statistical analysis of gold, mercury, arsenic, and antimony ______________________________ 8 Statistical analysis of gold, barium, copper, molybdenum, lead, and zinc __________________ 16 Statistical analysis of gold, boron, tellurium, selenium, and tungsten ____________________ 18 References cited ______________________________________________________________________ 20 ILLUSTRATIONS Page FIGURE 1. Index map of the Carlin gold deposit _______________________________________________________________________________ 3 2. Simplified geologic map of the Carlin gold deposit ___________________________________________________________________ 4 3. Diagram showing mineralization paragenesis at the Carlin gold deposit _____________________________________________ 6 4. Direct scatter plots of gold, mercury, arsenic, and antimony in unoxidized Carlin ores _________________________________ 10 TABLES Page TABLE 1. Data on elements studied and analytical procedures _________________________________________________________________ 2 2. Abundance of gold, mercury, arsenic, and antimony in fresh limestones and unmineralized carbonate host rocks and unoxi- dized gold ores at the Carlin deposit ___________________________________________________________________________ 8 3. Mean values and standard deviations for gold, mercury, arsenic, and antimony in the West, Main, and East ore bodies at the Carlin gold deposit ___________________________________________________________________________________________ 11 4. Linear correlation coefficients between gold, mercury, arsenic, and antimony for the West, Main, and East ore bodies at the Carlin gold deposit _______________________________________________________________________________________ 11 5. Abundance of gold, barium, copper, molybdenum, lead, and zinc in fresh limestones and unmineralized carbonate host rocks and unoxidized gold ores at the Carlin deposit ___________________________________________________________ 17 6. Mean values and standard deviations for gold, barium, copper, molybdenum, lead, and zinc in the West, Main, and East ore bodies at the Carlin gold deposit-____-________-__________________-____-_____-___-_____________-7 _______________ 17 7. Linear correlation coefficients between gold, barium, copper, molybdenum, lead, and zinc for the West, Main, and East ore bodies at the Carlin gold deposit _______________________________________________________________________________ l7 8. Abundance of gold, boron, tellurium, selenium, and tungsten in fresh limestones and unmineralized carbonate host rocks and unoxidized gold ores at the Carlin deposit _________________________________________________________________ 19 9. Mean values and standard deviations for gold, boron, tellurium, selenium, and tungsten in the West, Main, and East ore bodies at the Carlin gold deposit _______________________________________________________________________________ 19 10. Linear correlation coefficients between gold, boron, tellurium, selenium, and tungsten for the West, Main, and East ore bodies at the Carlin gold deposit _______________________________________________________________________________ 19 III STATISTICAL STUDY OF SELECTED TRACE ELEMENTS WITH REFERENCE TO GEOLOGY AND GENESIS OF THE CARLIN GOLD DEPOSIT, NEVADA By MICHAEL HARRIS and ARTHUR S. RADTKE ABSTRACT Linear regression and discriminant analyses techniques were applied to gold, mercury, arsenic, antimony, barium, copper, molybdenum, lead, zinc, boron, tellurium, selenium, and tungsten analyses from drill holes into unoxidized gold ore at the Carlin gold mine near Carlin, Nev. The statistical treatments employed were used to judge proposed hypotheses on the origin and geochemical paragenesis of this disseminated gold deposit. The West, Main, and East ore bodies of the Carlin deposit are in the upper 265+ m (869 ft) of the Silurian and Early Devonian Roberts Mountains Formation in the northeast corner of the Lynn window near the crest of a large northwest-plunging anticline. The intensely deformed carbonate host rocks are in an area containing many intersecting high-angle faults and breccia zones that were the principal channelways for the ore solutions. The major faulting and igneous activity in the mine area took place before the ore formed. The ore solutions dissolved calcite out of the limestone host rock and introduced silica, pyrite, gold, and various other elements. The initial calcite removal created permeable sections that were to be the more receptive sites for ore deposition. Most of the gold and mercury precipitated during this phase is either associated with organic materials or occurs together with arsenic and antimony as coatings on pyrite grains. A subsequent mineralization (possibly a later stage of the same mineralization) precipitated arsenic, antimony, and mercury sulfides followed by the deposition of copper, lead, zinc, and molybdenum sulfides and an overlapping formation of barite veins. The two sulfide stages and the barite veins were deposited independently of the gold. Geologic relations and statistical compari- sons showed that the barite veins were formed after the gold deposition and are either associated with arsenic and (or) antimony sulfides, base-metal sulfides, or have no associated sulfides. Many analyses of barite veins performed to date have shown no detectable amounts of gold. The linear correlation coefficients between element pairs and suites may or may not reflect the direct cause and effect relations of the mineralization sequence since it is most probable that many parameters controlled their precipitation. However, most of the correlations found agreed with geological observations, and some correlations revealed subtle geochemical relations that had not been previously recognized. The gold, mercury, arsenic, antimony suite of metals in the West ore body conformed quite well to the proposed linear model with an almost 80 percent reduction in variance and linear correlation coefficients as high as 0.87. We believe that the West ore body most closely approximates original conditions of the earliest mineralization phase. This opinion is substantiated by the scarcity of late arsenic, antimony, and mercury sulfides, by the lower organic carbon content, and by the simpler structural setting of this ore body. The distribution and interrelation of gold, mercury, arsenic, and antimony between and in the three ore bodies were compared by stepwise discriminant analyses. The results of these tests consistently showed the West ore body to be significantly different from the other two. The mean values in the Main ore body tended to be much more similar to the West ore body than to the East. The polarity of the element values in the West ore body from the other areas is in direct accord with the proposed mineral paragenesis model in which the late mercury, arsenic, and antimony sulfide mineralization occurred only in the vicinity of the Main and East ore bodies. Gold and mercury values showed high correlations in all of the areas tested. These correlations support the hypotheses that (1) gold and mercury were both transported in the same are solution and were contemporaneously precipitated in response to similar influences, and (2) more than 90 percent of the investigated occurrences of both elements are either associated with various organic materials or occur together on the surfaces of pyrite grains. Statistical analysis of a gold, barium, copper, molybdenum, lead, zinc suite in the West ore body gave results consistent with the proposed paragenetic model. The highest correlations were between the base metals, followed roughly by these elements to barium, and finally very low or not significant correlations of gold to barium and to the base metals. Gold was found to be negatively correlated to barium over the entire deposit; this relation reflects their paragenetic separa- tion. Discriminant analysis techniques employed for this suite recon- firmed that statistically meaningful geochemical differences exist between the three ore bodies. Mean values in the Main ore body were similar to the mean values in its neighbor bodies, whereas the mean values of the two peripheral areas were fairly polarized toward their own means. Linear regression analyses of gold with the independent variables boron, tellurium, selenium, tungsten suite show that data for these elements fit a linear model poorly. The only significant correlation found was between gold and tellurium; this correlation is strongest in the East ore body and becomes progressively weaker through the Main and West ore bodies. The mode of tellurium occurrence has not been firmly established, but tellurium and possibly selenium could be present in hydrothermal pyrite or could coat pyrite grains together with gold, mercury, arsenic, and antimony. Discriminant analyses of this suite again showed the West ore body to be significantly different from the other two. There are still a multitude of unanswered questions concerning the genesis of the Carlin deposit. The significant standard deviations found make it somewhat hazardous to draw exact conclusions from the statistical findings alone. However, the statistical methods employed were very useful in establishing guidelines to interpret data and were also valuable in lending support to theories on the ore genesis. 1 2 STATISTICAL STUDY OF TRACE ELEMENTS, GEOLOGY AND GENESIS, CARLIN GOLD DEPOSIT INTRODUCTION During the course of the geologic and geochemical study of the Carlin gold deposit, which began in 1968, a large amount of chemical data has been gathered on the host rocks and gold ores. Although most of these data were summarized in a paper by Radtke, Heropoulos, Fabbi, Scheiner, and Essington (1972), none of the data had been treated statistically. This paper presents the results of a statistical study made by Harris (1974), of certain minor elements in the unoxidized gold ores, together with basic geologic, mineralogic, and geochem- ical information provided by Radtke. Statistical data are applied to evaluate certain aspects of the chemical model of the Carlin deposit. Only analyses of samples of primary unoxidized ore were used, as it is likely that highly variable supergene oxidation processes would produce complicated elemental distribution patterns different from those produced by the original hydro- thermal conditions. Standard stepwise regression analyses, incorporating computer treatments shown in Dixon (1964), were used to determine correlations between the important ele- ments and the dependence of the gold content on the presence of these other elements. Linear correlations were emphasized in order to establish direct corre- spondences between the elements involved. Ahrens (1954), Lepeltier (1969), and Miesch (1967) gave reasons for treating geochemical data by using the lognormal equivalents of the numerical values found by analysis. We decided against using this approach for two main reasons: (1) Many of the analyzed values for some elements were reported as zero, so that by defini- tion no log value could be assigned. We considered add- ing a constant, such as +1, to all the values, but so far we are unconvinced that it would be statistically valid to add + 1 to zero and also add + 1 to a very large number in the same population; (2) An attempt to use logs on some of the more receptive populations did produce more accurate predictor equations and more dramatic correlations, but the relative magnitudes of the latter remained unchanged from the order given in this text. Since the main purpose was to use statistical correla- tions to delineate geochemical associations, we felt jus- tified in relying on linear functions. All data used in this study were obtained from various types of analyses of composite samples representing 1.53-m (5-ft) intersections in rotary drill holes that penetrated mineralized carbonate rocks below the zone of oxidation. The drill holes were made at 15.3-m (50-ft) intervals at the intersections of north-south and east-west gridlines and represent development drilling carried out to establish the volume and grade of ore throughout the entire deposit. In order to assure coverage of the entire deposit, individual drill holes TABLE 1.—Data on elements studied and analytical procedures [Spectrographic results are reported to the nearest number in the series 1.5, 1.0, 0.7, 0.5, 0.3, 0.2, 0.15, 0.1; analytical error is approximately one standard deviation at plus or minus one reporting interval] Number of Samples Element Methods of $12333? analys1s Analyzed Containing (percent) detectable amounts Au ........ Fire assay ,,,,,,,,,,,,,,,,,,,,,,,,,, 292 292 2— 50 Atomic absorption ,,,,,,,,,,,,, 292 292 2— 50 Hg ,,,,,,,, Instrumental meter __239 230 10— 50 As ,,,,,,,, X-ray fluorescence W 80 80 3— 10 Colorimetric ,,,,,,,, 159 159 10—100 Sb ........ X-ray fluorescence __ 80 80 1% 10 Colorimetric ,,,,,,,, 159 159 1(L100 Ba iiiiiiii Spectrographic ______ 292 281 , X-ray fluorescence __ 80 80 .3 10 Cu ________ Spectrographic , , _ 292 292 Spectrographic , , i 292 198 Spectrographic i. i 292 274 .,,Atomic absorption ._ 212 176 2~ 50 X—ray fluorescence .. 80 80 3* 10 B __________ Spectrographic ...... , 288 286 W ,,,,,,,,,, Colorimetric _____ “288 168 25»100 Se ,,,,,,,, Colorimetric ,,,,,,,,,,,,,,,,,,,,,, 288 145 20—100 Te ,,,,,,,, Atomic absorption ,,,,,,,,,,,,,,,,,, 288 89 5—100 used for geochemical studies were selected from combined geologic and drill location maps. For each of the 96 holes chosen, assay and lithologic logs were used to help select representative materials from the top, middle, and bottom of each ore zone to achieve a complete three-dimensional sampling of the entire unoxidized part of the ore deposit. In general at least two samples were chosen from each hole. Elements included in the study, methods of analyses, total number of samples analyzed, and the number of samples containing detectable amounts of each element are given in table 1. All analyses were performed in various laboratory facilities of the U.S. Geological Survey in Menlo Park, Calif, and Denver, Colo., as well as at the U.S. Bureau of Mines Experimental Station in Reno, Nev. ACKNOWLEDGMENTS We wish to thank the management of Carlin Gold Mining Co. for their cooperation and assistance in the study. Robert Akright, former resident geologist at the mine and presently associated with Occidental Miner- als Corp., and Dan Higley of Carlin Gold Mining Co., deserve special thanks for their efforts in preparing the samples and for assistance in the interpretation of rotary drilling information and logs of critical indi- vidual drill holes. We gratefully acknowledge the help of the many individuals in various laboratories of the U.S. Geologi- cal Survey as well as at the U.S. Bureau of Mines Experimental Station, Reno, Nev., who carried out the analyses that made this study possible. Numerous conversations with Professor F. W. Dickson, Department of Geology, Stanford University, helped to establish chemical parameters for the origin of the Carlin deposit. Professor P. Switzer, Departments of GENESIS OF THE ORES 3 Winnemucca n! s 0 n In 0 m :1 p. l‘ EUREKA COUNTY FIGURE 1.—The location of the Carlin gold deposit in north—central Nevada. Statistics and Geology, Stanford University, generously criticized the preliminary statistical work, for which we are very appreciative. GEOLOGIC SETTING The Carlin gold deposit, the largest of the dissemi— nated replacement-type gold deposits found to date, is located about 64 km northwest of the town of Elko in north-central Nevada (fig. 1). Carlin Gold Mining Co., a wholly owned subsidiary of Newmont Mining Corp., operates the mine, which currently produces about 2,000 tons of ore per day from three open pits. General information on the deposit has been pre- sented in papers by Hardie (1966), Hausen and Kerr (1968), and Radtke and Scheiner (1970). A detailed geologic map of the deposit was published by Radtke (1973) and additional information on the structural and stratigraphic setting was included on a regional map of the Lynn mining district (Radtke, 1974). For the purpose of this paper, a simplified map of the deposit (after Radtke, 1973) is shown as figure 2. Most of the Carlin ore bodies are in the upper 265 + m of the Silurian and Early Devonian Roberts Mountains Formation, which is overlain by the Devonian Popovich Formation (Akright and others, 1969). The Popovich Formation, composed of 80—120 m of alternating thick- and thin-bedded silty limestone, is truncated above by the Roberts Mountains thrust and is overlain by inter- bedded shale, chert, and limestone of the Ordovician Vinini Formation. Host rocks for the Carlin deposit, together with other rocks of the lower plate, are exposed in the Tuscarora Range. In the central part of the Tuscarora Range uplift, folding, and high-angle faulting accelerated erosion and removal of the upper plate rocks to form the Lynn win- dow. The Carlin deposit is located in the northeast corner of the Lynn windowinear the crest of a large northwest-plunging anticline where carbonate host rocks have undergone intense faulting and brecciation. Within the mine area, high-angle faults and breccia zones served as main channelways for ore solutions and as structural controls for ore deposition. Radtke (1973) showed three sets of high-angle faults including an early east-west-trending set, a later north- to north- west-trending set, and a subsequent northeast-trending set. The presence of Late Jurassic or Early Cretaceous granodiorite dikes along several northwest-trending faults and one east—west fault indicate that these faults predate the dikes; in many areas the dikes are offset by northeast-trending faults confirming the later age of these structures. These relations plus the intense hy- drothermal alteration of the dikes and the distribution of the ore, strongly suggest that the dikes and most of the faulting predate the ore formation (Radtke, 1973) and that the hydrothermal ore solutions were chan- neled along preexisting faults. GENESIS OF THE ORES The ore bodies at the Carlin deposit were formed where hydrothermal solutions moving along fault and breccia zones penetrated outward into thin-bedded ar- gillaceous carbonate rocks of the Roberts Mountains Formation. Reactions between the ore solution and the host rock composed mainly of calcite, dolomite, illite, and quartz, dissolved the calcite, deposited hydrother- mal silica and pyrite, and introduced gold and various other elements (Radtke and Scheiner, 1970). A generalized sequence for the hydrothermal develop- ment of the Carlin ores is shown in figure 3. Petrographic examination of host rocks and ores indi- cates that the initial penetration of the hydrothermal fluids resulted in the removal of calcite causing a marked increase in permeability of the rocks. This per- meability, in turn, had a strong influence on the de- Upper Jurassic and Silurian and Lower Devonian STATISTICAL STUDY OF TRACE ELEMENTS, GEOLOGY AND GENESIS, CARLIN GOLD DEPOSIT Gold ore Includes both oxidized and un- oxidized ore. Most unoxidized ore is not exposed Granodiorite dike Lower Cretaceous Unconformity UPPER PLATE 0F ROBERTS MOUNTAINS THRUST Vinini Formation Predominantly tan to gray, thin- bedded, interlayered siliceous shale, chert, and limestone EXPLANATION LOWER PLATE 0F ROBERTS MOUNTAINS THRUST Popovich Formation Dark-gray, medium- to thick-bedded siliceous dolomitic limestone. Roberts Mountains Formation Medium-gray, thin-bedded, sili- ceous, dolomitic limestone 11w TRUE NORTH >- M S H a: {1-} I— 3 a a 332% “<55 ‘3 a: o z 5 z o > [-1-] r: J 2 Egg $25 ._1<> a a 2 S. 2 >> o n a: o Contact .A. A. .A. .L. .A. Roberts Mountains thrust High-angle normal fault w Strike and dip of beds Outline of pit Solid line denotes top of pit wall; dashed line denotes toe of pit wall. FIGURE 2.—Simplified geologic map GENESIS OF THE ORES 200 400 600 800 1 000 F E ET |||||l|||| 0 | F | | | | | | o 100 200 300 METR ES of the Carlin gold deposit. 6 STATISTICAL STUDY OF TRACE ELEMENTS, GEOLOGY AND GENESIS, CARLIN GOLD DEPOSIT Barite Z Quartz + pyrite + 3 Au + H9 + As + Sb ; U <( 2’ As+Sb+Hg Cu+Pb+Zn+Mo E sulfides sulfides Lu I I— O I D >. I {—EARLY TIME LATE-> FIGURE 3.—Mineralization paragenesis at the Carlin gold deposit. velopment of zones favorable for the movement of later ore-bearing solutions. The main period of gold deposition followed as the ore solution removed additional amounts of calcite and small amounts of dolomite, and deposited silica, pyrite, arsenic, antimony, mercury, and gold. Most of the gold and mercury deposited from the solution during this phase either is associated with organic materials or, along with arsenic and antimony, occurs as coatings on pyrite grains (Radtke, Taylor, and Christ, 1972). On the basis of electron microprobe studies, Radtke, Taylor, and Christ (1972) distinguished four major modes of gold occurrences in the unoxidized ore: (1) as scattered particles along the rims of amorphous carbon grains, (2) as rather homogeneously distributed parti- cles in other kinds of carbon compounds, or as gold—organic compounds, (3) as coating on pyrite surfaces along with arsenic, antimony, and mercury, and (4) rarely as free native particles. They found the mercury in the unoxidized ore to occur in three principal ways: (1) in association with carbonaceous matter, (2) as a coating on pyrite grains, and (3) as fine-grained Cinnabar. Any mechanism postulated to explain the transport of gold must be compatible with properties of a hydro- thermal solution that carried mercury, arsenic, anti- mony, and quartz. Helgeson and Garrels (1968) used thermodynamic evidence to argue for gold transport as the aurous chloride species, AuC12_, in acid ore solu- tions. Later Radtke and Scheiner (1970) attempted to use this model to explain many of the observed chemical and mineralogical features of the Carlin deposit. Weissberg (1970) suggested that gold could be trans- ported in alkaline sulfide solutions, and Seward (1973) showed that neutral to slightly acidic solutions contain- ing HS‘ ion would be capable of moving gold under geologically feasible conditions. Recent work by Rytuba and Dickson (1974) indicates that in a reducing environment, the chloride ion is not effective in transporting gold. Their experimental studies also suggest that at elevated temperatures in a NaCl—H20 solution reacted with pyrite and quartz, gold is in solution as a gold-sulfide complex, presumably Au(HS) 2“. Arsenic and antimony sulfides and silica are trans- portable in neutral aqueous solutions lacking excess sulfur or chloride ions (Dickson and Tunnell, 1968; Weissberg and others, 1966; Kennedy, 1950). However, so far as is known, mercury sulfide is extremely insolu- ble in any naturally occurring solutions other than slightly alkaline (pH 8 to 9) solutions containing dis— solved sulfur (Dickson and Tunnell, 1968). Taking into account the restrictions imposed on the nature of an ore solution capable of simultaneous transport of gold, sulfides of mercury, arsenic, and antimony, and silica, it is reasonable to conclude that the hydrothermal solu- tion responsible for developing the main stage of the Carlin ores was weakly alkaline and contained excess dissolved sulfur. Although quartz veinlets are very uncommon at Car- lin, a few containing small amounts of pyrite, gold, fluorite, and frankdicksonite have been found. Frankdicksonite, BaFg, was found and described as a new mineral by Radtke and Brown (1974). Locally the ore solution was trapped in open frac- tures, and sulfide minerals of arsenic, antimony, and mercury were formed. Important minerals containing these elements are orpiment, realgar, stibnite, and cin- nabar. An apparent segregation exists between sulfides or arsenic and stibnite. In areas where orpiment and realgar are concentrated, only very small amounts of stibnite are present and antimony may substitute for arsenic in the structure of orpiment (Radtke, Taylor, and Heropoulos, 1974). In contrast, stibnite is concen- trated in places that are deficient in arsenic sulfides. The position of barite in the paragenesis of the deposit has been a subject of much speculation and uncertainty. Hardie (1966) initially concluded that barite occur- rences have no correlation with the presence of gold in the mine area. Hansen and Kerr (1968) regarded the barite, together with the sphalerite, galena, and part of the pyrite, as forming in the Cretaceous, but the gold and other minerals forming in the Tertiary. Radtke and Scheiner (1970) suggested that barite was transported in and deposited from the gold-bearing hydrothermal solution, thus making the formation of barite veins and the deposition of gold roughly contemporaneous. De- tailed geologic mapping suggests that most of the barite veins formed after the main phase of gold deposition and are either associated with arsenic sulfides or base-metal sulfides or are barren and deficient in sulfides (fig. 3). Results of stable isotope studies and considerations of the chemical model for ore genesis at Carlin support the GEOLOGY OF THE ORE BODIES 7 fact that barite formed late in the paragenetic sequence (Dickson and others, 1975). Laboratory studies have shown that small amounts of various sulfide minerals of copper, molybdenum, lead, and zinc occur both in the mineralized limestones and locally in barite veins. These four elements show no association with gold either on the surfaces of pyrite grains or chemically combined with carbonaceous materials. They apparently occur only in the form of discrete sulfide minerals that formed late in the hydrothermal episode after gold deposition. Small grains of scheelite randomly scattered through the mineralized limestones account for the tungsten in the ores. Boron occurs in the clay minerals. The form in which the low amounts of selenium and tellurium occur has not been determined. An important unresolved problem at Carlin is the paragenetic relation of the mineralization with respect to time. The locus of the mineralizing activity varied with time not only in its intensity and chemistry but also in space. For example, mercury, one of the most volatile elements, can be transported aqueously as Hg++, Hg2++, or Hg° but only as Hg" and HgClZ, in a gaseous phase. Silica and gold (so far as is known) can only be transported in solution and not as a vapor. Thus, to find an expression for the volatility of the elements in the rocks penetrated by mercury, it is necessary to examine the distribution pattern of mercury not as- sociated with gold (that is, mercury not introduced in the aqueous phase). Unfortunately, we could not find a feasible way to do this as almost all samples contained detectable amounts of both elements. The time distribution of the hydrothermal activity seems to fall into three rather broad categories covering early, intermediate, and late stages. This report deals with the unoxidized ores presumably derived from the early and perhaps part of the intermediate stages. How- ever, much of the oxidized zone may have formed from processes for which these early stages are responsible rather than exclusively from a secondary supergene weathering by downward percolating meteoric waters, although these certainly did have an effect. The domin- antly aqueous, ascending, early mineralization phase most probably ranged in temperature from a quite low value to the maximum value achieved. If boiling were induced, H28 vapor that would be generated would stream upward and become oxidized to H2804. This phase would cause the overlying rocks to be leached and altered. The liquors derived from this process would then descend, adding sulfate to the subsurface waters and acidifying them, causing the deposition of the late sulfide-bearing sulfate and carbonate veins. It is also possible that descending meteoric waters during this stage were superheated and charged with volatiles, and the waters thus also rose and joined in the oxidation action. GEOLOGY OF THE ORE BODIES The Carlin gold deposit is made up of three separate yet structurally and stratigraphically related ore bodies, each of which contains anomalous concentra- tions of the 13 elements studied statistically. The general shape and dimensions of the West, Main, and East ore bodies are shown in figure 2. The West ore body is almost veinlike in form, strikes roughly east-west, dips steeply to the north, and follows the hanging-wall side of a normal fault with the same attitude (fig. 2). This ore body has strong structural controls, contains the smallest amounts of organic carbon of any of the ore bodies, and lacks visible concentrations of arsenic, antimony, and mercury sulfides. The Main ore body is actually a northeast-trending zone roughly 915 m long within which the gold content varies widely. This zone contains numerous individual ore bodies, the positions of which are related to sets of north- and northeast-trending normal faults. In the northeast part of the Main ore body, stratigraphic control becomes more dominant; individual ore bodies up to 60 m thick with attitudes similar to the host rocks dip about 30°—33° to the north under Popovich Hill (fig. 2). The Main ore body is characterized by large amounts of unoxidized ores of widely varying silica and pyrite content.,The organic carbon content, which is not uni- form, ranges from about 0.1~0.8 weight percent. Sulfide minerals of arsenic, antimony, and mercury, and barite veins, some of which have associated base-metal sulfides, are locally abundant. The East ore body is a general term used to describe a gold-bearing zone made up of many individual ore bodies that occur within a stratigraphic interval of 100—150 In near, but not at the top of, the Roberts Mountains Formation. Northeast-striking faults and earlier, dike-filled, northwest-trending faults provide primary structural controls (fig. 2), and downdip to the west the East ore body may join the Main ore body under Popovich Hill. Primary unoxidized ores in the East ore body exhibit extremely wide variations in the contents of organic carbon, silica, and pyrite. The organic carbon content usually ranges from 0.5—1.0, and locally it is as high as 8.0 weight percent. Sulfide minerals are common, especially realgar and orpiment, and the element thallium, not included in the present study, has been found in amounts up to several tenths of a percent by weight. 8 STATISTICAL STUDY OF TRACE ELEMENTS, GEOLOGY AND GENESIS, CARLIN GOLD DEPOSIT STATISTICAL ANALYSIS OF GOLD, MERCURY, ARSENIC, AND ANTIMONY Several workers, including Akright, Radtke, and Grimes (1969), Joralemon (1951), Radtke, Dickson, and Rytuba (1974), Wells, Stoiser, and Elliott (1969), and Wrucke and Armbrustmacher (1975), have established that gold, mercury, arsenic, and antimony are ubiq- uitous constituents of disseminated replacement-type gold deposits. The general levels of concentration of these elements at the Carlin deposit are established by comparing the abundance of each element in mineralized rocks with that in normal carbonate rocks in general and the Carlin host rocks in particular (table 2). Statistical studies were carried out by stepwise linear regression analysis to determine the linear relations of these four elements and to examine how closely related the three elements are to gold by determining the statistical dependence of gold values on the values for the other three elements. Data used were taken from analyses of 239 samples from 96 rotary drill holes throughout the mine area. The data, in parts per million, had the following characteristics: Element Mean Standard deviation Au ____________________ 7.2 9.9 Hg ____________ __ 23 38 As _________________ 507 549 Sb ____________________ 126 282 Since these mean values are so much greater than those in similar unmineralized rocks near the deposit, we concluded that these elements were introduced and were not original in situ constituents of the Roberts Mountains Formation. The gold value of 7.2 ppm is lower than values usually cited for Carlin ores because it includes only unoxidized mineralized rocks and excludes certain areas of higher grade oxide ores and also because it includes large tonnages of mineralized rocks of lower grade now considered to be ore at gold prices of $140 to $160 per ounce. TABLE 2.—A bundance of gold, mercury, arsenic, and antimony in fresh limestones and unmineralized carbonate host rocks and unoxidized gold ores at the Carlin deposit [All values in parts per million. N, not determined; X, order of magnitude estimate] Carbonate host rocks, Mineralized carbonate Carlin deposit3 rocks, Carlin deposit“ Fresh carbo- nate rocks2 Fresh carbo- nate rocks‘ Element Average Average Average Median Average Median Au ______ 0.00X 0005—0009 <0.02 <0.02 11 10 Hg ______ 0.00 0.07 0.08 0.07 25 30 As ______ 1. 2.5—: 1,0? 4(?) N 480 360 Sb ______ 0.2 0.2i0.1? 0.8 0.5 130 90 lAbundance in carbonate rocks in the Earth’s crust (Turekian and Wedepohl, 1961). 2Abundance in carbonate rocks (Graf, 1960). 3Values for fresh unmineralized Roberts Mountains Formation (Radtke and others, 1972; Radtke, unpublished data). ‘Values for unoxidized mineralized Roberts Mountains Formation (Radtke and others, 1972). Krumbein (1959) suggested that it is reasonable to make the assumption that independent variables (mercury, arsenic, and (or) antimony in this study) that show the strongest mathematical correlation to the dependent variable gold also have the strongest physical relation to the phenomenon being studied. Obviously, many factors besides the ones used influ- enced the deposition of gold. However, as an indication of association, linear correlation coefficients may be valuable even though the variables compared may or may not be related by simple cause and effect. The nature of any relation is expressed by a correlation coefficient that ranges from —1 (perfect negative correlation) through 0 (absolutely no correlation) to + 1 (perfect positive correlation). The least-squares method finds a linear function that minimizes the sum of the squared deviations between 03served gold values (Y1) and the calculated gold values ( i), where Yi=ao+a1x1+a2x2+a3X3+ . . . +anx,’ where a1, a2, . . . an are the least-squares coefficients associated with x1, x2, . . . x independent variables, and a0=Y intercept. This method will minimize: 2(Yi—Yi)2. 7) The percent reduction in variance due to a linear model (R2) is given by: 2(Yi—Yi): 2(Y,_Y)2 where Y is the average of all gold values. R2=1 _ Comparison of the four elements showed them to have the following linear correlation coefficients over the entire deposit: (1) Au to Hg:0.44 (2) As to Hg:0.32 (3) Au to As:0.26 (4) As to Sb:0.22 (5) Au to Sb:0.11 (6) Hg to Sb:0.07 Upon reviewing the results of these correlations, we found that by eliminating approximately 1 percent of the samples from the entire population, much better correlations could be obtained. The eliminated samples were those that had created the greatest amount of variance from the proposed linear models. Scatter plots of these new populations with the anomalous values deleted are shown in figure 4. Inspection of the aberrant samples showed that some contained highly anomalous values, usually of mercury or arsenic, that probably STATISTICAL ANALYSIS OF GOLD, MERCURY, ARSENIC, AND ANTIMONY BNDOSD GENERAL PLOY - INCLUDING HISTOGRAN - REVISED JANUARV 301 ”EALYN SCKENCES COMPUTING FACILITYv UCLA 1970 GOLD-MERCURY REVISED DITA SCATTER PLOT FOR UNOXIDIZED ORE PLOT 0F VARIABLE 2 (VERTICAL AXIS) VERSUS VARIABLE(S) l (SYH50L=.), -7-500 7.500 22.500 37.500 52.500 67.500 -0.000 15.000 30.000 45.000 60.000 '0.D...II0..IIC'.COO‘COUI.I.CC...OC..IIO.IICI...IO.I.OI‘III...I'0‘...0......ICIIOI.I.‘c."‘l.l.ollll. . . 172.000 ° ‘ 172.000 . . I . I o . 152.000 ’ ‘ 152.000 . . I . I 132.000 ‘ ' 132-000 . . a . . . . . 112.000 ’ ‘ 112.000 ' s z . I 2 . o . j 93.000 ‘ ' ‘ 92-000 ~ I . I 1 . . . K G . 3" z o o a , .n 72.000 ~ 0 72.000 E . . < . ' . 0- . fi . ; . 0 . (x 52.000 ‘ ' ‘ 52-000 8 . D I a p I . a: a. 3 I 2 e . '5] : C u. o u a 0 . . 2' ' 2 ”2 ‘ . 32.000 ¢ ' ‘ 2" ' “ ' i’ ‘ 32-000 . 2 3 O. 0 u . . 35 2 a... e a O 0 In a . . 2.. I I .32 on o u , . 2‘. 2 2"2' i no . 12.000 0 262 3 . ‘ ‘ 12-000 . 3936?" 0 ' . . ":06" a o 9 u . o 29600.2 no Q o . . . -8.000 ‘ ‘ '3-000 o . . . . . 0...:‘.o..’-...'....‘....'....‘....‘....‘....'....*....‘....‘-...'....‘....'.-..‘....'..--‘....'....‘ -7.500 7.500 22.500 37.500 52.500 67.500 -0.000 15.000 30.000 05.000 60.000 GOLD. PARTS PER MILLION FIGURE 4.—Direct scatter plots of gold, mercury, C, Au to Sb; The following symbols represent ties of more than one point occurring at the same coordinates: Number of Number of Number of Symbol points Symbol paints Symbol points 2 2 8 8 E 14 3 3 9 9 F 15 4 4 A 10 G 16 5 5 B 11 H 17 6 6 C 12 I 18 7 7 D 13 / >18 were due to pieces of Cinnabar, realgar, or orpiment. Most of them contained combinations of values that were incompatible with the proposed function. The means found for this revised data set do not differ arsenic, and antimony in unoxidized Carlin ores. Plots are: A, Au to Hg; B, Au to As; D, Hg to As; E, Hg to Sb; F, Sb to As. significantly from the means for the original data, but the standard deviations were reduced for most samples. When the revised set of data was treated as one population over the entire area of the deposit, the results obtained in parts per million were: Element Mean Standard deviation Au ____________________ 6.6 8.9 Hg ____________________ 21 22 As ____________________ 502 5 15 Sb ____________________ 123 185 and the linear correlation coefficients were: (1) Au to Hg:0.55 (2) As to Hg:0.27 (3) Au to As:0.27 10 BHDOSD GENERAL PLOT - INCLUDING NEALTH SCIENCES COMPUTING FACILITYO UCLA STATISTICAL STUDY OF TRACE ELEMENTS, GEOLOGY AND GENESIS, CARLIN GOLD DEPOSIT HISTOGRAM - REVISED JANUARY 30: 1970 GOLD-ARSENIC REVISED DATA SCATTER PLOT FOR UNOXIDIZED 0H! PLOT OF VARIABLE 3 (VERTICAL AXIS) VERSUS VARIABLE(S) l (SYMBOL-‘)0 -7.500 7.500 22.500 37.500 52.500 67.500 -0.000 15.000 30.000 05.000 60.000 .OOOO.IIIJ.IICI...OI.OIII.0'Il‘lIOI.~O|0......OOIO‘UOI0".'0‘.tOI‘-Oon...00.0....OQOI.OI.OOOOIt‘tool. 3150.000 0 0 3759.000 2 . . I I . . . . 3315.000 0 v 3375.000 . . I O O O I 3000.000 9 . 3000-000 . . O O O . I . . 2025.000 . . 2025.000 I . I . . 2 o o 2 2250.000 . ~ 2250.000 3 . . .. . . x O . I g . . 0. 1315.000 0 . 1075.000 m , no . E a . n < l o I. n . ' n 5 1500.000 . o . 1500.000 ~ . O I . 2 U ‘ o m , a . ‘5 . . ~ . 1125.000 . 0 . 0 ~ . 1125.000 : '0 I I 0 i Z a ‘ o . Q. Q 2. D . 750.000 . 0 - M o 0 . 750.000 . O... O I I Q . . O. O Q .2. IO. 0 . . 3. on a D a o o n I o . . 3 2 02 u 2 . 375.000 0 502- 30 3 2 2 n 2 - . 375.000 . 305323» rt 90 . . 93325 - c 0 u u . . -563 2 n . ~ . . 0303 0 o . 0.0 . o . 0.0 9.0Io'-Ion‘nono‘oo-o‘o-oo'oco-‘I-o-‘onu0‘-o-o‘ooo0‘s.nu.-cog‘ogo-‘uoua‘uoc-‘oscn‘o-o-‘nca-‘no-o‘o...‘ -7.500 7.500 22.500 37.500 52.500 07.500 -o.000 15.000 30.000 45.000 60.000 GOLD. PARTS PER MILLION FIGURE 4.—Continued. Explanation on page 9. (4) As to Sb:0.22 the correlation is not significant at the 95-percent— (5) Au to Sb:0.12 (6) Hg to Sb:0.05 The revised means and standard deviations for each of the three ore bodies are summarized in table 3, and linear correlation coefficients for each pair of elements in each ore body are shown in table 4. These correlation coefficients (R) are strictly applica- ble only to the data used in the computations. It is our hope that they reflect the correlation coefficients (p) of the true population of these elements in the ore bodies at Carlin. We want to establish whether the range of values Within which p falls includes zero, in which case . confidence level. It is modeled that 1/2 ln (1 +R/1—R) can be considered to have an approximately normal distribution with a standard deviation of 1/Vn—3. Hence, the 95-percent- confidence interval for p is: i In 1+R _ 1.96 <1 ln l+p 2 1—R Vn—3 2 l—p 1 1+R 1.96 < — 1n — 2 1~R + vn—3 where n = number of samples. BNDOSD GENERAL PLO HEALTH SCIENCES CO PLOT 0F VARIABLE -7-500 .00. 0500.000 ~ . 4000-000 ‘ . . 3500.000 3000.000 0 o o o u a a o - 2500.000 2000.000 1500.000 ANTIMONY. PARTS PER MILLION l000.000 STATISTICAL ANALYSIS OF GOLD, MERCURY, ARSE T - INCLUDING HISTOGRAM - REVI MPUTING FACILITY! UCLA GOLD-ANTIMONY REVISED OAT t (VERTICAL AXIS) VERSUS VAR! 7-500 ‘0-000 15.000 o‘--.-‘----'----‘----'.---'--- 5E0 JANUARV 300 1970 A SCATTER PLOT FOR UNOXIDIZED DUE ABLF(S) l (SYMBOL=“)v 22.500 31.500 30.000 -’---- ’--oI‘----'-c-.‘o.-.‘I.--‘.--- NIC, AND ANTIMONY 52-500 45-000 60-000 *---.*----'-.--‘.-~-‘----‘---- 11 67-500 0 4500-000 0000-000 3500-000 3000-000 2500-000 2000-000 1500-000 1000-000 500-000 - - 6 n a o a o a o o o O o n o - 0 - o - a Q o . o n 6 a o o a 0 500.000 “ fl 9 2 a u a a 52332 22 #22 2&2 0* ° 2M *2" “' “ ° ” ZIF°6A35 153»2¢22v32 * 3v 2*»- ° 0.0 71/5 2 2 ° -5oo.ooo '-.-u‘----‘-o-u*--.- -7-500 '0-000 7-500 ‘----‘----‘----‘--.o 15-000 22-500 ‘---- 31.500 ‘oo--'.---‘---- 30-000 GOLD- PARTS PER MILLION FIGURE 4.-—-Continued. Explanation on page 9. ‘-.--‘----‘--.- 65.000 52-500 60-000 0-0 '500-000 ‘-.--‘.---'----‘-o--*----* 67-500 TABLE 4.—-Linear correlation coefficients between gold, mercury, and East ore bodies at the TABLE 3,—Mean values and standard deviations for gold, mercury, antimony in the West, Main, and East are bodies at the arsenic, and Carlin gold deposit [All values in parts per million] arsenic, and antimony for the West, Main, Carlin gold deposit West ore body Main ore body East ore body ____.______ West ore body Main ore body East ore body Element pair nggflfilxi Element pair cgggfigfi Element pair 0202;102:211; Element 5: d d Standard Standard an r Mean deviatzl‘on Mean deviation Mean deviation Sb to AS _______ 0.87 All to Hg ______ 0.62 Au to Hg ______ 0.48 :u to SAb ______ .85 All: to Sb ______ .62 Au to As ______ .23 A ____________ 75 8.4 6.5 8.2 6.4 94 u to S ______ .81 S to AS ______ .47 As to Hg ______ .19 H: ____________ 22 20 20 25 21 21 Au to Hg ______ .68 As to Hg ______ .41 As to Sb ______ .19 As ____________ 222 220 490 540 590 580 Sb to Hg ______ .56 Sb to Hg ______ .40 Au to Sb ______ .08 Sb ____________ 52 40 106 76 155 382 As to Hg ______ .54 Au to As ______ .34 Sb t0 Hg ______ .00 12 STATISTICAL STUDY OF TRACE ELEMENTS, GEOLOGY AND GENESIS, CARLIN GOLD DEPOSIT BMDOSD GENERAL PLOT - INCLUDING HISTOGRAM ~ REVISED JANUARY 300 1970 HEALTH SCIENCES COMPUTING FACILITY, UCLA MERCURY‘ARSENIC REVISED DATA SCATTER PLOT FOR UNOXIUIZED ORE PLOT 0F VARIABLE 3 (VERTICAL AXIS) VERSUS VARIABLEIS) 2 (SYMBOL3M10 -0-000 00.000 80.000 120.000 160.000 .-20.000 20.000 60.000 100.000 160.000 I....O.OOOI..OIO.OOOO.IIII.OC0".IO..OQIO’ICOIODOII,IO.O‘QIUl'lO.l‘.aI...Don‘t.il‘.cll.ll.l.0-IQ‘IIQU 3750.000 0 0 3750.000 . n M M 0 M 3375.000 ‘ ’ 3375-000 0 ‘0 0 3000.000 3000.000 . M . 2525.000 . ~ 2625.000 . . M . Z I I 3 2250.000 . . 2250.000 j . . H I I . M E. - . '1 1075.000 . . 1075.000 m . M M . E . M < O '1 . M M . .3 1500.000 . M + 1500.000 5 . M M . kl a X: - M < . M M 1125.000 0 2 M M . 1125.000 . M M M M M M . M . M214 14214 M M . 150.000 0 M 2MM M 0 750-000 . 2 M 2 M M M . o M MM? MM M M M M M . MMM 2 2 MM M M M M M M . MM 3M MM M M MM . 375.000 0 23323 2MM3 2 2 M 2 . 375.000 . 42149444 MZMMMMZ MM 14 . . 002214 MZMM MMM M M MM MM M . 76M3MM3M M an 2 M . 3531434 M M 2 M M M 0.0 . M 0 0 0 O. I ‘IOII'OII0...00...Io‘ooll’l-Io’ln.0,.II..OOIo‘IOoI‘Io 0..ICC...-0".Q“.I..‘.ICI‘I'."IIOI.IIII -0.000 00.000 00.000 120.000 160.000 -20.000 20.000 60.000 100.000 100.000 MERCURY. PARTS PER MILLION FIGURE 4.—Continued. Explanation on page 9. 1.96 is obtained from the normal distribution tables so functions with all the independent variables entered in that there is a 95-percent probability that p lies between parts per million: the two limits. (A) West ore body: The correlation coefficients calculated at the 95- E(Au) : _2_22959 + 0.11746 (Hg) + percent-confidence level were: 0.00833 (As) + 0.10364 (Sb) Sb to As, R=0.87 Au to Hg, R=O.68 (B) Main ore body: 0.77 < p < 0.93 0.46 < p < 0.82 E(Au) = —1.20076 + 0.14972 (Hg) — Au to Sb, R=0.85 Sb to Hg, R=0.56 0.00105 (As) + 0.04956 (Sb) 0.72 < p < 0.92 0.30 < p < 0.75 (C) East ore body: Au to As, R=0.81 As to Hg, R=0.54 E(Au) = 0.79597 + 0.19946 (Hg) + 0.66 < p < 0.90 0.26 < p < 0.73 0.00203 (As) + 0.00132 (Sb) The least-squares predictor equations for the mean The percent reductions in variance due to a linear gold content may be computed by the following three model [R2=1—2 (sample gold values-computed gold STATISTICAL ANALYSIS OF GOLD, MERCURY, ARSENIC, AND ANTIMONY 13 BHDOSD GENERAL PLOT - INCLUDING HISIOGRAM - REVISED JANUARY 30! 1970 NEALTH SCIENCES COMPUTING FACILITY! UCLA MERCURY-ANTIMONY REVISED DATA SCATTEP PLOT FOR UNOXIDIZED OFF PLOT 0F VARIABLE Q (VERTICAL AXIS) VERSUS VARIABLE($) Z (SYMBOL-M); 80.000 120.000 160.000 100.000 0 Ih0.000 o‘.ooo‘ou.o‘oca0.0.oo‘.nI-‘oo00'...o‘uoao‘ooon‘oong'I-n.'.-In goo-‘0-.n‘I-go‘uco00.noo’otnl‘oono‘u.uc -0.000 00.000 -20.000 20.000 60.000 0500.000 0 . C . 0000.000 . M . . . 3500.000 . . . 3000.000 0 O . z . O o 3 2500.000 0 g . i o 1 o g . 2000.000 . m L . a O 4 l o ; 1500.000 o M O o I x . 5 . ( - 1000.000 0 . M . . 500.000 0 M a M . M M MM 2 M . ZMzzzMMz MM337M 2MMMM M2 2 . 99975025162650 5M A233MM M M M 0.0 0 08049650 2M2? MM M an . . -500.000 ~ .Ogoo.‘-..a‘n.oo‘...a‘.uau‘oo-.‘o.n.‘n...’.-..’oo.~'ua..*oo.o a... . -0.ooo £0.000 -20.000 20.000 00.000 0 4500-000 0. - 0000.000 3500.000 3000.000 - o. o n - on - 2500-000 - o 0- .- 2000.000 1500-000 o n u 0- a n o 'n u .- l000.000 0- 500-000 0.0 9- o . o 0- . .- -500.000 ‘ ' un'...o‘c-un‘..a.‘.o--‘on.u‘.-.o 80.000 120-000 160.000 100.000 1‘0.000 MERCURY! PARTS PER MILLION FIGURE 4.—Continued. Explanation on page 9. values)2/2(sample gold values-mean gold values)2] using these functions are: (A) West ore body—79.22 percent, (B) Main ore body—54.90 percent, (C) East ore body—25.01 percent. A linear model fits the West ore body quite well. The standard error of prediction (SE) of using these three variables to predict the mean gold value is determined by SE=(Mean square of residuale/2 7) _ [2 i =1(Sample Au value—computed Au value)2]l/2 (number of samples —number of variables —1) where n is the number of samples and i=1 is the first sample considered. The standard errors found in the above linear functions (A, B, and C) are (A) West ore body=3.9 ppm, (B) Main ore body=5.6 ppm, and (C) East ore body=8.2 ppm- The multiple correlation coefficients, defined as the square root of the percent reduction in variance and used as a measure of the adequacy of the different values in predicting the mean gold value (Krumbein and Graybill, 1965), were found to be West ore body Main ore body East ore body As:0.89 As:0.74 Sb:0.50 Hg:0.88 Sb:0.74 As:0.50 Sb:0.85 Hg:0.62 Hg:0.48 14 BMDOSD GENERAL PLOT - HEALTH SCIENCES COMPUT INCLUDING HISTOGRAM - REV ING FACILITY! UCLA STATISTICAL STUDY OF TRACE ELEMENTS, GEOLOGY AND GENESIS, CARLIN GOLD DEPOSIT 1550 JANUARY 300 1970 ANTIMONY-ARSENIC REVISED DATA SCATTER P LOT FOR UNOXIDIZED ORE PLOT 0F VARIABLE 3 (V ERTICAL AXIS) VERSUS VARIABLF(S) 4 (SYMBOL-5). 0.0 000.000 1500.000 2400.000 3200.000 4000.000 400.000 1200.000 2000.000 2000.000 3600.000 'ooot‘onoa‘oo-o‘o000‘s.00‘s.00‘..-0’...0’.-00‘.nonooou.‘.-.-‘Ioon‘ouo-‘uouo’oooo’uouc‘nocv‘ac-o‘oocn‘ 3150.000 . . 3750.000 ' Q . s s . . s . . . 3375.000 0 . 3315.000 . . C I . . . . 3000.000 . 0 3000.000 ' I . . . s . . . 2025.000 . . 2025.000 . . . s . . . Z n a 2 2250.000 . 0 2250.000 J . I d . . I . s . m I C 5 1075.000 0 o 1015.000 m . ss . t; . s . § 0 a a . ss . 5 1500.000 0 s . 1500.000 u . s s . 5 . . w . s . 3 . s s . 1125.000 . s 55 s . 1125.000 . . . sssz s. . s . . 535 as s s . 750.000 . 5553 . 750.000 . 2 325 . . 5650 . . 553545 . . 263 s . 375.000 0 50905 s . 315.000 . 900 s s . . 01 2 . .2/3s . .ar . 0.0 .5 ‘ 0.0 ‘I....lII....I..OOOO‘OI.I.OIIO‘UOIC.IOIt...0...I’D....I‘ll.-....O..CO..IIO0.....‘IIOC.II.I‘.IIOOIIOI. 0.0 000.000 1600.000 2400.000 3200.000 4000.000 000.000 1200.000 2000.000 2800.000 3600.000 ANTIMONY’ PARTS PER MILLION FIGURE 4.—Continued. Explanation on page 9. By using the linear model with the revised data from the entire deposit, the variance is decreased by 31.8 percent with mercury accounting for 93.8 percent of the reduction. The least-squares equation is: E(Au)= 1.21723 +0.20225 (Hg) +0.00186 (As)+0.00209 (Sb). The standard error of estimate is 7.35 ppm. Multiple correlation coefficients for the variables in the equation are: Hg:0.55, As:0.56, and Sb:0.56. Some interesting relationships appear when correla- tions are compared to the mean values in the three ore bodies. Gold content is fairly constant between the three areas, but the arsenic content is much lower in the West ore body than in the other two. However, the correlation between the gold and arsenic is fairly high (0.81) in the West ore body, whereas the correlations in the Main and East ore bodies (0.34 and 0.23, respectively) show much less dependence of one on the other. Graf (1960) reported that carbonate rocks normally contain only 1 or 2 ppm of arsenic unless there is an abundance of carbonaceous material, iron oxides, or sulfides present. Information derived from samples of the fresh unmineralized Roberts Mountains Formation around Carlin (Radtke, Heropoulos, Fabbi, Scheiner, and Essington, 1972) indicates that this observation holds true here. They STATISTICAL ANALYSIS OF GOLD, MERCURY, ARSENIC, AND ANTIMONY reported that although only 2 out of 15 samples had detectable amounts of arsenic (10 ppm), some rocks contained as much as 0.5 percent syngenetic pyrite and as much as 1 percent organic carbon. The very high Au:As correlation may be partly explained in the following three ways: (1) Gold and arsenic appear to have been coprecipi- tated on the surface of pyrite grains. (2) The Main and East ore bodies are characterized by an apparent late buildup of arsenic as evidenced by the presence of arsenic sulfides in these two areas. (3) The host rocks for the West ore body contain relatively small amounts of the carbonaceous materials that tend to induce precipitation of gold and mercury, but not arsenic, elsewhere in the deposit. Consequently, the deposition of gold as well as arsenic in this area was apparently more responsive to physicochemical factors such as fluctuations in the compositions of the invading solutions and (or) changes in pH, temperature, and pressure than in other areas. The West ore body probably formed under a more uniform set of chemical conditions that existed during the earlier part of the mineralizing sequence. Since the late arsenic, an- timony, mercury sulfide stage is lacking in this area, the high AuzAs correlation in the West ore body suggests that deposition of gold and arsenic in early stages of mineralization occurred under similar physicochemical conditions. The content of antimony is also very much lower in the West ore body than in the other two. The correlation between antimony and arsenic is much higher in the West ore body (0.87) than in the Main and East ore bodies (which have values of 0.47 and 0.19, respective- ly). The relation of antimony to gold shows a similar trend. In the West ore body, where antimony has its lowest mean value, the correlation is strongest, whereas in the East ore body, which has the highest mean antimony value, the correlation is very weak. The poor As:Sb correlation in the East ore body is explained by the presence of antimony as late-formed and unevenly distributed pods of stibnite, whereas arsenic occurs as random scatterings of arsenic sulfides. Mineralogical studies show that there is almost no correlation between the occurrences of the antimony and arsenic sulfides within the Main and East ore bodies. In rocks that contain realgar and (or) orpiment, there is usually little stibnite, concentrations of which are almost invariably marked by the absence of arsenic sulfides. This observation was confirmed by the values of the correlation coefficients. The low correlations between gold and antimony and gold and arsenic in the Main and East ore bodies may be explained by the higher content of organic carbon in these areas that removed large amounts of gold and mercury from the 15 hydrothermal solutions. The correlations between gold and mercury are the strongest correlations between any element pairs in carbonaceous areas. The stronger gold to mercury correlation also reflects their coeval deposi- tion on the surface of pyrite grains. Within the host rocks of the West ore body, that had less carbon to fix gold and mercury, a simple paragenesis developed in which gold, mercury, arsenic, and antimony were all deposited together in response to physicochemical conditions that changed in a consistent direction. Relations between elements in the West ore body are those that occurred in the presence of limited or small amounts of organic carbon. In the Main and East ore bodies, larger amounts of organic carbon caused early deposition of gold and mercury, thus the paragenesis was complicated by depleting the ore fluid in these elements and leaving significant amounts of arsenic and antimony to be deposited in a later depositional stage. Distributions and relations of these four elements between the three ore bodies were compared by stepwise discriminant analysis. Discriminant analysis makes linear combinations of the four quantitative variables to produce two (number of qualitative variables, that is,- the three ore bodies, minus one) pseudovariables called the canonical variables. These are chosen so as to: (1) minimize the squared distance from a sample point to the plot of the mean of its respective ore body on a graph with the canonical variables as axes; and (2) maximize the separations of the pits; that is, maximize the distance between the plots of the pit means. The canonical variables thus produce two optimum linear functions (L1 and L2) from the original four variables so that when the samples from each site are evaluated with these functions, they are best separated from the samples of the other areas. Consequently, when the data points are plotted in a L1, L2 graph, the three sites have the greatest mutual separation. Listed below are the values found for each element in terms of percentages of samples nearest to the different ore body means, that is, the values are percentages of samples taken from the vertically listed ore bodies whose mean values are closest to the mean of the horizontally listed ore bodies. It is important to note that arsenic values would be included in the second step, gold and arsenic would be included in the third step, and all values are used in the computations for the final step below. Step 1, Arsenic: East Main West East ________________ 41.13 20.16 38.71 Main ________________ 27.14 20.0 52.86 West ________________ 5.4 8.11 86.49 16 STATISTICAL STUDY OF TRACE ELEMENTS, GEOLOGY AND GENESIS, CARLIN GOLD DEPOSIT Step 2, gold and arsenic: East Main West East ________________ 30.64 42.74 26.62 Main ________________ 20.0 30.0 50.0 West ________________ 0.0 8.11 91.89 Step 3, antimony, gold, and arsenic: East Main West East ________________ 32.26 43.55 24.19 Main ________________ 21.43 31.43 47.14 West ________________ 0.0 8.11 91.89 Step 4, mercury, antimony, gold, and arsenic: East Main West East ________________ 34.68 37.10 28.22 Main ________________ 20.0 42.86 37.14 West ________________ 0.0 10.81 89.19 When various combinations of the elements were compared between the ore bodies in this fashion, the West ore body consistently showed the greatest polari- zation. That is, values of arsenic, gold and arsenic together, antimony, gold and arsenic together, and mercury, antimony, gold and arsenic together in the West ore body were always much closer to the mean value in that ore body than to the mean values in the other two ore bodies. These values also reflect the much higher correlations between elements in the gold, mercury, arsenic, antimony suite in the West ore body. Values for these suites of elements in the Main ore body tend to be much more similar to the West ore body than to the East ore body. Strong polarization of the West ore body is probably due also to the relation of the mean values (given earlier) where, except for gold and mercury that show little variation between the three areas, the values in the West ore body are drastically lower. This reflects the different chemical conditions originally present in this area, which were previously described. The polarity of the element values in the West ore body support the proposed mineral paragenesis model (fig. 3) in which the late mercury, antimony, and arsenic sulfide mineralization occurred only in the Main and East ore bodies. The relatively high correlation between gold and mercury found for the entire deposit was probably caused by two conditions: (1) gold and mercury were both transported in the same ore solution and were removed from that solution simultaneously in response to similar influences; and (2) more than 90 percent of the occurrences of both elements either are associated with various organic materials or occur together on the surfaces of pyrite grains; the occurrences reflect a similar depositional process for both elements. Both of these conditions represent deposition on fine-grained materials that are fairly evenly distributed throughout the volume of mineralized rocks in the entire deposit. The removal of gold and mercury from solution probably reflects the extraction capability of various solids on complexes of these metals in highly supersaturated solutions, and this extraction capability led to their precipitation in fairly constant proportions. The proces- ses responsible for the deposition of gold and mercury in highly carbonaceous rocks were probably not the same as those leading to the precipitation of arsenic and (or) antimony. STATISTICAL ANALYSIS OF GOLD, BARIUM, COPPER, MOLYBDENUM, LEAD, AND ZINC The fact that anomalous amounts of barium, copper, molybdenum, lead, and zinc occur in areas of dissemi- nated gold mineralization has been pointed out by Akright, Radtke, and Grimes (1969); Radtke, Heropou- los, Fabbi, Scheiner, and Essington (1972); and Wrucke and Armbrustmacher (1975). Comparison of data for normal limestones and unmineralized host rocks with mineralized rocks (table 5) shows that, except for barium, these elements are concentrated from about 4 to 10 times in the ores. The content of barium is slightly increased in the mineralized unoxidized ores. These concentration factors are significantly less than the general range of 100 to 500 times for gold, mercury, arsenic, and antimony (table 2). Data from analyses of 292 1.53—m (5-ft) composite samples taken from 96 rotary drill holes in unoxidized gold-bearing rocks were used to find the correlations between gold, barium, copper, molybdenum, lead, and zinc. A stepwise linear-regression analysis was per- formed to find the dependence of gold on the other five elements throughout the entire deposit, and then it was done for each of the three ore bodies individually. Then the elements were compared between the ore bodies by means of a stepwise discriminant analysis. The data, in parts per million, had the following characteristics taken over the three ore bodies: Element Mean Standard deviation Au ____________________ 7.3 10 Ba ____________________ 400 200 Cu ____________________ 33 15 Mo ____________________ 5.9 6.9 Pb ____________________ 30 99 Zn ____________________ 165 665 These mean values compare well with those reported for Carlin ores cited in table 5. The lower computed mean value for gold was explained previously in this report. The difference in zinc values is not significant in View of the large standard deviation. Linear correlation coefficients found over the entire deposit are, in decending order: (1) Au to Zn:0.20 (5) Zn to Cu:0.07 (2) Cu to Mo:0.19 (6) Au to Mo:0.06 (3) Au to Cu:0.17 (7) Au to Ba:—0.06 (4) Cu to Ba:~0.08 (8) Pb to Mo:—0.05 STATISTICAL ANALYSIS OF GOLD, BARIUM, COPPER, MOLYBDENUM, LEAD, AND ZINC (13)Pb to Cu:—-0.02 (14) Zn to Ba:0.01 (15) Zn to Mo:—0.01 (9) Pb to Ba:0.05 (10) Au to Pb:—0.05 (11) Pb to Zn:0.04 (12) Mo to Ba:0.03 All correlations between these elements are weak. The weaknesses may in part be attributed to the fact that many of these elements were analyzed by semiquantitative spectrographic analysis rather than by the more precise methods used for the previous suite of elements. TABLE 5.——Abundance of gold, barium, copper, molybdenum, lead, and zinc in fresh limestones and unmineralized carbonate host rocks and unoxidized gold ores at the Carlin deposit [All values in parts per million. N, not determined or calculated; X, order of magnitude estimate} Fresh carbo- Fresh carbo- Carbonate host rocks, Mineralized carbonate Element nate rocksl nate rocks2 Carlin deposit.a rocks, Carlin deposit" Average Average Average Median Average Median Au ______ 0.00X 0.005—0.009 <0.02 <0.02 11 10 Ba ______ 10 1501-110 250 280 400 500 Cu ______ 4 14: 9 9 1 5 35 30 Mo ______ 0.4 1.1:0.7? <2 <2 7 5 Pb ______ 9 8:4 3 N 30 20 Zn ______ 20 26:5 14 1 1 185 120 $233322: l3 22:33:13: :33: 18525 $55in "‘1“ ”“9““ and wedepd‘l’ 1961" 3Values for fresh unmineralized Roberts Mountains Formation (Radtke and others, 1972; Radtke, unpublished data). , ‘Values for unoxidized mineralized Roberts Mountains Formation (Radtke and others, 1972', Radtke, unpublished data). Irregularities in the barium and associated base- metal distributions were predictable from geologic occurrences of the mineral phases involved. Small amounts of barium occurs as barite that is scattered through the mineralized carbonate rocks. Most of the barite, however, is concentrated in late veins that follow faults and fracture zones in the upper oxidized parts of the ore bodies. The erratic distribution and irregular shapes of the barite veins, only some of which contain sulfide minerals, account for local high anomalies in these elements. Genesis of barite in the Carlin ores was discussed in detail by Dickson, Radtke, and Rye (1975). Detailed electron microprobe studies of Carlin ores (Radtke and others, 1972), including analyses of pyrite grains and carbonaceous materials known to contain gold and mercury, showed that copper, molybdenum, lead, and zinc do not occur in this association. These four elements do occur, however, as discrete sulfides locally concentrated in barite veins and erratically scattered through mineralized rocks. This distribution of these base-metal sulfides tends to produce insignificant correlations between these four elements. Because the three ore bodies at Carlin are known to possess different geologic and mineralogic characteris- tics, the correlations between the six elements were examined for each individual ore body. Mean values and standard deviations for gold, barium, copper, molyb- denum, lead, and zinc in each ore body are given in table 17 TABLE 6.—Mean values and standard deviations for gold, barium, copper, molybdenum, lead, and zinc in the West, Main, and East ore bodies at the Carlin gold deposit [All values in parts per million] West ore body Main ore body East ore body Element Mean Sifilflfi Mean 323235233. Mean 33:233. Au ____________ 8.6 10 6.9 8.9 7.2 11 Ba ____________ 650 330 500 200 300 100 Cu ____________ 25 13 36 18 33 13 Mo ____________ 6.2 9.3 7.3 6.1 5.0 6.3 Pb ____________ 26 51 49 175 20 19 Zn ____________ 7 72 193 197 177 890 6. Linear correlation coefficients for gold, barium, copper, molybdenum, lead, and zinc within each ore body, listed in descending order, are shown in table 7. Correlation coefficients show that for the Main and East ore bodies most of the correlations between pairs of these six elements are very low. Only in the West ore body do the correlation coefficients show any consistent relations predictable from the geochemical paragenetic model. Correlations between pairs of base metals and between zinc and barium are relatively uniform and strong compared to those between the other base metals to barium, gold to the base metals, and gold to barium. These correlations reflect the fact that the base metals were deposited during a stage in the hydrothermal paragenesis separate from that of gold, and that although some barite formed as base metals were deposited, most of the barite was deposited later in the paragenesis (Dickson, and others, 1975). That the West ore body is best represented by a linear model is entirely consistent with the following known geologic facts: the West ore body has the most simple structural setting of the three ore bodies; the minerali- zation in this area included the entire hydrothermal sequence except for the late, postgold, mercury- TABLE 7.——Linear correlation coefficients between gold, barium, copper, molybdenum, lead, and zinc for the West, Main, and East ore bodies at the Carlin gold deposit West ore body Main ore body East ore body Correlation coefficient Correlation coeffiment Element pair Correlation Element pair Element pair coefficient Pb to Cu ______ 0.56 Zn to Pb _________ 49 Zn to Cu _________ 40 Zn to M0 _______ 31 Cu to M0 _______ 28 Zn to Ba _________ 28 Cu to Ba ______ —.23 Au to Zn ______ —.20 Au to Ba ______ —.17 Au to M0 ____—.15 Pb to Au ______ —.14 Au to Cu _______ 13 Pb to Ba _________ 10 Pb to M0 ______ —.09 Mo to Ba ______ —.07 Au to Cu ______ 0.27 Zn to Ba _________ 17 Mo to Cu _______ 16 Pb to Cu ______ —.15 Pb to M0 ______ —.14 Au to Zn _________ 12 Zn to Pb _________ 11 Zn to Cu _________ 10 Mo to Ba ______ —.09 Au to Pb ______ —.08 Zn to M0 _______ 07 Cu to Ba ______ —.05 Au to Ba ______ —.04 Au to M0 _______ 03 Pb to Ba _________ 01 Pb to Ba ________ 0.27 Au to Zn ___________ 26 Pb to Cu ___________ 18 Mo to Cu _________ 17 Au to M0 _________ 16 Au to Cu _________ 15 Zn to Pb ___________ 09 Zn to Cu ___________ 06 Pb to M0 _________ 05 Zn to M0 ________ —.03 Zn to Ba ________ — 02 Au to Ba ________ — 01 Au to Pb ___________ 00 Cu to Ba ___________ 00 Mo to Ba _________ 00 18 STATISTICAL STUDY OF TRACE ELEMENTS, GEOLOGY AND GENESIS, CARLIN GOLD DEPOSIT arsenic-antimony sulfide stage; and the West ore body has a much lower and more uniform content of organic carbon, creating simpler overall geochemical conditions as described in the preceding section. The correlation between gold and barium as deter- mined for the entire deposit and also for each separate ore body was the only consistently negative value; this fact indicates that the amount of gold in the ore is inversely correlated with the content of barium. Barite veins generally contain no detectable amounts of gold, and the correlations show that within the mineralized carbonate rocks the areas of higher gold content would contain the smallest amounts of barite. All these facts further support the idea that most of the barite formed during a different stage in the paragenesis from the deposition of gold. The least-squares predictor equation for the average gold content over the entire deposit with the indepen- dent variables in parts per million is: E(Au)=3.79151—0.00002(Ba)+0.09556 (Cu) +0.03908 (Mo)+0.00529 (Pb)+0.0029 (Zn). The percent reduction in variance from this linear model using only zinc values is 3.99 percent; using zinc and copper values 6.31 percent; zinc, copper and lead values 6.62 percent; zinc, copper, lead and barium values 6.81 percent; and for the full model 6.88 percent. Thus, the copper values improve the prediction using only zinc by 2.32 percent, the lead values reduce the variance by only an additional 0.31 percent, barium by 0.19 percent, and molybdenum by 0.07 percent. The standard error of prediction of this equation is :9.7 ppm gold. Statistical analyses carried out to predict the mean gold value in each individual ore body using these four base metals and barium gave only slightly better reductions in variance. These values are for the West ore body 8.20 percent, the Main ore body 8.75 percent, and the East ore body 11.09 percent. The multiple correlation coefficients of these ele- ments are: West ore body Main ore body East ore body Ba:0.43 M0:0.30 Ba:0.33 Mo:0.43 Ba:0.29 Pb:0.33 Pb:0.35 Pb:0.29 Cu:0.33 Cu:0.30 Zn:0.29 Mo:0.31 Zn:0.20 Cu:0.27 Zn:0.26 The equally low multiple correlations found for all these elements reflect the very limited value they have in predicting the average gold content. This conclusion is consistent with the paragenetic model proposed previ- ously. The distribution and relations of these six elements between the three ore bodies were compared by stepwise discriminant analysis. The values found are sum- marized subsequently. The figures are percentages of samples taken from the vertically listed ore bodies whose mean values are closest to the mean of the horizontally listed ore bodies with each step also involving the variables in the preceding step(s). Step 1, copper: East Main West East ________________ 75.38 _. 13.85 10.77 Main ________________ 43.05 33.33 23.61 West ________________ 32.5 10.0 57.5 Step 2, barium and copper: East Main West East ________________ 74.61 13.85 11.54 Main-___, ____________ 36.11 33.33 30.55 West ________________ 30.0 10.0 60.0 Step 3, molybdenum, barium and copper: East Main West East ________________ 60.0 26.92 1308 Main ________________ 27.78 40.28 31.94 West ________________ 22.5 17.5 60.0 Step 4, 'lead, molybdenum, barium, and copper: East Main West East ________________ 61.54 16.15 22.31 Main ________________ 25.0 44.44 30.56 West ________________ 32.5 12.5 55.0 Step 5, gold,.lead, molybdenum, barium, and copper: East Main West East ________________ 63.85 20.0 1615 Main ________________ 22.22 45.83 31.95 West ________________ 27.5 15.0 57.5 Step 6, zinc, gold, lead, molybdenum, barium, and copper: East Main West East ________________ 63.85 20.0 16.15 Main ________________ 22.22 47.22 30.55 West ________________ 25.0 15.0 60.0 Discrimination analysis using all the variables (step 6 above) shows that values differed less from the mean of their own ore body than from those of the other two. This result reaffirms the need to analyze the data separately for each ore body. Values in the Main ore body were very similar to the ore bodies on either side, whereas values in the peripheral areas were much more polarized towards their own means. These values compare with the mean values in the Main ore body which were highest for zinc, lead, molybdenum, and copper but lowest for gold. STATISTICAL ANALYSIS OF GOLD, BORON, TELLURIUM, SELENIUM, AND TUNGSTEN Boron and tungsten are present in highly anomalous concentrations in Carlin ores compared with their normal abundances in carbonate rocks. Tellurium and selenium are present in low abundance in both the ores and the host rocks, although selenium is enriched in mineralized rocks. Published data on the abundances of these four elements in primary unoxidized Carlin ores, the fresh carbonate host rocks, together with other average values in carbonate rocks are given in table 8. STATISTICAL ANALYSIS OF GOLD, BORON, TELLURIUM, SELENIUM, AND TUNGSTEN TABLE 8,—Abundance of gold, boron, tellurium, selenium, and tungsten in fresh limestones and unmineralized carbonate host rocks and unoxidized gold ores at the Carlin deposit [All values in parts per million. N, not determined or not given; X, order of magnitude estimate] Carbonate host rocks Mineralized carbonate rocks Element 1:35: 15:11:? 13:56:11.3):123- C arl in deposit“ C arli 11 deposit“ Average Average Average Median Average Median Au ______ 0.00X 0.005—0.009 <0.02 <0.02 11 10 B ______ 20 1 21- 8 1 5 1 5 70 7 0 Te ______ N < .2 < .2 < .2 < .2 Se ______ 0.08 0.1—1? <1 <1 2 N W ______ 0.6 0.5? N N 18 N ‘Abundance in carbonate rocks in the Earth‘s crust (Turekian and Wedepohl, 1961). 2Abundance in carbonate rocks (Graf, 1960). 3Values for fresh unmineralized Roberts Mountains Formation (Radtke and others, 1972; Radtke, unpublished data). ‘Values for unoxidized mineralized Roberts Mountains Formation (Radtke and others, 1972; Radtke, unpublished data). Data used to establish correlations between gold, boron, tellurium, selenium, and tungsten come from analyses of 288 composite samples selected from 94 rotary drill holes in unoxidized mineralized carbonate rocks. The data were treated statistically by stepwise linear—regression analyses to examine correlations throughout the entire deposit as well as within each individual ore body. Stepwise discriminant analysis was also used to compare the mean values of the elements between the three ore bodies. The data, in parts per million, averaged over the entire deposit have the following characteristics: Element Mean Standard deviation Au ____________________ 7.4 10 B ______________________ 79 53 Te ____________________ .02 .08 Se ____________________ 1.5 2.3 W _____________________ 12 44 With the exception of gold, as discussed earlier, the mean values for the elements correspond well to those given in table 8. Values for selenium of 1.5 ppm and tungsten of 12 ppm probably are better than queried values for these elements reported by Radtke, Her- opoulos, Fabbi, Scheiner, and Essington (1972), and the tellurium value of 0.02 ppm refines the figure of <02 ppm given in table 8. Linear correlation coefficients between these five elements for the entire deposit are, in descending order: (1) Au to Te:0.37 (6) Au to W:0.03 (2) B to Te:0.08 (7) B to Se:0.02 (3) W to Te:—0.07 (8) Au to B:—0.02 (4) W to Se:0.04 (9) B to W:0.01 (5) Au to Se:—0.04 (10) Se to Te:—0.01 Among all these pairs of elements, only the ratio of gold to tellurium is significant and even that correlation is much weaker than many others in the Carlin deposit. The very low correlations reflect the fact that the data fit 19 TABLE 9.—Mean values and standard deviations for gold, boron, tellurium, selenium, and tungsten in the West, Main, and East ore bodies at the Carlin gold deposit [All values in parts per million] West ore body Main ore body East ore body Element Mean Standard Mean Standard Mean Standard deviation deviation deviation Au ____________ 8.8 9.8 6.9 9.0 7.2 10 B ____________ 54 17 85 55 84 56 Te ____________ .0 .0 .04 .1 .02 .07 Se ____________ 1.5 1.7 .9 2.2 1.8 2.5 W ____________ 9.7 9.9 17 43 10 50 a linear model poorly: E(Au)=6.76239—0.00808(B) +46.84856(Te)—0.15657(Se)+0.01276(W) when the standard error of estimate is :9.4 ppm gold. The percent reduction in variance due to this function is only 13.95 percent with tellurium providing 95.63 percent of this reduction. Correlations between these five elements were assessed within each of the three ore bodies. Mean values and standard deviations for each element within each ore body are shown in table 9, and linear correlation coefficients for element pairs for the three ore bodies are given in table 10. Examination of the correlation coefficients within each individual ore body shows that the only significant correlation is between gold and tellurium and that this is strongest in the East ore body and becomes progres- sively weaker through the Main and West ore bodies. Detailed electron microprobe studies and spectro- graphic analysis of mineral separates show that tungsten occurs in small discrete grains of scheelite, and boron is concentrated in clay minerals. The form of the tellurium and selenium has not been established. We believe that small amounts of these elements could be present in hydrothermal pyrite or could occur together with the gold, mercury, arsenic, and antimony coatings on surfaces of pyrite grains. Multiple correlation coefficients for boron, tellurium, TABLE 10.—Linear correlation coefficients between gold, boron, tellurium, selenium, and tungsten for the West Main, and East ore bodies at the Carlin gold deposit West ore body Main ore body East ore body Correlation coefficient Correlation coefficient Correlation Element pair coefficient Element pair Element pair Au to B ______ 0.25 Au to Te ______ 0.34 Au to Te ______ 0.50 Au to Se ______ —.15 Au to Se ______ -.13 W to Se ______ .12 W to Se ______ —.13 Au to B ______ .09 W to Te ______ —.07 Se to B ______ —.12 W to To ______ —.09 B to Te ______ .06 Au to W ______ —.04 W to Se ______ —.09 Au to B ______ —.06 W to B ________ 02 Se to Te ______ .08 Se to Te ______ —.06 W to Te ______ 0 Au to W ______ .08 B to Se ______ .04 Se to Te ______ 0 B to W ______ — .06 B to W ________ .04 Au to Te ______ 0 B to Te ______ .03 Au to W ______ .02 B to Te ______ 0 B to Sc ______ .02 Au to Se ______ .02 20 STATISTICAL STUDY OF TRACE ELEMENTS, GEOLOGY AND GENESIS, CARLIN GOLD DEPOSIT selenium, and tungsten in each ore body are: West ore body Main ore body East ore body W:0.29 B:0.39 Se:0.51 Se:0.28 W:0.38 W:0.51 B:0.25 Se:0.37 B:0.51 Te:____ Te:0.33 Te:0.50 Although the multiple correlation coefficients are stronger in the East ore body than in the West and Main ore bodies, they are too low to be useful in predicting the average gold content. Results of discriminant analyses performed to evaluate the distribution of the five elements between the three ore bodies are shown below. The values are the percent of samples taken from the vertically listed ore bodies that are closest to the mean of the horizontally listed ore bodies. Step 1, boron: East Main West East ________________ 53.6 26.4 20.0 Main ________________ 36.11 41.67 22.22 West ________________ 35.9 0.0 64.1 Step 2, tellurium and boron: East Main West East ________________ 24.8 14.4 60.8 Main ________________ 31.94 18.06 50.0 West ________________ 0.0 0.0 100.0 Step 3, selenium, tellurium, and boron: East Main West East ________________ 28.8 48.0 23.2 Main ________________ 12.5 63.89 23.61 West ________________ 5.13 23.08 71.79 Step 4, gold, selenium, tellurium, and boron: East Main West East ________________ 24.8 46.4 28.8 Main ________________ 12.5 65.28 22.22 West ________________ 10.26 15.38 74.36 Step 5, tungsten, gold, selenium, tellurium, and boron: East Main West East ________________ 26.4 41.6 32.0 Main ________________ 12.5 62.5 25.0 West ________________ 10.26 15.38 74.36 These analyses again showed the West ore body to be significantly different from the other two ore bodies; the abundances of these elements in samples within the West ore body are consistently strongly polarized toward the mean values for that ore body. Any cause and effect relation between the statistical data and the paragenetic sequence can be much better understood when more information is available on the forms or occurrences of tellurium, selenium, boron, and tungsten in these ores. REFERENCES CITED Ahrens, L. H., 1954, The lognormal distribution of elements, I: Geochim. et Cosmochim. Acta, v. 5, p. 49—73. Akright, R. L., Radtke, A. S., and Grimes, D. J ., 1969, Minor elements as guides to gold in the Roberts Mountains Formation, Carlin gold mine, Eureka County, Nevada, in Canney, F. C., and others, eds., Proc. Internat. Geochem. Explor. Symposium, 1968: Colorado School of Mines Quart, v. 64, no. 1, p. 49—66. Dickson, F. W., Radtke, A. S., and Rye, R. 0., 1975, Implications ofthe occurrence of barium minerals and sulfur isotopic compositions of barite on late-stage processes in Carlin-type gold deposits: Geol. Soc. America, Abs. with Programs, v. 7, no. 5, p. 604—605. Dickson, F. W., and Tunnel], George, 1968, Mercury and antimony deposits associated with active hot springs in the western United States, in Ridge, J. D., ed., Ore deposits of the United States 1933—1967, Graton-Sales, v. 2: New York, Am. Inst. Mining, Metall. and Petroleum Engineers, Inc., p. 1673—1701. Dixon, W. J., 1964, Biomedical computer programs: Los Angeles, Univ. California, Dept. Preventive Medicine and Public Health, School of Medicine, 585 p. Graf, D. L., 1960, Geochemistry of carbonate sediments and sedi .nentary carbonate rocks—Pt. 3, Minor element distribution: Illinois Geol. Survey Div. Circ. 301, 71 p. Hardie, B. S., 1966, Carlin gold mine, Lynn district, Nevada: Nevada Bur. Mines Rept. 13, Pt. A, p. 73—83. Harris, M., 1974, Statistical treatment of selected trace elements in unoxidized gold ores of the Carlin gold deposit, Nevada: Stanford Univ., Stanford, Calif., Dept. of Applied Earth Sciences, M.S. thesis, 66 p. Hausen, D. M., and Kerr, P. E., 1968, Fine gold occurrence at Carlin, Nevada, in Ridge, J. D., ed., Ore deposits of the United States 1933—1967, Graton-Sales, v. 1: New York, Am. Inst. Mining, Metall. and Petroleum Engineers, Inc., p. 908—940. Helgeson, H. C., and Garrels, R. M., 1968, Hydrothermal transport and deposition of gold: Econ. Geology, v. 63, no. 6, p. 622—635. Joralemon, Peter, 1951, The occurrence of gold at the Getchell mine, Nevada: Econ. Geology, v. 46, no. 3, p. 267—310. Kennedy, G. C., 1950, A portion of the system silica-water: Econ. Geology, v. 45, p. 629—653. Krumbein, W. C., 1959, The “sorting out” of geologic variables illustrated by regression analysis of factors controlling beach firmness: Jour. Sed. Petrology, v. 29, no. 4, p. 575—587. Krumbein, W. C., and Graybill, F. A., 1965, An introduction to statistical models in geology: New York, McGraw-Hill Co., 475 p. Lepeltier, Claude, 1969, A simplified statistical treatment of geochemical data by graphical representation: Econ. Geology, v. 64, no. 5, p. 538—550. Miesch, A. T., 1967, Methods of computation for estimating geochemical abundance: U.S. Geol. Survey Prof. Paper 574—B, p. Bl—B 14. Radtke, A. S., 1973, Preliminary geologic map of the Carlin gold mine, Eureka County, Nevada: U.S. Geol. Survey Misc. Field Studies Map MF—537, scale 1:4,800. 1974, Preliminary geologic map of the area of the Carlin and Blue Star gold deposits, Eureka County, Nevada: U.S. Geol. Survey Misc. Field Studies Map MF—552, scale 1212,000. Radtke, A. S., and Brown, G. E., 1974, Frandicksonite, Ban, a new mineral from Nevada: Am. Mineralogist, v. 59, nos. 9—10, p. 885—888. Radtke, A. S., Dickson, F. W., and Rytuba, J., 1974, Genesis of disseminated gold deposits of the Carlin type: Geol. Soc. America, Abs. with Programs, v. 6, no. 3, p. 239—240. Radtke, A. S., Heropoulos, C., Fabbi, B. P., Scheiner, B. J., and Essington, G. M., 1972, Data on major and minor elements in host rocks and ores, Carlin gold deposit, Nevada: Econ. Geology, v. 67, no. 7, p. 975—978. Radtke, A. S., and Scheiner, B. J., 1970, Studies ofhydrothermal gold deposition (I)—Carlin gold deposit, Nevada: The role of car- REFERENCES CITED 2 1 bonaceous materials in gold deposition: Econ. Geology, v. 64, no. 2, p. 87—102. Radtke, A. S., Taylor, C. M., and Christ, C. L., 1972, Chemical distribution of gold and mercury at the Carlin deposit, Nevada [abs]: Econ. Geology, v. 67, no. 7, p. 1009. Radtke, A. S., Taylor, C. M., and Heropoulos, C., 1974, Antimony— bearing orpiment, Carlin gold deposit, Nevada: US. Geol. Survey Jour. Research, v. 1, no. 1, p. 85—87. Rytuba, J. J ., and Dickson, F. W., 1974, Reaction of pyrite+pyr- rhotite+quartz+gold with NaCl—H20 solutions, 300—500°C, 500—1,500 bars, and genetic implications [abs]: Internat. Assoc. Genesis Ore Deposits, 4th Symposium, Varna, Bulgaria, 1974, p. 312—313. Seward, T. M., 1973, Thio complexes of gold and the transport of gold in hydrothermal ore solutions: Geochim. et Cosmochim. Acta, v. 37, p. 379—399. Turekian, K. K., and Wedepohl, K. H., 1961, Distribution of the elements in some major units of the earth’s crust: Geol. Soc. America Bull, v. 72, no. 2, p. 175—192. Weissberg, B. G., 1970, Solubility of gold in hydrothermal alkaline sulfide solutions: Econ. Geology, v. 65, p. 551—556. Weissberg, B. G., Dickson, F. W., and Tunnell, George, 1966, Solubility of orpiment (A5283) in Nags-H20 at 50—200°C and 100—1,500 bars, with geological implications: Geochim. et Cosmochim. Acta, v. 30, p. 815—827. Wells, J. D., Stoiser, L. R., and Elliott, J, E., 1969, Geology and geochemistry of the Cortez gold deposit, Nevada: Econ. Geology, V. 64, no. 5, p. 526—537. Wrucke, C. T., and Armbrustmacher, T. J ., 1975, Geochemical and geologic relations of gold and other elements at the Gold Acres open-pit mine, Lander County, Nevada: US. Geol. Survey Prof. Paper 860, 27 p. fiGPO 695—905 #5] 7 DAYS p % Petrology, Mineralogy, m and Geochemistry of the :5“? East Molokai Volcanic Series, Hawaii GEOLOGICAL SURVEY PROFESSIONAL PAPER 961 ,-W‘ ,_ <«»F:tv-M»Hx .r , v — 7' N‘“ ’53; , « rx'.” I“ g; . N3 (“sill Ur Mimi/:7,” 1‘; \ :, “(a DOCUMENTS DEPARTMENT JUL 9 1976 LIBRARY UNIVERSITY OF CAUFORNM Petrology, Mineralogy, and Geochemistry of the East Molokai Volcanic Series, Hawaii By MELVIN H. BEESON GEOLOGICAL SURVEY PROFESSIONAL PAPER 961 UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON: 1976 UNITED STATES DEPARTMENT OF THE INTERIOR THOMAS S. KLEPPE, SMWHH)’ GEOLOGICAL SURVEY V. E. McKelvey, Director Library of Congress catalog-card No. 76—12795 For sale by the Superintendent of Documents, U.S. Government Printing Office Washington, DC. 20402 Stock Number 024—001—02810—7 CONTENTS Page Page Abstract __________________________________________________ 1 Phenocryst-free compositions ______________________________ 14 Introduction ______________________________________________ 1 Mineralogy ______________________________________________ 21 Previous investigations ____________________________________ 3 Compositional variation of olivine ______________________ 21 Purpose and scope ________________________________________ 4 Compositional variation of clinopyroxene ______________ 25 Analytical procedures ____________________________________ 4 Compositional variation of plagioclase __________________ 27 Stratigraphy and petrology ________________________________ 6 Compositional variation of opaque minerals ____________ 28 Whole—rock compositions __________________________________ 11 Summary ________________________________________________ 30 References ________________________________________________ 33 FIGURE 1. 2 3. 4 01 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. ILLUSTRATIONS Page Index map of Hawaiian Archipelago _______________________________________________________________________________ 1 . Geologic map of island of Molokai showing location of Kalaupapa section _____________________________________________ 2 Oblique aerial photograph of trail along which Kalaupapa section was sampled _______________________________________ 5 . Photograph of thin-bedded flows of tholeiitic basalt exposed along north coast of Molokai just west of mouth of Waikolu Valley _______________________________________________________________________________________________________ 7 . Histogram showing phenocryst content of lavas of Kalaupapa section ________________________________________________ 8 . Photomicrographs of a resorbed olivine phenocryst from Kalaupapa section showing intratelluric alteration and fresh overgrowth ___________________________________________________________________________________________________ 9 . Photomicrograph of olivine phenocryst from Kalaupapa section showing growth resorption, alteration, regrowth, resorption, and further alteration _____________________________________________________________________________ 10 . Photomicrographs of plagioclase phenocryst from Kalaupapa section showing tabular opaque (magnetite) inclusions oriented parallel to (010) twin planes _________________________________________________________________________ 10 . Total alkali-silica diagram of dry-reduced whole-rock compositions from Kalaupapa section ___________________________ 12 Olivine. nepheline, silica diagram of dry-reduced whole-rock compositions from Kalaupapa section _____________________ 12 AFM diagram of dry-reduced whole-rock compositions from Kalaupapa section _______________________________________ 13 Magnesia variation diagram of dry—reduced whole-rock compositions from Kalaupapa section _________________________ 14 Graph showing weight percent of oxides in dry—reduced whole-rock compositions from Kalaupapa section plotted against relative stratigraphic position _________________________________________________________________________________ 16 Total alkali-silica diagram of phenocryst-free compositions from Kalaupapa section ___________________________________ 17 AFM diagram of phenocryst-free compositions from Kalaupapa section _______________________________________________ 17 Magnesia variation diagram of phenocryst-free compositions from Kalaupapa section _________________________________ 18 Graph showing weight percent of oxides in phenocryst—free compositions from Kalaupapa section plotted against relative stratigraphic position _________________________________________________________________________________________ 21 Total iron-variation of diagram of olivine from Kalaupapa section ___________________________________________________ 22 Graph showing weight percent of oxides in olivines from Kalaupapa section plotted against relative stratigraphic position _____________________________________________________________________________________________________ 23 Diagram showing composition of clinopyroxenes in Kalaupapa section on pyroxene quadrilateral _______________________ 25 Diagram showing compositional variation of clinopyroxenes in subsections of Kalaupapa section plotted on a part of pyroxene quadrilateral _______________________________________________________________________________________ 25 Graph showing weight percent of oxides of clinopyroxene from Kalaupapa section _____________________________________ 26 Albite, orthoclase, anorthite diagram of plagioclase from Kalaupapa section _________________________________________ 28 Graph showing weight percent of oxides in plagioclase from Kalaupapa section plotted against relative stratigraphic position _____________________________________________________________________________________________________ 29 Graph showing variation of A1203, Ti02, MgO, and ‘FeO’ plotted against Cr203 in opaque minerals from Kalaupapa section _______________________________________________________________________________________________________ 31 Diagrams showing compositional variation of opaque minerals from Kalaupapa section with respect to FeAl2 04—FezTiO4 FeCr204 - MgA1204 — MngiO4 - MgCr204 trigonal prism _______________________________________________________ 32 III IV TABLE CONTENTS TABLES Page Chemical analyses of rocks from the Kalaupapa section ____________________________________________________________ 36 . Chemical analyses of rocks on a phenocryst-free basis from the Kalaupapa section __________________________________ 38 . Chemical analyses of olivines from the Kalaupapa section __________________________________________________________ 40 . Chemical analyses of olivine alteration rims from the Kalaupapa section ____________________________________________ 42 . Chemical analyses of clinopyroxenes from the Kalaupapa section __________________________________________________ 45 . Chemical analyses of plagioclases from the Kalaupapa section ______________________________________________________ 48 . Chemical analyses of opaque minerals from the Kalaupapa section __________________________________________________ 51 PETROLOGY, MINERALOGY, AND GEOCHEMISTRY OF THE EAST MOLOKAI VOLCANIC SERIES, HAWAII By MELVIN H. BEESON ABSTRACT The Kalaupapa section, a 305-m (1,000-ft)—thick section of mostly porphyritic lavas that range from transitional tholeiitic to alkalic to distinctly alkalic, is exposed along the trail that leads up the cliff southwest of Kalaupapa on East Molokai Volcano, Hawaii. On the basis of whole-rock composition, the section was divided into eight subsections characterized by enrichment upward in the section in NaZO, K20, and A1203 and depletion in MgO, NiO, and Cr203. En- richment in Na20, K20, and P205 and depletion in NiO and Cr203 are also evident upward in the section as a whole. Olivine and clinopyroxene phenocryst content of the lavas decreases and plagio- clase phenocryst content increases (with minor variations) upward in the subsections. Analyses of phenocrysts and some groundmass minerals also show variation up the section. Clinopyroxene increases upward in NaZO and T102 content and decreases in MgO; olivine increases upward in FeO, CaO, and MnO content and decreases in MgO; and plagioclase increases upward in NaZO and K20 content and decreases in A1203 and CaO. Plagioclase also shows an upward increase in K20 in the subsections. Phenocryst-free compositions were determined by subtracting phenocrysts from the whole-rock compositions on the basis of their modal amounts and compositions. The subsections defined on the basis of whole—rock compositions also remain intact when defined on the basis of phenocryst-free compositions; therefore, not all the chemi— cal variation in the subsections can be attributed to variation in phenocryst content. The subsections are believed to represent distinct magmatic batches similar to the ones that were proposed to explain the chemical variation of tholeiites and the eruptive histories of Hawaiian vol— canoes and that were recently confirmed from a detailed study of the lavas of Kilauea. Concentrations of phenocrysts in some lavas mask trends in composition. Calculation of phenocryst—free compositions has, to a first approximation, removed variation of the whole-rock compositions resulting from shallow fractional crystallization and has revealed earlier (and deeper) fractionation dominantly controlled by aluminous clinopyroxene. The more fractionated character of the subsections in the upper part of the section, starting with olivine basalt and ending with mugearite, perhaps reflects changing condi- tions in the zone of magma generation prior to phenocryst growth. INTRODUCTION The island of Molokai, a volcanic doublet, is the fifth in size of the islands of the Hawaiian Archipelago, which extends about 2,500 km (1,554 mi) southeast- ward across the central Pacific Ocean from Kure on the northwest to Hawaii on the southeast (fig. 1). The broadly linear chain of islands that make up the ar- chipelago caps the Hawaiian Ridge. The ridge is bor- dered by the Hawaiian Moat (or Deep) which is in turn bordered by the Hawaiian Arch (or Rise). These three topographic features, the ridge, moat, and arch, are superposed on a broad elongate topographic high, the Hawaiian Swell (Betz and Hess, 1942; Dietz and Menard, 1953). Dana (1849, 1890) recognized from geomorphic evidence that extinction of the chain pro— gressed from northwest to southeast, and McDougall (1964) confirmed the general progression with potassium—argon ages of lavas from Kauai, Oahu, Molokai, Maui, and Hawaii. The general progression of ages of the Hawaiian—Emperor chain and its relation to Cenozoic circumpacific tectonics have been described by Jackson, Silver, and Dalrymple (1972) and Dalrymple, Silver, and Jackson (1973). Fiske and Jackson (1972) investigated the influence of regional structure and local gravitational stresses on the orientation and growth of Hawaiian volcanic rifts. Macdonald and Ab- bott (1970) gave an excellent summary of the geo- morphology of the Hawaiian Islands and of the proces- ses that formed and are continuing to form them. Their summary contains a geologic sketch of each of the major 170° 160° 150° I | /Kure 0 vMidway 180° 30 1000 MILES 0 1000 KILOMETRES HAW’q/MN Nihoa éKauai . AMolokai Oahu/Q ‘mgMaui .Hawaii — PACIFIC OCEAN . J FIGURE 1.——Index map of the Hawaiian Archipelago (after Jackson, 1968, Fig. 1). 4 ”CH/PE 20 AGO 10° 2 EAST MOLOKAI VOLCANIC SERIES windward islands (extending from Kauai to Hawaii) and the leeward islands (extending from Kure to N ihoa) as a group. Molokai is an elongate island 78><16 km (48.5X10 mi), with its long dimensions oriented east-west (fig. 2). It is made up oftwo coalescing volcanoes, West Molokai1 Volcano, which rises 421 m (1,380 ft) above sea. level, and East Molokai Volcano, which rises 1,515 m (4,970 ft) above sea level. Both East and West Molokai Vol- canoes are elongate in plan reflecting the structural control of several major rift zones. West Molokai has three rift zones oriented due east, northwest, and west— southwest. East Molokai probably was controlled by stress patterns set up in the edifice of the older West Molokai Volcano whose sloping apron it pierced (Fiske and Jackson, 1972). West Molokai is believed by Fiske and Jackson (1972) to have grown as an “isolated” edifice that was influenced by the regional structure of 1West Molokai Volcano is also referred to as Mauna Loa, but this name is not used here to avoid confusion with the better known Manna Lna of the island of Hawaii. the Pacific basin. However, if the prolongation of the west-southwest rift of West Molokai Volcano (Penguin Bank) is an older independent shield built along the same trend as the main rift zone of West Molokai Vol- cano (Macdonald and Abbott, 1970), then the orienta- tion of the rift system of West Molokai reflects gravita- tional stress patterns set up in this older independent shield. The north coast of Molokai, especially its eastern part, is bounded by a spectacular pali (cliff) that is locally more than 915 In (3,000 ft) high. Several deep amphitheater-shaped valleys, notably Wailau, Pele- kunu, and Waikolu, have been cut into the pali. Two of these, Wailau and Pelekunu, dominate the drainage system of the northern coast, probably because they head in the highly altered, and therefore easily eroded, rocks of the caldera complex. The drainage system on the south (leeward) flank of East Molokai is not so de- eply incised as the drainage system of the northern coast, because it receives far less rainfall. Flat-topped interfluves along the south flank permit reasonably re- ?) a>; llio Point ¢ E (:5 1); KALAUPAPA > > E SECTION 3 3 > o C : Kalaupapa i‘ i g m 1) (U E g 3 Halawa Valley .'. Yr / ° (9 Mokuhooniki Volcano l 05' 056° 45' Kaunakakai 21 05' P 157 00' “40/170 , . 0 55 Kamalo 50 0 2 4 MILES 06:4]‘7 t—r‘l—l——' 0 2 AKILOMETRES EXPLANATION SEDIMENTARY ROCKS VOLCANIC ROCKS East Molokai Volcanic Series ‘W’W ent >4 7/ Vent M q ., “v a . a ' M 3 W” a: 0/”: 5 Beach sand 8 13 § E03 Lithified dunes 8 Kalaupapa Basalt E Upper andesite member :53) § 5% 72 73> L; V t o e I. a 3: Younger and E a < H n >. older alluvium . _ , 3 , ‘ a: Dunes Marine limestone Mokuhoomki tuff cone Lower basalt Caldera complex _ 5 member E-¢ % EROSIONAL UNCONFORMITY l-‘ Contact West Molokai Volcanic Series a——— Fault Basalt J Note: Dikes not shown Bar and ball on downthrown side FIGURE 2.—Geologic map of island of Molokai showing location of Kalaupapa section (after Stearns, 1946, fig. 18; Stearns and Macdonald, 1947). PREVIOUS INVESTIGATIONS 3 liable reconstruction Of the preerosional edifice, which shows that, disregarding subsidence, East Molokai was never much higher than its present 1,515-m (4,970-ft) elevation. However, considering possible subsidence Of 2,300 mi500 m (7,550i 1,640 ft), calculated from data of Moore (1970), it may have reached an elevation of 3,815i500 m (12,500:1,640 ft), nearly as high as Mauna Loa and Mauna Kea on the island of Hawaii. West Molokai, the smaller and Older of the two Molokai volcanoes, is made up of thin-bedded flows of tholeiitic basalt with many thin (<1/3 m) interlayers of vitric tuff. It never formed a caldera or had significant eruptions of alkalic basalt (Macdonald and Abbott, 1970). East Molokai, on the other hand, has a well—developed caldera stage and erupted about 490 m (1,610 ft) of transitional and alkalic lavas in the Kalaupapa area some 70 m (230 ft) of which are mugearite and hawaiite (the upper member of Stearns and Macdonald (1947)). McDougall (1964) dated the mugearites of the upper member at 1.31-1.41 m.y. (million years) and several flows in the upper part of the alkalic basalt flows (the lower member of Stearns and Macdonald (1947)) at 1.48—1.49 m.y. The tholeiitic lavas of East Molokai have not yet been dated, and so the absolute chronology of the major parts of the two edifices remains in doubt. PREVIOUS INVESTIGATIONS The earliest work on the petrology of Molokai was done by M'Ohle (1902), who studied 50 samples from various localities scattered over the island but concen- trated mostly along the northern pali (cliff) in the vicin- ity of Kalaupapa. M'Ohle (1902) considered the lavas of Molokai to be Of typical basaltic character consisting of plagioclase, augite, Olivine, and magnetite. He distin- guished three groups connected by intermediate rocks: (1) Olivine-bearing basalts in which olivine content is not high, (2) rocks in which Olivine is rare or entirely lacking, and (3) rocks in which Olivine is abundant and augite rare or lacking. He also reported a nepheline- bearing rock from Kalae. Stearns and Macdonald (1947) reported rocks corresponding to the first and second groups but found none that correspond to MIOhle’s third group on Molokai or any of the other Hawaiian islands, Because they did not find any nepheline-bearing rocks on Molokai, they concluded that some of the rocks studied by MOhle (1902) were mislabeled and actually came from Oahu. Lindgren (1903) made the first field study of Molokai to evaluate the water resources but paid little attention to petrology, except to note that the rocks were normal somewhat glassy feldspar basalts containing Olivine and a few plagioclase phenocrysts. While exploring the headwaters of Wailau Stream, he did note a coarse-grained intrusive rock that he termed a very coarse grained Olivine diabase. Powers (1920) visited all the windward islands and judged Wailau (East Molokai) Volcano to consist largely of feldspar basalt with rare alkali trachyte. He was the first to recognize that the Kalaupapa peninsula, north of the pali, is a young olivine basalt cone and not part of a sunken area as Lindgren (1903) believed. The most comprehensive petrographic study of Molokai was made by Stearns and Macdonald (1947). Although their report was primarily concerned with ground water resources of Molokai, it gives the most detailed account of the geology and the only geologic map of Molokai to date. The lavas of West Molokai Volcano are described as basalts, Olivine basalts, and picrites of primitive type. Because the flows are very thin, less than 1 m thick, and are poorer in Olivine phenocrysts than is common in Hawaiian tholeiites, they are considered by Stearns and Macdonald (1947) to have been erupted in a very fluid condition. Only a thin discontinuous cap of alkalic olivine basalt and hawaiite is reported on West Molokai (Macdonald and Abbott, 1970). Stearns and Macdonald (1947) reported the petrology of the stratigraphic section measured along the trail that leads down the great northern pali southwest of Kalaupapa—the subject of this study. This strati- graphic section, considered by them to be characteristic of the East Molokai Volcanic Series, is designated its type section and is herein referred to as the Kalaupapa section (fig. 2). Stearns and Macdonald (1947) reported that augite phenocrysts are about as abundant as olivine pheno- crysts and that these phenocrysts constitute 40—60 percent of the rock. However, no flows Observed in the Kalaupapa section that contain more than 5 percent Olivine phenocrysts have nearly equal abundances of augite and Olivine phenocrysts. Augite phenocrysts are generally only about a third as abundant as olivine, and in only one sample does the augite phenocryst content amount to about 60 percent of the Olivine phenocryst content. Phenocrysts constitute more than 20 percent of many rocks, but none in which modes were made exceed 30 percent Of the rock, and none that I Observed ap- proach the 40—60 percent reported by Stearns and Mac- donald (1947). The general life history of Hawaiian volcanoes has been divided into four stages (Macdonald and Abbott, 1970; modified from Stearns, 1946): 1. Youthful stage, characterized by a shield, consisting almost entirely of thin basalt flows, that has been built by frequent eruption of very fluid tholeiitic lavas. The lavas of this stage range from basalt almost free of Olivine phenocrysts to oceanites con— 4 EAST MOLOKAI VOLCANIC SERIES taining more than 50 percent olivine phenocrysts (Macdonald and Katsura, 1962; Macdonald and Abbott, 1970). 2. Mature stage, characterized by a caldera complex that formed by repeated collapse of the summit area and infilling of the resulting caldera by lava. Activity during the mature stage was still Vigor- ous, and many of the tholeiitic lava flows produced were thicker than in the youthful stage because they were ponded in the caldera. 3. Old stage, characterized by a cap of alkalic basalt that was built over the top and upper flanks of the shield and that thins away from the summit. Erup- tions at this stage were less frequent and more explosive and produced thicker flows and steeper slopes than the earlier tholeiitic eruptions. The lava flows of this stage are commonly separated by weathered zones and a few soil zones or local ero- sional unconformities. 4. Rejuvenated stage, following a long period of vol- canic quiescence. The flows of this stage are sep- arated from those of the previous stage by a profound erosional unconformity. Many occur as intracanyon flows. Most lavas of this stage are critically undersaturated and contain nepheline or nepheline and melilite (Winchell, 1947; Jack- son and Wright, 1970; Macdonald and Abbott, 1970). This history is an idealized scheme, for not all vol- canoes show the entire sequence and the sequence may be terminated at any stage. West Molokai Volcano, for example, ceased eruption before reaching the mature stage and has no caldera and few alkalic flows. East Molokai Volcano, on the other hand, has as complete a sequence as any of the volcanoes in the Hawaiian chain. It is probably just entering the rejuvenated stage. Post— erosional eruptions have formed the Kalaupapa Penin- sula and Mokuhooniki Island, but unlike most of the posterosional lavas of Kauai (the Koloa Volcanic Series) and Oahu (the Honolulu Volcanic Series), these contain no nepheline. PURPOSE AND SCOPE This study was undertaken to determine what, if any, systematic changes in chemistry and petrology of the lavas could be detected through the stratigraphic sec- tion and how removal of phenocrysts from the whole- rock compositions affects the fractionation trends. All too frequently, chemical analyses from an area are plot- ted with little or no stratigraphic control, making it impossible to relate the presumed fractionation trends to stratigraphy. Because analysis of all phenocryst min— erals was required to determine the phenocryst-free compositions, an evaluation of the variation of mineral compositions in different parts of the section was also possible. Most of the original plan to sample lavas of the tholeiitic suite and work upward into the alkalic suite has been accomplished. The flows in the lower part of the section are petrologically alkalic but are composi- tionally tholeiitic. They are probably best classified as transitional between tholeiitic and alkalic. The flows in the upper part of the section are unquestionably alkalic. Several stratigraphic sections were sampled, but de- tailed work has been done only on the Kalaupapa sec- tion (fig. 3). A reasonably complete sampling of the upper 305 m (1,000 ft) ofthe section was made, but only sparce sampling was done in the lower 180 m (600 ft), which is largely covered by talus. Flows presumed to be tholeiitic have been sampled in Pelekunu and Waikolu Valleys, but limited time and difficult topography prevented sampling of a complete section. None of the rocks of the caldera complex that are exposed in Pelekunu and Wailau Valleys have been included in this study, but they merit a study in them- selves inasmuch as the structure of this exhumed cal- dera complex appears to be similar to the one developing on Kilauea. Acknowledgments.—Grateful appreciation is ex- pressed to those in US Geological Survey who provided technical assistance during this study. V. C. Smith and R. L. Rahill performed chemical analyses, and Keith Bargar separated and X-rayed olivine alteration prod- ucts. Helpful discussions and critical reviews of E. D. Jackson, R. G. Coleman, K. J. Murata, and T. L. Wright ofthe US. Geological Survey, and ofA. C. Waters, R. S. Coe, and O. T. Tobisch of the University of California, Santa Cruz, are sincerely appreciated. Field assistance by my wife, Naomi K. Beeson, is gratefully acknowl- edged. ANALYTICAL PROCEDURES After preliminary petrographic examination of all the samples taken from the Kalaupapa section, 26 sam- ples from 24 flows were selected for further study. Whole-rock compositions were determined in the analytical laboratories of the US. Geological Survey in Denver, Colo., and phenocryst and groundmass miner- als were analyzed by the author. In order to assure that the phenocryst modes would be representative of the sample analyzed for whole-rock chemistry, two slabs were taken from opposite sides of the samples submitted for whole-rock analysis. These two slabs were fine ground with 600-mesh abrasive, sprayed with plastic, and inscribed with a 1-cm2 grid. Using a 10X10 mm reticle, 121 points were counted in each of the 1-cm squares. A drop of distilled water between the rock slab and the reticle significantly improved the Visibility of ANALYTICAL PROCEDURES papa peninsula showing trail along which Kalaupapa section pward. Pali is about 490 m (1,610 ft) high. View to south. FIGURE 3.—-Oblique aerial photograph of pali near Kalau was sampled. A~Z, sample localities from base of section u Photograph by U.S. Navy. 6 EAST MOLOKAI VOLCANIC SERIES the phenocrysts. Following the procedure of Jackson and Ross (1956), a minimum of 1,000 points was counted for each sample, and with few exceptions, an area at least 100 times that of the largest phenocryst was co- vered. The volume percent mode was converted to weight percent using an estimated average density of 3.5 for olivine, 3.35 for clinopyroxene, 2.7 for plagio- clase, and 2.9 for the groundmass. The mineral densities were calculated from end-member densities and aver- age phenocryst compositions, and the groundmass den- sity was estimated from densities of crystalline rocks (Daly and others, 1966). X-ray diffractograms of olivine alteration rims were made with CuKa radiation using a Phillips X-ray generator and diffraction unit equipped with a graphite crystal monochrometer. Mineral analyses were made with an ARL EMX-SM electron microprobe using natural mineral standards. All data were corrected for drift, background, matrix absorption, characteristic fluorescence, and atomic number effects (Beeson, 1967; Beaman and Isasi, 1970). During analysis of pheno- crysts, the beam was scanned rapidly over a 10>< 10 ,um raster, which helps to even out inhomogeneities and gives a better average composition. The mineral analyses are believed accurate to :2 percent of the amount present for major elements and :10 percent of the amount present for minor elements. The phenocryst compositions were subtracted from the whole-rock composition in proportion to their modal abundance. Because some of the olivine phenocrysts are altered, it was deemed advisable to make modes for the amounts of fresh and altered olivine and subtract ap- propriate amounts of fresh and altered olivine. This procedure appears to have been justified because two samples, I and M, with relatively abundant and the most severely altered olivine phenocrysts, do not depart appreciably from the trend established by the samples with relatively few olivine phenocrysts and those in which olivine phenocrysts are only slightly altered. It is recognized that subtraction of phenocryst com- positions derived by microprobe analysis yields only a first approximation to phenocryst-free lava composi- tions because of the inhomogeneity of the phenocrysts as a whole. Neither chromite inclusions in olivine, magnetite inclusions in plagioclase, nor other inclu- sions in the phenocrysts have been taken into account. STRATIGRAPHY AND PETROLOGY Three stratigraphic sections of the East Molokai Vol- canic Series were sampled on a flow-by-flow basis: one along the trail to Kalaupapa, one along the road leading into Halawa Valley from the south, and one up Lama Loa Head north of Halawa Valley. The Kalaupapa sec- tion was selected for detailed study because it is thickest and exposes rocks lower in the volcanic pile than the other two. The two sections in Halawa Valley contain fewer olivine phenocrysts relative to plagioclase and are much like the upper part of Kalaupapa section in that respect. No attempt was made to trace the flows laterally because of the steep slope, but samples were taken up the trail on the assumption that vertical position can be directly correlated with stratigraphic position. Such an assumption is reasonable because no evidence of ero- sional unconformities was seen and in general, even in a late mature stage, the frequency of eruption of Hawai- ian volcanoes is high enough to prevent development of extensive stream channels. Flows are commonly sepa- rated by weathered aa clinkers and in a few places by baked red soil zones, usually less than about 1/3 m (1 ft) thick. Compared with the West Molokai Volcanic Series, there seem to be fewer vitric ash beds separating lava flows, which is curious in light of the fact that the flows in the Kalaupapa section are mostly alkalic, and they tend to be erupted more explosively than tholeiitic ones. Because sills are present in some Hawaii sections, coarsely crystalline dense rocks without clinkers were looked for, but none were found. Presumably none of the flows sampled are really sills. The Kalaupapa section lies about 12 km (6.8 mi) west-northwest of the caldera complex, as shown by Stearns and Macdonald (1947) (fig. 2). On the basis of gravity contours (Moore and Krivoy, 1965) and extrapo— lation of dikes that crop out on the pali due south of Kalaupapa, the section lies about 1 km (.62 mi) south of the west-northwest rift zone. Wright and Fiske (1971) presented a model for the conduit system of the east rift of Kilauea, in which magma from the summit area migrates outward along separate subparallel conduits. The magma is stored in pockets along the rift zone where it may be fractionated to various degrees and later flushed by batches of un- fractionated lava migrating outward from the summit. Lavas in the Kalaupapa section may have originated in this general setting. As a result, the compositional vari- ations may not be typical of summit lavas but rather of lavas from rift-zone eruptions. Nonetheless, any long- term variation in the lavas, not related to fractional crystallization in the shallow rift-zone chambers, should provide information on deeper (more fundamen- tal?) changes in magma composition. It cannot be defin- itely established that all or part of the lavas in the Kalaupapa section were derived from rift—zone erup- tions, but by analogy with Kilauea, most of the flows near the rift zone are probably not summit eruptions. The gentle (2°) slope of the flows away from the rift zone in the Kalaupapa section argues for this rift-zone origin. STRATIGRAPHY AND PETROLOGY 7 It is unlikely that the Kalaupapa section records the _ entire history of the East Molokai Volcanic Series. The period represented by accumulation of this section probably also includes many hiatuses resulting from a statistical distribution of the lavas from a volcano erupting frequently enough to maintain a nearly sym- metrical form. Stearns and Macdonald (1947) divided the East Molokai Volcanic Series into two members: (1) a lower member composed of basalts, olivine basalts, and picrite-basalts of the primitive and ankaramite types (2) and an upper member composed of andesine ande- sites (hawaiites), oligoclase andesites (mugearites), and trachytes. Picrites of the primitive type are reported to be present in the lower part of the lower member, and picrites of the ankaramite type are reported to be pre- sent in the upper part of the lower member. All the “little differentiated” lavas of the lower member are not tholeiitic as McDougall (1964) presumed, though the lavas of the lower part of the lower member, which Stearns and Macdonald (1947 ) described as containing picrites of the primitive type, probably are tholeiitic. The lavas of the upper part of the lower member, which McDougall dated and which Stearns and Macdonald (1947) described as containing picrites of ankaramite type, are distinctly alkalic. No picrites of the primitive type were observed in the upper two-thirds of the Kalaupapa section, which was sampled in detail, or in the lower third, which was only sparsely sampled because it was covered by talus. Sam- ples from the base of the pali about 1 km (0.62 mi) east of the Kalaupapa section were found to contain augite as phenocrysts and in the groundmass but no pigeonite or hypersthene. Petrologically they, and presumably the rest of the flows in the section, are alkalic. Rocks that are apparently tholeiitic crop out at the east side of the mouth of Waikolu Valley. The flows are very thin bedded, 1/3—11/3 In (1—4 ft) thick, with thin unweathered clinkers separating them, and they con- tain almost no olivine either as phenocrysts or in the groundmass. In these respects they resemble the tholeiitic rocks of West Molokai but differ in that they contain no vitric tuff beds between flows. This section of tholeiites(?) (fig. 4) is unusual and may be unique in the Hawaiian Islands. Other basalts, more typical of the Hawaiian tholeiites, are exposed outside the caldera complex in Pelekunu and Wailau Valleys. Approximately 80 flows are present in the Kalaupapa section. The flows of the lower member range in thick- ness from about 1 to 20 m (3—70 ft) and average about 6 m (20 ft). Those of the upper member range in thick- ness from about 6 to 24 m (20—80 ft) and average about 9 m (30 ft). McDougall (1964) obtained potassium-argon ages for t . , FIGURE 4.—Thin-bedded flows of tholeiitic basalt exposed along north coast of Molokai just west of mouth of Waikolu Valley. Average thickness of flows is about 0.7 m (2 ft.) three flows in this section. One hawaiite flow from the upper member at an elevation of 457 m (1,500 ft) gave ages of 1.44—1.46 m.y., and two adjacent flows from the lower member at an elevation of 350 m (1,150 ft) gave ages of 1.47 and 1.49 m.y. for the upper and lower flows, respectively. Extrapolations based on these ages indi- cate that the 490-m (1,610 ft) Kalaupapa section could have accumulated in as little as 0.05 or as much as 0.22 m.y. If it is assumed that the section accumulated in about 0.14 my. (the average of the two extreme ages given above) and that there were 80 flows in the section, then the frequency of eruption was less than 2,000 years. Further extrapolations based on the ages pro- vided by McDougall (1964) indicate that the base of the section could be as young as 1.5 or as old as 1.65 m.y., though the maximum age is not tenable because the natural remanent magnetism polarity of the entire sec- tion is reversed (Sherman Grommé, Edward Mankinen, and Monte Marshall, oral commun., 1972). It is improb- able that the Gilsa event of the Matuyama epoch (from 1.64 to 1.79 m.y., Cox, 1969), a period of 150,000 years, would not be recorded by flows erupting with a 2,000— year frequency. The section was probably built in less than 0.22 my; this agrees with the work of McDougall (1964) and McDougall and Swanson (1972), who con— cluded that any of the volcanic edifices in Hawaii could have been built in 0.5 m.y. or less. Twenty-six samples from 24 flows were selected for detailed petrographic examination, whole—rock chemi- cal analysis, and modal determination of phenocryst 8 EAST MOLOKAI VOLCANIC SERIES content. Electron microprobe analysis of olivine, clinopyroxene, and plagioclase phenocrysts, chromite microphenocrysts, and olivine, clinopyroxene, plagioc- lase, and opaque minerals in the groundmass was made on 25 of these samples. The 26 samples analyzed have been lettered from A to Z from the base of the section upward. The section is not continuously exposed. Sev- eral gaps are present in the record between flows A and B, E and F, and Q and R where one or more flows may be covered by talus. Most of the analyzed samples were taken from near the base of the flows. Two exceptions are sample G, taken from near the top of a flow about 1.2 m (4 ft) thick, and sample I, taken near the center of a flow about 9 In (30 ft) thick. Samples were analyzed from two horizons in two of the flows. These are samples L and M from near the base and top, respectively, of a flow about 9 m (30 ft) thick, and samples 0 and P from near the base and top, respectively, ofa flow about 3.5 In (12 ft) thick. One of the most striking aspects of the lavas in the Kalaupapa section, in comparison with alkalic basalts of most other Hawaiian volcanoes, is the abundance of plagioclase phenocrysts relative to augite and olivine. Most of the flows in the Kalaupapa section are porphy- ritic except for a few in the upper third of the section. The porphyritic flows commonly contain more than 20 percent phenocrysts, but none in which modes were made contained more than 30 percent phenocrysts. This is contrary to the assertion of Stearns and Macdonald (1947) that augite phenocrysts are generally about equal in abundance or slightly more abundant than olivine and together constitute 40—60 percent of the rock. From the geometry of packing of equal-sized spheres alone, these abundances seem too high. For cubic open packing, the phenocryst content would be 52 percent, and for cubic closed packing, it would be 74 percent. If a lava contained 40—60 percent of pheno- crysts, its mobility would be substantially reduced be- cause the phenocrysts would be touching rather than free floating. The section studied was divided into a series of sub- sections on the basis of the whole-rock composition, but division on the basis of phenocryst content would have yielded the same subsections. Olivine phenocrysts are most abundant in the lower flows of the subsections, decreasing upward to almost zero in the uppermost flows of the subsections (fig. 5). In a broad way, the amount of augite phenocrysts decreases with the de- crease in amount of olivine phenocrysts, but augite phenocrysts are usually only about a third as abundant as olivine phenocrysts. Of the 19 samples containing more than 5 percent phenocrysts, only one has appreci- ably more than a third as much augite as olivine phenocryst content; in it, augite is about 60 percent as abundant as olivine. The plagioclase phenocryst con- tent does not vary with olivine content in as systematic a manner, although, in a very crude way, it does in- crease upward in both the entire section and the subsec- tions. None of the flows in this section merit the designation of “ankaramite”2 as defined by LaCroix (1916), because the augite phenocryst content is not equal to, nor does it exceed, that of olivine in any of them, and plagioclase phenocrysts are abundant. Olivine phenocrysts almost invariably have deep-red to reddish-yellow alteration rims usually accounting for 10—20 percent of the grain. These alteration rims follow the outline of the phenocryst even when it is strongly 2LaCroix defines ankaramite as being ”olivine-bearing and melanocratic, rich in pyroxenes up to 1 cm (0.4 in) in diameter, and somewhat fewer and smaller olivines. The compact groundmass is composed of large microlites of'augite and titanomagnetite, and a little biotite and labradorite" (translation from Johannsen, 1938). Olivine Clinopyroxene Plagioclase I 120 L 7< w._|_ Alkalic field Tholeiitic field J——W ’— D ’fl<_uo < __-O§.——— Nepheline Silica FIGURE 10.—Olivine, nepheline, silica diagram of dry-reduced whole—rock compositions from Kalaupapa section. Solid line, field boundary proposed by Poldervaart (1964); dashed line, field bound- ary proposed by Irvine and Baragar (1971). A~Z, samples from base of section upward. tween the petrologic and chemical classification of Hawaiian basalts, and especially of basalts of the Kalaupapa section. The boundary lines proposed by Macdonald (1968) and Poldervaart (1964) are therefore preferred for Hawaiian basalts. A similar discordance between chemical and mineralogic classifications based on feldspar compositions was noted by Keil, Fodor, and Bunch (1972) in lavas from Maui, which they considered to be transitional between tholeiitic and alkalic suites. In general the basalts in the lower part of the Kalaupapa section defy a common chemical and pet- rologic classification and are better considered as tran- sitional between tholeiitic and alkalic. The trend that the lavas from the Kalaupapa section define (fig. 9) is at a steeper angle to the alkalic- tholeiitic field boundary of Macdonald ( 1968) than is the general trend of Hawaiian alkalic lavas that he pre- sented. Conservatively, the trend of the Kalaupapa lavas is about 30° to Macdonald’s alkalic trend. It ap- pears that there is less olivine control3 for the lavas of the Kalaupapa section than for the Hawaiian alkalic l lavas in general and that the trend of the lavas is r0- l tated toward Macdonald’s nephelinic trend. Macdonald 3Powers (1955) drew lines that he called “olivine control" lines, in an oxide-oxide diagram ’ from plots ofolivine phenocrysts into the field of basalts. These olivine control lines indicate 1 the straight—line Variation that would result from changing the amount of olivine in any l plotted rock composition through which the lines are drawn. Comparable control lines can be r drawn for other minerals. Wright (1971) noted that if more than one mineral controls the chemical variation, unique control lines cannot generally be defined. However, the sense in l which deviations occur from lines that define dominantly a single-mineral control may i indicate which additional mineral species is causing the chemical variation. WHOLE-ROCK COMPOSITIONS (1968) compared trends of alkalic and nephelinic lavas from Hawaii with experimental ones determined by Green and Ringwood (1967), pointing out that progress- ing from high to low pressures, the fractionation trend shifts from nephelinic through alkalic to tholeiitic trends of the Hawaiian suites. In terms of mineralogy and considering only Na20+K20 and SiOZ, the control shifts from clinopyroxene through orthopyroxene to oli- Vine+plagioclase. Figure 11 shows the Kalaupapa lavas on an AFM (alkali, iron, magnesium) diagram on which the gen- eral field of Hawaiian lavas has been outlined (Mac- donald, 1968). The Kalaupapa lavas fall to the iron-rich side of the field but totally within it. The compositional range of olivine and augite phenocrysts from these lavas, indicated by arrows, shows that in an AFM dia- gram olivine control cannot be separated from augite control. It was shown earlier that the amount of augite phenocrysts varies directly with that of olivine, and this type of plot emphasizes the trend. It is apparent (fig. 11) that the fractionation trend does not progress continuously from “primitive” lavas with a high MgO/‘FeO’ ratio to a “fractionated” lava with a low MgO/‘FeO’ ratio going from flows A to Z, but rather several successive flows plot along a trend only to be interrupted by a discontinuity marking an abrupt increase in the MgO/‘FeO’ ratio to start anew the frac- tionation trend. Also the talus-covered gaps discussed earlier must be considered when the fractionation pat- tern is being examined. FIGURE 11.—AFM diagram of dry-reduced whole-rock compositions from Kalaupapa section. Outlined area, field of Hawaiian basalts (Macdonald, 1968). A—Z, samples from base of section upward. Ar- rows, compositional range of olivine and clinopyroxene (augite) phenocrysts from baseof section upward. 13 The section has been divided into several subsections on the basis of where abrupt increases in the MgO/‘FeO’ ratio occur. The subsections that have been delimited are A, BCD, EF, GHIJK, LMNOP (L and M, and O and P are samples from different levels of two flows), QRS, TUVW, and XYZ. The gap between A and B is consider- able, and so A must stand alone as a subsection. A small gap between E and F leaves only two samples in this subsection. The samples in subsection QRS show very little variation and conceivably represent surges of the same flow. Subsection XYZ has a discordance at Y marking a regression from hawaiite back to alkalic basalt and probably represents segments of two subsec- tions. This regression is significant because it shows that regression to more “primitive” lavas can occur in all stages of fractionation and not only within the fields of transitional and alkalic basalts. In an approximate way, the lowermost flows of the progressively higher subsections are more fractionated than those of the lower subsections; also the uppermost flows of the progressively higher subsections are more fractionated than those of the lower subsections. The fractionation pattern is thus one of several steps for— ward and a jump back, starting at rocks transitional from tholeiite to alkali olivine basalt and culminating in mugearite. Magnesia variation diagrams of dry-reduced whole— rock compositions (fig. 12) give some idea as to what minerals control the fractionation. The minerals that occur as phenocrysts (olivine, augite, and plagioclase) control the major elements, especially in the lower part of the section. The abrupt drop of ‘FeO’ and T102 in sample Z indicates control by ilmenite or ulvéspinel late in the section. Chromite, which occurs as inclusions in olivine and as microphenocrysts in the groundmass, controls the Cr203 content almost entirely. The sub- sections that were delimited from the AFM diagram correspond precisely to those delimited from magnesia variation diagrams on the basis of the elements that typically are used as fractionation indicators, that is, NaZO, K20, and Ti02 as well as Cr203 and NiO. Variation diagrams have been criticized as indicators of fractionation in recent years, especially by Chayes (1971). Each oxide in the analyses was therefore plotted against relative stratigraphic position (fig. 13) to bypass the “closure restraint” discussed by Chayes. In the sub- sections, SiOZ, A1203, Na20, K20, T102, and P205 in- crease rather consistently and sometimes dramatically. If an envelope were drawn to include all the analyses, it would show that there is a general increase of these oxides up the entire section, substantiating the conclu- sion, indicated by the AFM diagram (fig. 11) that the subsections become progressively more fractionated up section. 14 EAST MOLOKAI VOLCANIC SERIES 4o _ I I I I I I I I _ I I I | I I 20 — _ E 30 - '-‘ .31 3° 15 - — Y=M and N n. C=A Om Y_=M and N By 'rJ IQ§exo N20 _ _ b R—X HL C F DKP 3 wz 0: E G vw x V ”- FQSKPO ‘ 1o _ o _ ,_ c I RD 0° 2 Z HGBY AT’U -\1 LLI E L ‘0 U 10 _ _ d\ I e, 0 Lu 0“ .1: u. *1» _ N‘ _ o 5 O I- A‘ 3 I 'OOQ/ 3 9 Olivine 0“ 8 w 0 ~53 3 0 SI 2 +21“ 0 l3. _. iso I- V Z 20 — -<~° — 55 — I5 o“ I=F “g? g 3:5 %- .3 ° Y=N 3) 1 g» a a E _ 15 - - 5° ' 0 x C=A 3 Vi» [lie/w 0 O I=F 3 $0 c S 8 5:151 d N 81 oN G Y ArJ 1Q X = an .... T D 59 27) E HIB. M u 10 _ HGLY CJ 1 K 9.,— 45 ._ _ E v sx pw 0 z 5 _ _ 4o — _ O - . WK Olivine 0 I l I | I I 35 I I I I I 1 l I 40 30 20 10 0 4O 30 20 10 0 M90 M90 OXIDE CONTENT, IN WEIGHT PERCENT FIGURE 12.—Magnesia variation diagram of dry-reduced whole-rock compositions from Kalaupapa section. A—Z, samples from base of section upward. Italicized symbols indicate coincidence of two or more points. Coincident points shown in left part of each diagram. Arrows, ranges in olivine, plagioclase, and clinopyroxene (augite) compositions from base of section upward and are not to be confused with “control lines.” In a similar manner, three oxides (MgO, NiO, and Cr203) decrease upward in both the subsections and the whole section. NiO and Cr203 may show as much as a 75 percent decrease from the lower to upper flows in a subsection. NiO and Cr203 in these lavas are very sensi- tive indicators of fractionation involving olivine, in which most of the NiO resides, and chromite, in which most of the Cr203 resides. The close parallelism in vari- ation of these two oxides indicates coprecipitation of chromite and olivine in early stages of fractionation. Because of the close relation between the variation of the phenocryst content and the compositional variation of the lavas, it is important to examine the composi- tional variations of the lavas with the contribution made by phenocryst content removed (phenocryst-free compositions) to see if any significant compositional variations in the lavas are masked by phenocryst con- centrations. PHENOCRYST-FREE COMPOSITIONS The phenocryst compositions can be effectively sub- tracted from the whole-rock compositions with mag- nesia variation diagrams (Powers, 1955) if the only phenocryst mineral in the lava is olivine. The lavas from the Kalaupapa section, however, contain as many as three kinds of phenocrysts, and they cannot be so conveniently removed diagrammatically. The pheno- crysts therefore were subtracted from the dry-reduced whole-rock composition on the basis of their modes and compositions. PHENOCRYST-FREE COMPOSITIONS 15 I I I I I I I I 0/. l . \ I=F L=B S=R .2 — V=O — Y=M .9 ,_z Z w E w 1 — E - n. *— I HG (_D N uJ I» 3 LY CT E z AJUIQ 8 —. SD‘ ’50 E Clinopyroxene )CKPVWZE? N 0 o I— P 2 O sx 0 UI RKVW LU J D 9 3 — =M F _ >6 LN AT BY c z 0 c; 2 _ 5 HG _ I— 5:; S Q. o .S 5 1 — O m 2 0 .9 OD Olivine .2 O ——|———I——> I l I I I I IL 45 40 35 30 25 20 15 10 5 0 M90 I I I I I I I I z 1.0 - _ 6“ S=R a Y: PO “ u X ‘tv .5 — IQSK 0 _ M A1 D g LY CT _ 0 H68 :2 E '50 . - . 2 0 Olivme Clinopflgxene m 2 — Z a o S=R a N=M >4 po x vw 1 - U _ QSK 8 LN F o 3 O G AT: 1 a Olivine Clinopyroxgne HE C E 0 ' S=R z :0 a =N 2 4 - .‘e’ pvw an O XK a «6‘ UFQS Z A 1 D LM CT 2 — HGB - 2/ E .15,“ o 9“ o Olivine '6‘ / 0 ———‘I———1—. I I I C} I I I 45 40 35 30 25 20 1 5 10 5 0 M90 OXIDE CONTENT, IN WEIGHT PERCENT FIGURE 12.——-Continued. Compositions and normative minerals of these phenocryst-free lavas are given in table 2. The total alkali-silica diagram for phenocryst-free compositions (fig. 14) shows that six of the samples plot in the tholeii- tic field as compared to eight in the corresponding whole-rock diagram (fig. 9), and most of these six sam- ples have shifted toward the field boundary line of Mac- donald (1968). It is significant that the data show less scatter than in the whole-rock diagram and the trend is at a steeper angle to the alkali trend (rotated more toward the nephelinic trend) of Macdonald than the trend of the whole-rock compositions. The trend of the phenocryst-free compositions can be attributed almost entirely to variation in the amount of clinopyroxene in the lavas and shows little or no olivine or plagioclase control. The AFM diagram (fig. 15) shows that the phenocryst-free compositions, like the dry—reduced whole-rock compositions, plot toward the iron-rich side of the field of Hawaiian rocks as given by Macdonald (1968), with only two falling just outside of this field. Removal of the phenocrysts has not produced any rock compositions for which there are no natural counter- parts in Hawaiian lavas. The main effect of phenocryst removal has been to shorten the range of MgO/‘FeO’ ratios from the high side (compare fig. 11 with fig. 15). This is expected because the part of the trend in the AFM diagram (represented by these compositions) re- sults mainly from augite plus olivine control, and phenocrysts of these two minerals have been removed. The magnesia variation diagrams for phenocryst—free compositions (fig. 16) provide a more complete picture of how removal of phenocrysts has affected the lavas (com- pare the dry—reduced whole-rock magnesia variation diagrams (figs. 12, 16) ). Both Sl02 and A1203 exhibited 16 EAST MOLOKAI VOLCANIC SERIES SAMPLES IN RELATIVE STRATIGRAPHIC POSITION Z 9 I: U) 0 IL 9 I n. < I 9 __-__o__.____ __L__ _____o_ I- ” P P < o o 0 II N N N I- M M M (I) ___L___________ ____ ‘_ _.__ w J '< TIE ‘L K J .I 2 . I I i— H H H < ___G______~_ ____ _G_ __ .J F F ‘T Lu II E E Z _—-fi3_‘fi—_‘*- “T_*_ “—__D_'— _ C C C m __B__________ L___‘ ___a____ m .1 IL 2 A A A < L 4.4 I I I I I I I tn 1 2 3 4 5 1 2 2 3 4 z Na,o K20 Tio2 9 I z I I z I 2! 1712 I I I2 ’— v V v L Y v a X___ ___>$_.‘ __2(__ __L__h __ O W w w w n. v v v v v u u u u u U 4___ _._T_______T_____J________T E S 5 S s s a. R R Fl a a < II 52 J___ ___q___i___o.____o____ __L_ |_ OF P P P P o o 0 g N N N 0 N N |_ M M M M M U) L.____ _L‘__‘___;___-.__I-___.___L_ Lu K K K K K > Jl J J J _ l I I I— H H H I H H < _Q___ __G.______G__-____G_______G_ J F F F F P Lu I L_ _ L__-____E.______‘__E_ ____E_._ Z c D— D D D _ D ‘7, L‘__ __L_____‘_CB______C_B_______E_ u.l _I n. A A E L 1 I l I I A I I A! 1 I Li 1045 50 10 15 20 10 15 0 1O 20 5 10 SiO2 AI203 'FoO' M90 CaO OXIDE CONTENT, IN WEIGHT PERCENT FIGURE 13.—Weight percent of oxides in dry—reduced whole-rock compositions from Kalaupapa section plotted against relative strati- graphic position. A~Z, samples from base of section upward. Dashed horizontal lines, subsections delimited on basis of abrupt increase in ratio of MgO/‘FeO’ and phenocryst content (fig. 5). PHENOCRYST-FREE COMPOSITIONS Alkalic field Tholeiitic field OXIDE CONTENT, IN WEIGHT PERCENT 4o 45 50 $50, OXIDE CONTENT, IN WEIGHT PERCENT 55 60 65 FIGURE 14.——Total alkali-silica diagram of phenocryst—free com- positions from Kalaupapa section. A—Z, samples from base of section upward. Solid arrows, trends of olivine, clinopyroxene (augite), and plagioclase from base of section upward. Open triangle, clinopyroxene in a garnet clinopyroxenite xenolith from Salt Lake Crater reported by Beeson and Jackson (1970). Field boundary from Macdonald (1968). F \/ V A \l V V v V FIGURE 15.—AFM diagram of phenocryst-free compositions from 1 Kalaupapa section. A—Z, samples from base of section upward. l Outlined area, field of Hawaiian basalts from Macdonald (1968). \ Arrows, compositional range of olivine and clinopyroxene (augite) from base of section upward. marked olivine control in the whole-rock lavas, but re- moval of the phenocrysts has telescoped the trend to \ such an extent that it ceases to exist, leaving a trendless cluster of points. Thus any systematic variation in Si02 1 and A1203 in the whole-rock chemistry is primarily the 1 result of the relative amounts of phenocrysts, especially olivine, in the lavas. 1‘ The contents of‘FeO’, CaO, NaZO, K20, P205, and Ti02 “ . tom of the magma chamber. In both the 17 in the whole—rock magnesia variation diagram (fig. 12) all form trends indicating control by one or more of the phenocryst minerals, olivine, clinopyroxene, or plagio— clase, with olivine generally dominating. The abrupt decrease in ‘FeO’ and Ti02 in the uppermost flows indi— cates late-stage fraction of ilmenite, ulvospinel, or other Fe-Ti oxides. The trends of the phenocryst-free composi- tions for ‘FeO,’ CaO, N320, K20, P205, and TlOg (fig. 16) leave little doubt that clinopyroxene is the dominant controlling mineral, except the late decreases in ‘FeO’ and TiOg. It is significant that the trends for ‘FeO,’ CaO, K20, and P205 in particular are at high angles to what they would be if olivine were the controlling mineral. The narrowness of the trend defined by these elements indicates that there is very little residual olivine control in spite of the possible additive errors in the modes and microprobe analyses. Variation in NiO and Cr203 content are completely controlled by olivine and chromite, respectively, and therefore need not be considered in the phenocryst-free compositions. Stearns and Macdonald (1942) envisioned a magma body beneath Hawaiian volcanoes that is zoned from hawaiite at the top to picrite at the bottom. Differentia- tion in the magma body was considered by them to be primarily the result of settling of olivine, clinopyroxene, and plagioclase, aided by volatile transfer of certain substances (particularly the alkalies) to the upper part of the magma body. Tapping of the magma body at different levels would yield magmas of different com- positions and phenocryst contents. Murata and Richter (1966a, b) showed for the 1959—60 Kilauea Iki eruption that the olivine phenocryst content of the erupting lavas was directly related to the rate of discharge. They suggested that strong currents of magma erode beds of previously sedimented olivine crystals lying on the bot- Stearns and Macdonald (1942) and the Murata and Richter (1966a, b) models, fractionation occurs in a shallow magma chamber. The phenocryst content of the lavas depends on the level at which the magma chamber is tapped or on the rate at which it is discharged. In either case, mechanical processes can introduce chemical variation in the lavas that masks more subtle, but possibly more significant, chemical variation. Subtraction of the phenocrysts from the lavas in the Kalaupapa section has, to a first approximation, removed compositional variation resulting from shallow fractional crystalliza- tion, the rate of discharge, and postextrusion settling of phenocrysts. The compositional variation of the phenocryst-free lavas allows us to discern an earlier fractionation event, one that occurred before fractional crystallization in the shallow magma chamber. At pre- sent there is no evidence to suggest whether the earlier 18 l I I I I I I I 3 «I E 30 ‘ :01 2 D- 3 20 — H=G z - < it P was = and E Y=A CLJIEK/KW “J 10 — S=F — O 92: I A $0 In ¢° a. [zoo '— Olivine (y I o 9 g 25 — — E (be ’é ‘90" Lu 20 - F=D 409/ — ’5 M=H c)‘ o N=G 3 E=R and Q A o 15 — _ — 0 E 8 BLfiY 8‘0 10 _ C J Kill a _ S W Z 5 — _ L Olivine 0 W l I I I I l 45 40 35 30 25 20 15 1o 5 0 EAST MOLOKAI VOLCANIC SERIES I l I I l I 20 — _ 15 ‘ N=M and H w ‘ ~ R=D 5 U0 9 =V BJngPL 5L Y=A €11,va IO _ “91; E G 2 __ 4' A o s A 5/ .¢ 5 — 0‘ 2 — 3 o '5" 0 °‘ I _ 0 55 M=J and F g = 9 U nhv ’9 Z .9 S=R S: 2 so - “3 D Q' ‘4; EstKPo a ’0 A LMSIYTXWI 9 B QU ”’ 45 — — 4o — , — 35 I I I I I I I I 45 40 35 3O 25 20 15 10 5 0 M90 OXIDE CONTENT, IN WEIGHT PERCENT part of each diagram. Arrows, ranges of olivine, clinopyroxene (augite), and plagioclase from base of section upward and are not to be confused with "control lines.” Open triangle, clinopyroxene in a garnet clinopyroxenite xenolith from Salt Lake Crater (Beeson and Jackson, 1970). fractionation event involved fractional fusion or frac- tional crystallization. However, it is evident from the variation shown in the phenocryst-free compositions that the mineral controlling the earlier fractionation was aluminous clinopyroxene. It should not be inferred that the phenocryst-free compositions represent primary lavas. The primary lavas are almost certainly richer in magnesia than these (Wright, 1971). It is clear, however, that most of the compositional variation imposed on the lavas by oli- vine fractionation was removed in the calculation of phenocryst-free compositions. Beeson and Jackson (1970) showed that the Inephelinic suite on Oahu, the Honolulu Volcanic Series, could have been derived by partial fusion of a garnet pyroxenite at a depth of about 100 km and that clinopy- roxene would be the last mineral to melt. The lavas in the Kalaupapa section may also be the product of frac- tional fusion of a garnet pyroxenite. If all minerals in the garnet pyroxenite except clinopyroxene were en- tirely melted, then the trends in the phenocryst-free lavas could be produced by the varying amounts of the clinopyroxene that is melted. Clinopyroxene from a garnet pyroxenite parent would be considerably more aluminous than the phenocrysts; in fact the lack of a trend ofAl203 and Si02 (fig. 16) demands that the Al203 and Si02 content of the controlling clinopyroxene ap- proach that of the lavas. The composition of clinopyroxene from a garnet pyroxenite xenolith from Salt Lake Crater, Hawaii (Beeson and Jackson, 1970), is plotted in figure 16. Except for Na20, it falls between the composition of the clinopyroxene phenocrysts and that of the phenocryst-free lavas. Clinopyroxenes, such as the ones analyzed from Hawaiian garnet pyroxe- PHENOCRYST-FREE COMPOSITIONS I \ I | l l I I O]. 1 . 17,16 .2 ' _ H=G S=D, F, and R X=K Y=A Q W=V and P 2 I- Z .1 ~ _ Lu U I ‘i’ I- 8 I E 2 0 HT 8 E C ”as “I 3 0 Clinopyglxene BLMYUX E Z 1 v E w E 4 50 z N=M and I Q 1 — O = K O Y A sx |.I.I mm D L D >—< 3 FY O H _ c G n E .9 z I- u E 2 - >< - 8 >. A a. o .5 5 1 - 0' tn :3 B .9 DD Olivine 5 0 I I I | | I | I 9" 45 4O 35 30 25 20 15 1O 5 0 M90 19, l I I I I I I I 1.5 — _ z u. 1— _ o H=G IL" =5 3, S=R u T=A X"0 mom .5 — _ LN‘iTT 31’ D m CBH '5 o E" Olivine Clinopyroxene '5; 0 _. 2 “ F=D Z - N=M O S=R " T=A V ¥ W=P (5° 1 _ mg; 8 _ Y m FT 3 J I .9. G to . . BH 2 0 Ohvme Clinopfloxene EC cu H=G M=J z S=R ” V=U and P 3 4 — 75 _ v0 .39, 0 Fox]? 2 é“ MSA I 9* LNY Z A BHDT 2 ~ EC _ e92, 0+ ‘9 . . 09 Ollvme «9/ O ——I——+——- l I l I I I 45 40 35 30 25 20 15 10 5 O OXIDE CONTENT, IN WEIGHT PERCENT FIGURE 16.—Continued. nites, may be responsible for the variations seen in the phenocryst-free lavas. Green (1971) devised a petrogenetic grid from which pressures of origin and degrees of partial melting of mantle-derived basalts can be determined from magma compositions. This grid shows that alkali olivine basalt, such as that in the Kalaupapa section, could originate from 10 to 20 percent partial melting of this pyrolite mantle at depths of 45—60 km (23—37 mi) and that clino— pyroxene would be a residual phase (along with olivine) remaining after extraction of the magma. As an alternative to the fractional fusion hypothesis, the compositional variation could result from the frac- tional crystallization in a deep (45—75 km, 28—465 mi) magma chamber in which clinopyroxene could be a liquidus mineral (Green and Ringwood, 1967). The composition of the liquidus clinopyroxene would proba- bly be similar to that from the garnet pyroxenite de- scribed above. My evidence does not discriminate be- tween fractional fusion or fractional crystallization. A necessary product of fractional crystallization of the lava at pressures of 15—25 kilobars would be clino- ‘ pyroxene cumulates on the floor of the magma chamber, and later eruptions probably would incorporate frag— ments of the clinopyroxene cumulate and transport them to the surface where they would be found as xeno- liths. The work of Jackson (1968), however, shows a decided lack of clinopyroxene cumulate xenoliths from the alkalic lavas of Hualalai and Mauna Kea. Many of the wehrlite xenoliths from these lavas, however, con- tain abundant veins of clinopyroxene, and some of the wehrlites appear to have been soaked in clinopyroxene. The possibility therefore exists that the clinopyroxene cumulates could have been destroyed by a metamorphic 20 event. On the other hand, Hualalai has not yet reached the stage of eruption that the Kalaupapa section repre- sents, and clinopyroxene xenoliths therefore should not necessarily be expected. Neither East Molokai Volcano nor Kohala Volcano on the island of Hawaii, which are more like each other and less like some of the other Hawaiian volcanoes, contain many xenoliths. It might therefore be argued that they are unique and an absence of xenoliths is characteristic of them. Many of the lavas in the southwestern United States (Wilshire and others, 1971), Australia (Binns, 1969; Binns and others, 1970), and Hawaii (observed by me) that contain ultramafic xenoliths also contain black, aluminous, iron-rich clinopyroxene megacrysts. There is no great difference between the composition of these magacrysts and the clinopyroxene from the garnet pyroxenite xenolith (688AL—11) described by Beeson and Jackson (1970). Both are thus prospective products of deep fractional crystallization. Murata (1960) indicated that fractional crystalli— zation of clinopyroxene appears to be the principal mechanism by which tholeiitic magmas are converted to alkalic magmas. Yoder and Tilley (1962), though recognizing the importance of pyroxene fractionation, objected to Murata’s scheme and example. Later Tilley and Yoder (1964) proposed a “pyroxenite stage” in basalt genesis, suggesting that at pressures correspond- ing to upper mantle conditions, extraction of ortho— pyroxene and clinopyroxene from an olivine-rich tholeiitic liquid leads to the generation of an undersatu- rated residual liquid of alkali olivine basalt. The lavas in the Kalaupapa section range from basalts transitional between tholeiitic and alkalic to basalts that are unquestionably alkalic. Clinopyroxene clearly exhibits marked control in the fractionation of these lavas, but whether the transitional lavas were derived from a tholeiitic parent by clinopyroxene frac- tionation alone is an open question. Certainly desilica- tion of a tholeiitic magma to produce the transitional basalts could not have been accomplished by fractiona- tion of clinopyroxene, because the silica content of the controlling clinopyroxene is very near that of the transi- tional and alkalic lavas. If desilication is necessary to derive an alkalic magma from a tholeiitic one, then another mineral (possibly orthopyroxene) or a different mechanism must be sought to accomplish it. Removal of the phenocrysts has allowed us to look back beyond the fractional crystallization in a shallow chamber to a fractionation event controlled by clinopy- roxene. If we concede that an alkalic lava may be de- rived from a tholeiitic parent magma through a process involving desilication, then there must be a fractiona— tion event farther removed from the ones we have been able to define. If, on the other hand, alkalic magmas EAST MOLOKAI VOLCANIC SERIES have been derived by fractional fusion of some primary mantle material, then we have identified clinopyroxene as the mineral controlling the compositional variation during partial fusion. If a magma produced by partial fusion of a garnet pyroxenite is fractionated at a depth at which clinopyroxene is the liquidus mineral, then little disturbance of the trend will occur because aluminus clinopyroxene would be the controlling min- eral in both events. N ow that it has been established that the Kalaupapa section can be divided into subsections either chemi- cally or petrographically on the basis of phenocryst con- tent and that much of the chemical variation can be attributed to phenocryst content, the question arises as to how the subsections originated. Powers (1955) pro- posed a “batch” concept for explaining the difference between historic and prehistoric tholeiitic lavas of Kilauea and Mauna Loa. Wright and Fiske (1971) con- firmed and further refined the batch concept for the tholeiitic lavas of Kilauea. They identified three groups of summit lavas from Kilauea (pre-1750, 1750—1895, and 191 l—present day) that can be distinguished chemi- cally with little overlap. They also show on a lesser scale that each Kilauean summit eruption in the 20th cen- tury as a chemistry that is distinctive with respect to the chemistry of every other summit eruption. Briefly the batch concept supposes that magma is generated in the mantle on an episodic rather than a continuous basis and that each batch has a characteristic composition. It is suggested here that the subsections delimited in the Kalaupapa section may represent batches of magma injected into the rift zone and subsequently fractionally crystallized and erupted and that the batch concept is applicable to alkalic as well as tholeiitic Hawaiian lavas. Because much of the compositional variation in each subsection can be attributed to phenocryst content, it is necessary to evaluate how removal of the phenocrysts affects the interpretation that subsections are the ex— pression of magmatic batches. This can best be seen in figure 17, in which the phenocryst-free compositions are plotted against relative stratigraphic position. Several oxides, Si02, A1203, and MgO, that show especially strong subsection variation in the whole—rock composi- ' tions (fig. 13) show little variation in the phenocryst- free compositions (fig. 17.) All these oxides are espe- cially sensitive to olivine control, and A1203, to plagio- clase control. In the whole-rock compositions (fig. 13), ‘FeO’ shows little subsection variation but in the phenocryst-free compositions (fig. 17) shows distinct subsection variation. Subsection variation of Na20, K20, and P205 in the phenocryst-free compositions (fig. 17) is nearly the same as it is in the whole-rock composi- tions. Two oxides, MnO and CaO, that show almost no MINERALOGY 21 2 E I z I IY z I ll I I z I Iz I I ‘ __ __ .L._ __._ ._..__ § g T v v 9 _§_____ UT__._ISU___.__ ____ L— E _E R R T E g _Q_.—_—______PQ——_._.Q__ ___o_______q____ < 0 0 o 0P 0? n: N N g N L: J.“ M _L 31. m K——_—_—_IE;—-— _$__ _ ___—'_KJ —— Z i I I 1 I- H H H H H ‘1 G___________L_ i__.__G____._._—_G____ 3 “F F F F “’- ___. _. __ ___.___E_ z JEF—'—_D—E 1215‘ ‘55 DC ; _B..___.____B_____B___ ___IL______B__ § 2 AI I IA I I A I I A I I IA g 45 50 10 15 20 1O 15 0 1D 5 10 15 3.0, A|,O, no mo CaO z YI I I z Y I z I ZY I I 0 __ ____.______2L___ _________x___ l_- + w V ‘7, v v 0 U U U I-L T__________l_ __ _____T__ __ u ———S S"— T i R R R 2% 5 __.&________9_____ _.____‘L__ I: 0" 5 l’0 g N N N I- M M M "’ ___L__.________L_______ ____L___.__ In K K K 2 1 g I, 2 HI HG H _l :1 —'_Q'7_——_—__'_F——_— ___G________ 2 E ___.______.E_ _.__ _E____ __ m CT C ‘15 c _D 5‘ _._B_______.___§_______ _____B__.__ n. 3 m I AI I I A I I I A I 2 3 4 5 1 2 2 3 4 Na,o K10 110, I Iz I z I IZ I I Y Y Y Z .____)L______.____2L_____X_______ 9 w w w I— v U V V ‘ U 8 ____L_U_________L_____L______ m s s s 2 R R R I Q. 4 g __.__._Q_.—_.___.__.Q___-_Q._______ .—_ P P P o o E N N O N ,_ M M M m _.___$_._.__-____L._____ _L._____._____ In K K K 2 1 J J '- I 1 I j H H H In _._G___.__.___G_______._£’__..____ n: F F F 3 ___E_____._____.___L__________E,______ m D D D u—‘J C C C n. .___B_____________B_______§________ 2 g I AI I l A I l 1i I 0 5 1 1 2 3 0 1 .2 no, MnO NiO OXIDE CONTENT, IN WEIGHT PERCENT FIGURE 17 .—Weight percent of oxides in phenocryst—free compo- sitions from Kalaupapa section plotted against relative strati- graphic position. Dashed horizontal lines, subsections deli- mited on basis of abrupt increase in ratio of MgO/‘FeO’ of whole-rock compositions (fig. 13) and phenocryst content (fig. 5). A—Z, samples from base of section upward. variation in the whole-rock compositions (fig. 13) show marked subsection variation in the phenocryst-free compositions (fig. 17). CaO variation in the phenocryst- free compositions is the result of its concentration in clinopyroxene, which controls compositional variation in the phenocryst-free lavas. It is not obvious why MnO shows so little variation in the whole-rock compositions but does show marked variation in the phenocryst-free compositions. MnO, unlike CaO, is not concentrated in clinopyroxene; instead, like NiO, it is concentrated in olivine. Nonetheless, discontinuities in the MnO varia- tion delimit the same subsection as do variations in the whole-rock compositions and the phenocryst content. Likewise, the CaO variation in the phenocryst-free compositions defines the same subsections; each subsec— tion starts at the higher CaO contents and decreases upward in the subsection, then abruptly increases at the base of the following subsection. The distinct subsec— tions, and accordingly the batches, can be seen in the phenocryst-free compositions as well as in the whole- rock compositions. No modes were made for the chromite micropheno- crysts because, judging from Cr203 content of the lavas, less than 0.5 percent of chromite would be present. As a result, variation of Cr203 has not been considered in the phenocryst-free compositions. MINERALOGY COMPOSITIONAL VARIATION OF OLIVINE Olivine occurs in the Kalaupapa lavas as pheno- crysts, in clusters (usually of smaller grain size than the phenocrysts), in the groundmass, and as fresh over- growths mantling altered olivine grains. The analyses of olivines are given in table 3, and the analyses for alteration rims of olivine are given in table 4. During the electron microprobe analysis of the phenocrysts and clusters, the electron beam was scanned rapidly over a 10 X 10 um area to help even out inhomogeneities and to obtain, as nearly as possible, an average composition. No special attempt was made to avoid tiny inclusions or microfractures that may have trapped polishing com- pound (alumina), and as a consequence the values for A1203 are on the order of .OX rather than nearly zero as Smith (1966) and Simkin and Smith (1970), among others, have shown for most olivines. The A1203 values for olivine in no way affect the conclusions reached in this paper. The plots of CaO, MnO, NiO, and T102 versus ‘FeO’ (fig. 18) and the plots of olivine composition versus rela- tive stratigraphic position (fig. 19) show that composi- tion does not differ consistently in the central part of the phenocrysts, the rims of the phenocrysts, or in the olivine of the clusters.4 The few grains identified opti— 4In the remainder of this report, the description of olivine phenocryst Compositions also applies to the olivine of the clusters. 22 EAST MOLOKAI VOLCANIC SERIES '8 I I I I I I I I I I I I o 08 s 0K 00 .3 — ' — g o g o 0 0 g o .. 6 s g 0 O 0 g tp 0 0 0 C g r p c r p 0g 8 D 0 g N 0 pp cpp p p on 0.2 _ — O 4 g r r C I; p I: s 0 (U . — _ o r peppp c p 0 g o p p p p so 3 p r o o P z z o 0 E I— p .1 - o g o 0 O ' Z .2 - _ g 030 o o “0" g gs cc r p p 3 ° 0 II ppiargppppé rrr p LIJ p P P P PC’PVC p 0. I i I’ I I I I- 0 I I I I I I I 0 o I I I I I I E s p p .4 — p — 2 }_~ 1.0 - - r 2 E p as z p pp 0 8 _ ° _ 3 _ pop 11 r - o p 0 ' p p p g LIJ 8 0 ”f p c r t D o c' PC P o s _ 0 0 0 C p 30 ° 0 g .9 g p r p at g g o g §.6_ oooogo ‘2 fr p 0000 o g 0 o 03 o 2 _ g 0 o _ g o ' ' P as so I; s o o o o 1- B o oo .4 — g — o o p c r p pcpcr g KPP’CD 1’ PVPP P pp pp“ 1 - p — DP 3 '2 _ P _. I I I I I I I I I I I I 10 15 20 25 3O 35 40 45 1O 15 20 25 30 35 40 45 TOTAL IRON, IN WEIGHT PERCENT FIGURE 18.—Total iron-variation diagram of olivine from Kalaupapa section. 0, central part of an olivine grain in a cluster; g, groundmass olivine; 0, fresh overgrowth on olivine phenocryst, in cluster, or in altered groundmass grain; p, central part of phenocryst; and r, rim of phenocryst but inside of alteration rim, if present. Italicized symbols indicate coincidence of two or more points. cally as groundmass olivine that plot in the field of phenocrysts in all probability are corners of olivine phenocrysts rather than groundmass olivine. For the purpose of subtracting phenocryst compositions from the whole-rock compositions, some olivine in clusters has been averaged with phenocryst compositions, espe— cially where data on phenocrysts in a particular rock were scarce. The lack of a consistent chemical difference between the groundmass olivine and the fresh overgrowths indi- cates that they are coeval.5 With increasing ‘FeO’ in the olivines, MnO content increases and NiO decreases (fig. 18). Apparently CaO 5In the remainder of this report, the description of groundmass olivine composition also applies to the fresh overgrowth. increases with ‘FeO’ also, but because of the scatter of the data, the trend is not so evident as for N i0 and MnO. There seems to be no variation of T102 with ‘FeO’ in the phenocrysts (fig. 18), even though there is little scatter in the data. Shaw and Jackson (1973) noted that 100Mg/Mg+Fe for early-formed phenocrysts in Kilauean and Mauna Loan tholeiites determined by Wright (1971) are in fair agreement with those of dunite xenoliths. They believed that the tholeiites are compositionally as well as spa- tially linked to the dunite xenoliths. However, Shaw and Jackson (1973) did not indicate whether minor- element concentrations in these two types of olivines are also similar. White (1966) reported earlier that olivine phenocrysts from Hawaiian alkalic lavas tend to MINERALOGY 2 ol p p I I lg I p p I I I I Y sp 3 p 25“, °__o°._________.._ _°_°°_____________.______ 9 V o o o E U o 0 pr 0 o o ”P 8 T __________c_r__rs____ __._________|‘£C_&’__I________ l S g go 9 R 8 z I 2 I o r o r c 2% T——4L———r————J———— " O 0: g 0 EN ”7 r I- M a p p 0 P P "’ L o __8____L__ _°_8_____P_______ E K 00 r 00 r I: J o g r c o g In < I c Pr 1) c r p A H g G __g_.____I?.r r .L_._ __3_______L________. 2 F pc pc 3, E __.________E_P______ _________E_L_______ “1 D 0E o g c “-1 C g p p s p p E B _.____.______"_P____ ________“_P__L._____ < (n A 31"” l P pI I°g °I I Ipp I I 34 as as 40 25 an :35 4o 45 50 sao2 M90 2 pip lg I I I I ppl 3 I I y p p c 2 0 0 0 9 x -—-———-——-a—-———-———————————-——o—-—- '1 v 00 o o 3 U rp o oo pnp o o o n. T .__L‘£______________‘.fl°£_____._____ 9 5 3° 33 I a s n. E w 0 ____J°___.°______8L£_______°__g_____ ‘ s :5 I; 0 o g If. I' p r 5131‘ pp 0 I) P 0 w L so ____ p ____z___0__ > K ‘_ _ pr 0‘6 ' ' — fp— 00 T. J I’g 0: Cr 5 0 j A P! p re g G ___.r_P_L_____________ur______z__ 2 F cp cp ;, E _P_._L_.____._____E_L_._.______.______._ In D 0 0; c g o -I c p p g p p g E B __P_P________________P__P_r_________ < (I) 0 00 A PpI 00 I g I | Ppl I l 5I l 2 4 .6 a 15 20 25 30 35 40 45 C30 F90 Z Z — T I I I I lp I I I 9 Y m; 3 0 I7 p p: g p I- x o o oo __ ____0 ______________ 5W ——‘_T'—_’—_—;T 7" BYJ VI“ 000 VP 0 DO OOI‘I'OPP o T ’ ____._££L_.._._._.___o_g_9fl._'__ _ 0 3‘ 1% 03 s 83 s I; E 9 9 LL!_.___<§’____ _L°__'__T£_________£_P_'_C____B____ :0 ,0 8 [p so 0% r p Ell pp 0 p p p 0 0 oOpg 13p :12 L s g___ ___ __ ____ m K _‘r%“(fi5—_————"—‘n§—_‘ fis—cr—ns > J o s CD so 0 scr P ; I-l ppr prc c p rp ‘1 r s rp __ _________g_r_r_p_______ , '9 4r ———-—.— A ,. , I E P ______ __fl_________.__________p__p____ 3 3 ”g g r p p ggo 30p c g P m B 517 _______LP__'____ _____.____ T ______L a‘ 00 P 3" pp°°1°g I Ipp I 80°? I I I g I I V’ o .1 .2 .2 .4 .6 .s .1 .2 .3 .4 no, MnO Nio OXIDE CONTENT, IN WEIGHT PERCENT FIGURE 19.—Weight percent of oxides in olivines from Kalaupapa section plotted against relative stratigraphic position. A—Z, samples from base of section upward. c, central part of an olivine grain in a cluster; g, groundmass olivine; 0, fresh overgrowth on olivine phenocryst in cluster, or in altered groundmass grain; p, central part of pehnocryst; and r, rim of phenocryst but inside of alteration rim, if present. One MnO value of 1.10 for sample Z is not shown. 23 be richer in CaO, ‘FeO,’ TiO2, and MnO and poorer in MgO, Cr203, and NiO than the olivine from associated inclusions. He did not indicate whether any variation in CaO, MnO, or NiO other than that related to the ‘FeO’ content of the olivine could be detected. Murata, Bas- tron, and Brannock (1965) showed that olivine pheno- crysts and olivine of periodotite xenoliths from Hawaii increase regularly in MnO but irregularly in CaO con— tent. They indicated that the CaO content may be partly a function of K _ e ee egg g e ge c c cc c ee e ; J ee pg e gpe p c p e e < l 8 1' g P T CC g P C C g —' H '5'“ G __ 1' pg rp 1' gm) 1' p c gp r r 2 F 8 r p P W D: P r s g E pp p g g prp pp p p p r s g D _ g p SC p p c C i5 i5 g L C C g 8 EB_ Cpgr r rpcvpr g pppc r s g A 8I 1 CD I p l ctgp r l l I? ll- ng l l l 46 48 50 52 2 4 6 6 8 10 12 14 SEC; AI,O3 'FeO’ Z I g g' ' ' ' l g ' ' ' c g Y e g PeCPC 9P pgge PC'P g e X __ 83 3'8: 83 W e g P 8 eg 3 p g p c e z V l' g P 8 r p g Pr 9 U 6 r g p e 8 pr pg e 5T__ r eecgpc gee cp gerpe o s g g’ g g gg 0- R g 88 38 2 I n. < 5 0_ s g es s z E P g w s r p r 3P < 0 eg c g c ec g E N g or p p r g p c p p or 3 (AM 1' rpcgpc grcpcp ccgp g L _ r gcr c p grcc prcrg p :K e ee g ccc eege cg egéee ( J e e p c e e p pc c ee 4 | g r cc p g 1' pct: p to g u a; H z G _ P “SP ['1' g c p rat 3 ‘7’ F g pp p z p pr 1; Lu i‘ E _ 39 p p p g r pp p 3 mm: 2 D 8 P PC C 8 p p cc gccp < C 8 8 8 m B __ Q r r r r s pcp pm __ A l I g p? c 1 I cg] r p p: l Cp 3: 12 14 16 18 16 18» 20 .2 .4 M90 Cao N830 FiGURE 22.—Weight percent of oxides of clino samples from base of section upward. p, c, central part of a clinopyroxene grain in a cluster; e, roxene. Dashed horizontal lines delimit subsections. OXlDE CONTENT, IN WEIGHT PERCENT pyroxene from Kalaupapa section plotted against relative stratigraphic position. A—Z, central part of a phenocryst or microphenocryst; r, rim of a phenocryst or microphenocryst; rim (edge) of clinopyroxene grain in a cluster; and g, groundmass clinopy- Italicized symbols indicate coincidence of two or more points. MINERALOGY positions in individual subsections (except fig. 21A in which subsection A and BCD are combined). The trend in figure 21A is toward subcalcic augite, and one of the clinopyroxenes plots in the field of subcalcic augites. The clinopyroxenes in subsection EF (fig. 213) are split; these from sample E (the lowermost flow of the subsec- tion) trend toward subcalcic augite, and those from sample F trend toward ferroaugite. The clinopyroxenes in the remainder of the subsections show trends toward ferroaugite, with the exception of those in subsection QRS which show no clear trend because there are no clinopyroxene phenocrysts in these flows. Poldervaart and Hess (1951), Wilkinson (1956), Muir and Tilley (1964), and Brown (1967), among others, showed that clinopyroxenes in saturated (tholeiitic) basalts trend toward subcalcic augite. Other workers (Wilkinson, 1956; LeMaitre, 1962; Brown and Vincent, 1963; Barberi and others, 1971) showed that clinopyroxenes in alkalic lavas trend toward ferro- augite. On the basis of clinopyroxene trends in the subsec- tions, one can argue that lavas in the lower part of the Kalaupapa section, if not tholeiitic, certainly must be transitional to tholeiitic basalt. The absence of modal hypersthene or pigeonite and the presence of olivine in the groundmass lead me to consider them transitional between tholeiitic and alkalic rather than tholeiitic. Smith and Lindsley (1971) determined that a trend 27 toward subcalcic augite was produced metastably as a “quench trend” in chilled margins of a Picture Gorge Basalt from Oregon, while the trend for clinopyroxenes from the center part of the flow parallels that in the alkalic Skaregaard intrusion. It is unlikely that the trend toward subcalcic augite developed in subsections A and BCD is the result of quenching, because the samples analyzed are nearly holocrystalline like most of the other samples in the section. Figure 22 shows clinopyroxene compositions plotted against relative stratigraphic position. The increase in Na20, TiO2, and A1203 (and a decrease in MgO content) upward in the section is apparent. The margins of the clinopyroxene phenocrysts, clinopyroxenes in clusters, and groundmass clinopyroxenes are enriched in ‘FeO,’ NaZO, Ti02, and M110 and depleted in MgO, CaO, and Cr203 relative to the cores of clinopyroxene phenocrysts and clinopyroxenes in clusters. COMPOSITIONAL VARIATION OF PLAGIOCLASE Plagioclase feldspars that occur as phenocrysts or microphenocrysts, in clusters with olivine and clinopyroxene, and as laths in the groundmass were analyzed with the electron microprobe. Interstitial groundmass feldspars were not analyzed. Table 6 gives analyses of plagioclase minerals in the Kalaupapa sec- tion. Figure 23 is an An, Ab, Or diagram plotted from the Z I g g I c Ig I I I Y [J g r 86 e 8 YePSP e e rg e P 0P z x 8 8! $8 38 w C e P P g 5 £8 3 V 8P 1' p gr 3 .. U r ge p r g e c l' p 8 T ceg rc e r pcc ee 3 ge F c c “- S 8 g 83 8 9 R g z 8 E 8 8 I n. E o Q g 8 8 s 88 i: P p f 8 8 Pl' PIE < 0 c e g ge g c I: N p r r c g c g l' l' rgr c p 5M pqpcrrg pcrcg p pl' :3 c c p In L p c It CCD CE 8 r g c g P 2 K ec cg c cc 3 ea c ee 63 c c I— J c p ece p pccp e e e e c p c p j I p gc r r pcc g g c c pr LU H u: G 2 cg r I' Le p r r 3 gr: p c 2 2 F pp 1' g gpr r g p 3 E 25 pp PPP 1' 8 '8 P PP J P .I D c g p 0P P 8 8 PPCC “' c z z 2 < B Cpl' 1' 8 P! I“: 7 8 P0 P P m A Ic gpr r I I I p g c l1 I I I [p c l 8I I p 1 2 3 4 2 .3 .4 0 .2 .4 6 Tioz MnO ego, OXIDE CONTENT, IN WEIGHT PERCENT FIGURE 22.——Continued. 28 EAST MOLOKAI VOLCANIC SERIES FIGURE 23.——Ab, Or, An diagram of plagioclase from Kalaupapa section. p, central part of phenocryst or microphenocryst; r, rim of phenocryst or microphenocryst; c, central part of plagioclase grain in a cluster; e, rim (edge) of a plagioclase grain in a cluster; and g, groundmass plagioclase lath. Italicized symbols indicate coinci- dence of two or more points. plagioclase analyses. Most of the plagioclase grains that occur as phenocrysts and microphenocrysts, or in clus-, ters, range in composition from A1180, Ab20, Oro to An 53, Ab40, 01‘2, although two microphenocrysts (both from sample Z, a mugearite) are as low as An49, Ab48, 01'3. One grain in sample X that was first classified as a microphenocryst contains An34, Ab61, Or5, but reexami- nation shows that it may be a groundmass lath. Groundmass laths range in composition from Anew, Ab36.5, Or2 to An39, Ab57, Or4, overlapping the composi- tion of phenocrysts, microphenocrysts, and plagioclase of clusters. The lowest An and highest Or and Ab groundmass plagioclase (fig. 23) is from sample Z and is not included with the basalt ranges. In the entire Kalaupapa section, the compositions of phenocryst and groundmass plagioclase overlap consid- erably, but in single samples, the compositions, with a few exceptions, do not overlap. Despite the scatter of data points in figure 24, the Na20 content increases and the 030 content decreases upward in the section. Sample Z is very close to the composition of the aver- age mugearite given by Macdonald and Abbott (1970) and, by definition, should contain oligoclase, but it con- tains andesine instead. A similar discrepancy was noted by‘Keil, Fodor, and Bunch (1972) in the lavas of West Maui and Haleakala volcanoes. . The groundmass, rim, and edge plagioclase composi- tions emphasize the more fractionated character of the lavas in the upper parts of the subsections. Na20, K20, and BaO increase and CaO and A1203 decrease upward in the subsections (fig. 24). COMPOSITIONAL VARIATION OF OPAQUE MINERALS A wide variety of oxide phases showing extensive solid solution among the various end members is pres- ent in the lavas of the Kalaupapa section (table 7). Some chromite grains that occur as microphenocrysts in the groundmass as well as inclusions in olivine contain more than 50 percent Cr203. Many of those not entirely enclosed in olivine phenocrysts are continuously zoned from a core of chromite to margins of ulvo'spinel (usually on the titaniferous magnetite side of stoichiometric ulvo'spinel). Brown (1967) indicated that “there is no evidence from the large fractionated intrusions, par- ticularly the Bushveld, of a continuous series of spinels from chromites to titaniferous magnetites.” However, continuous solid solution between chromite and ulvo- spinel has been reported in the tholeiitic lavas of Kilau- ea Iki and Makaopuhi, Hawaii (Evans and Wright, 1972), from the lunar rocks by several investigators (Agrell and others, 1970; Haggerty, 1972; and others), as well as that reported here. The inicrophenocrysts range in diameter mostly from 100 to 250 Mm, but a few are larger than 250 ,um. The groundmass minerals are ilmenite, ulvé‘spinel to titaniferous magnetite, and (rarely) rutile. The opaque minerals in the groundmass, though generally smaller than the microphenocrysts, are in places as large as some of the smaller microphenocrysts, and opaque min- erals tend to cluster so that clots of tiny octahedra or other crystal forms are easily mistaken for mi- crophenocrysts. Hence some that were assumed to be microphenocrysts on the basis of their size turned out to be groundmass in their composition. Some of the points plotted as microphenocrysts in figures 25 and 26 should more reasonably be plotted as groundmass grains. The groundmass opaque minerals are usually un- mixed, as a lattice network of ilmenite lamellae in titaniferous magnetite, as titaniferous magnetite with patches of ilmenite (usually toward the edge of the grain), or as half ilmenite and half titaniferous magne- tite sandwiched between two ilmenite patches. Several tiny TiOé-rich areas (approaching rutile) were ob- served, one of which is included in table 7 and figures 25 and 26. The exsolved phases usually have gradational con- tacts with the host. In order to obtain analyses that total roughly to 100 percent, at least two areas were selected in each grain, one giving a maxium Ti02 and the other giving a minimum Ti02 content. This procedure biases the analyses toward TiO2-rich and TiO2-poor end mem— bers, but in spite of this bias, the solid solution shown between ilmenite and titaniferous magnetite and be- tween ulvo'spinel and chromite is complete and must be real. In fact, complete solid solution was detected in a single grain containing more than 30 percent Cr203 and less than 4 percent Ti02 in the chromite core and more MINERALOGY 2 'pp g ' I pp I I I I I p I g' Z Y Pg P8 g 0 X Pg r P rfg P Pg P ;w ppr gr j g rpr I) pp rr g J) V PP rpr YE g H TDDP PPP rr T g 0U P PPIF g g r P Pp p p pr r g n. T P p rg gr P p P 13 TE 9 S gp {:10 g I R P g P p g (L <( u: 9 Op PC g r IE 17 p p Cp 2 r I- P P P H g ggr P P P P rrg <( O r p e g g e r pp cpp re g EN 9 g g p p g m M D g g P P g LU L ppg r r ED 1’ PP g I" 2 K cp g e r eg p c C p g e I— J p g g r p p pr g 34m) I) rg g rpr PP PP P rgf ‘3: G pCrp eg ge pc rp p c eg 3 F prp g g rrpe 1) PC N g m E p g g p p g L_IJI D pc r ge e g cpr C r g e l c p g g p p g 2 BC eg 6 g r c cp eg <1: U) A p or g g 013‘ P p D“ E I I I I I I I I I I I I 50 5 60 65 26 3O 32 3 4 5 7 sao2 A|203 NaZO I I I I I I I I I Z PP g g P Y Pg g X p g P P r E!) W prr g r g r p V prrr g g rr rpp U pppr g g r I’P P P 2 T Pg gr I3 P Q 8 pg 1’ g I; R p g g p m 0 0. 9 :1: 2 Q PPS!” 1' _g_ PC 13 I; P Pp rg I g r p P Q 0 P V 8 g 1' e PPC EN pg g a: I—M Pg g P zn m L pg r F g P 2 K 6p g8 e g pc EJ pprg Dr D d ' PPT g rg pr P I H 26 peg g P C 1' P U, Fcrg g rpr ec p UJ ._I (L g E pg g P gopcg e e grc p C 98 g P chg TE 8 A CPg g 0P P I I I I l I I I L I 0 1 2 3 4 8 12 16 K20 30 FIGURE 24.—Weight percent of oxides in plagioclase from positions. A-Z, samples from base of section upward. p, phenocryst or microphenocryst; c, central part of a plagioclase in a cluster; and g, groundmass plagioclase lath. Italicized s OXIDE CONTENT, IN WEIGHT PERCENT Figure continued on following page. Kalaupapa section plotted against relative stratigraphic central part of phenocryst or microphenocryst; r, rim of grain in a cluster; e, rim (edge) of a plagioclase grain ymbols indicate coincidence of two or more points. 29 30 than 22 percent T102 and less than 3 percent Cr203 in the ulvospinel rim. Major elements in the chromite show marked varia'- tion with Cr203 content (fig. 25). A1203 and MgO de- crease almost to zero as Cr203 decreases to zero, and Ti02 and ‘FeO’ increase as Cr203 decreases, becoming more scattered near zero Cr2O3 content. At zero Cr203, ‘FeO’ and Ti02 range from values expected for titanifer- ous magnetite almost to rutile (fig. 25). Because it is not possible to separate Fe3+ from Fe2+v with the electron microprobe, no evaluation of mineral variations on the FeO-Ti02-Fe203 diagram has been attempted. The total iron could be partitioned between FeO and Fe203 if it is assumed that there are no defect structures present, but such an exercise is beyond the scope of this paper. SUMMARY To unravel the detailed compositional variation of a volcanic sequence, closely spaced sampling and firm stratigraphic control are necessary. This type of sam- pling of the Kalaupapa section revealed that the frac- tionation trend is discontinuous; it proceeds with sev- eral steps forward toward more fractionated composi- tions and then reverts abruptly to a less fractionated EAST MOLOKAI VOLCANIC SERIES composition. The Kalaupapa section can be divided into eight subsections on the basis of the discontinuous frac- tionation trend. These subsections are the same as the ones that would result from dividing the section on the basis of phenocryst content of the lavas. The upper sub- sections are generally more fractionated than the lower ones. Petrologically the lavas in the lower part of the sec- tion are alkalic, but chemically they are tholeiitic and are best classified as transitional between tholeiitic and alkalic basalt. The lavas in the upper part of the section are, by any classification, alkalic. The subsections are believed to have resulted from distinct magmatic batches that are generated in the mantle on a periodic rather than a continuous basis. Similar batches were first proposed by Powers (1955) to explain chemical variation of tholeiites and eruptive histories of Hawaiian volcanoes and were later used by Wright and Fiske (1971) to explain the compositional differences between historic and prehistoric lavas of Kilauea. Subtraction of the phenocryst content from the dry- reduced whole-rock compositions removes composi- tional variation that results largely from shallow frac- zl l i) | g 1 l I I lp fi lg I I Y g D D8 Xl' g D D prg w r p g P r g V 1' P P l' 1' W P gr ZUV P P g prg P OT g PP 1' g 1:8 g pg 0) OR g Pg 0. 2 I 0. (OP 8 pp 31' 5P1: r g r gp FOP g rpgp e x o¢~..~ omo.c mnm.m ooo.m -c.- omc.mu oom.cH omo.n onm.¢fi ~p~.h m_¢.- m~>._~ mm”.m 2m»: mec.- mpo.o moc.°~ o-.o .ss.». omo.o~ o~n.n~ mm¢.m ~e~.»~ ~ea.» doo.¢~ mem.n~ moe.- >x oom.o ¢~c.m mo~.. mmo.~ one." odo.e nnm.~ °=~.~ o-.~ 0.0." emu.“ mo~._ com.~ muuo ohm.» mom.» opm.c o~¢.m no... mfip.o ~mm.o o~°.m .mm.o ¢mo.o o-.m pom.o _mo.e zuaa .co.o mmm._~ -¢.m ~mm.m mco.o mgn.w ~no.o “an.“ _co.o -c.o mmo.» oc°.o oem.o ox~o -~.m~ Noc.- ~om.- flc~.- ~nn.n_ omo.m. ego.p~ sao.m~ ~>°.- moo.a_ nmo.¢~ «mo.»— mm¢.m~ Ho «ofl.¢m m~>.mm omm.mc noc.mc »_o.oc oe_.mm mwm.¢m ouo.m¢ o.m.~o mmo.~¢ we"..m o¢_.mm -o.fim umeu ¢~o.mq mam... oo~.pm om~.~m oco.~m .mn... «up... moo.om mom.»n pm=.pm ope.oe mom.¢< oo~.m¢ u~4o.~ cno.n ¢sm.m ope.” ms~.n «u mam.o owu.m~ enm.~ oem.o n-.~ omn.s oom.°~ ~mn.o oo~.¢~ ~H0.m sm~.o on ~m«.~ ocm.n omo.o nm~.o wow.» nco.~ omc.. oom.n nop.n oom.m mfls.m cmm.m c-.o mu mwo.ofi oco.m -_.o~ ~m<.- ego.o~ ~m~.m~ mmn.s~ moo.m cam.~m om_.¢_ nmc.m~ ~«o.m~ psfi.¢~ zm .«o.o m-.- -o.o ~m¢.o nao.o mam.m flmo.c -m.~ ~eo.o ~p¢.o emu.» ooo.o o¢m.o o: mz gmfi.m~ mn>.o~ Noo.m~ cem.m~ °~+.m~ _on.o~ omc.cm -c.p~ oflo.n~ mmo.m~ mefi.o~ coo.o~ ~H>.mm z< m-.o~ Fm<.m~ oeo.om eoh.- no~.~m cmo.o~ >n~.~fi ~.c.mm co..~_ moo.- ~c0.#~ o~¢.o~ o~n.- sq amo.e mmo.c mmo.¢ ”mm.~ shn.m mon.H eem.H ~no.¢ oma.o oo~.m mfie.~ nm~.— sad.m mo soo.m nmc.e a $502 SMHU xq.oo sm.oo am.oo No.00 .e.oo om.oo .p.oo 00.90 mm.oa o~.oo o~.oo eo.coH m~.oo 4 “29:55. i532“ OS on WHEN—«Ego: :9: . OU dam Os: wiuumhnsm ”me mm :oh 153 mczfizflmowa NE 35550.2“ 953 $19 ,uwumhmaom on 8.58 mOamrm «Ea Ooh €53 hem £9855 no mmmbmnm 30323:. 5.; :omimafis 33:63 8 @353 Em mcosmmomicu uwuzumHJSQ .m .mc 5 523m 93 NI< 33:59: :Efivmm skaaxgaN 3% Sci $33 33?: kc wmméafia NGQNESNDIA mum—<8 37 TABLES 1—7 nocoe cmwua nmo.m coauo opv.m cm~.e (LAO Mac-a NMoo# n~.o ofiooa Ntflomu man.» 40 co—.~ muh.~ one.“ Non-m Nmmoe omo.m mooam mam-m etc." mm>x obn.o “cocoa how-m ~ccom omm-m 00°.~ wooom— nonoum o#N.o nm¢.o~ mcc.o end-m com-ca 2m»: com.» mam.m~ §o~.o Hoe-m Mahob occa— mec.m~ 000.0” mem.N~ amt-on mop-o ¢#~.m omm.H~ >1 oo~.o Hamcu cho.~ mecca com-N 0mm.“ Nap.“ m~m.m com-c mmno N0o.~ mwc.p ochcw moo.n -~.c baa-m QFMoh hoo.n mmm.m nhh.m om¢.v {no.¢ moo-m Zuno ¢c¢- «Nhom ommo¢ m°¢.¢ chm-o mas.» onm.m oom.m 000.5 Oho.c Nhnuh ooh-m mno.c— Oxuo moo-N ca¢.w~ tetoo com-o onn.N~ cos.mn «go-mu ~hnoc~ amnomn Nm¢.m~ {No-nu N¢o.o~ coo.mH no ommocN chhumm och-rm cmaoon M@0.Qn atm.o¢ boo-cc "Ho-av ONo.m¢ omh.~< ~¢N.om ougumn no~.¢m Unxwm pcmoch omo.mv age-Ac mow-00 mo~.mo omn.nm mom.om oomohm ooh-om omr.hm mno.oo www.co hm>.m¢ UHJdm 0#o.oo mm¢.oo awn-00 nee-00 nnmooo moo-00 N>~.oo Gem-Go ans-mo ccm.oo aka-ow moh.oo mam-00 J new-m Nuo.o mom-o mom.o on nom.~ mNo.N honoe tn~.¢ Qua-n mom.o hdh.¢ hmoom mcw.~ mm mmc.# nmo.m~ m~<.o oom.m 000.0 Nma.o nonomm com-«H nahuca cowooa 000.0" weaoc— woo-w“ 2m co¢.~ QNhoo oNoce nc<.v annoo mvnom onmcm con.m {00.5 aha-c ~hauh mop-m mMo.o~ 0: m2 ace-0H men-mm caoubw m¢~.°n men-5N no~owm chm-hm hcc.pm ¢¢¢.mN m~¢.om wh<.~m Nco.mw Nmo.~m 2‘ FON.M¢ mom-o” boo-cm mohoom ecu-Hm awaomN cmc.m~ Moo-cm moo-am hem-mm HMO-cm Non.on o¢h.o~ n< coo.ua anhcm mnmoo Nan-o omo.o coo-m moo-N cam-m bm~.m Ava-h mod-h owe-q 1° «mo-m n~o.~ chmoc nwwoo ocm.e ascoo can.n a may 02 SmH U swuoo cmooo hc.oo €~.oo ¢¢ooo nhooo c—.oo moooo mm.oo omcoo ncooa om.mo oo.mo J0.m om.o 0—.0 Hm.m {mom mo.o o¢ ¢m.ht cm.m< Noun Auzooumm unwfimsv mmmzamam HmuHEuau N r x 3 > a H w m o m o z MMI NMM.N— oom.c~ 09m.m~ nmm.o~ om~.n m>m.s ocm.oa mom.oH Jo ism-m mmo.~ 0~h.c hue.m n~n.m 000.0 ~Nm.¢~ mmh.m n~0.~ mom-m mu): m¢~.N cao.~ "co-c hum.n 00¢.n eoN.> ¢~¢.m mom.0 mac-N 0°0.N zw>z 0~mom mma.n onh.c~ «no.0 cam.e ¢0~.m~ mmm.0~ 0mm.m~ mm0.m mo:.o >I oe¢.¢ hmhum ewe-m «mmoc cho.c 0Ne.< mvo.v m~c.q ~00.M ®0¢.m 0N>.m 000.9 mn.m #¢~.¢ Ho0.¢ csh.m ash-m mum.¢ cam-e mmo.¢ Mmm.c Zuno ~0¢.0 ohm.Nu Nma.0 0mc.0 £NM.> >¢n.m Ho~.0 umm.w m0~.o~ moh.0 mc~.m mnn.0 o-.o~ 03—0 ecs.m~ hmv.¢m mm¢.m~ MoH.m~ ash.¢~ MNm.oH hon.0~ on<.o~ mc0.0~ som.0~ mo~.o~ 50¢.mfi mo~.om no monowv «-.h¢ 0h~.¢c amm.mc 0Nm.~¢ mon.mv 0h$.~# n0gomt omh.mc com.0c ne¢.mv hmo.c¢ nes.<¢ u—xwu swh.>m o~0omm th.mm m0s.cm ~o~.mm {mo.bm oc~.hm wa.pm www.mm 0mo.cm osm.~m m0m.mm www.mm 0u4m~o moN.m ~N¢.o~ m0c.c new." nms.m 0No.m «em-o cmo.~ om0.m ~c°.m QOo.m oo~.o m~m.~ omm.n 0mm.m ~N0.¢ Oh oov.¢ hmh-m 50p.m om¢.o 00n.~« 00¢.» mm0.h N_¢.< ~60.c 0~m.o~ ¢m¢.mu mom.o so~.0 mm 0am.¢ o~n.0 mao.o mom.m hhh.o moo.h p0n.m ekh.n ch0.~a 0mo.~_ ~mo.¢~ m0o.c m~¢.5 zw Hoe.0 chm.N~ Nm~.0 00e.0 own.» hon.o Achoo ~mm.m m0~.ca mmh.0 mo~.m mnm.0 o-.e~ ox Mum.o Macon 0QH.~ wz mhc.mm Ham.mN tug-mm 0c0.0~ 050.cm www.mm can-mm cmm.hN mhc.mn m~o.em Nom.~m mmm.~m amo.ow z< m~m.n~ mo0.om 00¢.hm 00°.MN m~0.0m wam.~N hkumm m0h.om ccm.»~ moo.- o-.0~ n0m.oN mo0.a~ m< 0mm.m MNo.m add-m hm0.N 000.N ~00.” >-.N m~m.¢ mmu.~ 0m0.m 00o.“ rho.” «co.m mo 3202 Ba _ u ~o.oo~ ec.oau no.°o~ 00.00 00.00 co.oo~ oo.oo~ ”c.9cu 00.00 eo.oo~ 00.00 me.eo~ 00.00 J—.c m~.o oz: mm.e n¢.e om.o hc.c om.o wn.o on.c o¢.o 0~.o N¢.o mn.o qm.o ov.c mom: neon hm.n ~h.n neon N0.n :0.N mmom mm.n om.~ cm.n 0#.N hn.n a~.m NOuh 0moo mm.o 50.6 $0.: 00.0 mn.o on.o nh.c 0~.e pc.e ®~.o mmcc No.3 omx v0.m «Q.N mN.m 00.N ~0.N oo.m eh.m 0u.m to.~ no.~ >~.m ~v.m m0.m om0.- c~.o~ m0.- cm.o~ ~h.o~ ~m.- om.o~ o«o on.o “0.0 mc.m mn.o MN.‘ 00.0 o~.o mu.o nh.0 mm.m mw.> —¢.# no.m oat 00.9“ cm.- 0°.m~ @0.N« H°.n~ mo.- om.c~ o¢.m~ cm.o~ oc.n~ m».~— ov.N~ mo.mH 0mm 0cooa mm.c~ chocfi mm.m~ om.o~ ~0.o~ No.0g n~.0~ no.m~ «c.v~ c¢.m~ 0h.m_ 0N.mu MONJ< 00.50 cc.hv 50.0. m¢.#c mo.ho wh.fi¢ ¢M.mc om.h¢ m0.#< mm.0v >¢.me 0m.o¢ 90.x: mean qumokwm pnwfimzv mmmhamnm HQUfifiwno 2 q x w H m u m m n u m < @756” . 37¢QO «Enid fiaufifim fluid $4.32 mmémd Enid fluid. 4-..qu TEE «Lag H-319 Elm mafia”: gate—~23 of we mzoflwmomfis «Ea Am. .9: mundoim H355 we Emma of :o mnogmomfioo xuonégw—B umoduwhlfu mfi Eek wamfiuonwam Max:523 N3 wwnmEuwumu wuim> .m (ma E E59? 93 N4< mmfizmoofi :SBmm aQGDSEBN mi EEK Swag mmfigmigfimfim a :0 $30.. kc mmméata NduNSmQDIN mam—<9 39 TABLESl—7 omo.o .-.w nas.o~ awo.- »»~.~— map.o oo~.o noo.w ooo.o omc.c~ ¢~>.o wme.o poo.» «udo c~m.¢ omo.m mom.» fisc.m omo.c nps.m coo.e can.w o»~.p ace.» emm.m mmp.q ~0a.o cugo osn.c~ ~np.n~ moo.m~ moc.o~ can.sn phc.m. mpo.o_ cm~.m_ mmw.o~ o»o.p~ oo~.mH mpH.¢H osc.¢‘ 4o m¢¢.o "co.~ pofi.m oo¢.¢ pom._ m~¢.° o_~.~ mu>z ~en.° mac." oma.m cqm.m moo." oo~.o m-.~ zu>x amp.o coo.n moo.o «eo.m ohn.~ mho.o .om.n >1 omo.o oo..o mom.m mon.m mmm.m oom.m mc~.c -o.n a—c.c moo.n p~o.a m~m.n om~.m mung cm¢.o oeo.c NmH.m pmm.~ oso.n pnm.n >mm.n moc.m ego.m om~.m som.~ oo~.~ m-.m zmua c°<.~ Afic.- one.m mn~.m no~.o ¢op.a ~5c.» mc¢.m moo.o ooo.> a~¢.~ >o>.m ~¢~.- cane pcm.m omm.~m poc.og can.oH moo.o~ opo.s~ ¢no.¢~ emm.e~ soo.o~ omH.oH mmo.¢~ o~>.- -~.~m Ha no¢.<~ moe.¢¢ nuo.om mem.o¢ nmm.cc www.mc moo.mc mmo.ac Fo=.m¢ oop.o¢ mn~.c« smm.mn nao.mq uaxmm pom.mp mmm.mm n~o.mo onp.om mwm.mm ome.pm cmo.pm moe.mm coo.om c¢~.om “co.om nma.fio ~¢o.om u~4~o.oo~ pma.oofl ono.oo~ one.oe~ ono.oo~ mmo.eo~ 4(ho» nmo.~ meo.o onm.~ mom.H goo.” oom.~ n__.~ flog.” Hod." mom.“ mos." ooh.~ coo.~ a4 mpm.e ”No.0 «pm.o ohn.m oc>.o onw.» cnm.o oom.o nmm.o mfim.h ”mm.h mmo.h Aec.o A” ¢¢o.o -_.o ccc.c omc.o omo.o oo°.o omo.o cco.o «ea.o «po.o eac.e «qo.o mo~.o so omo.o H_H.m nos.c~ emo.- >p~.~— mo».o oo~.o noo.c omo.o omc.o~ o-.o omo.o pmo.h «u c~m.¢ omo.m mon.~ ~>¢.m o~o.o mup.m no... oom.o avg.» ooc.p onm.m mm».c ~oo.o on omo.o pdo.o mmx.~ eon.m m-.m oom.m -n.o Fae.» o~>.m Noo.n Nno.m mno.m om~.m mm ¢m¢.o ‘m~.m ~m~.m ~mo.~ osc.n sam.n nn~.p owo.m m»a.¢ am..m hma.n om~.n m~p.m zu cog.” ~d<.- eno.m mm~.m no_.m oo>.w m><.> o¢¢.m moo.m moo.» m—¢.> >o>.m ~e~.H~ ox oop.o ahn.o pso.° coo.~ osm.o som.e mz mno.e~ ~o¢.»~ meo.m~ oep.~m noo.»~ Non.- m~¢.~m mm~.>~ pco.m~ omo.>~ mwo.- ooo.~m ~mo.p~ z< o~o.mc owm.n~ mon.>~ .~—.o~ ~—¢.o~ "mm.¢~ oop.~m moo..m Hn~.o~ m~o.om mom.um mmo.~m cmm.n~ ma a-.mH "mm.¢ mop.e can.» ~m¢.o hon.» mmo.n ppm.m Noo.m ego.m mom.~ can.» mmo.m mo megoz zmHu mo.oo oo.oo _e.°o~ oo.oo oo.oo oo.ea~ oo.oo oo.eo~ oe.oe~ oo.oo oc.oo~ 00.00 oo.¢o Jgho» “9.: ”e.o ~e.o ~o.o no.o mo.o ~o.o No.3 Ho.e ~o.o ~o.° on no.e -.o no.9 «c.o +o.c oo.o oo.o nc.c no.o mo.o me.o no.9 Hfi.o mowzu o~.o o~.° om.e «m.o m~.o o_.e o~.o o~.o a~.o »~.c _~.o ~m.° m~.o oz: -.A o¢.o mo.o om.o «o.o o>.e >¢.o oc.a oc.o mm.c ~p.o ~>.c oc.o mowa .c.~ pfi.m no.» A... mm.c on.c c«.n pc.m co.n mm.n co.m no.» pn.n Noah mo.m h».c mfl.g cm.“ me." m~.~ om.o ”o.e om.o mo.c p~.~ hm.~ mm.o ow: -.m m~.~ ~m.n c«.m pm.n ~m.m oo.~ mo.m o~.m c~.n a~.n po.m oa.~ o~.o om.ofi >o.m .>.o oo.o mp.m on.a m¢.- o(o m~.~ en.m mo.m mn.¢ oo.¢ ~p.c ~o.m mo.o oa.m mm.m cc.c mo.¢ om.o om: ¢¢.o -.- <~.~H ¢¢.<~ ho.c~ on.n~ on.- m~.m~ cm.~_ mo.- mo.- om.m~ >~.c~ own nd.w. >«.mfi m~.»~ n¢.m~ oo.n~ mm.ma pm.o~ "e.o_ ne.m~ c~.o~ mfl.n~ om.m~ mm.m~ nom44 ~¢.~m mo.hq on.p« ~e.p« om.»o mm.oc om.po 05.”. oo.~¢ mo.>q ¢~.m¢ om.m¢ ~o.mq Noam Auzwoumm uswwmsv mwmwamum Hmowfiwso N w x 3 > a a m m o m o z mmu.ov chasm cuufi< oN-dN omocc oo.c¢ em.#~ oo.N¢ oN.om oHooN oo.om cc-mm ab.m~ oo.m~ aeomn cm.m— coohm om-mu cc.o~ cmuen oco>~ po.o ec-o so.e meoc oo.¢ eH-o mnge cuuo mc.c 0c.c co.e c~.c me.o om..hm coshm om.mn e~.on cc.om cmocm ocuom co.om ownom u u m w n m M a o a a a w 00ml; wnfila UA—I4 wMIJ mamlq UQNIJ unalg om~lJ UQNIJ cuwlg uaMIJ 01”-; — am! _ u a; = a 3 a omooo cacao Hm-oo ~muoo cm-oo Oh.00 no.00 ouooo~ ow.cc~ m».co~ o~.~o~ oo.cc~ mcooo «coo moco mace No.0 No.0 te.e #o.o 09.0 99.9 «c.o «coo cc.o Mo-o mmoc cm.c omuc ONoc ~¢.c Hooo hN-c ON.c «m.c om.o omuo mm.o once oW-c once sec: Ew.e cN.o 0N.e ~¢.o an.o «moo hm.c hN.c om-o om.o mN-o mooc hooc m¢oe Mesa «cue Qcoc no.9 ¢a.o cgoa Meco oouc mace m§.o once Nm-c 00.9 ~n.o mmoc nmcc mnoo hcoo m#.o ce.o h¢.c anuo cc.mm cm.hn en-sm coooc oo-ov ooomc cmohm oc-o¢ egn>N cm.om c~.Nc oe-ON chuce oooon cm.~N onodn om.ou omym~ e¢.oH cN.NN om.o~ ccomm om.mm encmn ov.om enca— -oo «cue H—.c no.9 Nuoo no.0 «ace Qe.° acuo hcoc co.c pc.c cc.c eh.mm c~.¢m c¢.om o~.om ch.on enuom om.¢m am.on ce.cm cm.om o>oom omoom oN-mm o u w a a m M a o o a o m amu- cuT. _ on: uam- on? _ cam: mm? on:- uazmn ounm- uawn own? 9:-— n o m < J mowzu On: 02! mom» o.o op.o mo.c na.° mm.° on.o mn.o -.° oc.o m°.a oo.o no.9 n~.o o_.o H~.° -.o m~.o o~.o cu.a mo.o mo.c oo.e cc.o ~¢.° mo.o mn.c co.o co.o No.9 am.c .sm.° 00.9 om.e ~¢.o e¢.o am.o~ e~.~n c¢.nn om.~¢ o~.m¢ ao.- o¢.n~ oo.- on.- om.- eon- coupw c‘ucm ouuon eo.mm eo.o~ cH.a~ oo.p~ on.¢~ oo.~¢ efi.oo c~.o¢ am.~a ae.oc om.om op.«m o~.- on.- o~.° -.o c~.o -.o mo.e #~.e fi_.~ ~..o ce.° an.o a" o ea 9 we a he a eo.mn o~.sn om.>n oo.on oo.om on.mn o_.om cc.mn oo.mm cm.mn oo.mn am.on o..mn o~.wn m a A M a o o o o o o o H a _ emu; az~na ax_u4 uwwug uafiud monnJ ma~14 mowua uwmp4 max—4 moMIJ uaN-J ma~u4 uawu4 7 N x 3 2.2.: = 1_ S ,0 mm on can pH.co_ o~.oc~ .o.eo_ ~n.°oH ~¢.co_ om.oe— c¢.oe_ -.oo~ -._o~ on.~o~ c~.~o~ o~.oa nu ro.o n°.o ne.o .c.o m°.o ~°.o no.o no.0 co.a co.o n=.e ~o.o mo.o .A ¢~.o -.° e~.o -.° «n.o pm.o o~.c pn.° ~n.c .~.° #~.o °~.° ¢~.o m. sp.o on.o an.c on.o ~n.a nn.¢ on.c qn.° ~n.c om.c 00.9 oo.o fio.c m_.c 00.: oc.o 09.9 ce.o mc.c mo.° mo.o co.o me.° pc.c oH.o 5—.9 none «to we... 3... mm... mm... ~¢.c 3... pm... we... we... 3.: ac... co.¢~ omusm oo.~n oo.on oo.~. cc.pn em.sn am.~. oo.oc ec.>~ eo.- oo.m~ c>.e~ c~.~m cu - o~.m~ om.- am.m~ cc.n~ eo.- an.- oe.o~ ¢~.on ch.mn om.pn co.cm po.o po.o mo.o o~.e -.o c~.a co.° oH.o .c.o no.9 me.a m».o ¢~.o co on oo.mm o~.¢m on.mm o¢.om om.mn om.mm on.om em.mn an.om em.om ap.mn eo.cn ‘ o u a u m u a u m w o w o mn~-4 umfiu4 cad-4 vac-4 emu; meuJ ua~p4 mmfi-4 ungu4 _ eNIJo max~4 efi-4o _ womdo = m 25:: a eo.mo mo.oo ¢¢.co_ -.oo_ No.o°~ o¢.°°— oa.ao eo.ec~ ee.~ofi pa.oo °~.oo -.oo p~.oo ~°.e no.o no.o No.9 .c.o co.o w pa.oo he.~o >~.mo h¢.~o «p.00 ve.mo nm.oo~ ma.mo c—.uo cm.¢o 4~.o h~.o mm.o mm.c n~.c m~.o man» mccc mo.e mm.o mm.c mo.c mo.c mo.c co.o mo.o we.c sc.o so.o o.o c>.o om.o em.o mom;< c¢.~m cm._n c<.mm ccomn eo.¢m oo.~ an.>m om.o~ om.cN oe.m~ oe.m~ cox on.mw o~.sm om.oN cm.am o¢.m~ oo.~m 0mm cc.cm o_.en om.- an.~m om.Hm o~.om nummu mw.o mm.c hH.o h~.o sm.o pm.o Ho.o Hc.o mm.o mm.o mm.e mm.o mcm4< crown o>.cm oe.on =¢.on o~.pm o~.hn cm.mm cm.mm o~.om o>.cm oc.¢m oo.¢m Noam H H H H H H H H H H H H HuHo oo.co mn.oo m~.oo oc.mo om.¢o mo.co« n<.mo om.ho oqoco 4.on oh.om om.om om.cm oc.om oo.cm oc.mm co.wm co: eh.nm cm.o~ o~.p~ oc.om om.mm om.>~ ouu c«.>m om.o~ om.em oo.mn om.mN o¢.om nomum mo.c mo.c no.9 00.: 05.9 os.c om.o mm.o cm.o om.c om.o om.o MONJ< oc.cn oc.¢m oc.on ec.on om.on cm.cm cm.om co.om oo.hm oO-pM om.>m om.hm mouw H H H H H H H H H H H H nannn Ha—an Hawum umwum HAHIN ~a~s~ _ Hame menH ammnH uawnn uaHIH HuHIH_ m Twat/So E :ommEuEH_ 15.5mm mEfioE .m .mm E :33? gm NI< wwdzmood S593 aaangak 3: SEK 32.: £83336 3:23 kc wmmSSS NSHSSHDIQ MAME? 43 TABLES 1—7 oococ 00.00 sm.ro hn.¢o oo.ue~ 05.50 ~o.¢o mo.mo -.~o Nc.wm mm.ho mo.ao mH.o mg.o um.o w~.c -.o ~N.o -.o mm.o H~.o Hm.c h~.o sm.o mcoo mc.o no.o ma.a no.c mo.c no.o no.o mc.o mo.o ac.o ee.c cn.o on.o >n.o hn.e >«.o rc.= en.o .o o#.o m>.o m>.c co.mm ca.c~ oo.m~ ao.mm om.e~ om.cm o~.~n o~.Hn co.m~ oc.m~ on.om an.om c_.>m o~.>~ oo.mn cc.o~ om.n~ om.am o~.cn o~.on co.mm o¢.o~ cc.Hn anonm mm.m mn.m .o~ cm.n~ cm.nN cm.¢~ om.<~ op.mm o>.n~ o¢.mm o¢.mm eh.~m op.Hm cm.o~ ah.mm cm.cm oo.mn oo.om o>.om oH.nm on.cn om.¢n om.mm om.nn o>.om cc.o cw.° MN.” n~.~ ~0.e no.e oh.c oh.c ~0.c mw.o mm.e mmoc c—.mm 0—.mm c~.~m auomm o<.¢n o¢.cm ao.¢n cc.¢m oc.vm oc.~.mo >m.~o oo.mo oc.~o Hc.mo He.om 0—.9 0—.9 cm.° ¢~.o o~.o o~.c m~.e m~.a s~.o 5—.o hm.o s~.o Ho.o «e.¢ mo.o mo.° «o.c *e.° vo.o ao.c m¢.a mo.o Ho.o co.e m¢.e n¢.o mm.o mn.o om.o on.c mn.c mm.o om.o cm.o mm.c mn.o c~.e ¢H.c o~.e ed.o h~.c h~.o cm.c c~.o on.o om.o -.o hnco nc.o no.c om.o om.o mm.o mm.o cm.o cm.c No.9 mo.o oo.o oo.o cm.s~ on.hm oo.om eo.o~ om.n~ om.n~ eo.H~ oo.HN o0.¢~ eo.cm on.>N em.>m cm.m~ oe.o~ oo.cn an.¢n c¢.Hn c¢.>m co.mn em.~m c—.¢m om.mn oo.¢m oq.on mc.o m¢.o c¢.o 00.9 o¢.e ceoo om.o om.o em.e cm.e ao.o o¢.o cm.mn cn.mm ep.¢n eh.¢m o¢.¢m o¢.¢m om.nn om.mm oo.mm ao.nm o¢.~m o¢.mm H H H H H H H H H H H H _ anxw‘ Huxw< unaam ~n~Oh Onz nommu oz: womb Odo ow: Own nomum MONJ< Noam 4«h0h on: mowxu ozx wouh o.o mwoc ch.¢n cb.¢m oh.~n oh.Nm cm.on omoom H H H H H H -az~n Hazmm anHm anHm _ wamm wawm 25:2: thcmo co.mo ~m.—ou Hm.~a Ncomcu ~5.no no.ho mm.no N~.co m~.mo m~.mo mooco ¢—.e mH-o m~.o mn.c 0H.o o_.e mH.o ca.o 0H.e o~.o Hm.o -.c nc.e no.o ncoc no.c no.9 n°.o ce.o vo.o mc.o me.c Mcuo no.0 hn.o pnoo 50.: hc.c 04.: oq.¢ mm.c rm.c mm.c mm.c mm.o mm.o oH.c o_.c nH.° nH.c -.° -.e oo.° oo.e 0H.o OH.° oH.c 0H.° No.0 mo-e mm.c nm-o s¢.o ~¢.o amuo "moo smuo sm.o mo.e mo.c chumm cp.¢~ eo.¢~ oo.c~ o¢.om o¢.cm om.HN om.~m e~.m~ am.m~ om.~N om.sm ooumm om.om 0—.on co.on o¢.en o¢.mN ch.~n emooc ov.om oc.~o oM-«M eo.~n mo.c mo.o m¢.o mc.c HH.~ “H.~ os.~ ch.H Hh.c H#.e hm.e sm.o cm.en em.(n ou.mn o~.mn om..m on..m am.Hn ¢¢.Hn 6H.mm oH.mn oo.om oo.om H H H H H H H H H H H H HnHNn Ha_~m HuomH Huon_ HuamH HuamH _ HuaH° _uaHo axHoN azfia~ _ Hemcw Hwnmm 25:3; ~0.ho ~n.co mo.oo mm.no 90.00 oh.no m~.mo n¢.¢o ho.co hm.no 00.50 oo.eo .~.° H~.° H~.° _~.o m~.° m~.o a~.° ¢~.c m~.° m~.o oH.o oH.a mc.e nc.e ma.c No.9 nc.o nooo cc.c co.c co.o {c.o nc.o ne.c cm.o an.c m¢.e mU.e Fm.c No.c ~0.e Ho.o no.9 nc.e «coo mc.o no.c oc.o ecoo o#.om oh.¢~ oc.o~ oo.om o».h~ ohapm emohm co.»~ o~.¢~ c~.¢~ oo.om cc.¢~ co-om oo.om oo.cm ce.om on.mm oo.~m cH.nn on.nm om.Hm om.~n oo.mm om.om emce cm.c mcoo m¢.o oo.o coo: oo.o 00.0 mm.o mm.o No.~ Noc~ cc.an 0°.cn cx.mm on.mn o..mn c..mm cm.mm cm.mn om.em om..m as.qn op.~.o m~.c m..c c~.o ~_.o -.° c~.e m~.= w~.c ofi.o m~.o oo.~ ¢..— oc.~ mm.“ so.e nn._ -.g »¢.~ mc.~ cm.~ mo.o ~¢.~ no.” ~m.° on.e on.o mn.e ~n.c Ho.o mn.c mn.o o¢.c pn.o mn.o ~..e on.o o~.- eo.o~ o~.- oo.c~ om.o~ on.o~ op.o~ oc.o~ o~.- en.o~ e~.- om.om co.o~ o~.o~ e¢.m~ co.m~ on.«~ on.o. cc.o~ cm.m_ cp.m~ co.m_ co.m~ e~.c~ cm.¢~ on.c~ om.m o>.e e~.~ c~.~” oh.m o¢.p oo.o on.o co.» co.» co.m cm.o co.m oo.m o¢.m mo.n m».m mo.n m~.n m_.n ~m.~ m¢.n po.n mm.m em.q mo.m on.~m c~.~m co.om cm.°m co..m om.~m on.—m oo.—m oo.flm om.om c».Hm om.o¢ cm.a< a u u u n o w a H a m u a_-xa _ um-x o~ax mmfiux on.nx uumnx mam-x cam-x magnx uafiux _ wmnx m1~nx . 25:3: .n.~eg oo.oo o‘.~o~ c~.~o~ m~.»a mo.oo op.po >~.oo~ mm.oo o¢.m¢ oo.oo~ oo.ec~ ¢~.Ho~ cm.c mm.c m°.o ~o.o .~.o ne.e oc.o Hm.o .o.c m~.c .~.o e~.c o~.° o~.c p~.o ~n.° m~.a o~.e H~.o m~.c m~.o o~.° «~.o zfl.o hfi.e «H.o o~.~ m~._ ca.o <_.~ oa.o c..~ m».° ma.c ~m.~ um.~ ~o.~ ma.~ cc." mm.c mm.c «~.c o~.° o~.e o~.e p~.o ¢~.o -.o om.c ~n.o om.o m~.e o_.- cm.om cm.m~ oo.o~ oc.o~ o~.o_ oo.o~ eo.e~ em.- co.m~ o~.o‘ om.o~ o¢.e~ co.o~ o~.o~ oo.e~ ap.h~ eo.o~ eo.o~ op.o~ ofi.p_ oo.efl cH.o~ am.m~ om.o~ oo.e~ op.» om.5 om.o~ cH.c o°.o co.» efi.o oc.o o~.- c~.o om.o oo.> am.F 9°.a o~.m nm.~ oo.m p~.n o¢.n om.n m~.n o».~ me.¢ mm.¢ mm.~ oo.~ em.flm eq.em cc.~m e~.~m o_.em oo.om o».o‘ ce.~m °~.mc co.¢. an.w« eo.~m e~.~m & a w a a H a a w a a u o cam-x end-x w¢ux oucux unnux ma~ux uamux egg-x om- oa‘- uan: uumn uufiwg u a mn.mo 65.nofi po.oo m~.oo oo.oo co.oo o~.~oH ca.oo ~p.oo oo.oa p».oc~ m~.ec~ pm.‘o~ cH.o ca.° -.o ne.e m..c c_.o mo.e o~.o ¢~.o no.9 mo.c c~.o #0.: m~.o .~.o cfi.° n~.o m~.a 0H.° c~.e m~.e m~.= o~.e c~.c -.o nn.o m~.~ ec.~ mm.“ n¢.~ c~.~ m~._ o~.. mn.H co._ no.“ Hm.” as.” :0.” m~.e rm.° «n.c mn.e nn.c en.o mn.o .m.o ~m.= on.o nn.c on.o ~m.o om.o_ oc.e~ em.c~ oo.m~ cc.“~ cp.o~ o~.o~ eo.o~ cm.a_ o».o~ c~.°~ co.ofl on.°~ om.mH ce.m~ oo.w_ e~.o~ o>.m~ oc.m~ on.>~ om.m~ an.p~ °~.m~ om.m~ an.m_ o~.o_ oc.- cp.o an.» o>.o oo.p e¢.p cg.» cm.» co.o co.» o>.m oo.m ca.c am.~ mm.m nm.n ~m.~ 99.. oo.~ hm.¢ om.n wo.~ Hm.~ F~.m °~.n «m.n oo.o¢ o~.cm co.oc om.om eg.om om.¢¢ om._m oe.em co.om oo.o¢ c».¢m cm.cm cc.~m m w o u a n. H a o H a u a _ wan cm- cue- man- can- ua~u may- un.u on- muwu uamu ma~n uagu ; u m < J< mwfizwood EAST MOLOKAI VOLCANIC SERIES 46 o~uco~ caomo Omuwo Hmuocm oo-ocu 00-00 mcoco~ tm-ec~ hhcoo mm-ecn Oc.°o~ n¢.0¢~ ouuccu mace mace mo.o No.0 Acne meao mono m—uo {Ono mono Neuo choc ha.c ~N-c WNuc mmno hwnc omoo «Noe Nm-c oaoo maoo knee cm-o N~.e mmoc NQ.N omon Boom cmom m¢oH chnN mm-N BN.~ Mb-N {mom qu— cone 0h.~ Ofioc mvoc hdoo hMoc Gene {moo Ecta Otoc OMoo Nvao Afloc mM-c mono oN-om otucm cnoom omu0~ ecoum cech cauom cunnm emoom cco~N oN-ON emu—N eoooN OO-MA o00a~ GN-N~ omoen om-a— c~om~ ecomfi cOoQ— anon" om-in ovocn afloon OQon omoo c0o0 oN-—~ ohoOA oeum c¢uo~ cc-mn eouh 00-0 oN-5 Ofi-ofi cmum o~no om-Q Cm.“ knoé 0¢nm Mean oc-Q 00cm No.m HG-m hO-m OO- ONoQ mO-N ocomt Gaga. omufiv ofino§ oeuom oMom§ ctumt cm.ofl ccnhe ON-QQ ONon ODaCm afloem m w w H a w m o m u M m M — @NIXQ cfilxa @NIXQ wauxn UQ~XQ whixn NUHXQ UUAXQ . 0¢lxa UMIXQ mawxm UQNXQ ma—xm n_ a 33:3qu {MoOOA #moOO “0-00 evooo {0.00 MQoOO {docc~ mw-OO 00-00 no.90 OH-mo 00.00 manna mcoc onuo mmuc MNoo m{.o acne #~.c uN-o v¢.c mo-o méco do-o Face eN-c mn-o mace {moo mm-c mN-o Nmoo {Hoe cave NNcc c~uc nm-o cm- meow *0." mm.~ m¢oH om.“ Nm- Hm-N onN ecn— ##on couN oou~ omue None Nmoc AQ-c hmoo mayo mcoc mfioe otno ovoe omnc mfi.c N¢oc cO-AN cfi-o— oG-oN cocoa oo.~N eo-ON om-cm oeoeN OOoHN cmaom cm-ON o¢oom cmucm ooomn omomn om.¢~ ecoc~ oc-mg on-Qn cm-QH econ" oOom~ ccumu oQom~ cc.<~ cHo¢~ cine cN-Q ceofi oQoh oM-O 90-0 chum 00.0 cM-h omoo oM-O each cmum mma¢ hmum ckofl oeom OO-M each ON-N om.m omoo OQoQ omoN cuofi ccum cmoom om-oc ah-04 on-om oQ-om cm-om o—onm canoe Ovohc cocom OF-Oc cN-hc ce-Hm m m o u a u a w u u M a u Ulfixn wmlxa UQIxa unlxn UQan wL—Xm Uthl wmlxl Ufllxn UMIxm mmwxa UQNXQ kmfixa z 2 9:3: {knead NQ-ccq 09.00 Maocofi anaco— wn.00 wm-mo mmnoa cacao" mmnocH mmuco~ omoho mm-oo hone mace Ncoo Ofioo mmoo ecuc ~Coo Q~.o enuo 00.0 acne Choc 00.5 myno mmoo mNao om-o omoc owoo mmoo "Nye NN-o onc "Mco «mac 0~nc o~.~ cc.“ om-H -u~ mung he." Geog ON.“ co.~ bu." Md.“ om.~ «Mgu once hm.o Nooo hm.c QMoc wenc mmoo once Qmoc mmo° NM.° cm-c omnc oo-QN ooom~ omooa oc-ON ccnom omoma omomu oicoN choofl on-N oh.®~ c>co~ chncw 09.0” cocma om.m~ acacfiv om-m~ emonm a¢.N~ choma co-mH co.m— oeuMH ccomH ocomH om-c emaom oo.mn coop cdoh entwa om.¢~ om-h enuv oo.o cm.m~ cuoc cone 0ma¢ hN-N hecm och {mun ovum ~¢o~ Nm-N vm.¢ whom deoN oN-Q ocum ocoam cm.nm omu¢c om-mm or-cm cfi-om omccm o-~m emoom enufim cQ-om cN-om oo-om a m w u u w w u w o w u a Ul~xl wmlxa muwxa oumxa UNIXQ mmU—x MUHXQ UUnxa — woexn UUQKQ wumxm cumxa unmxu a x 25.23: J.e om.e mm.o om.e o~.o gm.o mm.c mn.o no.9 om.o~ cm.oH co.o~ oo.o~ =~.o~ om.o~ oa.~m c5.o~ o«.o~ cm.om eo.o~ oc.c~ ec.°~ cm.H~ o¢.m~ on.m_ om.n_ o¢.m~ om.mH o~.m~ cm.m~ am.m~ o_.c~ on.m_ co.c~ oo.m~ oo.n~ om.~fi oo.h cw.m cm.o om.w °~.o om.c o“.o eo.o o¢.a cc.» oe.- >~.m oc.m h~.¢ no.~ co.m ~s.o o~.c mm.~ oh.e s«.« on.m so.¢ mm.m om.o< oo.o« oo.w¢ oo.~m co.nm oe.»< co.hc oc._m oo.o¢ oo.oc °«.o¢ co.oe 95.x: m M a m w w m a m u u a m _ wmnxu mngxa uafixa ¢mnx wnux wmnx mazex umzex manlx uunnx mam-x oam-x muavx N Au.“==uv> o~.oo~ oo.oo ~m.oo o¢.oo cm.mo m_.oo oo.oo oo.oo oc.oo pm.oo co.oo mp.oo om.oo oo.c mo.o no.9 ea.o «c.o mo.o me.c mo.c No.9 mc.o mo.o ~c.c Ho.c a~.o o~.c mm.c c~.o hm.o cm.e -.o ¢~.o sm.o m~.o 9—.e cm.o mm.o mm.~ oo.n mo.¢ ao.m an.” me.m pc.m om.~ mo.~ mm.~ oc.m m~.~ o~.~ mm.o mm.c pm.o om.c ~¢.c mc.c -.o ma.o hn.o m..c o..° m¢.c mm.c os.cm oo.fim oo.o~ cc.cn em.o~ c¢.o~ o_.o— 9°.AN cp.a~ c«.c~ on.- co.o~ om.e~ eo.m~ on.~H oo.- on.- co.n~ c~.m~ o~.~H om.mH cc.n~ on.m_ oe.<~ o~.efi oo.- on.» oo.o~ e~.a~ cp.o om.o~ c~.- eo.¢~ o_.o cm.o. o».¢~ om.m oc.o om._~ «m.¢ mp.e ”h.m co.m oc.~ o>.m mm.m oa.m He.m hm.» ~¢.n c~.m m¢.m om.o¢ cm.p¢ eH.o¢ co.c« cc.om cm.qe o~.o« cm.h¢ c».e« om.ce o~.oe co.o¢ ao.m¢ u w m m m m m a w » a w m uuaux emux cmnx o~ux cm-x um-x wazfix uazfix annx mazfix omxAx onux mumnx > x > 25:: oo.wo m~.ao oc.mo co.oo >c.oo~ op.oo a~.oe~ pe.co. m~.ee~ -.oo Hm.oo -.~°H ¢~.oo~ c~.o pm.c ~o.e mc.o mm.o mo.c pm.c o~.° ~m.° m~.o m~.c ~o.o om.o s~.o m~.c mm.c sm.o s~.o mm.c m~.: m_.o o~.o -.o ~m.c ¢~.o -.c -.~ mp.~ ow." mm.~ "m.” Hm.~ ~o.. mm.. cm.“ me.~ n~.~ an.” om.~ ac.o o¢.c mm.o n<.o mm.c om.o ~¢.e cn.o ~¢.c ~<.o nc.o m<.o ¢c.o o~._~ eo.H~ cc.o~ on.o~ o~.M~ om.an o~.H~ o¢.o~ an.- om.o~ on.o~ cm.o~ o~.o~ oc.n_ oc.¢~ c‘.m~ e¢.¢~ oo.wd oo.m~ o~.m~ cq.¢~ oo.wH co.m~ co... o~.¢~ on.¢_ o~.m oc.o co.¢ oo.m cm.o o¢.o om.» cw.c~ cc.» om.c oo.o c>.c~ o~.o -.m po.¢ cm.m o~.~ H>.m mm.~ oc.m mm.~ oo.m ‘o.m o~.< ~o.¢ oo.o em.oa om.o¢ co.cw oc.~m om.om om.om om.ev cm.cm cm.om c¢.em o«.oc oo.cm e¢.me H a w w u m u a a m m m w mmgux unanx _ oeux mum-x com-x muwnx uumnx mafiux un~nx _ wNIxa a_nxa _ am-xa eAIxn_ H J<»°b momzu oz: mom» o~.m o~.N mm.m oc.¢ c~.m~ oo.- cmomu on.~n cm.4~ cm.m_ ooocu cs.o~ oo.N~ o>.° co.” hm.e cN.~ «0.: wo.o 0h.c mo.o «0.0 nowmm onomm co.wn on.uN o>.~m oh.>m omoo~ om.s~ om.on c¢.~m o~.mm om.b~ ch.o~ mom4< e¢.nm co.o¢ oo.mm om.o¢ o~.om oo.~m oo.mm cm.om cc.~m o«.0c cm.mm oo.Hw Ncum h a H a w a m u u a w u wamlu oamlu unalu on—Io menu aznuu uwnu uumlu unalu umqtu h wwvo wanna G .H __ o a _ oo.Hc~ mp.oo oo.oc_ mo.~o~ uncgou mm.oo~ oh.oo mw.mc~ om.mo cm.ocH mm.mc~ moucofi 4.¢ H0.m c<.m acuc m<.n 0N.m~ o~on~ om.¢~ emcm~ c¢.c~ om.- cm.m~ cm.o~ ch.m~ om.m~ on.¢~ ouc so.c mo.e hs.o «cue mu.” co.o No.“ co.o Nm.c oo.~ ¢h-c nomuu ch.~m oc.om om.on oQ-cn 50.nn oo.cm oo.m~ c<.¢m oo.mm om.om oo.~m oo.om nom4< coocm on.~m oo.mm em._m oo.am o~.~m c~.nm cw.«w oéoom ow.~m cm.mm om.~m Nonm a M a w u a w w w o u a unncu wamlu uamlu wufilu UO—IU AZNIU w~lu an: wuml uumt wnal un~|_ .._ a mm.con mc.oo m~.oo~ ~¢ono 9—.ao oaooou Nmovo nw.‘o~ Hm.ooH sn.oo ~F.cc~ ecuoo AF.o nowmm c~uom c~.en om.mm ocohm counn em.om om.~m oh.b~ cc.om oo.om c~.on oN-on mom4< ononm o~.—m on.nm o>.~m oo.mc oc.mm o~.0§ co.mm om.~m cN-mm ch.mm om.cm moum w a m m u u a w a u u a _ ¢mo ax? on- mum: cum. uaT ufiu of .27 um» mo:- um? _ u m < Eng mmflocEm—Q mmmEuzszmum “Ea magma—u w 5 52m 48 mmflofimfln m mo $3.3 ECHm “sums? w E Emum $32:me m b5 ”Ian Hana—Soup .amioozwsmosEE yo emioonmia a we EEHp ,amiuozwgno»u_5 ,5 Hague—~23 «:8 fan Eganwuwm mum flan—EN? wigs—m ,m ad 5 525m 95 NI< $53501: £0.28.» URGQSEGM 3% ESQ mmmSoomeRKo mmwbufia NSNSmQDlw mqmflH 49 TABLES 1-7 c~oco mmnoo ~n.cc~ 0c.~o~ sm.cc~ mw.oo o¢.oo~ hm.—c~ mm.coH o¢.co— m¢.oo an.pm no.9 nc.e Ha.e ac.o cc.o co.o ~q.o an.c whoa m~.o om.o cwoc o~.o mm.c -.o ~o.~ Mm.e oo.o ¢(.< .m cm.~ ~m.o hm.e coom om.N~ cm.- econ eo.- o~.- eo.m~ om.m— co.o om.c~ oo.¢ e¢.N~ om.o o~.~ ¢~.~ hm.o nmoe hm.o m>.o hm.o ~0.o m>.c ~m.o om.e ew.o cc.mm c¢.¢~ co.m~ coucm oo.om on.om om.om c¢.m~ om.~m an.mm o_.o~ c¢.mm cw.mm o¢.nm ccchm o~.cm om.¢m om.~m cw.—m co.om em.m¢ o¢.hm o~.~m o>.¢m w a w a m a u u a w a M ~0mtua azuun wmluu azaua wcuun nxmua uwtua ua—ua unyua emnoa .axmua uu~ua z a 255;: No.00 oo.Hc— m~.~o~ No.0o— Ho.mon mo.oo~ na.ow m¢.>o wm.~oH em.oc~ chcoo~ om.oon mo.c Nc.o No.c No.c ~.c c~.e cw.m cm.e oo.n p>.m m>.m mm.m o~.m mp.m Nm.m mo.m ‘¢.¢ so.m cc.q~ om.m cm.m~ oc.n~ em.o~ og.o~ oh.o~ cn.mn oc.o~ om.m~ cm.~A cm.m~ c¢.c m>.c eh.o cm.c mo.o n~.o om.“ éO-o "m.“ »>.c mo.o ¢~.o ow-Hm om.cm o~.on co.on o_.c~ co.om oo.m~ co.o~ om.- oooom oo.mm om.on cc.cm o».¢m em.~m cm.~m oo.cm om.~m ao.nm em.0< om.~m oo.mm oo.em om.mm a m a a m u w a w a m a unwun wouua minua cumua mu~un uudua _ omvun azaua ewlua axgou ‘ onion nxmua ._ 9:3: ~m.~o~ s~.oc~ oo.oo~ 0c.no_ Nn.co~ mc.oofl No.~o~ oo.eo~ om.oo~ Nm.oc~ no.00 ~h.>o ~c.c «o.c No.9 ho.o ~c.e mo.c ~e.o no.9 mo.o mm.c em.o mh.o ph.o o~.c o~.o mm.o ¢~.c mm.c scoc me.e o~.o o¢.m Ho.n mc.m nn.c eo.~ c¢.n o~.m sm.¢ 50.: cm.~ “0.0 ma.n cm.o cm.m~ omoo oo.h om.¢~ co.n~ ch.o~ onown o¢.m~ oooma eo.- oeo- Ncuc mm-o co.o 00.9 Nm.o ah.e 00.9 00.9 oc.o n#.c ho.c Mpoo cc.hm oc.om cmco~ on.o~ c~.Hn oo.om oo.h~ cm.m~ cmumw co.—m couh~ eo.om co.mm em.mr coopm em.~o ec.~m oo.Hm o~.>m co.¢m o¢.¢m o~.oc o~.o em.eo~ se.o e°.c sc.o ~o.o mc.o me.o wo.e om.~ cm.c do.o me.c mn.e mm.c ~m.o 50.0 ~m.m cm.m ~¢.a mn.< cm.¢ mm.c om.s e~.o~ o~.c~ o¢.- o~.mn om.- 03.0 «m.o o¢.o oc.c pc.o co.o mm.o Hm.o o~.nm om.mm c~.mm oo.om o~.om om.>N o~.om oo.>m oo.om oh.wm c>.em co.nm o~.nm c~.om m a a w a w a P omuu azmnu uzfinu omxu azdnu _ omuu azmno N > 9.2:: cc.~e~ oc.oo mm.ec~ ~c.~e~ p~.mo No.»o c~.oo cc.oo om._o~ Fm.oc‘ ~o.co~ e~.~c_ co.c mo.o nc.o mo.o "0.: ~o.o e~.e No.0 ~o.o No.9 ~c.c fin.c c¢.o oc.° mm.o om.o o~.o «9.: pc.o pm.c mm.e cm.c e>.c ¢«.¢ nH.o ms.m -.4 om.m mn.o «~.e m>.m me.¢ mo.m .c m».o c>.o ne.o «n.a Nc.o mo.c sm.o mp.o c¢.p~ o».om oc.- om.»~ oc.m~ oc.mm cm.o~ cm.mm oo.mw o>.om op.»m om.om oc.cm on.~m ec.>m om.~m ee.mm om.+m oo.mm cm.»m oo.cm cm.nm ofi.mm oo.em M a w u m M m m H a u a wnznu uqzfiu wcuu wamnu cam-o ma~|u Una-o _ menu waxmo unzmu mamuo uamlu ; 25:3: eo.mo n>.mo fimqmo~ «n.eo— m~.~o_ mn.eo~ o¢.oo ~¢.oo~ mc.~o~ o¢.mo~ ~m.~o~ ao.~o_ NH.c no.c oo.c ~c.o Ho.a mo.c ~o.o mn.o om.c om.e o~.o mn.c n~.o wm.o m~.c om.e m~.c om.c pc.e Fu.m ce.¢ em.o do.c ce.m ~o.m ~o.< c~.m »¢.4 m~.m he.c nm.~ om.o oo.- on.m co.m~ om.c~ 95.." eo.- em.m~ cc.- om.¢~ o«.m_ o~.m~ nc.o mo.c oc.e mm.o mm.o 00.: mc.o ~o.o o~.~ oh.c cm.o oh.o c».om o¢.o~ co.>~ oo.o~ om.o~ oh.~m co.m~ ca.om o¢.mm oo.Hm am.o~ o~.mm c“..m oc.~w eo.mm oo.nm co.mm o~.om co.nm co.mm om. o 910 ON: ON.~ hco~ oo.~m oc.m on.r oeohm acowm mmoc oo.mv mmum um.m mc.n No.» He.» Nb.¢ mc.m oo.m hdom MN.N 9—.0 «moo ocnhn oc.mm sooth om.~n oc.~m ch.mm cmocm e~.¢< c~.mc coomc oo.m¢ co-oe om.o mm.c no.” cooun ch.- 90.0 oc.- c¢.m~ Nocn mm.~ oo.m~ oN.o m m w a a H a a w H m w wM|Ju cnlhz war»: mambo QINFU nxnbu azgho nighu amipz aid»: azahu _ wHIJHb ._ x e<.co oNoma ocomo mc.oo <~.oo em.~o owchm ~¢.~o oo.mo mm.mo eo.eo wonmo ——.o none meoe no.o mnuo oc.o cm.~ om.- o~.hH c~.h~ om.hm co.N~ cmohm coouu c~.nm oc.m~ co.mm o~.n~ cc.mm c¢.m~ oo.mm heum no.n Wm.~ co.o #0.: oouc ao._ on.~ so." he.“ cm.~ m~am p«.~ much oh.ck owonc cM-OO eoocb eN.- cm.mo oc-oc oh.mo o¢.mm o¢.mc o«.¢o om.m¢ cm.” omum ccncfi cc.” Nc.~ m~.~ «cow om.m mm.¢ on.» cc.m~ coca“ w H H w m u u a a H H H omlbz -|A~ -|ht wmlh wNI» aINIH grad» aznlp aiglh “nihu “Niko amnho _ 9.2:.qu #c.ho 00.50 ch.mo mc.¢0 m~uocn ¢m.Ho~ ”N.om ooumo m¢.co— Hm.oa ncomo ooowo no.o cN-c co.o em-c ccccn c~.cm ~c.c (9.0 oh.>~ oo.¢ o~.o m~.o cm.b¢ o¢oh< oouum owom~ mo.N cm.N em.cm cm.mm MH.¢ o¢.m~ om.0m oc.uo Fenm woua mo-o Nc.~ cm.m oo.c Mb.” me.— once“ «o.m om.m om.m om.» ecocw onomp om.mh cm.mn om.on 0—.Nm cm.~> c~.~¢ oc.me o~.o< omofim mm.c Hcco mm.~ n¢-N oQ-m~ ohaha hc.~ oo-o oo.o~ h~.n 0m.o mm.o w m m m a a m m H a m m @NIJu o~04~ omlhx ¢Nlhz azahu mzfihu wGMIJ cwmlh HNIP azanp _ mwmla ONIA a £5.33“ >~.mo cooho 90.50n No.50 Ho.m°n nt.ocn nm.oo 0m.mo m~.oo mm-ocu w—omo ooumo wmuc ~o.c Neua mace oeoNN mnoh co.c c¢.c cc.~ °¢o~n 0N.N cM-Mn count on.cN chumc coomm ¢c.o coon“ cmoac cmoeu ev.o~ mc.¢ cmuom oecm m¢.m choc hh.~ anot mnuc coon ocum ON." nocm o¢.w Noon 00.0 couma ecoch oo.~m chumm ecoeo choc» c~.m¢ cm.mk ocooo oo.m¢ cacao eacfic cmcc Neom Am.a NM.~ on.c moum c¢~o co." m¢.~ e¢.m~ ho.m on.e~ m w w w a a w w M a .H a wulJ _ 0—! on! — an! ASNI $2"! an! emu waxwl UQINI wax—l oar”!— m a u m J4h°h momma monk out out mo~4¢ 4o ”1.00 mma¢o o~.m0 omuco scumo A.m ocom omcm cc.m~ owom~ on.m~ momcu couca atom z~.x cmucm cM-MN om.hm cc.mm cmqu o#.m— cc-w¢ cmarc cm.m_ Noah cr.e ouch c~.o Hocm whum mm.c 0#.m —~.¢ mace Nc.~ we." >~.m ow! Gen—M Omonm enofit oN-AO co-NO Decom °>on0 cm-Ew acumm mum-me omorw¢ ofinqm OUL cuom cm.N~ ctuwd ocum cmuN xnom Nm.m cmuo oooo «m.c om.c can: mo~4< H m a a a u H a a w a fi nzrku Q:Mhu mehu arm»: AIM»: min»: AKA»: azuhz min»: — wMIJH QENJH unlbu a 9.: =3 “— ccn¢a eh.ma ofiomo 00.90 mo.mo chumw odofia >m.xo hooom mm.¢o oo.co mo.mo Jakob cs.o hm.co cm.oo mm.co w~.mm «c.c «o.o :~.o oo.o «coo cm.a¢ o¢.- cc.mm om.mm o<.~m ~¢.m o>.N hx.m mm.m cm.~ c¢.m< cm.0F co.oo o~.oo 35.3: «<.c hm." oc.~ mm.~ mo.— w w a m w _w¢-; am-p 12mph 1:..» _ on-» 2:: Z omoro om.ow mo.~c~ no.~m oo.co~ ¢¢.mo nc.ho mm.mo cm.¢o ca.mo mo.mo mm.mo mc.c eo.< o>.o¢ ~m.m co.mm Hc.c n~.o No.c «co: ao.c >o.e n~.m c<.~m om.mn mc.m c~.c~ o>.~ cm.o¢ o».m~ co.or om.om om.nm om.o~ on.mm mc.~ m cm.~: o<.:c c¢.o< c~.m¢ om.eu o:.~> om.¢c .m.~ an.m cm.h~ <~._ oo.o~ mm.c m~.~ cm.c mn.c ~¢.o m~.o :0.“ m u a H a w w m w w w m cMI» azNIJ arm-» azfilh gran» oNIJ can» wNIJ egng omuh cnlp azmu» > 9.2:: ao.po co.~o anorm nh.¢o mm._o on.~o mouho ec.eo sc.mo no.00 m¢.¢o c~.ho c<.m rm.m mc.o hm.~ mo.n o~.m mo.o No.9 no.0 so.H cm.o~ ocoqm oo.o oc.m~ o>.o~ ccucg cM-en om.o¢ co.om on.¢ erode oo.- on.om cc.m m~.fi mm.~ hm.n mh.e c>.¢ oa.m o>.m m~.~ mm.m mc.m >_.~ e~.mm cm.m> emcee o~.«c om.Nc o¢.cm a—.¢c c~.~c :m.mm om.o¢ oc.>m om.mn cm.m ~0.m mn.~ oo.¢ rm.¢ o>.m >~.o am.c mm.m mc.o om.c om.o a a m a m a w m w H a w ar—n» axfinb emu» azwuh azalr midi» @{IJ uni» ¢m1h waz~4 ntfilh — wguquxh p m J £8 53> gunman: :sm .mmmfi.mb.g.HH.> £8 30> gammamm .88 .82 .E a A: s .20 £8, fififim dam .BmmHV £855 Rivas Hmmmww .8 9H 8 a: mauwfigofi HSCON 88: Eggm £18m H9: -8: £33: £880 H5580 .88ng sfiooxovH .8 89558: 83. H8583 «Em m5 £83m >32 .9898 H5595 638088 98 Z .9888 H5580 Amy HEN Amy .E wlm 3 an 83wa .88 532% 38 .mfidmm 888: .9898 H2595 AHV llllllllllllllllll Quill} IIIIIIIIIIII WMUfihU @HHDOHU .Eo wv ”.32 an .8 dam: H28 3858 H.580 I! 55.38 mm 8.8 wfimm wwanuwww 282 .839? 35% 93M 58mg HEN 98s £2 ummw HERD woniumww 982 .839? aHsmm 08M .8ng 98 28a at ummé 8&3 Hume—Comma. 982 8.8820 Hasnomxmz HEN. wfl< .88: 6:8 fit ummm 8v .8320 8H< :8 895558 Ex N «33m “8893 2:8 98M Amy .dmfidemem 89a Ex ov .ESHDESM 980 8c: 98a ti ummm AHV .8393 :33 wqu BE 823me 58515 cm 8m magma 8235.8 85 59c Emkumwzfifidow .Ewamzm «18m mmovH 9:: Ex H :93 8.85 8H p850 Hgmomx .mSH 50¢ EMBumoBfigw d: «5&2 Ow SNESNEQMNWH EOhm .38 :3 35on 89G 833533 Eu Nw 882 8 mo ouawEwmsm 855$ 26: mo mfinwmo N3 Esmnfiouum 8.88 8 E H .8 $888meme 88838: 38.393 ”3398 H5580 £5 @8an .8 0% m 523 mme bnmaoam .38 :5 85.8% 89¢ $33318 Eu om ammoH am no wonwgmnsm .gwuwfioEmem 8338 -qoa no 2m .8 £583 9m 628233 £583 Ego? 3:96 32 8 bomb: 26 in £08 .98 av .8 258me23 38:3: mmmH banana 8 vmaH awnfioamm 8m Saw wEHw>oH 339.555 SHE 8.59% we manofiohsmmofi \3 H8833: ounwEmnsm waniomww 98 Z .E «H mm £058 mm NHEHmmonH 20:883me 13 $378: wmaowhwéawafi ”E H» umme am no woqogmnsm .E o: Bonn Bwawfits :mfiszmem E :EdHoo «Emma ”:5 “wen—Comet 282 .8 com ammoH as 32938, :mfismeHmm E 5:38 wEmmE pan 83883. 982 .E E. ”Hoax 398%?» SufismeHmE 5 55:8 “~8me ad; 8038088 08 Z .mommocEmHm 5593 .280 9898M 8 5128858: Ewan 50¢ 6:8 at 8mm 9%: $547385;me .dmfidemem mo 58 .fi8n Ex H 55:5 $me3 c8 8cm HEM dmfiswEonm vmmH ~32 llllllllllll :mEEwEmem NmmH was.» vvvvvvvvvvvvvv :23ng OZ ommH HwQEwaQ llllllllllllll £2358 02 vva 835889 wmmH ummflzxw ”05% £32 vmmH ngEwumom SE “3583 83 388962 5:358 oZ ............ :mfiswEonm mNmH 35H“ mNmH Nuwsfiom ummH NEH" vaH NEH“ 4822882 vaH -mm 5 E88038 389:5 KANSHJHEEQ .3320 Esmommmz Ham owH< £09503 dag £2 ammo SHED mNmH uwflw=< @8850 :mmmZ Him Esmomxmz NNmH ~32 .8: «582 massage 658 ”ct amoafisom .mfiuom £2855“ 9.9. season .ADVQHV ANMMNW llllllllllllllllll emu IIIIIIIIIIIIIIIIIIIIIIII Ow IIIIII .HHHH mo mfiwu Q~£mmh®>®h =Hwem -HHOO ~A=m5uhm> .SNESNEGHME wuNmHIhomH 838m .88: :32 E; 83:83: .......... «.8588 282 .......... 383% 2oz .......... @2538 262 £88.20an .swasmewam 82.82 9582 83884 «$8 “1555 859.8 .wuwQ 8288QO $5 fififidw 83350 398 E :ofiucfiomwfl E ”85.25an yo :ofiaooq :Hbmd 33$» HEN $va 328982 9.8 98.35 9% £85358 .8 .8353“ HE“ 2:39» 8,": VAQNIQQQN Kofiaofis SSSNM “a 329:» 8.83:5 fimxxmkfi N35 mtednfxmrtH Hamfib STRUCTURAL SUBDIVISIONS OF KILAUEA .oD 88:: .3883 :8: .8830 0:815 :mmaamm .GEC :8 -:8>m 6:8 6:883” 685:9 .09 6Q .938 85:3 :8: IN>H®w£O oflfio~o> gammwafim .5”.va 36:8 ME“ :om:mam .oQ .68: 89:8 98 838:» .823 86:8 min 8885M .Awwmd «mangovw «:8 mama .88 :8: 98 .38an .283 .830 Raging «:8 6832 .6me 6832 8:: .Emig £96m .mme .3 m8::::w $8: 856.30 0:815 :mfiwgmm $me 3385M .Qmad >0>CM «Em 9802 da— .me0 @880 6:: 833m .8qu A3: :3 888,830 0:88.? 5:95“: ”:86: E8: 6an 82:8 “0:: 8a£65m .88: 22:8 is .823: x88 :8: IIIIIIIIIIIIIIIIII ovll‘l IIIIIIIIIIIIIIIIII ovllill llllllllllllllllll ovlnull llllllllllllllllll evil} i=1 lllllllllllllll 3...: I329: ~58 3:88 @5850 IIIIIIIIIIIIIIIIII cull}- llllllllllllllllll owl}: doe—Sm 6:898 @5880 “83:88“: 8:0 Z 8:? 48:88 mama: 63:3 12:8: 6x88 @5880 -I 3 858m .8888 “6:580 llllllllllllllllll ovnllu llllllllll mvmomho WCSOHO $5285 862 mwfifiaa 8.88: 6888 “658:0 $180 888: «:8 9:88 380:0 .88 .38: 6x688 825:0 .8 w 3 a: 3:88 8818:: 8638:, .338 888: MEN 5888 888:0 12:8: .......... coniowwv 682 8:8,: 888: 6898 8880 A8 .8388 6888 «658:0 AC “8:886 8:02 , |IIIII| llllllllllllllllll evil} IIII GOmaQH—hw mfl wwhm wawm 8828 .8 98: «5:8 8 “:68?“ 6:8 pm: 83: I--- :oE::8 mm 8.8 688m .8880 82 .8 8858: :88 38oz IIIIIIIIIIIIIIIIII 36-1-: llllllllllllllllll 311:- .................. evil} :1 88:58 8 88 888m .......... 80:88:: 68 Z 2:: «8:888: 8 Ex om 8:: 839$ :83 68M mmo8m 88$ 0 EZ 68.x 88m llllllllllllllllll och}: llllllllllllllllll evil} :1 :2858 mm 88 688w 58:58:88: We 885:8 mam 5:8 8:: m 3 v £888 ”58.: 68M mo :8: :888 «Em 85:60 .smfisafiflmm no 863858 :88 :88 8:: w 8 m .8688 2:8: 68M .8 :8: 5383 «in 8.580 .8280 87w .8 :83 E: H :88? #8»: 83: A6 .8858 mm 88 888m 30 {l 883:8 mm 88 68mm IIIIIIIIII “88.88: 6:02 .:m:::m:8fimm mo :88 EM N. “:88? “18.: 68M :3 .:oE::8 mm 88 286 Ad 85:88:: 6:0 Z .......... 66:66:38 :wEm .8588 .86 mam 8:66:88 58% .8838 .8 fin: 8:8 E :8 am we 86:88:: -mm. 888:8: 886:8 883:8 MEN :8 w: .8 £33 llllllllllllllllll 36:1: IIIIIIIIIIIIIIIIII owl‘s- defiant :8 HEM 686388 :«Ew :8 S .8 88:88:: .mfi: 888:8: 886:8 8.83:: ”:8 NH .8 8:83mnsw .858: :8 88888: 6M8: ”:8 ma :8: 68:: e8 60:6Emn5m :8 E .8 82:83: .mE 888:8: 886:8 683:: ”:8 m: .8 683.88% ...... :8 m8 mo 8:6Ewndm 3% 8 6558 :3 8868:: 6688586 so on 8 8 a. 85338 llllllllllllllllll owllii .................. 36:1: lllllllllllllllll ,‘Ofiv‘ll‘ll ‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘‘ owlily‘ IIIIIIIIIIIIIII ow??? 88m. :5 @888 ma awn—88:: 66:8Emn:m \\\\\\\\\\\\\ floaomuwfi GCOZ .38 :5 889% :8: 881.686 .8 mA :88 :8 8:8686 .8488 @888 #8:: ow 0o :5 @8388 :3 8:88:65 :28 «.632 -8 8.3 m: :0 8 183:: 53:33: 35 .8 5:8 6:8 3:: 88m .8820 82 We 8858: 8:8. 58: 6:8 :5: pmmm ............. :ofiad8 SA .8880 8?. .8 88> 6:8 E 6:8 at: 83: 888:0 83w mam 8:» 8.62563 “ED 8:532 6:8 ab: :80 .88:0 83w 8 88:0 83w mo :83 EM ma 88.: 6:8 ”E: ammm .mofimofimvw 55m 3 88:0 83w 6:8 at: 8mm .moEonmM :dm 8 9:83 o :52 856M 6:8 :5: ammm 6:00 mgflavw 88: 8 88:0 SEEM 6:8 ”E: ummm 8258::me .88: o 82 68M 8 88:0 82. 6:8 9:: 8mm 6:o0 «338M :8: 8 88:0 £588wa 6:8 ti 8mm 6:o0 «EMEM E8 88:0 :8: .82 86380— 688 a: 83: .8820 68144 :8: v:m E 6:8 ab: 8mm ................ owlyll‘ ............ :8358 e Z 8::me o EZ SEE 3 88:0 8:» 6:8 at :80 6:00 3: Eofiwm 86: 8 88:0 :8 «Z 6:8 at: :80 Dwaimamwflm 680 onommvw :8: 6:8 ti ammm ‘‘‘‘‘‘‘‘ 88:0 8: «adamx HR: 58:83» 82 baa :8: :32 83 E8 82 $886: .82 $2 $8 8256: $2 .8238 :68: :88 ”82 8:832 mom: 8868: mmm: £082 83 3:88 mme um:m:< m8: :8” m8: 32 $2 83:88Q Sm: 8988:6w 3m: Eda .882 £82an own: 58:56.”: JCSENH $2 8832 DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO #595 ENE 59:5 55» $83 3555 c855 biumam .32 33. En: 9:2: 555:”: .HbmH 53800 H355 neumfl :oHun5uMHv .33 bi. Ea: 3:2 555:? m 505555835 03635 $5395: Hog 255555. 323 we amcE ”5235.5... 9: 525555585 2:: 355:3 6255000 503555855 «mo 52,55 5 mwuaozui :33 45%: 93 E 5553 9:556 55% 25,5 woos mwawu _a.~v>wm 035$ Jaw; 02m .5uE5wEmfim E 853 @5qu £398» 525me o: “.55 won—nomwfi 35H. «6:: :5: R355 «o mmnafl 5235.535 .8 55555.5 «0 555 .5 503558 we EwEmgafifiou .Su 3mm? .oQ .oQ .oQ .335 55553 is» 52830 25815 5555353 .HmbmHv 055m 555 “.895 .mmaaom $8: 3255 .5 .oQ A33. #5553 be”. $3.830 25815 5535353 $052»wa [#555 555 335.8 55520 352085 @502 IIIIIIIIIIIIIIIIII 051:: Ila—.255 555 935.8 @5586 ............. 530355 wnoZ 13:35 H58 mxowuo 5558.0 52:5me @502 2552 II 50555.8 m5 598 255m 595,885 982 1:- 50355.8 mm 355 wfimm .Eoumkm £55m mqu 85 5mm? $5.8m $525 635 3.83% 5555358 053 doEQEm m5 N98 255m llllllllllll 53033. 282 I1 55355.8 mm 525 085m Hume—€085 282 5555qu 598 £5555 £5355 manta E noEwEueuwD .55 cm mm 5055 mm no 505055 .58 Ranouiox €551 OOH 593 325 we M553 5.835 .58 wH mo 5385.558 Hmunofigo: 28355.8 9:555 @551 CH no @535 5.82:: 503553 283 «mi. 55.3 mH 555 325 mo M553 5553550 5035598 Han—5.5.8: 593 M88 mH 5553 985 mo fizmb 5235.5 .58 H3528: 525 ”55 5 55.: 958 {8 woqumn5w 55.3 mH 555‘ 988 no 95:» 55535 -1- 853535 Emzm 53> .E H 555 0.88 mo 558qu5 .me 135818: 538.55 Emkzg ”58 wv mo 935D .Eo Hum no “595353 .25 338535 538.55 555253 HEM So mm mo QED «593 35:55 E 50:55:0me .wmovH 55nH mo 855 -5555 £53 a! $53558 5935 no 935 amwofi5om 5.8350 5055=M ma €55 E95558 H58 555555253 58550 Hoxmxmsmwm so: 555 E “53250 aw5m=vH .«o «.35 ummmfi5om 535353 55m mo £55335 Ex H 8 535.5 Emafim 658 $5 ummm 58550 Em5m® H53 585.8 mo “mm? Ex H 8 .5330 Nxmmm “839$ 255$ 33M “585.85 555 058 at ammm IED «5552 £53 £5 ummm .EH «5552 we 5mm? 5.558 55H m 3 mqu a? 325558 555 5.8350 $553M mo €35 «mo355om 558555 .25: HEN 58550 8wa .mqmmvH 58553 £53550 $55sz no flan pmwmngom 555555 go 523004 ES 30.8085 53H 538355@ 32 55;. ES 358952 32 $2 mg: 92:85 H hmH .HoQEwumwm E: 55mg .33 5555Hu500lvmm~leem~ 6:855 355ka “B £55m 3.3455.» @9285: H55 maccaSLWlH 5554a. EARTHQUAKE DISTRIBUTION 7 155°15‘ 155°00‘ W | i l Mauna Kea .Kaiwiki New T '\ ___/ Mauna L03 Kaloli and Kaloli 2 Kulani 19° ' Waikah/ekafie 30’ _ Strip Volcano House / / Honu' . Kapoho KAOIKl Uwekahuna . ' POhOIkl FAULT Ohaikea » IQ ' ,. Kahuwal SYSTEM - . . . alama Ki Moana Hauae Keakapulu Hakuma d p, 00$ j 19° 15' ——Kapapa|a , 1C 0 PACIF 15 20 MILES 0 5 10 15 20 KILOMETRES l EXPLANATION Structural subdivisions of Kilauea Summit area [:3 Active part B a Koae fault system Unfaulted part East rift zone South flank Inactive part Hilina fault system E Southwest rift zone Boundary between volcanoes o Triangulation or trilaterution station FIGURE 3.—Structural subdivisions of Kilauea Volcano and locations of triangulation stations referred to in this paper and established before 1970. Boundaries of subdivisions are gradational in most places. Compare with principal structures shown in figure 2. subdivided into upper, middle, and lower parts. The upper part extends from the summit area to about Makaopuhi Crater, the middle part from Makaopuhi to Heiheiahulu, and the lower part from Heiheiahulu to the seacoast at Cape Kumukahi. The rift zone can also be divided into active and inactive parts on the basis of data presented in this paper. The east rift zone forms a prominent submarine ridge that extends far beyond the coastline (fig. 1). 3. The southwest rift zone, about 4 km wide, is characterized by structures similar to those of the east rift zone. It is the site of many eruptions, though fewer than the east rift zone. The southwest rift zone extends only a short distance beyond the shoreline (fig. 1; Moore and Fiske, 1969). 4. The Koae fault system is an east-northeast trend- ing, 3-km-wide zone of open cracks and normal faults connecting the east and southwest rift zones. 5. The south flank is that part of the volcano seaward of and boundéd on the north and west by the east and southwest rift zones and the Koae fault system. The northern part of the subaerial section of the south flank is characterized by unbroken, seaward-dipping lava flows. The southern part is disrupted by the Hilina fault system, a set of predominantly south- dipping normal faults. 6. The north flank is bounded by the summit area and the east rift zone. This flank is characterized by east— to northeast-dipping lava flows that abut and in- terfinger with flows from Mauna Loa. The rift zones are believed to have a structure simi- lar to that outlined by Fiske and Jackson (1972). Chiefly on the basis of seismic evidence, the dikes are considered to be shallow, bladelike bodies, largely if not entirely confined within the volcanic edifice and fed either directly from the summit reservoir system or from secondary high—level reservoirs in the rift zone itself (Swanson and others, 1976). The dips of the dikes are believed to be steep to vertical on the basis of (1) observations in older, deeply eroded Hawaiian shields such as Niihau (Stearns, 1947) and Koolau (Stearns and Vaksvik, 1935), (2) seismic evidence, which shows that earthquakes and tremor directly associated with the intrusion of magma occur in a narrow zone directly below the point where lava reaches the surface, and (3) interpretations of ground deformation data related to the intrusion of magma on the upper east rift zone in August 1968 (Jackson and others, 1975). Contrary in- terpretations regarding the dip are examined later. The flanks of the volcano are assumed to consist of lava flows above sea level and pillow lavas below sea level, with a layer of hyaloclastic debris between; data for this interpretation are given by Moore and Fiske (1969). EARTHQUAKE DISTRIBUTION The distribution and timing of crustal earthquakes provide probably the most unequivocal evidence for in- stability of the south flank related to magmatic events in the rift zones. Koyanag‘i, Swanson, and Endo (1972) summarized this seismic evidence, some of which is shown in figure 4. Numerous earthquakes take place beneath the summit area and are presumably related to stress release in and near the main reservoir system. Almost all other earthquakes occur at shallow or in- termediate depths beneath the south flank or its bound- ary zones, the east and southwest rift zones and Koae fault system. Thousands of earthquakes ranging up to 8 DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO Kilauea Caldera 19°25’ PACIFIC OCEAN O 10 MILES O 10 KILOMETRES EXPLANATION Earthquake epicenters o o Earthquake during initial Earthquake during middle period-of activity or late period of activity FIGURE 4,—Location of earthquake epicenters from brief seismic swarms related to eruptions and ground-cracking events between 1963 and 1969. Earthquakes plotted are of M22 and occurred at depths less than 15 km. Note that earthquake epicenters migrate south-southeastward across south flank following initial seismic- ity along east rifi; zone and Koae fault system and that earth- quakes are virtually absent from north flank, a distribution also typical of periods between eruptions (Koyanagi and others, 1972, fig. 1). Slightly modified from Koyanagi, Swanson, and Endo (1972, fig. 6). magnitude (M) 5.0 take place each year in these areas away from the summit, furnishing clear evidence of instability within the volcanic pile. Study of the focal mechanisms of south-flank earthquakes shows that most maximum stress axes are oriented southeast and are horizontal or plunge gently seaward (Koyanagi and others, 1972; Endo, 1971; E. T. Endo, oral com- mun., 1974). In contrast, the north flank of Kilauea has few earthquakes, and most of those that do occur emanate from upper mantle depths well beneath the base of Kilauea. Thus, an immediate conclusion is that the north flank is relatively stable seismically compared with the south flank. The high seismicity of the rift zones is not surpris- ing, for these are areas where intrusion and eruption take place. The reason for the high seismicity of the south flank, on the other hand, is not so readily appar- ent, but the timing of earthquake activity relative to rift eruptions and intrusive events appears pertinent to an explanation. Figure 4 shows that earthquakes related to magmatic events occur on the south flank only during the middle and late stage of seismic activ- ity, after the intrusive event has largely or completely ended, whereas earlier earthquakes cluster along the east rift zone and adjacent parts of the Koae fault sys- tem, near where intrusion is taking place. This se- quential relation, discussed in more detail by Koyanag'i, Swanson, and Endo (1972), indicates that intrusion is a precursor to major south-flank seismic- ity, which in turn suggests a cause-effect relation be- tween them. The inference is that stress is built up in the south flank by intrusion along its northern edge, then relieved over a period of a few days during the late-stage earthquake activity. One effect of forceful intrusion of magma is lateral displacement of wallrock on either side of the dike. This may provide a mechanism to account for the south-flank earthquakes. If the observed south-flank seismicity is caused by such displacement away from the rift zone, it should be possible to detect this dis- placement by geodetic means. For this reason we undertook a geodetic study of Kilauea, results of which demonstrate that large ground displacements on the south flank and elsewhere have indeed taken place in a remarkably systematic fashion. The following presentation of the horizontal and ver— tical deformation data is long and detailed as we at- tempt to document our case. Many readers may wish to examine figures 5—14 and then proceed directly to the three summary sections on long-term horizontal de- formation, south-flank horizontal deformation, and vertical deformation on the flanks. The principal interpretive part of the paper begins with the section, “A Structural Model for the South and North Flanks of Kilauea.” LONG-TERM HORIZONTAL DISPLACEMENTS DURING THE 20th CENTURY EXISTING DATA AND THEIR QUALITY Horizontal ground displacements over several periods a decade or more long have been derived from six regional horizontal control surveys. Four of the surveys are triangulations, in 1896 (1914)1 and 1949 by personnel of the US. Coast and Geodetic Survey, and in 1958 and 1961 by personnel of the US. Geologi- cal Survey. The other two surveys are trilaterations conducted in 1970 and 1971 with a model 8 laser-beam geodimeter by personnel of the Hawaiian Volcano Ob- servatory; the general procedures and measured dis- tances for the 1970 survey, the trilateration most used in this paper, are given by Swanson and Okamura (1975). The regional triangulation surveys by Wilson (1935) and Wingate (1933) are not used in this paper because neither was tied to a baseline that we consider 1Most of the 1896 (1914) survey took place in 1896, but stations on the south flank were not triangulated until 1914. See text discussion of figure 5D for details. LONG-TERM HORIZONTAL DISPLACEMENTS DURING THE 20TH CENTURY 9 to be stable. Figure 3 shows the location of all triangu- lation stations occupied at least once prior to reoccupa— tion in 1970. Each of these surveys took place over a period of weeks to months, and the 1896 (1914) survey was split into two intervals separated by 18 years. As Kilauea sometimes deforms greatly over much shorter periods of time, the effect of possible contemporaneous ground deformation on the quality of each survey must be con— sidered. This is done in the section “Supplemental In- formation,” with the conclusion that the problem of contemporaneous ground deformation is minimal, with the possible exception of the summit area in 1961. The general quality of the four triangulation surveys is given in table 2. The 1949 and 1958 surveys are adequate to define reliable long-term displacements of the magnitudes shown by Kilauea. The 1896 (1914) and 1961 data are poor, owing to large closure errors and, in 1961, very poor network geometry on the south flank. The precision of the 1970 trilateration is on the order of 5—8 mm/km, far superior to precisions of all the triangulations. The problems of imprecision and the generally poor geometry of the survey networks make the triangula— tion data unrealiable for strictly quantitative studies. Nonetheless, most long-term ground displacements have been large enough to overshadow these problems, so that the triangulation data can be compared with the trilateration data to derive general patterns and semiquantitative amounts of displacements. The selec- tion of stable baselines and the procedure used in de- termining ground displacements are outlined in the section “Supplemental Information.” RESULTS Horizontal displacements derived from comparison of the four triangulation surveys with the 1970 trilat- eration survey are shown in figure 5. The net dis— placements for most stations, especially those on the south flank, are large and systematic for the survey periods, which range from 9 years to 74 years in dura- tion. In this section we correlate these displacements with magmatic (eruptive or intrusive) events that oc- curred during the survey intervals and find that the TABLE 2.—Precision of the triangulation surveys Closure of triangles used(sec) Probable error in length 1 Order less than1 Triangulation Average Maximum 1896 (1914) 3.9 12 3 1 in 5,000 (20 cm/km) 13:9 f 1.19 3.09 2 1 in 20,000 (5 cm/krn) 1 8 ri t 4.0 1 .6 3 1 in 10,000 (10 cm/km) summit 2.3 } 3'6 5.6 2}3 (good) 1961 4.8 9.6 3 1 in 5,000 (20 cm/km) 1Taken from Gossett (1959, p. xiv—xv). Results of the present study suggest that the 1896 (1914) and 1961 surveys have probable errors of 1 in 10,000 (10 cm/km) or less. displacements on the south flank are systematically related to the opening of new ground cracks in the east and southwest rift zones and in the Koae fault system. However, the exact timing between specific magmatic and displacement events is unknown because the sur- vey intervals are long and may include several events; later we examine shorter survey intervals in which specific magmatic and displacement events can be cor- related. In this section, the results are presented in the order of decreasing reliability instead of chronologi- cally. 1949—70 Displacement vectors on the lower east rift zone gen- erally point perpendicularly away from nearby fissures that opened in 1955, 1960, and 1961 and are of mag- nitudes close to the probable aggregate amount of opening of these fissures, judging by observations of the fissures soon after they formed (fig. 5A). Kaliu and Honuaula are both located within the rift zone, less than 400 m northwest of some of the 1955 fissures (Macdonald and Eaton, 1964). Kapoho is southeast of the easternmost 1955 vent and the main 1960 vents (Richter and others, 1970). Heiheiahulu is southeast of the nearest 1961 fissure (Richter and others, 1964). The radial pattern of displacement defined by the Volcano House Flag, Keanakakoi, Uwekahuna, and Ohaikea vectors indicates tumescence of the summit region centered northwest of Keanakakoi. Such tumescence is reasonable in View of the similar loca- tion of swelling centers documented for many later periods of inflation (for example, Jackson and others, 1975). This radial pattern is defined by comparatively small displacements, 30 cm or less, and would probably have been distorted or even completely obscured if the baselines had moved. For this reason we have confi- dence in the stability of the baselines between 1949 and 1970. The displacement of Keanakakoi is larger than at the other summit stations, perhaps partly the result of the opening in 1954 of a fissure oriented N.75°E. about 1.5 km north of the station (Macdonald and Eaton, 1957). Puu Huluhulu shows a large northwest displace- ment of about 1.7 m that is not perpendicular to the trend of nearby eruptive fissures. This displacement is possibly the net result of major north-northwest movement directly“away from the rift zone during the nearby eruptions of 1962, 1963, 1965, and 1968—70 combined with lesser westward movement away from an area of uplift centered over a magma reservoir near Makaopuhi Crater (Swanson and others, 1976). Ohale, in the Koae fault system, was displaced about 2.6 m south-southeastward, perpendicular to cracks that formed in the area during major faulting events in 10 DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO Kulani Kaakapulu KIh-hl 2 KID-Doll A 1949-70 B 1958-70 EXPLANATION ‘p Keakagulu . Displacement vector — Eruptive fissure C 1949—58 Area of ground cracking D 1896(1914)—1970 Trilateration station Kulani Heiheiahulu and Wm Kan-h. 2 I I, I I l I | Fun Ulau ’ I E 1896(1914)—1958 ,‘ 1 V F 1961—70 1’ | 0 40 80 120 INCHES O 5 10 15 MILES I I }_1—‘_|——l_|‘[—I | ‘I I 0 100 200 300 CENTIMETRES 0 5 10 15 20 KILOMETRES ¢ VECTOR SCALE MAP SCALE LONG-TERM HORIZONTAL DISPLACEMENTS DURING THE 20TH CENTURY 1 1 1950 (Finch, 1950), 1963 (Kinoshita, 1967), and De- cember 1965 (Fiske and Koyanagi, 1968), and several minor events as recently as May 1970 (Duffield and others, 1976). The pattern of displacement directly away from the southwest rift zone suggests rift dilation, for Koae and Kamakaia are southeast of the axis of the rift zone and Puu Ulaula and Kapapala northwest. The several episodes of faulting in the nearby Koae fault system could also have contributed to the displacement of Koae and Kamakaia. 1958—70 All stations within the lower east rift zone were dis— placed southeastward, directly away from the rift zone, in a pattern consistent with rift dilation during the 1960 and 1961 eruptions (fig. 5B). The 1960 vents were near Kapoho, and the eastern 1961 vents near Heiheia- hulu. The four stations between Kapoho and Heiheia- hulu are not located near vents active since 1958, but their displacements may record the effects of intrusion accompanying the two eruptions. Stations on the south flank (Pohoiki, Malama Ki, YY66, and all stations southwest of YY66) were dis- placed by large amounts directly away from the east rift zone. These displacements are largest (2.3 m at Panau) at stations opposite the most active part of the rift zone between 1958 and 1970, even though the nearest vent areas are more than 5 km away. The dis- placements generally decrease away from the rift zone; for example, compare Panau with Laeapuki and Pulama with Kupapau. FIGURE 5.—Horiz0ntal ground displacements (solid vectors) at Kilauea Volcano derived from comparison of 1896 (1914), 1949, 1958, and 1961 triangulation surveys with 1970 trilat- eration survey. Procedures used outlined in the section “Sup- plemental Information.” Base stations considered stable are underlined. Southwest rift zone and active part of east rift zone (fig. 3) are shown by shading. Locations of rift eruptions and major episodes of ground cracking that occurred during a given survey interval are indicated. Vectors in C were ob— tained by subtraction of vectors inB fromA; vectors inE were obtained by subtraction of vectors in B from D. Dashed vec- tors in C and E were computed by Lloyd (1964) using stand- ard least-squares analysis with no geologic constraints. Dot- ted vector for Honuaula inA indicates station was intersected but nbt occupied during 1949 survey. Alii in F shows no dis- placement greater than survey error. See text for discussion of the vectors at Panau and Laeapuki in F. Parts of network common to both surveys are indicated for each survey inter- val; further information can be obtained from the US. Geologi- cal Survey, the National Geodetic Survey, or the senior au- thor. Waikahekahe and Halona, located a comparable dis- tance from the axis of the rift zone as Laeapuki and Kupapau but on the north flank, registered little if any displacement. The radial pattern of displacements in the summit area during this period indicates expansion around a center 1—2 km southwest of Keanakakoi. This center, south of that for the 1949—70 interval (fig. 5A) corre- sponds to an area of uplift defined by leveling between 1958 and 1971 (Okamura and Swanson, 1975). The fact that these small horizontal displacements define the center of deformation quite precisely is evidence that the assumption of baseline stability is justified. Puu Huluhulu shows about the same displacement for the 1958—70 interval as the 1949—70 interval; this indicates that most displacement took place after 1958, consistent with the absence of upper east rift eruptions between 1949 and 1958. The displacement at Ohale during the 1958—70 interval has about the same azimuth as between 1949 and 1970 but is somewhat smaller, probably because it does not reflect the 1950 ground-cracking event in the surrounding Koae fault system. Koae moved away from the southwest rift zone, and its magnitude of displacement suggests that about half of its 1949—70 displacement took place after 1958. 1949—58 Vector subtraction of the 1958—70 displacements from the 1949—70 displacements yields the 1949—58 values. Kaliu, Honuaula, and Kapoho, all within the rift zone, clearly responded to dilation accompanying the 1955 eruption (fig. 5C). Heiheiahulu, Puu Huluhulu, and Uwekahuna did not move measurably, Ohale reflects the 1950 ground-cracking episode near it, and Keanakakoi was displaced away from the 1954 vent fissure nearby in the caldera. The dashed vectors in figure 5C are displacements calculated by Lloyd (1964, fig. 3A) using data from the 1949 and 1958 triangulation surveys adjusted by least-squares methods with no geologic constraints. The pattern of these displacements is rather similar to the one we derived, except for Puu Huluhulu and Heiheiahulu, where Lloyd’s data show large displace- ments along azimuths nearly parallel to the east rift zone, and ours show none. This discrepancy clearly re- lates to the methods used in the derivation of the dis- placements. We believe it important that there is no geologic evidence for deformation that would give rise to displacements at Puu Huluhulu and Heiheiahulu of the azimuths and magnitudes indicated by the least- squares adjustment. 12 1896(1914)—1970 In 1896, stations Kulani, Olaa, Kaloli, West Heiheiahulu and Heiheiahulu (100 m apart), Kaliu, Honuaula, Kapoho, Moana Hauae, and Hakuma (1891) were triangulated, and our derivation of their displacements up to 1970 is shown in figure 5D. In 1914, Heiheiahulu-Kaliu was used as the base for triangulating Hakuma (150 m south of Hakuma (1891)) and all stations farther west. Assuming that this base was stable between 1896 and 1914, we de- rived the shortest vector shown in figure 5D at Hakuma for the period 1914—70. The Hakuma and Hakuma (1891) vectors are so similar in magnitude and azimuth, with the vector for the longer period somewhat larger, that the assumption of stability of the Heiheiahulu-Kaliu baseline for the period 1896— 1914 seems justified. Furthermore, this period was one of almost continuous weak eruption in the famous Halemaumau lava lake at the summit of Kilauea (ta- ble 1) and on the basis of recent work at Kilauea, it is unlikely that significant deformation took place along the rift zone during such summit behavior, especially as far as 30—50 km downrift. We therefore assumed baseline stability and derived the 1914—70 displace- ments for Kupapau, Pulama, Laeapuki, and Panau. The general relations between displacements and eruptive fissures were the same for the 1896(1914)— 1970 period as for the previously discussed examples. Heiheiahulu, Kaliu, Honuaula, and Kapoho were all displaced at right angles away from nearby vent fis- sures and ground cracks. Moana Hauae and other south flank stations farther west were also displaced away from the nearest active parts of the east rift zone, and the seaward stations (Laeapuki and Kupapau) were displaced less than landward stations (Panau and Pulama) (compare fig. 5D with 53). The largest dis- placement, nearly 4.4 m at Panau, was opposite the most active part of the rift zone. 1896(1914)—1958 Displacements for this period (fig. 5E) were derived by subtracting vectors of figure 5B from corresponding vectors of figure 5D. Lloyd (1964, fig. 3A) also com- puted vectors (dashed in fig. 5E) for this period, again by a least-squares analysis. The displacements derived by Lloyd and by us are rather similar, although ours are smaller and more systematic, lacking the clockwise swing suggestive of adjustment or survey error. Both sets of displacements show that the south-flank sta- tions moved seaward by large amounts, most likely along azimuths nearly perpendicular to the trend of the east rift zone. Comparison of figures 5E and 5D shows that only about half of the displacement of the DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO south flank from 1914 to 1970 took place in the 44 years before 1958. Thus the rate of displacement has been almost four times greater in the last 12 years, a time during which eruptions along the middle and upper east rift zone have been far more frequent than during the earlier period (table 1). Displacements of Heiheiahulu, Kaliu, and Honuaula between 1896, 1958, and 1970 are reasonable in light of the eruptions in these areas. However, the displace- ment of Kapoho between 1896 and 1958 is unexpec- tedly small, considering the major episode of ground cracking and graben subsidence that took place just north of Kapoho in April 1924 (Finch, 1925; Jaggar, 1924). In fact, the overall displacement at Kapoho be- tween 1896 and 1970 can be largely accounted for by the 1960 eruption (compare figs. 53 and 5D). We can- not eliminate the possibility that significant displace- ment during the 1924 event has gone undetected owing to unrecognized survey errors. However, the mode of ground deformation in 1924 (maximum subsidence along the north edge of the graben and northward tilt- ing of the graben floor: Finch, 1925; Jaggar, 1924) suggests the development of an asymmetric graben similar to that which Cloos (1968, fig. 18) modeled in clay. In this model, the north side of the graben would have moved northward, and little if any ground dis- placement south of the graben would be expected. 1961—7O Three major episodes of faulting (May and July 1963 and December 1965) in the Koae fault system (fig. 2) and thirteen upper and middle east rift eruptions (De- cember 1962, August and October 1963, March and December 1965, August and October 1968, February 1969, and five events associated with the Mauna Ulu eruption in 1969 and 1970) were the dominant struc- tural events of this interval. Ohale moved southeast- ward more than 2 m, and Puu Huluhulu moved north- westward more than 1.5 m, perpendicular to new ground cracks and vent fissures (fig. 5F). Dilation of the southwest rift zone is suggested by displacements at Koae, Kamakaia, and possibly Puu Ulaula. South flank stations within, and inland of, the Hilina fault system (fig. 2) moved seaward. The displacement at Kaena Point is similar to that at nearby Laeapuki for the 1958—70 period (fig. 53), not surprising since the two stations are equidistant from the east rift zone, only 4 km apart, and are not separated by intervening ground cracks of faults. Laeapuki itself was not triangulated in 1961, as shown in figure 5F, but Panau was triangulated from Kaena Point and Laeapuki. In order to estimate the displacement of Panau between 1961 and 1970, it is necessary to assume a displacement vector at LONG-TERM HORIZONTAL DISPLACEMENTS DURING THE 20TH CENTURY 13 Laeapuki. If we assume that the displacement at Laeapuki for 1961—70 is the same as that measured at Kaena Point for the same period, the displacement de— rived for Panau compares closely with that for the 1958—70 period (compare figs. 5F and B). This close agreement seems reasonable and constitutes evidence that the 1961—70 displacements are indeed reliable, despite the poor quality of the triangulation (table 2) and the weak network geometry. Just as for earlier periods, south flank stations oppo- site the most active parts of the east rift zone and Koae fault system were displaced more than other stations (for example, compare displacements west of Goat with those east). The north flank apparently remained sta- ble, as the displacement at Alii is well within probable survey error. AUGUST—SEPTEMBER 1970 TO OCTOBER 1971 This period is bracketed by two trilateration surveys conducted on the western half of Kilauea. A more closely spaced survey network was used than during the previous periods discussed. Major eruptions and related ground cracking occurred in August 1971 in Kilauea Caldera and September 1971 in the caldera and along the southwest rift zone. In addition, a small eruption took place in late January and early February 1971 from new cracks south of Fun Huluhulu (fig. 6) on the east rift zone; this episode was related to the con- temporaneous eruption at nearby Mauna Ulu (Swan- son and others,_1971). The effects of the August and September eruptions dominate the displacement map. HVO 34, about 1.5 km west-southwest of HVO 10, shows the largest dis- placement, more than 1.25 m, which is best interpreted as the net result of deformation attending the two eruptions combined with lesser inflation of the summit area. Most stations in the summit area show some ef- fect of this inflation, which took place largely before the August eruption. Displacements decrease abruptly down the southwest rift zone beyond the eruption site. The south flank was markedly displaced during this period, although far less so than areas closer to the eruption sites. Displacements are directed away from the general trend of the east rift zone and Koae fault system and are oblique to the southwest rift zone. The magnitude of displacement is greatest in the sector di— rectly south-southeast of the area in which the three eruptions took place and decreases both to the east and west. In contrast to previous periods, the eastern part of the Hilina fault system dilated slightly as indicated by direct measurement of extension between Laeapuki and Panau, and Kupapau and Pulama; this extension is not shown clearly in figure 6 because of the scale. The western part of the Hilina system was displaced l O 10 20 30 40 50 60 CENTIMETRES VECTOR SCALE N Kea kapuluo Kaena Point EXPLANATION O Trilateration station. Open (lot if no dis- placement detected ——> Displacement vector MAP SCALE Eruptive fissures and O 5 10 75 MILES associated ground +___1__L_r_._'_J————l cracks O 5 10 16 KILOMETRES FIGURE 6.—Horizontal ground displacements between August 24- September 23, 1970 and October 3—18, 1971, derived from trilater- ation surveys. Holei was first occupied on November 18, 1970. Stations held fixed (Kulani, Strip, and Keakapulu) are underlined. Eruptive fissures and associated ground cracks that opened during survey interval are shown. Southwest rift zone and active part of east rift zone (fig. 3) are indicated by pattern. Stations west of Kamakaia and Nali show no displacement greater than survey error. Diagram of network showing lines and stations occupied is given by Swanson and Okamura (1975, fig. 2). seaward as a unit, with little or no buildup of horizon- tal strain. SUMMARY OF LONG-TERM HORIZONTAL DEFORMATION Long-term horizontal displacements, as measured over periods of years to decades, are of considerable magnitude at Kilauea. The summit and southwest rift zone showed small net expansion for the periods 1949—70 and 1958—70 and considerable dilation related to the 1971 eruptions. Throughout this century, sta- tions within the east rift zone and Koae fault system were displaced along azimuths nearly perpendicular to the rift zone and fault system; stations southeast of active fissures were displaced south-southeastward, those northwest of active fissures, north- northwestward. The entire south flank moved south- eastward in a direction almost exactly perpendicular to the east rift zone and Koae fault system; this direction is more nearly perpendicular to the trend of the rift zone as a whole than to individual fissures within the rift zone. The direction of displacement is similar to that of maximum stress axes found from the study of 14 the focal mechanism of south-flank earthquakes. The displacements on the south flank are large, as much as 4.4 m at Panau between 1914 and 1970 and 2.3 m at Panau between 1958 and 1970. Displacement of the south flank is greatest south-southeast of the most ac- tive part of the rift zone for a given survey interval, and observations show that the magnitude of this dis- placement approximately accounts for the probable total width of new fissures and ground cracks in the rift zone. For at least the 1960 eruption, north-flank sta- tions about 10 km from the vents did not definitely move, whereas south-flank stations as far as 10 km from active vents moved tens of centimetres south— ward. This is consistent with the seismic evidence indi- cating stability of the north flank relative to the south flank. In contrast to north-flank stations, those close to, but west of, the southwest rift zone were displaced to- ward Mauna Loa during the 1970—71 period. Most of this evidence implies a direct causal relation between magmatic events and ground displacement. However, the periods of measurement between these regional surveys were so long that they contain more than one magmatic event; consequently the measured displacements are the net effect of several episodes of deformation. These displacements alone are therefore not adequate to correlate a specific eruptive or intru- sive event with a specific episode of ground deforma- tion, and we must turn to shorter survey periods to demonstrate such a correlation. HORIZONTAL DEFORMATION DURING SPECIFIC INTRUSIVE OR EXTRUSIVE EVENTS In this section, we present examples of short-term deformation events keyed to specific eruptions or epi- sodes of ground cracking associated with intrusions. It is well known that filling and emptying of the magma reservoir system beneath the summit of Kilauea is generally accompanied by largely reversible ground deformation (Wilson, 1935; Eaton, 1962; Decker and others, 1966; Fiske and Kinoshita, 1969). We focus at- tention here on other types of events that apparently reflect the forceful intrusion of magma as dikes into the rift zones. Most of our examples are controlled by geodimeter data of excellent quality, indicated by both repeated measurements and small closure errors. The triangu- lation data for the 1955 eruption are poor (third order), but the pattern of displacement is compelling. FEBRUARY—MAY 1955 EAST RIFT ERUPTION The early part of the February 28—May 26, 1955 eruption took place in an area where a new highway was under construction. Fortuitously, triangulation DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO surveys were carried out just before and just after the eruption; all stations in this local network are located along the east rift zone. The resurvey indicated that substantial ground movement took place during the eruption. The ground displacements shown in figure 7 were derived using two different assumptions. The preferred assumption is that all displacement of Honuaula, Kapoho, and Kaliu between 1949 and 1958 (fig. 50) took place during the 1955 eruption. The other as- sumption, used by Macdonald and Eaton (1964) in computing displacement, is that Kapoho did not move during the eruption. Both interpretations indicate northwest displacement of stations north of the fissure zone, and displacement vectors based on the preferred assumption indicate southeast movement of stations south of fissures. Macdonald and Eaton (1964, p. 107) noted that the pattern of displacements is similar to that produced by “a simple pulling open of the fissure zone,” with no evidence of strike-slip movement. Sta- tions opposite the central part of this zone, where erup- tive activity was concentrated, moved more than those at either end. The displacement at Honuaula is less than that at 154°55' 154°50' I I , 19° o 20 40 so so INCHES 30' i—i—t—i—4 O 50 100 I50 200 CENTIMETRES VECTOR SCALE Heiheiahulu I 0 1 2 3 4 5 MILES 0 1 2 3 4 5 KI LOMETR ES EXPLANATION ,/” . / Displacement vector Vent fissure Triangulation station FIGURE 7.—Horizontal ground displacements during February 28—May 26, 1955 east rift eruption. Solid vectors were derived on preferred assumption that all displacement of Honuaula, Kapoho and Kaliu between 1949 and 1958 (fig. 5C) took place during 1955 eruption. Dashed vectors were computed by Macdonald and Eaton (1964) on assumption that Kapoho did not move during eruption. Data for triangulation stations north of Honuaula-Kapoho baseline were obtained in late 1954 (before eruption) and May— June 1955 (after eruption) by the Territory of Hawaii Highway Department. HORIZONTAL DEFORMATION DURING SPECIFIC INTRUSIVE OR EXTRUSIVE EVENTS stations farther northwest. This may result from the combined effect of horizontal dilation and uplift along the fissure zone (Macdonald and Eaton, 1964, p. 105— 107), for such uplift should in theory cause the maximum horizontal displacement to be located some distance away from the crest of the uplift (Mogi, 1958; Dieterich and Decker, 1975). Alternatively, the intru- sion of northwest-dipping dikes could have caused such a displacement pattern, although first-motion studies of earthquakes during the eruption suggest normal faults dipping 70° to 90° southeast (Macdonald and Eaton, 1964, p. 122). FEBRUARY—MAY 1969 EAST RIFT ERUPTIONS Eruptions on February 22—28 and May 24—29, 1969 on the upper east rift zone took place within the newly established geodimeter network (fig. 8A). Surveys over the February eruption were conducted on February 11—13 and 24—27; most ground deformation occurred between February 22 and 24 (Swanson and others, 1976). Surveys over the May eruption were conducted on April 21—23 and June 2—4. The two bench marks closest to the eruption sites, Puu Huluhulu (within the rift zone) and HVO 117 (along the northern margin of the south flank), moved away from the vent fissures and associated ground cracks. The displacement vec- tors are nearly perpendicular to the new fissures and reflect the divergence in trend of the fissures for each eruption. SEPTEMBER 1971 ERUPTION An eruption took place on September 24—29, 1971 from a line of fissures between Kilauea Caldera and a point along the southwest rift zone nearly 12 km away (fig. SB). The eruption site lies within a trilateration network that was occupied August 16—19 and October 4—7. Most of the stations are either within the south- west rift zone or near the caldera. Horizontal dis- placements were directed away from the new fissures and ground cracks and were of roughly equal mag- nitude on both sides of the eruptive zone. The average extension across this zone, indicated by the displace- ment vectors, was about 1.4 m, virtually the same as the total amount of opening on new cracks as deter- mined by direct measurement. Stations HVO 111 and HVO 109, on the north edge of the south flank of Kilauea about 6 km from the eruption zone, were dis- placed an order of magnitude less than stations nearer the zone. This dramatic decrease may largely reflect the partial closing of preexisting open cracks in the intervening Koae fault system. Such closing was documented by detailed measurements of deformation related to the forceful intrusion of magma into the southern part of the caldera in May 1970 (Duffield and others, 1976). 15 I N O l 2 3 MILES U 1 2 3 KILOMETHES MAPSCALE IT 0 5 10 I5 20 INCHES ‘\ 0 I0 20 3) 40 50 CENTIMETRES VECTOR SCALE Pauahi Puu Huluhulu / // Crater , XI C t ’ ae ra er w / / (£9 / ‘09’6 HVO 117 pa EXPLANATION /§/ Eruptive fissure 5 February 1969 displacement vector May 1969 A displacement vector EXPLANATION Displacement vector 0 I 2 3 MILES N O l 2 3 KILOMETRES MAPSCALE Kilauea o 5 10 I6 INCHES /\ Caldera 0 I0 20 30 CENTIMETHES ‘3‘; 3V VECTOR SCALE /\ \/ Q3“ B HVO 109 FIGURE 8.—Horizontal ground displacements related to specific recent eruptions at Kilauea. A, February and May 1969 upper east rift eruptions; base stations are HVO 135 and HVO 136 (fig. 6). February 1969 data from Swanson and others (1976). B, September 1971 summit and southwest rift eruption; base stations are Kulani, Strip, and Keakapulu (fig. 3). Kalanaokuaiki Pali is southernmost fault scarp in Koae fault system. SPECIFIC DISPLACEMENT EVENTS ON THE SOUTH FLANK The foregoing examples demonstrate that horizontal deformation accompanying single eruptions is similar in style to that suggested by the long-term geodetic surveys. We feel confident that the long-term dis- placements are the summation of several smaller episodes of deformation. All of the examples of specific displacement events, however, dealt with the rift zones themselves. The behavior of the south flank of Kilauea —J 16 must be related in some way to the rift zones, but what is this relation? The 1970—71 survey period (fig. 6) is the only reason- ably short period for which measured ground dis- placements are available for the south flank. Even this period, however, includes three eruptions, a prolonged episode of summit inflation, and more than a year dur- ing which other processes could have affected the south flank. The measured displacements for this period are consistent with the interpretation that they directly reflect the effects of the eruptions, but we now present more specific examples of deformation events on the south flank. Strains between March 6 and August 27, 1970.— Displacement vectors on the south flank are nearly parallel to one another, so that there is little, if any, extension or contraction along lines perpendicular to them. However, there has been measurable extension or contraction across the Hilina fault system, parallel to the direction of displacement, for all survey periods, so that coastal stations are displaced by different amounts, generally smaller, than inland stations. Thus we can use the observation of strain along a south-southeast direction, with little or no strain per- pendicular to that direction, to infer displacement of the type indicated by the more complete survey. This is not the only interpretation, but we believe it reason- able in the context of Kilauea deformation. As an example, figure 9 shows the results of geodimeter measurements for three triangles on the south flank between March 6 and August 27, 1970. Most lines that cross the Hilina fault system at high angles contracted, whereas lines approximately paral- lel t0 the trend of the system did not change length significantly. Contractile strain characterized the area. The principal axis of contraction (E2) of the calcu- lated strain ellipse is oriented approximately perpen- dicular to the east rift zone and Koae fault system and parallel to the azimuth of overall south-flank dis- placement. The minor axis of the ellipse is very small, probably within the limits of survey error. These rela- tions suggest that the south flank underwent seaward displacement during this short survey period. During this period, two ground-cracking events took place in the upper east rift zone and adjacent part of Kilauea Caldera (fig. 9). On April 9, wide ground cracks opened in and west of Aloi Crater, and the cra- ter was filled by new lava. On May 15, a major earth- quake swarm (Endo, 1971) accompanied the inferred forceful intrusion of magma into the southeast part of the caldera; this intrusion resulted in severe uplift, dilation, and some ground cracking (Duffield and others, 1976). A reasonable interpretation is that the measured contraction and inferred displacement of the L— DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO south flank were caused by one or both of these defor- mation events. Linear contraction of a geodimeter line between 1965 and 1970.—We can extend the observation that strain accumulates across the Hilina fault system during dis- placement events to the interpretation of long-term measurements of a geodimeter line oriented nearly parallel to the azimuth of displacement of the south flank. Figure 10 shows the location of this 5.4—km-long geodimeter line, which was measured 30 times be- tween August 1965 and December 1970; measure- ments were effectively halted in December 1970, when the line began to skim the top of a newly erupted lava flow, and soon thereafter the inland station was cov- ered by lava. In the 51/3-year period of measurements, the distance contracted a net amount of 32 cm (an ex- tensional strain of —5.9 X 10—5). The rate of contraction, at least during 1969—70 when measurements were most frequent, was related in a systematic way to major magmatic and structural events on the upper east rift zone and adjacent areas in the caldera and Koae fault systems. Each new eruption or ground- cracking event in 1969—70 was followed 2 to 4 weeks later by an episode of rapid contraction. Measurements were infrequent before 1969 but are consistent with the later pattern. The period of maximum contraction includes three upper east rift eruptions, all of which were accompanied by substantial ground cracking and large measured horizontal displacements (Jackson and others, 1975; Swanson and others, 1976). Beginning in early 1970, the geodimeter line began to show periods of extension alternating with periods of contraction. Most of these extensions could be corre- lated in time with earthquakes of M 2 3.5 having epicenters within a few kilometres of the geodimeter line and focal depths between about 5 and 9 km. All such earthquakes between February and November 1970 are noted with respect to changes of line length in figure 10. After the earthquakes in July and Sep- tember, we predicted extensions later confirmed by measurements. There was no large earthquake on the south flank before the November extension, but a 30- km-deep earthquake with M of 4.0 took place beneath the southern part of the caldera at the end of October. Summary and interpretation of horizontal deformation on the south flank—The data presented in this section show that the south flank responds to specific rift eruptions and related structural events by undergoing displacement away from the rift zones. The resulting deformation is similar in style to that shown by the long-term survey periods. Almost certainly, the long- term displacements are the cumulative effect of sev- eral magmatic and structural events along the rift zones and within the adjacent summit caldera. VERTICAL DISPLACEMENT OF SOUTH AND NORTH FLANKS OF KILAUEA 17 155°15' l Kilauea Caldera ' .1 May 15-16, I970 EAST RIFT Q Aloi Alae Crater Crater KOAE FAULT SYSTEM ‘k‘ April 9, |970 30‘5"“ I EXPLANATION {I} Pit crater or caldera O E Area of ground ZON deformation on indicated date Makaopuhi {3 A: -7.6 Crater Dilatation, in units of 10‘6 CE Napau Crater >____< H Maximum Maximum contraction extension Axes of strain ellipse Trilateration station LE FOR AXES OF STRAIN ELLI OTC 3 4 SMILES | l 4 I 4 5 KILOMETRES FIGURE 9.——Changes in line length, extensional and dilational strains, and orientation and magnitude of principal axes of strain ellipse for three triangles on south flank between March 6 and August 27, 1970. Changes in length are in millimetres; numbers in paren- theses are extensional strains in units of 10‘5. Dilation and center of strain ellipse are plotted near center of gravity of each survey triangle, following usual convention. Kalanaokuaiki Pali forms s The response of the south flank to rift events is not immediate; instead, there is a lag time of 2 to 4 weeks before the wave of deformation reaches the area 5 to 10 km from the rift zones, an average propagation veloc- ity on the order of 100 m/day. The lag time shows that the displacement is a result, not a precursor, of the magmatic event in the rift zones, a conclusion consis- tent with the timing of seismicity on the south flank relative to the rift events. Throughout this century, deformation of the south flank has resulted in net contraction across the Hilina fault system (fig. 5D), and this contraction was closely outh side of Koae fault system. monitored between 1965 and 1970. Such long-term ac- cumulation of contractile strain may have reached a critical level by early 1970, so that comparatively small but nearby earthquakes could trigger brief episodes of extension. We return to this contention in a later section on the Hilina fault system. VERTICAL DISPLACEMENT OF SOUTH AND NORTH FLANKS OF KILAUEA Repeated leveling surveys on Kilauea during this century document substantial vertical displacement of 18 5404.20 .16 .14 .10 .08 .06 .02 5404.00 5403.98 LENGTH, IN METRES .96 .94 .92 — .90 - .88 - .86 r .82 - §§§§§ 5403.80 lllllllllllllgll DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO .90 .88 .86 .84 LENGTH, IN METRES I965 1 I966 lllllllll’llllllllllllllllllllll 155°15' Kilauea o 5 I0 MILES Caldera 5 10 KILoMETRES 54 s E r R I FT 10“ . \\ okualw Pa Geodimeter line A Sept Oct Nov Dec I970 llllllllllllllllllllllJIIIIIIIIllllllllllllll l967 l l968 l I969 l I970 FIGURE 10.—Contraction of 5.4-km-long geodimeter line (upper right inset) on south flank of Kilauea between August 1965 and early De- cember 1970. Major magmatic and structural events are: 1, December 1965 eruption and ground cracking; 2, November 1967—July 1968 summit eruption; 3, August 1968 eruption; 4, October 1968 eruption; 5, February 1969 eruption; 6, beginning of May 1969—Octo- ber 1971 Mauna Ulu eruption; 7, new fissure north of Alae Crater, December 1969; 8, ‘new fissure and cracks in and west of Aloi Crater, April 1970; 9, intrusion and cracking in southern part of Kilauea Caldera, May 1970; 10, new fissure east of Mauna Ulu, July 1970. Inset in lower left shows changes in line length relative to times of all south-flank earthquakes of magnitude (M) 2 3.5 between February and early December 1970 and D, M=4.1; E, M=3.5; F, M=4.3; Cal, M=4.0. one caldera (Cal) earthquake at the end of October: A, M=4.1; B, M=3.8; C, M=3.9; VERTICAL DISPLACEMENT OF SOUTH AND NORTH FLANKS 0F KILAUEA the south flank and minor, if any, displacement of the north flank. In general, these displacements correlate in time and space with nearby eruptions, a relation also found true for the horizontal deformation. We are not concerned here with deformation in the summit area, although it is well known that this area under- goes uplift between eruptions and, generally, sub- sidence during eruptions (Wilson, 1936; Fiske and Kinoshita, 1969). We also do not examine vertical de— formation within the rift zones themselves, for several workers have previously shown that the area near an eruption site is uplifted, commonly with the develop- ment of a keystone graben along the linear crest of the uplift (Macdonald and Eaton, 1964; Fiske and Koyanagi, 1968, fig. 8; Jackson and others, 1975; Swanson and others, 1976). The leveling data on the flanks were obtained over periods of several days during more extensive surveys. All available information indicates that no deforma- tion occurred during the periods of leveling. The level- ing data were evaluated and tabulated by Okamura and Swanson (1975) and Karren (1959). In particular, Okamura and Swanson (1975) recomputed data from a 1921 survey in order to correct for errors in rod length and instrument collimation by somewhat more accu— rate methods than previously used (Wilson, 1935). CENTRAL PART OF SOUTH FLANK Leveling surveys were completed across the central part of the south flank in 1921, 1958, 1965, and 1971 (fig. 11). The estimated qualities of the surveys are: 1921, poor third order; 1958, good second order; and 1965 and 1971, poor second order or good third order (see US. Geological Survey, 1966, for definitions of orders). Bench mark (BM) 10 (inset, fig. 11A) is the only sta- tion common to all four surveys and is taken as the local datum point for comparison purposes. However, it is located in the area of deformation and may itself have undergone some vertical displacement. In fact, surveys suggest that BM 10 was uplifted about 17 cm relative to a tide gage at Hilo between 1921 and 1958 and 9 cm more between 1958 and 1971 (Okamura and Swanson, 1975, table 3). Although the leveling data, particularly in 1921, are of relatively poor quality, it nonetheless seems safe to conclude that BM 10 has been progressively uplifted at least several cen- timetres relative to the tide gage since 1921. Con- sequently, all displacements given in figure 11 are minimum values and would probably be several cen— timetres larger if computed relative to the tide gage. The Hilo tide gage is sinking at a rate of 4.1 mm/yr relative to sea level, but this is considered as reflecting islandwide isostatic, rather than local, subsidence (Moore, 1970). Comparison of elevations for the 1921, 1958, and 19 1971 surveys, which followed the same route but in— cluded more bench marks in the later surveys, indi- cates uplift and seaward tilting of the south flank, in- creasing in magnitude toward the east rift zone (fig. 11A). The amount of uplift near the east rift zone is large, amounting to at least 2 m at BM 2728 since 1921; most of this displacement, more than 1.65 m, took place since 1958. The maximum recorded uplift since 1958 is 2.1 In at BM YY35, near the south edge of the east rift zone; this bench mark was not occupied in 1921. Most bench marks were displaced much more between 1958 and 1971 than between 1921 and 1958, indicating an accelerated rate of uplift in recent years, during which volcanic activity along the east rift zone has been intense (table 1). The rate of horizontal de- formation likewise has increased since 1958 (compare figs. 5E and 5B). The apparent subsidence of BM 2503 for the 1921—58 and 1921-71 periods is anomalous (fig. 11A) but could be removed if the 1921 elevation were about 0.9 m too high. A blunder of this magnitude (about 1 yd) would have been easy to make if the leveling rod were mis- read, as yard rods were used, but should have been caught in field computations. We rechecked the 1921 field notes but found no error, so that, if our sus- picion is correct, an entry error must also have been made. The 1921 elevation of BM 2728 would also have to be lowered by the amount, if any, by which BM 2503 is in error. An alternative interpretation is that the 1921 eleva- tions are good and the displacements are those indi- cated in figure 11A. In this regard, the subsidence of BM YY37 (2302) between 1958 and 1971 should be noted. This area may actually experience local sub- sidence at times, perhaps in some way related to the rather abrupt change in slope 5 km southeast of Makaopuhi Crater (inset, fig. 11A), which may reflect an unrecognized fault in the Hilina fault system. Re- gardless of which interpretation is correct, the overall increasing uplift toward the rift zones is still the dom- inant pattern of displacement. The 1965 leveling, and part of the 1971 leveling, followed a route (inset, fig. 113) slightly south and west of the 1921 and 1958 routes, and none of the older bench marks except BM 10 and BM YY35 was oc— cupied. The 1965—71 displacement profile (fig. 113) in- dicates uplift much like that shown by other profiles (fig. 11A) but at an accelerated rate. For example, BM YY35 was uplifted about 1.6 In between 1965 and 1971 but only 0.5 between 1958 and 1965, and BM HVO 53 was uplifted about 0.75 In between 1965 and 1971 but only about 0.4 In between 1958 and 1965 (extrapolat- ing from BM 2503, very close to BM HVO 53). Volcanic activity along the upper east rift zone has been espe- cially intense since 1961 (table 1). The displacement profiles of figure 11 (except the questionable 1921—58 profile) mimic in a general way 20 DISPLACEMENT OF THE SOUTH 11‘ 7/ // " Mk h'Ct _ 2,0_ 18‘ X Egg/a/mpul raer 1000 “9 2728 19— A - 2 1'8_YY35 ml]— $1.6— '— ”41.5— E 21.4— 51.3— 51.2— n: O”— 51.0— 2 ,9— LE .1 '8— : .7— 5 ._ I.IJ E 5— 8 .4— < —| .3— n. 9 ,2— D .1 .1— < Sea E 0— _0level I D: -.1— \ I Lu \/ > -,2— \ —100 I __ \I .3 D A -.4 200 lllllllllllllllll 012345678910111213141516 FLANK OF KILAUEA VOLCANO 2.1 — 2.0 _ YY35 Makaopuhl Crater _ 1000 1.9 — 1.8 — — 900 a, 1.7 — u.I o: 1.6 — 80° [.3 1.5 ~— 2 35 700 E 1.4 — 0‘11 __ C; 1.3 — ’ ,_ E 1.2 — —— 600 m U) u) _ Lu 2 1.1 g E L“ 1.0 — — 500 ,_ DJ > LLI E l: .9 — E g 3 _8 _ —— 400 Z . .u z‘ E I '7 — <2 '2 E _5 _. —~ 300 1:: 3‘ L“ > LLI E ‘5 _ Lu .1 g 4 _ — 200 ‘7'” u.| . < _| _ o. .3 ‘13 .2 _ — 100 Q _l .1 — 6 bench marks HVO 5 >1 km off profile 78 Sea ; 0 — level 0: -.1 — u.I > .,2 — — 100 “3 _ B 200 -.4 | | l | | l l l l | l l 0 1 2 3 4 5 6 7 8 9 1O 11 12 DISTANCE FROM BM YY35, IN KILOMETRES FIGURE 11.—Vertical displacement and topographic profiles across central part of south flank between 1921 and 1971. Datum point is BM 10. A, Leveling surveys of 1921, 1958, and 1971, along Kalapana Trail. B, Leveling surveys of 1965 and 1971 along Chain of Craters Road. Steep parts of topographic profiles are flow-mantled fault scarps, such as Holei Pali. Bench mark YY35 was destroyed in 1969, but its 1971 elevation can be estimated to within 0.1 In by comparison with a nearby Okamura and Swanson (1975, tables 5 and 7). Inset maps show location of bench marks and leveling ro point. All data are from utes in A and B and line of topographic profile in B. See figure 2 for location of Makaopuhi Crater. Contour interval in inset maps is 75 metres. the shape of the topographic profiles. Both leveling routes cross the Hilina fault system, and both the topo- graphic and displacement profiles show abrupt changes in slope across the fault system, suggesting recent fault movement. This effect is especially notice- able for the 1965—71 profile (fig. 113), which is nearly perpendicular to the trend of the fault scarp. The na- ture of another abrupt change in the 1965—71 profile, between HVO 56 and C/L 21, is not known because lava flows covered the area in 1970—71, prohibiting reoccupation of intermediate bench marks and obscur- ing evidence of possible surface rupture. EASTERN PART OF SOUTH FLANK AND LOWER EAST RIFT ZONE The 1921 and 1958 leveling surveys extended north of BM 10, across the lower east rift zone and the north flank (fig. 12). For comparison purposes, BM 359.3, in Keaau, is used as the datum point. Leveling from the tide gage established in 1926 at Hilo indicates that BM 359.3 has been stable within survey error relative to the gage (Okamura and Swanson, 1975, table 3) since 1926; we cannot document such stability between 1921 and 1926 but think it reasonable. The poor quality of the 1921 leveling must be considered in any quantita- tive interpretation of the displacement data. The 1921—58 profile indicates uplift of the south flank and lower east rift zone, with a maximum dis- placement of nearly 25 cm near the topographic crest of the rift zone. The uplift is asymmetric, the area south of the crest showing substantially more displacement than that to the north. BM 10, the datum point used in figure 11, was uplifted about 17 cm. Lower east rift eruptive and intrusive events be- tween 1921 and 1958 include the 1955 eruption (Mac- donald and Eaton, 1964), the fissure zone of which is crossed by the leveling profile, and the 1924 ground- cracking episode centered 10—15 km east of the level- ing route (Finch, 1925). The section between BM 10 and BM 655 was re- leveled in 1973 (Hawaiian Volcano Observatory, un- pub. data). Assuming BM 655 did not change elevation VERTICAL DISPLACEMENT OF SOUTH AND NORTH FLANKS OF KILAUEA 0 4 8 12 16 20 24 0_7 | | I I I I EXPLANATION 0 Bench mark 0.6 r 737 0.5 — 0.4 1 ~ 400 0.3 — m E: — 3005-, 2 0.2 ~ 2 _ 200 —. Z 9 0.1 — I- v 100‘; LU _I Lu VERTICAL DISPLACEMENT RELATIVE TO BM 359.3, IN METRES 497 I I 0 4 8 12 16 20 24 '0-1 f I I I I PROJECTED DISTANCE FROM BM 359.3, IN KILOMETRES FIGURE 12.—Vertical displacement and topographic profiles across lower east rift zone and eastern part of south flank between 1921, 1958, and 1973. Datum is 1958 elevation of BM 359.3. Inset shows location of bench marks, leveling route (dashed line), line of topo- graphic profile (light solid line), and generalized eruptive fissures for 1955 eruption (heavier solid lines). Displacement data are pro- jected onto line of topographic profile. Contour interval on inset map is 100 m. between 1958 and 1973, displacements relative to it can be used to indicate the total displacement since 1921. This assumption is dubious but serves to show that a striking change in the pattern of uplift took place (fig. 12). The crest of the uplift shifted southward, with BM 737 being displaced nearly three times as much as BM 1065 between 1958 and 1973. Thus the overall uplift since 1921 is centered 2 to 3 km south of the topographic crest of the rift zone, in an area where no past eruptive activity is known. The 1958—73 uplift can be examined more closely by using data from surveys conducted along the same route in 1964 and 1969 by R. W. Decker (1965, 1969; Dieterich and Decker, 1975) (fig. 13A). The uplift grew sporadically between 1958 and 1973. The rate of uplift was greatest between 1964 and 1969 (maximum of 3.2 cm/yr) and least between 1969 and 1973 (maximum of 1 cm/yr). The crest of the uplift did not shift laterally during this time. A grabenlike zone developed on the north limb of the uplift, progressively deepening with time; new cracks cutting a paved road apparently define the south side of the graben (R. W. Decker, oral c0mmun., 1969). Both the crest of .the uplift and the graben are located south of all known eruptive fissures 21 and vents. BM YY71, only 100 m from a 1955 eruptive fissure (vent R of Macdonald and Eaton, 1964), sub- sided, possibly owing to thermal contraction during so- lidification of the 1955 dike and cooling of its wallrock. In contrast to the profile in figure 13A, displacement profiles about 8 km farther east show progressive sub- sidence relative to BM YY80 since 1958 (fig. 133). The subsidence is of much smaller magnitude than the up- lift, amounting to a maximum of about 12 cm at BM YY195. The rate of subsidence was most rapid between 1958 and 1964 and subsequently slowed. Maximum subsidence is centered along the 1955 fissures and may reflect thermal contraction. Removal of this effect by smoothing the profiles leaves a broad basin with a small but sharp uplift centered at BM YY198; this up- lift persists on all three profiles and therefore must be real although of small amplitude. WESTERN PART OF SOUTH FLANK AND KOAE FAULT SYSTEM Leveling surveys in 1958 and 1971 permit construc- tion of a displacement profile across the Koae fault system and western part of the south flank (fig. 14). The profile is complicated by displacements caused by several faulting events in the Koae fault system, prin- cipally in 1965 (Fiske and Koyanagi, 1968). The south flank shows uplift and southward tilting, increasing in magnitude toward the Koae fault system. NORTH FLANK The north flank of Kilauea and adjacent part of Mauna Loa between Hilo and the summit were leveled three times between 1926 and 1971. Okamura and Swanson (1975, table 3) tabulated the measured eleva— tions and showed that changes are generally unsys- tematic and close to or within survey error. Changes are significant only near the summit, presumably as a result of the documented summit tumescence (Oka- mura and Swanson, 1975, table 3). Displacement on the eastern part of the north flank between 1921 and 1958 was similarly small (fig. 12). SUMMARY AND INTERPRETATION OF VERTICAL DEFORMATION ON THE NORTH AND SOUTH FLANKS The following statements characterize vertical de— formation of the flanks of Kilauea during the 20th cen— tury: (1) The north flank is virtually stable, as also indi- cated by both seismic and horizontal displacement data. (2) In most places, the south flank has been increas— ingly uplifted toward the rift zones and Koae fault sys- tem, with the maximum displacement within or just south of the east rift zone. DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO 400 l I I l I l ELEVATION, IN METRES EXPLANATION 0 Bench mark VERTICAL DISPLACEMENT, IN METRES _ 05 I I I l I I 200 I I 100 ELEVATION, IN METR ES EXPLANATION c\ YY174 Bench mark VERTICAL DISPLACEMENT, IN METR ES _.15 I I l I 2 4 6 8 10 12 PROJECTED DISTANCE FROM BM YY80, IN KILOMETRES FIGURE 13.—Vertical displacement and topographic profiles across lower east rift zone of Kilauea between 1958 and 1973. A, Profiles between BM YY80 (Pahoa) and BM 10 (Kalapana); B, Profiles between BM YY80 and BM YY174 (Pohoiki). Datum is 1958 elevation of BM YY80. Inset maps Show locations of leveling route, key bench marks, and line of topographic profile. Displacement data are projected onto line of topographic profile. Contour interval of inset maps is 150 In. RELATIVE MAGNITUDE OF HORIZONTAL AND VERTICAL DISPLACEMENTS 23 Kalanokuaiki Pali Koae f'ault system East rift zone South flank — 1100 A ——1 — 1000 — 900 — 800 YY139 - 700 ' 600 VERTICAL DISPLACEMENT RELATIVE TO HILO TIDE GAGE, IN METRES ELEVATION, IN METRES Displacement profile — 500 — 400 -2 .2 I I I I I ' I I 300 0 2 4 6 8 10 12 14 DISTANCE FROM BM YY26, IN KILOMETRES EXPLANATION 0 Bench mark __1_ Fault Bar and ball on downthrown side FIGURE 14.—Vertica1 displacement and topographic profiles across Koae fault system and western part of south flank between 1958 and 1971. Datum point is Hilo tide gage. Inset map shows location of bench marks along Hilina Pali Road and Kalanaokuaiki Pali, a normal fault that forms south boundary of Koae fault system. Contour interval is 60 In. Hilina Pali is a prominent fault-line scarp. (3) The rate and amount of displacement in areas of large uplift correlate directly with the number and proximity of rift eruptions or related faulting events for any given time period. The uplift of the south flank can be ascribed to the effects of forceful intrusion of magma into the rift zones. Such uplift was suggested by Macdonald (1956, p. 286) on the basis of geologic arguments. Along the upper and middle east rift zone, intrusion has appar- ently been followed quickly by eruption, so that a di- rect relation between the timing of uplift and eruption is evident. Along the lower east rift zone, intrusion appears to have proceeded more slowly or over longer periods of time, or both; consequently, uplift during noneruptive periods may occur (fig. 13A). Subsidence, such as that shown in figure 133, may indicate migra- tion of magma to another site. The displacement profiles that completely cross the lower east rift zone (figs. 12 and 13) are particularly instructive because of their asymmetry. The profiles suggest that either the south flank is virtually de- coupled from the north flank along the rift zone or that the uplift is the result of intrusion of magma as south- dipping dikes into the rift zone (Dieterich and Decker, 1975). South-dipping dikes would explain the location of maximum uplift 1—2 km south of the lower rift zone (fig. 13) but would not account for the location of maximum uplift within the central and western parts of the rift zone, where vertical or nearly vertical dikes seem most likely. We examine the possibility that dikes dip southward in the section. “A Structural Model for the South and North Flanks of Kilauea,” concluding that some dikes may dip southward at com- paratively low angles where gravity faults of the Hilina fault system impinge on the rift zone, but that most dikes along the length of the rift zones are steeply dipping to vertical. The profile across the Koae fault system and adja- cent south flank (fig. 14) is more complicated than the others. Fiske and Koyanagi (1968, p. 19) suggested that the uplift immediately south of the Koae system during the December 1965 eruptive and faulting episode could be attributed to “* * * elastic rebound due to the sudden release of accumulated stresses along Kalanaokuaiki Pali,” the fault that forms the southern margin of the fault system. Another possibil— ity is that magma was laterally intruded as a dike into the Koae from the site of concurrent eruption near the intersection of the fault system and east rift zone, caus- ing uplift similar to that elsewhere along the northern edge of the south flank. In May 1973, such a dike ap- parently was intruded several kilometres into the Koae system from its source along the upper east rift zone (Duffield, 1975; Koyanagi and others, 1973). RELATIVE MAGNITUDE OF HORIZONTAL AND VERTICAL DISPLACEMENTS The relative magnitude of horizontal and vertical deformation can be estimated by comparing measured horizontal displacements with measured or extrapo- lated vertical displacements for approximately the same time intervals at six south-flank triangulation stations (table 3). The horizontal component of dis- placement is at least several times larger than the ver- tical component. This generalization appears valid across most of the south flank, judging from compari- sons of figures 11—14 with figures 5—8. The relative importance of vertical displacement probably increases near the rift zone, but no triangulation stations are located in the critical area to test this suggestion. Within the rift zones themselves, vertical displace— ments (both uplift and subsidence) are commonly large 24 TABLE 3,—Comparison of horizontal and vertical displacements at six triangulation stations on the south flank of Kilauea Volcano [Vertical data have been extrapolated from nearest bench mark ifthe triangulation station was not included in the leveling survey.] Time interval Displacement, m Ratio of Station horizontal to Horizontal Vertical Horizontal Verticall vertical 1961470 195&71 08 +002 240 1914—58 1921758 2.1 + .30 a7 1958~70 195&71 2.3 + ,30 8 Kupapau ,,,,,,,,,, 191¥58 1921758 1.2 + .18 37 195&70 195&71 0.8 + ,07 11 Hakuma ,,,,,,,,,, 191L58 1921758 1.4 + ,17 38 195&70 195&71 0.7 + .09 8 Laeapuki ,,,,,,,,,, 1961770 196i71 1.3 + .08 316 Kaena Point ,,,,,,,, 1961770 1965»71 1.3 + ,05 326 1Vertical displacement relative to Hilo tide gage or, for the 1921 survey, BM 359.3. The 1965 survey was based on BM 10 and has been adjusted by assigning half the uplift of BM 10 between 1958 and 1971 to the 1965—71 interval. 2Probably minimum ratio, because time between horizontal surveys is shorter than that between vertical surveys, aProbably maximum ratio, because time between horizontal surveys is longer than that between vertical surveys. but smaller than horizontal ones at a given location (Macdonald and Eaton, 1964, fig. 34; Jackson and others, 1975; Swanson and others, 1976), although ex- ceptions associated with the formation of major gra- bens are known (Jaggar, 1924; Swanson and others, 1972, p. 114). The predominance of horizontal displacement on the south flank probably results from dikes that are force- fully wedged into the rift zones, causing horizontally- directed dilation that manifests itself by the opening of gaping fissures and the corresponding displacement of the south flank away from the rift zones. A STRUCTURAL MODEL FOR THE SOUTH AND NORTH FLANKS OF KILAUEA The nature and timing of seismic events and ground displacements prompt us to propose a model in which the south flank of Kilauea moves in response to force- ful intrusion of magma into the rift zones (fig. 15). In- trusions, in the form of dikes, literally shoulder the south flank upward and southward and are effectively tearing the volcano apart along the two rift zones. From the timing of seismic and deformation events, we conclude that intrusion is the immediate cause of south flank displacement, not the reverse. The data further suggest that the north flank is rela- tively stable and does not respond notably to intrusive events. This lack of response may be explained by the following two interrelated reasons: (1) the north flank is buttressed by the huge mass of Mauna Lea and the other volcanoes on the island, whereas the south flank is free to move southward; and (2) the development of the rift zones of Kilauea is dominated by the gravita- tional stress field imposed by Mauna Loa, on which Kilauea is constructed, and this gravitational field favors seaward displacement away from the center of the volcanic pile (Fiske and Jackson, 1972). DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO Figure 16 illustrates the present great difference in size between Kilauea and the rest of the island, par- ticularly Mauna Loa. It is reasonable that this large mass of older rock might buttress Kilauea, especially since intrusion in the rift zones is a shallow process, probably taking place within 8 km, chiefly 5 km, of the ground surface (fig. 15). The buttress effect would be important even if the contact of Mauna Loa and Kilauea were vertical and extends to the old sea floor on which the volcanoes are constructed. In all probabil- ity, however, the contact interfingers as shown schematically in figures 15 and 16. In fact, we consider it probable that Mauna Loa was already a large vol- cano when Kilauea first began forming, because of the present large difference in size. Koae fa n in South sysiem Caldera North LU : n: A A ,_ LU 2 Sea 9 level 5‘ + + z 5 + + + + + + + «- — t t .. 2‘ E .10 Pre-Mauna Loa oceanic crust < i Mantle fl -15 Magma conduit East rift zone r~%\ Active Inactive part part (I) LLI I: i— Lu 2 O :‘ :1 Z 2‘ 9 I— <( > Lu _i LLI u: in u: i— LU E O :‘ x E z~ 9 i— i Lu Mantle _l “J 0 5 10 15 MILES 0 5 10 15 20 KILOMETRES EXPLANATION +l+ + 1 _ + Kilauea Mauna Loa r > I 44 v Kilauea Mauna Loa Clastlc rocks Subaerial shield Pillow lava (— Direction and relative amount of displace- ment caused by forceful intrusion of magma as dikes. Length proportional to amount of displacement A STRUCTURAL MODEL FOR THE SOUTH AND NORTH FLANKS OF KILAUEA Not so easily visualized, but perhaps of greater im- portance, is the control exerted by Mauna Loa on the stress field in which intrusions at Kilauea take place. Fiske and Jackson (1972) demonstrated that gravita- tional stresses govern the direction of dike propagation in gelatin models free of other stresses, whether the dikes are injected directly beneath the top of the model or at some place on its flanks. If this modeling can be generalized to naturally occurring volcanic edifices, then a younger shield built on the sloping flank of an older shield should come under the influence of the latter’s gravitational stress field, which should in turn control dike orientation and overall shape. Fiske and Jackson (1972, p. 314—317) showed how this concept could explain the relation of Kilauea to Mauna Loa. Assuming that Kilauea inherited the gravitational stress field of Mauna Loa, intrusion of magma into the rift zones as dikes should produce dila— tion in a direction consistent with this stress field, that is away from the central part of the edifice and toward the free slope. Thus, the Fiske-Jackson model predicts that the south flank of Kilauea would be far more mobile than its north flank, a relation indicated by the seismic and geodetic evidence. Much of the east rift zone of Kilauea probably ex- tends beyond the effects of Mauna Loa, however. The submarine part of the rift zone extends at least 110 km beyond the east tip of the island (Macdonald and Ab- bott, 1970, p. 313). The submarine rift zone forms a prominent ridge standing high above its base (Moore, 1971; fig. 1), and extends well beyond the bulk of Mauna Loa; consequently, it may have its own high- FIGURE 15.—Diagrammatic cross sections, with no vertical exagger- ation, through Kilauea depicting our interpretation of relation be- tween east rift zone and rest of volcano. Locations of sections shown in figure 1. A—A ’, Summit region. Magma is intruded, primarily as dikes, into east rift zone from upper part of magma conduit or reservoir complex. Magma only rarely enters Koae fault system. B—B’, Middle east rift zone. Magma intruded as dikes from beneath summit region wedges south flank seaward and upward. North flank moves little if at all. Dikes in inactive part of rift zone are older, extend to greater depths, and terminate upward at slightly lower elevations than dikes in active part. C-C’, Sub— marine part of east rift zone. Displacements caused by intrusion of magma are assumed to be nearly symmetric across rift zone. In A—A’ and B—B ’, gravity sliding is interpreted to take place chiefly in layer of hyaloclastic rocks above and interbedded with pillow lava, displacing seaward part of south flank downward. Hyaloclas- tic material is generated at shallow depth but sloughed downward by slides and current action (Moore and Fiske, 1969), possibly forming a deposit much thicker than indicated in A—A’ and B—B ’. Depths to base of volcanic pile and to mantle from Hill (1969), and bathymetry and rock type from Moore and Fiske (1969). Depths to contact of Kilauea and Mauna Loa are speculative; rocks from the two volcanoes are shown as interfingering only at relatively shal- low depths, consistent with our belief that Kilauea did not begin to form until Mauna Loa was already a large edifice. Flows of pillow lava from Mauna Kea may underlie Mauna Loa pile (fig. 16). 25 level gravitational stress field. If so, dike-induced dila- tion should be directed equally northward and south- ward as C—C’ in figure 15 portrays. The most recent intrusion into the submarine part of the rift zone may have taken place in 1924 (Finch, 1925), and earth- quake activity in this area has been minor since then (Koyanagi and others, 1972, and references therein). Thus it is unlikely that this part of the rift zone is presently undergoing much deformation. The north side of the submarine part of the rift zone is steeper than the south side (Moore, 1971), possibly because of gravity-induced slides, although why slides are con- centrated on only one side is not evident. The south flank is assumed to be mobile everywhere south and southeast of the rift zones, but the depth to which it remains mobile is not easily determined. The flank is probably mobile to a depth of at least 5 km because seismic evidence during eruptions suggests that some intrusion takes place at such a depth. Al- most all earthquakes on the south flank have focal depths of less than 15 km, and most have depths of 8 km or less, calculated using a crustal model devised by J. P. Eaton and E. T. Endo, based on Hill’s (1969) re- fraction study (Koyanagi and others, 1975). The mobil- ity may therefore extend to a depth of at least 8 km, which is almost certainly well within the edifice of Mauna Loa and near the level of the old sea floor on which Mauna Loa is built (fig. 15; Hill, 1969). Seismic evidence suggests that the lithologic break at shallow depths between subaerial flows capping the volcano and pillow lavas forming the submarine bulk of the volcano (fig. 15; Moore and Fiske, 1969) does not influ— ence the overall mobility of the south flank, although the Hilina fault system may largely bottom out at this level, as discussed in a later section. It seems unlikely that the base of the mobile flank would be defined by a single plane of dislocation. In— stead, we favor a model in which the displacement gradually decreases below the depth at which intru- sion of magma is concentrated. Thus the base of the mobile south flank may actually be a zone several kilometres thick, and much of the displacement may be taken up by many local adjustments within the pillow complex. Moore and Krivoy (1964) believed the east rift zone to be a southward—dipping structure forming the sole of a large landslide block (the south flank). They attrib- uted dilation of the rift zone to “* * *southward move- ment of the flank of the volcano south of the rift zone under the influence of gravity and inflation of the dip- ping rift zone by magma from the summit reservoir” (1964, p. 2043). However, the documented uplift of the south flank associated with rift dilation and eruption during recent years argues against landslide-type faulting along the rift zone, for subsidence would be 26 DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO Sea level __l:g ELEVATION, IN Kl LOMETR ES oceanic Mauna Loa Mauna Loa . Hualalai Mauna Kea Kohala Volcanic Pile crust Mauna Kea 3 0 Sea level -3 ...... -6 """""" Volcanic .9 Original oceanic crust 3 C 0 Sea level Kilauea Mauna Loa *3 ...... .6 _9 Original Oceanic 0 6 12 O 6 12 18 FIGURE 16.—Diagrammatic cross sections, crust 18 24 MILES 24 KILOMETHES with no vertical exaggeration, across Kilauea, Mauna Loa, Hualalai, and part of Mauna Kea. Depth to base of volcanic pile is from Hill (1969); submarine topography off south flank of Kilauea from Moore and Fiske (1969). We as- sume that each volcano grew to substantial size before its next younger neighbor began erupting, as indicated by interfingering only at relatively shallow depth. Other volcanoes that never reached above sea level may contribute to volcanic pile. Contour inverval in index map 600 m. expected, at least on the south side of the rift zone, from gravity faulting during eruption. In addition, seismic studies show that earthquakes associated with intrusion emanate from directly below the surface trace of the rift zone, not south of the zone (Koyanagi and others, 1972). Finally, intrusion and eruption along the east rift zone precede seismic activity and displacement on the south flank, suggesting that force- ful intrusion is the immediate cause, not the effect, of the rift dilation and displacement. We agree with Moore and Krivoy (1964) that gravity faults characterize the Hilina fault system (see later section) but believe that these are only relatively superficial, secondary structures resulting from in- stabilities generated by injection of steeply dipping dikes into the rift zones. A broad unfaulted area lies between the rift zone and the Hilina system, except locally along the lower east rift zone, and this alone suggests different origins for the rift zones and Hilina fault system. Gravity plays an important role in our model, just as it does in the model of Moore and Krivoy. Were re- gional tectonic stresses dominant over local gravita- tional stresses, the rift and flank structures would doubtless be much different, as Fiske and Jackson (1972) found them to be in isolated Hawaiian shields. However, from the observed timing of events we con- tend that intrusion, not gravity-inducing sliding, opens the rift zones. Ryall and Bennett (1968) interpreted the east rift zone to be a north—dipping structure related to a large normal fault that displaces the crust and upper man- tle. This interpretation was largely based on gravity data in the east rift area, which, as we show in a later section, can better be interpreted as indicating south- ward growth of the east rift zone with time. Their model presents a host of other difficulties not easily rationalized with seismic, geodetic, and geologic data largely gathered subsequent to their interpretation. Dieterich and Decker (1975) suggested that dikes in the lower east rift zone dip 45° 8., on the basis of com- parison of the 1958—64 leveling profile in figure 13A with finite-element models of dike emplacement into an elastic medium. The critical feature of this and other leveling profiles along the same route that suggests such a low dip to these workers is its marked asymmetry, the south flank showing substantial uplift and the north flank almost no uplift. In their interpre- tation, the graben just north of the crest of the uplift formed at the locus of maximum stress, as located by the model. Dieterich and Decker’s (1975) interpretation con- flicts with the magnitude of horizontal displacements in that area. For example, the horizontal displace- ments at Iilewa and BM YY66 between 1958 and 1970 are each about 50 cm (fig. 58), whereas the maximum displacement predicted from the finite-element model for about the same time period (1958—69) is only 5 cm, an order of magnitude less. The large horizontal dis- placements do not fit any of the several models given THE RIFT ZONES AND KOAE FAULT SYSTEM AS ZONES OF DILATION by Dieterich and Decker (1975). Instead, the large ex- tension may reflect the intrusion of a family of dikes, not just one dike. In his helpful review of this paper, Decker (oral commun, 1973) suggested that the hori- zontal displacements resulted from ground rupture, a process not covered by the finite-element model. This may be true, but no new ground cracks were observed in the critical area along the rift zone during this time, and seismic activity in this area, especially since 1961, has been so slight as to suggest no surface rupturing. Another problem with the interpretation by Dieterich and Decker is that the most definitive data, those north of the rift zone, lie 3 to 4 km off the profile used in their model comparison, whereas most of the other data are much closer to the profile. Whether the data can be projected so far and still be subject to quan- titative interpretation is open to question, particularly in View of the observed rapid change from uplift to subsidence within a short distance along the rift zone (fig. 13). In the absence of data for the north flank closer to the profile, the displacement profile can be interpreted to indicate almost any dip desired. As a result of these difficulties, we regard as un- proven the existence of dikes in the lower east rift zone that dip much less than 90°. Nonetheless, we believe it is possible that some dikes in the area crossed by the leveling line in figure 13A do dip southward at angles of 50—70° because of the structural setting of this part of the lower east rift zone. In this area, the Hilina fault system intersects the rift zone (figs. 2 and 3), and magma supplied from steeply dipping dikes farther up— rift could conceivably be intruded along some of the preexisting south-dipping fault planes. If such intru- sion has taken place, we would view it as a perturba- tion from the otherwise steep dips of the dikes indi- cated elsewhere along the rift zone by the patterns of seismicity and uplift, and found on eroded Hawaiian shields (Stearns, 1947; Stearns and Vaksvik, 1935). NUMERICAL COMPARISON OF VOLUMES OF INTRUSION AND DEFORMATION If south flank deformation is caused by injection of dikes, then there should be a specific relation between the volume of deformation (defined principally by the amount of uplift) and the volume of intrusion. This relation will not be one-to-one, as the deformed rocks should decrease in internal volume to some degree. Still, a rough balance between the volume of intrusion and the volume of deformation should be maintained. For the period 1958—71, a cross section of the south flank southeast of Makaopuhi Crater shows an in- crease in volume of about 107m3/km owing to uplift, 27 computed from data in figure 11A assuming negligible deformation of the north flank. A dike 0.5 m Wide, a width consistent with many observed Hawaiian dikes, has a volume of 5 X 105m3/km if it extends from the surface to a depth of 1 km, 1.5 X 106m3/km to a 3-km depth, and 2.5 X 105m3/km to a 5-km depth. The calcu- lated volume of deformation could thus be balanced by 4 to 20 such dikes. Table 1 indicates that 8 to 13 intru- sive events occurred near Makaopuhi during this time; the uncertainty in number reflects the unknown de- gree to which events near Alae Crater and Mauna Ulu could have influenced the deformation. At least one of these intrusive events, in August 1968, involved very shallow magma injection—possibly only to a depth of 500 m (Jackson and others, 1975)—and seismic evi— dence for others suggests that the dikes bottom at depths of 5 km or less. Thus, a reasonable agreement between the volumes of deformation and intrusion is evident. Similar calculations for the 1965—71 period (fig. 118) show a volume of deformation of 8X106m3/km, which could be balanced by 3 to 16 dikes. During this time there were five to eight intrusive events. Again, the two volumes are in reasonable agreement. Quite obviously these calculations are crude, for the existing data and theoretical models are inadequate for detailed analysis. Nonetheless, the approximate balance between the volumes of intrusion and defor- mation is of some value, as our model predicts that the dikes make room for themselves by displacing the wallrock up and to the side. THE RIFT ZONES AND KOAE FAULT SYSTEM AS ZONES OF DILATION If the south flank of Kilauea is being displaced southeastward because of the forceful intrusion of magma as dikes into the rift zones, its contact with the less mobile part of the volcano should be marked by a zone of dilational opening. We postulate that such a zone of dilation is defined by the two rift zones and the Koae fault system. The east and southwest rift zones have long been recognized as containing normal faults, grabens, and gaping cracks with matching opposite walls, and in recent years such features have been observed and measured in the process of formation. Fissures that form during any given rift eruption tend to be ar— ranged in an en echelon pattern, with either a right- or left-hand sense of offset depending on their location along the length of the rift zone (Macdonald and Eaton, 1964; Moore and Krivoy, 1964; Duffield and Naka- mura, 1973; Jackson and others, 1975). Analysis of 28 these en echelon fissures by the method of Nakamura (1970) and Duffield and N akamura (1973) suggests di- lational opening perpendicular to the rift zones. How- ever, new lava erupted along the rift zones repeatedly covers the fissures, so that the net amount of dilation over long periods of time cannot be estimated adequately. Thus we must look elsewhere for pertinent quantitative information regarding the amount of cumulative dilation. The Koae fault system (fig. 3) is not the site of erup- tions, except for small, infrequent outbreaks near its intersections with the east and southwest rift zones. Further, it has, in the recent geologic past, seldom been covered by lava erupted from vents farther up- slope, as the geologic maps of Peterson (1967) and Walker (1969) show. As a result, the Koae fault system serves as a window through which the net effect of displacements produced by many episodes of deforma- tion are well displayed, and it is the best place to find direct structural evidence that displacements such as those measured in this century have characterized the recent geologic past. Duffield (1975) found that the Koae fault system is characterized by gaping cracks arranged in en echelon fashion, some of which coalesce to form long, sinuous zones of normal faults, generally with north-facing scarps, that define the south margins of asymmetric grabens (fig. 17). Symmetric grabens are uncommon. The fault scarps are as high as 20 In, although most are less than 5 m, and single cracks are as much as 2 m wide. Most cracks and faults dip vertically, strike N. 75° E., and open perpendicular to their strike, as deter- mined by numerous measurements of matching crack walls. A similar direction of opening was determined by Duffield and Nakamura (1973) from a study of the pattern of en echelon cracks, and this direction virtu- ally parallels that of horizontal displacement vectors on the south flank (fig. 5). This evidence suggests rather definitely that similar intrusive events cause both the south-flank displacement and the Koae (and rift zone) dilation. Knowledge of the amount of dilation recorded by the Koae fault system and the time during which this dila- tion occurred is helpful in evaluating the history of past displacement events. The total amount of dilation along two profiles across the fault system was meas- ured to be 18.69 In (fig. 17,A—A’) and 32.55 In (fig. 17, B—B’) (Duffield, 1975). Our field observations suggest that the total dilation decreases westward from profile A—A’. We interpret the greater amount of opening in the eastern part of the fault system to reflect proximity to the principal source of dilation—intrusion within the east rift zone. The measured dilation is that accumulated since the DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO Kilauea Caldera E457 Q RIFT ZONE Puhimau _ Crater Puu Kane NuuoHamo g0 Hiiaka Huluhulu O ‘9 Crater [1’ f/ / Mauna Ulu _/ EXPLANATION Pitcrater O 1 2 3 4 5 MILES Eli} l*—v—L“I—i*v—Ly—‘—*j Cone 0 1 2 3 4 SKILOMETRES __1_ Fault Bar and ball on down thrown side FIGURE 17 .—Principal zones of normal faults in Koae fault system. Note that most fault scarps face north. A—A’ and B—B’, traverses along which total dilation was measured. Modified from Duffield (1975). Kalanaokuaiki Pali is southernmost fault scarp in Koae fault system. last lava flows covered the area. Duffield (1975) broadly estimated this time to have been 500—2,500 years ago from evidence involving, respectively, ex- trapolation of 20th century rates of displacement on the south flank and the interpreted age of the summit caldera (Powers, 1948; Rubin and Suess, 1956), in which lava erupted at the summit tends to pool rather than flooding areas farther downslope. Hawaiian legends concerning the latest flow in the Koae area suggest that the 500-year age may be more nearly cor- rect (H. A. Powers, written commun., 1974). An an- cestral Koae fault system apparently existed before the latest flows because there is field evidence that these flows poured into cracks along the trace of present faults, particularly Kalanaokuaiki Pali. Thus, the available data suggest that displacements similar to those during this century have occurred over at least the past few hundreds to thousands of years. Fault-bounded blocks in the Koae system appear to have been tilted south-southeastward during deforma- tion, important evidence bearing on the origin of the faults. Such tilting was directly measured over the De- cember 1965 deformation event, when one tiltimeter on a fault block showed more then 200p.rad of seaward tilt (Fiske and Koyanagi, 1968, fig. 3). Unusually steep southward slopes on some fault blocks suggest a sub- stantial cumulative effect of such tilting, as least locally. SOUTHWARD MIGRATION OF THE RIFT ZONES AND SUMMIT RESERVOIR 29 The tilting and general configuration of the fault blocks are remarkably similar to structures obtained in modeling clay by Cloos (1968, fig. 18) during exper- iments that produced asymmetric grabens. Such a gra— ben forms when one side of a clay cake resting on two overlapping metal sheets is pulled laterally while the other side remains stationary. The major fault scarps face toward the stationary block, and the part of the model near the zone of separation sags and tilts in the direction of movement. It is interesting that the direc- tion of tilt is opposite to that which generally accom- panies landsliding, when slopes are rotated toward the stationary block, not away from it. The observed struc- tures in the Koae fault system are just those expected from Cloos’ experiment, if movement of the south flank away from the less mobile part of the volcano is analo— gous to pulling apart the clay cake. Duffield (1975, fig. 5) discussed in more detail the relation of Cloos’ model to structural problems at Kilauea. The interpretation of the Koae fault system as a complex asymmetric graben produced by seaward- directed dilation in the rift zones accords with our postulate that the system is part of the tear-away zone separating the south flank from the rest of the volcano. Intrusion in the rift zones causes dilation, which then propagates into the Koae system as the south flank moves seaward, gradually dying out away from the site of intrusion. In this sense, the Koae is a relatively pas- sive area, reacting to events in the rift zones without generating activity itself. Intrusion within the Koae seldom takes place, presumably because it lacks a di— rect connection to the summit reservoir complex. Erup- tions that do occur are thought to be fed by magma that enters the fault system during rift eruptions near the intersection with the Koae. Such an eruption occurred in 1972, when the migration of earthquake foci into the Koae indicated that a dike was being intruded from the site of initial eruption on the east rift zone (Koyanagi and others, 1973). Such a rift eruption apparently causes dilation, which propagates into the Koae, creat- ing stress conditions favorable for magma to leave the rift zone and intrude the fault system. In summary, the Koae fault system and rift zones together define the zone of dilation that, according to our interpretation, necessarily bounds the north edge of the mobile south flank. RELATIVE IMPORTANCE OF THE TWO RIFT ZONES IN GOVERNING MOBILITY OF THE SOUTH FLANK The horizontal displacements from the September 1971 eruption (fig. SB) indicate that the south flank responds to dike intrusion in the southwest rift zone, just as in the east rift zone. However, the southwest rift zone has apparently been of subordinate importance in governing south-flank structures. It is much shorter, lower, and narrower than the east rift zone (fig. 1, 2) presumably because it contains fewer dikes and thus has fewer eruptions. Furthermore, Duffield (1975) found that both width and total dilation of the Koae fault system decrease westward away from the east rift zone, a pattern suggesting dominance of the east rift zone in the development of the Koae. Duffield (1975) and Duffield and Nakamura (1973) also showed that the direction of dilation across the Koae parallels that across the east rift zone and the south flank but is in most places oblique to the direction of dilation along the southwest rift zone. Additional evidence suggest- ing dominance of the east rift zone is the east-west extent of the Hilina fault system; this system extends south of much of the length of the east rift zone and Koae fault system but apparently ends, or becomes much less conspicuous, near the southwest rift zone (fig. 2; Stearns and Macdonald, 1946, pl. 1). From this evidence we conclude that events in the east rift zone largely govern the mode of deformation of the south flank. SOUTHWARD MIGRATION OF THE RIFT ZONES AND SUMMIT RESERVOIR EAST RIFT ZONE Several lines of evidence suggest that the active part of the east rift zone has migrated southward together with the south flank. The topographic crest of the rift zone is as much as 4 km north of the recently active upper and middle sections of the rift zone (fig. 18A). This crest is probably a constructional ridge built by eruptions along it, by flows from a parasitic shield cen- tered near the east end of Kilauea Iki (fig. 2; see also the topographic map of the Volcano quadrangle), or both. Relics of possible vent areas along the crest are hidden beneath dense jungle cover or younger flows erupted from the parasite shield. No eruptions have taken place along the crest in late prehistoric or his- toric time, suggesting that, if the ridge was built by rift eruptions, the active part of the middle and upper rift zone shifted southward from the crest before this time. The relation of the complete Bouguer gravity anom- aly to the recently active part of the east rift zone likewise suggests southward migration (fig. 183). Kinoshita and others (1963; W. T. Kinoshita, oral commun., 1973) interpreted this positive anomaly as caused by a swarm of dikes of greater bulk density than the vesicular flows or loosely packed pillows that make up most of Kilauea and underlying Mauna Loa. These dikes must be the feeders for lava erupted on the 30 DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO 155° 00' CONTOUR INTERVAL 60 METRES 19 3,0 A 155°15' COMPLETE BOUGU ER GRAVITY ANOMALY CONTOUR INTERVAL 10 MILLIGALS 10 15 MILES I I 15 20 KILOMETRES FIGURE 18,—Maps of Kilauea Volcano. A, Topography B, Complete Bouguer gravity anomalies. Locations of young cones, vent fissures, and pit craters along both rift zones (black areas on both maps) from Stearns and Macdonald (1946) with additions for later eruptions. Main belt of recent volcanic activity lies along southern part of fissure zone. Location of axis of gravity anomaly, taken from B, shown in A by shaded band. Gravity data, in milligals, are from Kinoshita and others (1963), with numerous additions along rift zones by W. T. Kinoshita (unpub. data); an average density of 2.3 g/cm3 was used in gravity reductions. SOUTHWARD MIGRATION OF THE RIFT ZONES AND SUMMIT RESERVOIR 31 rift zone. However, almost all of the recently active vents are along the south side of the gravity anomaly (fig. 183), not along its crest. This suggests that young dikes have been added preferentially to the south side of the swarm. The gravity anomaly on the east rift zone is strongly asymmetric, with steeper gradients south of the rift zone. Thus the dike swarm may be thought of as defin- ing a wedge-shaped prism, with a gentle north face and a steep south face (W. T. Kinoshita, oral commun., 1973). Such a prism is the expected product of southward-directed growth of the rift zone with time, because successively younger dikes would on the aver- age be intruded to higher elevations than older dikes to the north as the rift zone grew southward and upward (fig. 19). This interpretation differs from that of Ryall and Bennett (1968), who used the gravity data of Kinoshita and others (1963) to infer a north—dipping rift zone. The gravity axis is located between the topographic crest and the active part of the upper east rift zone (fig. 18A). This is consistent with the hypothesis of south- ward migration if the rate of migration has recently been sufficiently rapid to prevent construction of a newer and higher constructional ridge. The 1969—74 eruptions of Mauna Ulu created a new, low ridge along the active part of the rift zone. If the active part of the east rift zone is moving southward, that part of the rift zone nearest the sum- mit should show more displacement because of the likelihood of more frequent dike injection near the summit magma reservoir. This may account for the greater divergence between the topographic and grav- ity axes, and between the location of young vents along the western and eastern parts of the rift zone. The gravity data (fig. 183) support the concept of greater dike frequency near the magma reservoir, for the Bouguer values are higher along the upper east rift zone than along the lower east rift zone. The conspicuous bend in the active part of the east rift zone, defined by the location of young cones, pit craters, and fissures (figs. 1 and 18), may reflect the interplay between the greater degree of intrusion near the source reservoir, the southward migration of the rift zone, and two stress systems, one radial to and confined to the reservoir area, the other the gravita- tional system favoring south—southeastward displace- ment. Evidence of an approximately radial stress field at the summit comes from studies of ground deforma- tion accompanying episodes of inflation and deflation (for example, Fiske and Kinoshita, 1969; Swanson and others, 1976). Near the summit area, dikes tend to be radial because of the dominance of the radial stress field. Away from the summit, however, the gravita- tional field controls the orientation of dikes. Figure 20 Rift zone South 6—— North h 5 w 7 Lava flows / A FIGURE 19.—Schematic cross section showing how continued lateral growth of a rift zone generally results in successively younger dikes intruding to higher elevations. Dike 1 is oldest; dike 5 is still erupting lava that may eventually cover flows from dike 4. Arrow, direction of lateral growth. In a less schematic diagram, dikes would be so closely spaced that they would define a wedge—shaped prism of higher density than wallrock, leading to an asymmetric gravity anomaly. \W shows schematically the stages in the possible de- velopment of the bend in the east rift zone. At an early stage in the history of the rift zone before much dis- placement of the south flank (fig. 20A), dikes near the summit area are radial to the source reservoir and tend to cluster in a zone perpendicular to the direction of south-flank displacement. As the rift zone migrates southward because of dike accretion (fig. 203), dikes remain radial to the reservoir near the summit, but curve farther downrift to accommodate themselves to the gravitational stress system. This accommodation, combined with the presence of more dikes near the reservoir than far from it, causes the bend in the rift zone. Figure 200 shows the present stage, in which the bend has been accentuated as more radial dikes were intruded near the summit area. Possible southward migration of the magma reservoir itself at a slower rate than that of the south flank, as discussed later, could also augment the development of such a bend. During the 1960’s and early 1970’s, a fairly un- obstructed conduit apparently transported magma from the summit reservoir down to and beyond the bend in the rift zone (Swanson and others, 1976). This conduit became periodically plugged, leading to erup- tions near the bend and in the summit area. Such a conduit is thought to have developed along the trace of the rift zone at some unknown, but very recent, time. Jackson, Swanson, Koyanagi, and Wright (1975) pointed out the infrequency of eruption along the southeast-trending part of the east rift zone, between the caldera and the conspicuous bend. The southeast trend of this part of the rift zone nearly parallels the direction of horizontal displacement in the zone of dila- tion and in the south flank, so that intrusion may be more easily accommodated without surface rupturing and resultant eruption than elsewhere along the rift zone. Krivoy, 1964; Swanson and others, 1976, pl. 1); in other words, northerly cracks are more nearly radial to the summit than those farther south. Such a pattern may be accounted for by the southward growth of the rift zone (fig. 203 and C). Even fissures of historic erup- tions, such as those in August 1968 and February 1969 (Swanson and others, 1976, pl. 1), have somewhat dif- ferent trends depending on their location, suggesting that older dikes may, to some extent, act as guides for later intrusions, at least beyond a distance of several kilometres from the source reservoir. Thus, several lines of evidence suggest that the east rift zone is migrating south-southeastward, parallel to the direction of south flank displacement and away from the rest of the island, including the north flank of Kilauea itself. This migration is consistent with our model suggesting that rift intrusion takes place in a predominantly gravitational stress system and that Kilauea is buttressed by the rest of the island and can accommodate intrusion chiefly by moving away from this buttress. In this model, each dike injected into the rift zone effectively shores up and stablilizes the zone, replacing loose pillow lava and highly jointed subaer- ial flows with relatively dense and coherent dike rock. Successively younger dikes then tend to seek out the contact between dike and flow rocks, accreting along the south margin of the rift zone where south- southeastward dilation is easiest. In this way the ac- tive part of the rift zone migrates seaward. This model is derived to explain the net long-term result. Actually 32 DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO N Summit magma reservonr A B C Dikes FIGURE 20.—Sketches in plan View showing possible development of bend in east rift zone of Kilauea Volcano owing to accretion of dikes along south side of east rift zone near summit magma reser- voir. A, Early stage. Dikes near summit area are radial to source reservoir and tend to cluster in a zone perpendicular to direction of south-flank displacement. B, Intermediate stage. As rift zone mi- grates southward because of dike accretion, dikes remain radial to reservoir near summit but curve farther downrift to accommodate themselves to gravitational stress system. C, Present stage. Bend accentuated as more radial dikes intruded near summit area. Size of reservoir, from which dikes are fed, is exaggerated for clarity, and shape is greatly simplified from its probable amoeboid form. Model assumes a constant stress field favoring dilation in a south- southeast direction of rift zone, and a radial stress field im- mediately surrounding reservoir. Trend of dikes near summit res- ervoir is largely governed by radial stress field, on the east rift zone by south- southeast dilational field. A slight change in the strike of the fissures within the east rift zone may reflect the southward migration of the rift zone. Recognizable prehistoric and historic fissures north of the main belt of activity trend more westerly by 5°—10° than do those in the main belt (fig. 18A; Stearns and Macdonald, 1946, pl. 1; Moore and the active part of the rift zone is 2—3 km wide, not razor thin, but the net result of intrusion into this zone is that predicted by the model. SOUTHWEST RIFT ZONE The southwest rift zone occupies a more complicated structural setting than the east rift zone. The south- west rift zone is bounded on the west by scarps in the Kaoiki fault system (fig. 2), a system of normal faults cutting the southeast flank of Mauna Loa. This fault system is still active, generating many hundreds or thousands of earthquakes yearly (Koyanagi and others, 1972, fig. 2; Koyanagi and others, 1966). One interpretation is that the Kaoiki is a system of landslide-type faults, analogous to the Hilina fault sys- tem on the south flank of Kilauea (p. 33), that de- veloped during an earlier stage in the growth of Mauna Loa before Kilauea was very large. The southwest rift zone of Kilauea may overlie part of the Kaoiki system, and it is conceivable that at least some of the south- west rift fissures merge at depth with Kaoiki faults. Further discussion of these possibilities must await de- tailed mapping of the southwest rift zone and the Kaoiki fault system. THE HILINA FAULT SYSTEM Although little is known about the southwest rift zone, available evidence is consistent with its seaward migration. The axis of the southwest rift gravity anomaly is offset toward Mauna Loa from the active part of the rift zone, as it is on the east rift zone (fig. 18B). The geologic maps of Stearns and Macdonald (1946, pl. 1) and Walker (1969) show that young vents tend to be located along the seaward side of the south- west rift zone, although this is complicated by vents within the western part of the Koae fault system (for example, Puu Koae and Cone Crater on the map by Walker, 1969). The vents of the September 1971 erup- tion also lie near the southeast edge of the rift zone. Thus we believe that the evidence is consistent with seaward migration, possibly less pronounced than along the east rift zone because of less frequent episodes of magma intrusion. Figure 6 shows that several stations on the Mauna Loa side of the southwest rift zone were displaced away from the zone between 1970 and 1971. We suggest that this displacement was largely absorbed within the Kaoiki fault system by adjustment of fault blocks, in a manner similar to the way in which the east rift zone north of sites of intrusion responds to displacement events (figs. 5A, 5C, 7, and 8A). SUMMIT RESERVOIR The summit reservoir system may also be migrating seaward, though at a slower rate than the east rift zone. Such migration is suggested by: (1) the northeast-southeast elongate shape of the caldera (fig. 2); (2) the location of the present Halemaumau pit cra— ter in the south part of the caldera; (3) the general southerly offset of the present caldera from the high area along its north margin; (4) the location of the present-day reservoir system beneath the south part of the caldera (Fiske and Kinoshita, 1969); and (5) the general southerly migration of the centers of ground deformation during periods of inflation and deflation related to reservoir filling and emptying (Fiske and Kinoshita, 1969; Jackson and others, 1975). Moreover, the positive gravity anomaly is centered somewhat north of the area of maximum ground deformation, a relation explained in a manner analogous to our rift- zone argument. This migration presumably is closely related to the same processes affecting the rift zones, but may be more complex. The summit system is fed by conduits rooted in the mantle and hence is susceptible to lateral movements of the Pacific plate and the asthenosphere. Nonetheless, the postulated seaward migration is an expectable consequence of intrusion in a gravitational stress system favoring southeastward-directed dila- tion. 33 THE HILINA FAULT SYSTEM Leveling surveys since 1921 document uplift in the Hilina fault system (figs. 2 and 3), yet abundant evi- dence indicates net subsidence of this area over the past few thousand years. Many geologists (for example, Stearns and Macdonald, 1946; Moore and Krivoy, 1964), have commented on the seaward-facing fault- line scarps, which in places are more than 500 m high and bounded by gaping cracks, as evidence of normal faults. Walker (1969) measured a seaward dip of about 50 degrees on one exposed fault plane in the Hilina system and noted that the sense of displacement on most of the faults is down toward the sea. The faults have been recently active, for they locally offset the youngest prehistoric flows in the area (Macdonald, 1956). Two large subsidence events were recorded along the south coast during the 19th century. In 1823, a series of strong earthquakes accompanied ground cracking and subsidence of at least 46 cm at the village of Kaimu (Ellis, 1827, p. 195—196), a short distance from BM 10 (inset, fig. 13). In 1868, much of the south coast- al area subsided 1—2 m or more during severe earth— quakes accompanied by a local tsunami (Brigham, 1909, p. 103—113). These two events occurred far from sites of eruption and thus differ from the well-known development of a graben along the rift zone in 1924 near Kapoho, where the east rift zone enters the ocean (Finch, 1925; Jaggar, 1924). The weight of geologic evidence and the two major recorded subsidence events compel us to believe that subsidence is the dominant sense of displacement in the Hilina fault system. If so, then the measured uplift since 1921 within the fault system is acting against the long-term trend and consequently may be decreasing the stability of the fault system. We suggest that this uplift, probably related to tumescence along the rift zone, combines with large seaward-directed displace- ments to produce instability, which ultimately triggers faulting along the unbuttressed south coast, such as took place in 1823 and 1868. We interpret many of these gravity faults to bottom out at shallow depth (fig. 15 A—A ’, B—B’), merging into bedding planes within the hyaloclastic deposits and underlying loosely packed pillow lava that form the basement to the subaerial part of Kilauea (Moore and Fiske, 1969). The displaced blocks are therefore simi- lar to large landslide blocks, as suggested by Stearns and Macdonald (1946) and Moore and Krivoy (1964). Macdonald later objected to the landslide hypothesis, primarily because he believed that “* * *piles of vol- canic rocks with average slopes of 4° to 7° should be very stable” (Macdonald, 1956, p. 286). More recent work by Moore and Fiske (1969, fig. 3), however, 34 suggests that a significant thickness of elastic deposits dipping more than 10° seaward underlies the south flank, and we think that these deposits form an adequate medium in which lateral and downward slip- page could occur. Macdonald (1956) offered a challenging alternative to the landslide hypothesis, suggesting that tumes- cence in the summit region, and by implication the rift zones, causes the formation of nearly vertical fault zones paralleling the rift zones with the inland side upthrown. If this were true, rather straight, linear faults might be expected. Instead, however, faults in the Hilina system curve and branch (fig. 2), and some are arranged in en echelon sets on the limbs of large, arclike master faults (fig. 21). Macdonald’s suggestion also does not explain the lack of such faults on the north flank of Kilauea, whereas landslide-type faults would only be expected on the south flank. The sea- ward dip of 50° that Walker (1969) measured on one fault plane is also inconsistent with Macdonald’s suggestion. Finally, broad areas at the base of the Hilina, Holei, and the Poliokeawe fault-line scarps are nearly flat or even slope gently inland (fig. 22), consist- ent with rotation accompanying landsliding but not with faulting related to rift tumescence. On the other hand, evidence from the 1965—71 level- ing profile (fig. 11B) suggests that the inland side of Hilina faults can at times be displaced upward relative to sea level. That profile shows a sharp offset across Holei Pali, with uplift of 9 cm relative to BM 10. This reference point itself probably moved upward, if at all, during the survey period, so the true amount of uplift relative to sea level may be more than 9 cm. If the fault forming Holei Pali dips seaward, as we believe, then the indication of reverse movement is clear. We con— clude that Macdonald’s suggestion of uplift on the in- land side of Hilina faults is indeed correct for certain periods of strong tumescence of the rift zones, such as between 1965 and 1971, but that this uplift is small and contrary to the dominant long-term sense of dis- placement within the fault system. We estimate that the depth to the base of the hyalo- clastic layer, within which we interpret many of the Hilina faults to die out, is between 1 and 3 km below sea level. The depth increases with distance from Kilauea. This estimate is based on the results of Moore and Fiske (1969, fig. 3) and considerations of the effects of isostatic subsidence (Moore, 1970), which may have lowered the layer several hundred metres since it formed. It is also possible that some of the Hilina faults bottom out in a deeper hyaloclastic layer formed dur- ing growth of Mauna Loa (fig. 15A and B). Walker (1969) mapped several thin dikes in the western part of the Hilina fault system; others occur DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO FIGURE 21.—Aerial view of Hilina fault system near Puu Kapukapu (lower left), highest scarp at coastline (fig. 2). Note en echelon arrangement of Puu Kapukapu and two scarps to northeast (mid— dle and upper right). These en echelon fault-line scarps define west limb of a large fault block. En echelon faults also occur around small fault block (lower right corner) southeast of Puu Kapukapu. Puu Kapukapu is 320 m high. Scarp (upper left) is Hilina Pali. View to north-northwest. FIGURE 22.—Nearly flat ground surface, which may reflect landward rotation during faulting, at base of Holei Pali (background) and unnamed fault-line scarp in Hilina fault system. Part of the area at immediate base of unnamed scarp is shown as virtually flat on Makaopuhi Crater 71/2-minute quadrangle; lava ponded in this area in 1970—73 because of the low gradient, which in fact seemed slightly reversed in places when visited in 1970. Lava flows in 1969—74 have greatly modified topography from that shown on quadrangle map. View to northwest from near Kaena Point (fig. 2). elsewhere in the system. On the basis of field relations, Walker interpreted these dikes as crack fillings fed from overriding flows. The numerous open cracks in the fault system make his interpretation seem likely to us, and we add that neither gravity data (fig. 183 ; Kinoshita and others, 1963) nor seismic refraction data (Hill, 1969) suggest a dike swarm in this area. Thus we interpret the dikes as bearing no genetic relation to the Hilina fault system. APPLICATION TO OTHER OCEANIC VOLCANOES We interpret the shape of the coastline of the south flank to be controlled principally by where and how often lava flows erupted from the rift zones reach the sea, not by faults of the Hilina system. Only locally, as at the horsts near Kaimu (fig. 2) and Puu Kapukapu (figs. 2 and 21), do faults define the trend of the coastline. Most of the coastline is vulnerable to flows from the rift zones, but the area centered near Puu Kapukapu lies between the two rift zones and is less frequently inundated with lava; consequently the rest of the coastline has grown seaward on either side of this sheltered area, leaving a broad embayment. The area inland of this embayment is higher than most other parts of the fault system and includes Puu Kapukapu, the largest and most prominent horst in the Hilina system (fig. 21). This area may stand high because it is pinched between two segments of the south flank moving at slightly convergent azimuths away from the southwest and east rift zones, and so may be kept from subsiding as drastically as elsewhere by this relative compression. Some of the data suggest that the Hilina fault sys- tem is poised for another episode of subsidence, with only minor additional ground displacement needed to trigger it. Figure 10 (lower inset) shows that, in 1970, every south-flank earthquake of M235 except one (fig. 10D) was immediately followed by a brief period of extension of a geodimeter line that otherwise showed strong net contraction in response to rift intrusion. No such correlation could be made in 1969. Possibly the amount of contraction across the fault system was reaching a critical level by 1970, so that earthquakes ' of even moderate size could trigger brief periods of strain release. Measurements in 1971 (fig. 6) and 1974 (Hawaiian Volcano Observatory, unpub. data) showed small extensions across the Hilina system, continuing the trend shown in 1970 and suggesting a further amount of strain release. Most of the strain acquired since 1965 (fig. 10) remains, however, and we antici- pate a subsidence (strain-release) event of unknown magnitude in the not too distant future. (A major subsidence event took place in the Hilina fault system on November 29, 1975, 15 months after this paper was submitted for publication. The subsid- ence, accompanied by an earthquake of M=7.2, af— fected much of the south coast of Kilauea and was as much as 3.4 m in the Puu Kapukapu area. Large ex- tensions were measured across the Hilina system, and many fault scarps showed renewed displacement, with seaward side down. Two campers were killed by a tsunami generated by the earthquake and subsidence. These deaths emphasize the fact that the unstable na- ture of the south flank of Kilauea has broad social as well as geologic implications.) 35 APPLICATION TO OTHER OCEANIC VOLCANOES We doubt that Kilauea is a unique volcano, and many of the principles gained from this study should be applicable to other oceanic shields. In Hawaii, for example, large landslide blocks are probably present along the west coast of Mauna Loa, bounded by the Kealakakua fault and unnamed faults between Hookena and Milolii (Stearns and Macdonald, 1946, pl. 1). We consider that the Kaoiki fault system on Mauna Loa (fig. 2) may also be a gravity-controlled landslide area, now partly shored up by Kilauea but still influencing the way in which the southwest rift zone of Kilauea is deformed. The landslide—type faults of Mauna Loa are downslope from a large rift zone and may well have developed in response to intrusion in that zone. Moore (1964) described possible giant land- slide masses from Koolau and East Molokai Volcanoes that may owe their origin to dike intrusion in rift zones parallel to the coastline. The flanks of many older Hawaiian shields, such as Haleakala, West Maui, and East Molokai Volcanoes, dip 11 to 18 degrees seaward, much steeper than the average shield slope of several degrees. It is possible that these steep slopes partly reflect the net effect of doming caused by the intrusion of numerous dikes, although all of these volcanoes are at a more advanced stage of evolution than Kilauea and consequently may be subject to processes not ob- served during the young, tholeiitic stage. Any oceanic shield volcano that grows on the flank of an earlier shield in the absence of a dominant tectonic stress system should behave similarly to Kilauea. Con- sequently, there are probably numerous volcanoes in the ocean basins that are characterized by asymmetric dilation of rift zones, a mobile seaward-facing flank, and secondary landslide blocks on the mobile flank. Recognition of these features in deeply eroded shields may be difficult, but there are possible clues to look for. Dikes in rift zones of buttressed volcanoes should on the average become younger from the landward side toward the seaward side; this might be detected in favorable circumstances by paleomagnetic or radiometric age data, temporally related differences in chemistry, or simple stratigraphic evidence. Examples in the Hawaiian chain where rift zones could be examined in this way are East Molokai and Koolau Volcanoes; both of these grew adjacent to an older shield (Fiske and Jackson, 1972) and are dissected suf- ficiently to expose dikes. On the other hand, unbut- tressed shields such as Niihau show no such tendency and when eroded reveal symmetric rift zones in the central part of the edifice. We hope that examination of eroded shields with the model of Kilauea in mind can 36 shed more light on their internal structure, as well as feed back important clues for a better understanding of Kilauea itself. SUPPLEMENTAL INFORMATION PROBLEM OF GROUND DEFORMATION OCCURRING DURING THE HORIZONTAL SURVEYS The 1971 trilateration was carried out between Oc- tober 3 and 18; the 1970 trilateration took place chiefly between late August and late September, with a few supplementary distances measured in November and early December. The volcano was quite stable during both surveys, as indicated by measurements of tilt and strain at the summit and the absence of major earth- quake activity elsewhere on the volcano. The 1961 triangulation was conducted between March 22 and July 25, spanning the last three days of the March summit eruption, the entire July summit eruption, and the intervening period of substantial summit inflation (about 90 microradians at one tiltme- ter at the summit) (Richter and others, 1964, fig. 3). These events, especially the inflation, almost certainly caused some of the poor triangle closures reported in the summit region (table 2) but probably did not greatly affect the rest of the volcano. The 1958 triangulation lasted from February to June. Only slight summit deformation took place dur- ing this time, as judged from records of tilt (Eaton, 1962, fig. 13) and seismicity (Eaton and Fraser, 1958a, b), and probably little if any deformation occurred on the flanks of the volcano. The 1949 triangulation spanned the March-July interval. A horizontal pendulum seismometer used as a tiltmeter at the summit recorded little deformation during this time (Volcano Letter, 1949). Virtually nothing is known about ground deforma- tion between 1896 and 1914, for no measurements were made then. Mild summit eruption in Halemaumau Crater continued throughout most of this time. Similar, but more recent, Halemaumau eruptions have been accompanied by very little ground deformation anywhere on the volcano, particularly on the flanks, so we infer that little took place between 1896 and 1914. BASELINE SELECTION The selection of baselines and stations considered as stable is one of the most critical factors affecting the study of horizontal displacement. In general, we tried to select stations as far from eruptive activity as feasi- ble. Table 4 shows the stations and baselines that we assumed to be stable. Keaau, Olaa, Kaloli, and Kaloli 2 are far removed from areas of known instability (fig. 3), DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO TABLE 4.—-Baselines used for deriving horizontal displacements in figure 5 Survey interval Stable stations Baselines (lengths) 1896(1914)-1970 Olaa, Kaloli1 OlazlJQalolil (11,745.71 In). 1949—70 Kulani, Keaau, Kaloli 2, Kulani—Keaau (27,984.81 m); Keakapulu, Kahaha 2, Keaau—Kaloli 2 (10,929.93 m). 195&70 Kulani, Keaau, Kaloli 2 Kulani—Keaau (27,948.81 In) KeaauiKaloli 2 (10,929.93 11)). Kulani—Strip (9,559.14 m). Kulani—Strip (9,559.14 In). Keakapulu. Kulani, Strip, Kahaha 2 Kulani, Strip, Keakapulu 1961770 1970471 1Kaloli has been destroyed by wave erosion. Distances were measured to nearby Kaloli 2 and reduced to Kaloli using offset data determined by the US. Coast and Geodetic Survey in 1949 before the destruction of Kaloli. and the distance between Keaau and Kaiwiki New, a station on the lower east slope of Mauna Kea, did not measurably change between 1949 and 1970. Thus the assumption of their stability is reasonable. Kulani, on the southeast flank of Mauna Loa, is closer to the summit of Kilauea and could conceivably have been affected by the great summit subsidence in 1924 (Wilson, 1935). For this reason, Kulani was not used as a fixed point for any interval spanning 1924. In fact, however, Kulani probably was not measurably affected by the great subsidence, for its apparent dis- placement over the event is negligible (fig. 5E). Kulani did not move measurably between 1949 and 1970 be- cause the distance to Kaiwiki New and the astronomic azimuth to Keaau are unchanged and because dis- placements derived using Kulani-Keaau as a baseline agree within probable error with those using Keaau- Kaloli 2. Strip, also on the slope of Mauna Loa, was estab- lished in 1961. No unusually large structural events affected the summit of Kilauea in the following dec- ade, and Mauna Loa was quiet, on the basis of the monitored seismicity of the volcano and leveling and geodimeter data from the summit (Hawaiian Volcano Observatory, unpub. data, 1961—73; Decker and Wright, 1968). Kahaha 2 is on Mauna Loa and is approximately along strike of, though southwest of and possibly higher than, the known limits of the Kaoiki fault sys- tem (fig. 3), a seismically active system on the south- east flank of Mauna Loa. Displacements for nearby stations are small and geologically reasonable, assum- ing stability of Kahaha 2. Keakapulu is located on the northwest edge of the Kaoiki fault system, and most distances from it had to be shortened by amounts approaching triangulation error to achieve geologically reasonable results. These adjustments suggest possible southeastward displace- ment of the station, although within probable survey error. In any case, the stability of Keakapulu is not critical if Kahaha 2 has remained stable. In summary, all available information, including the consistency of our results, indicates that the base sta- REFERENCES CITED tions have remained stable within the probable limits of survey error during the 20th century. PROCEDURE FOR TREATING TRIANGULATION DATA We used originally observed angles, adjusted only for triangle closure to 180°, in this study. No use was made of published lengths or angles resulting from previous network adjustments, as such adjustments were made in various ways for each triangulation sur- vey and involve different baseline lengths, numbers of stations, and other constraints. The original angles were obtained through the courtesy of B. K. Meade (US. Coast and Geodetic Survey data) and J. P. Church, C. R. Lloyd, and R. F. Thurston (U.S. Geologi- cal Survey data). Each triangulation survey was compared with the 1970 trilateration survey by reference to one or more baselines (table 4). The 1970 geodimeter length of each baseline, reduced to sea level by standard techniques, is assumed to have been the length during each preced— ing triangulation. Using this length and originally ob— served angles, the triangulation was then recomputed using the law of sines. In this way, distances derived from the trilateration and triangulation surveys are directly comparable. In general, corresponding sides of triangles differ in length for the two surveys, indicat- ing apparent displacement of one or more stations. The ground-displacement vectors were derived by a graphical method described by Kinoshita, Swanson, and Jackson (1974). This method was selected in pref- erence to the standard least-squares approach nor- mally used for triangulation data because it allows visual inspection and evaluation of the displacements as they are derived. In this way geologic input could be made as to what displacements seem reasonable, and what distances should be given more weight than others on the basis of the displacements they produce. This approach seems justified, given the quality of the data. After geologic evaluation, distances were then changed by trial and error within permissible limits (table 2) to give geologically reasonable displacements. The graphical derivation was then resumed along the chain of triangles until a displacement had been de- termined at each station. When a geologically pre— ferred, graphically derived length differed signifi- cantly from the original computed one, the new length was used for subsequent law of sines computations de— pendent on it. Finally an acceptable degree of internal consistency and geologic reasonableness was obtained for all triangles in each survey interval. This time-consuming approach is partly subjective, but the basic data were not violated. All graphical ad— justments were made within stated error limits (table 2), and most were far less than the maximum permis- 37 sible amount. In essence, then, this approach is analogous to a standard least-squares adjustment with an overriding geologic monitor on each operation. Experience has shown that the basic pattern of dis- placement could not be changed greatly when adjust- ments were kept within the limit of survey error, but that, with minor adjustments, the displacements could be better fitted to known geology. An example is shown in figure 5E, in which displacements derived by us are compared with displacements calculated by Lloyd (1964) using the standard least-squares approach. A similar basic pattern is shown by both displacement sets, but the progressive clockwise swing of the least- squares vectors (a pattern typically indicative of sur- vey errors) may be eliminated by small adjustments in favor of displacement directly away from open ground cracks that formed during the survey interval, a reasonable geologic constraint. In short, the basic pat- tern is there regardless of how the data are manipu- lated, and it is with this basic pattern that we are most concerned in this paper. REFERENCES CITED Brigham, W. T., 1909, The volcanoes of Kilauea and Mauna Loa on the Island of Hawaii: Bernice Pauahi Bishop Museum Memoir, v. 2, no. 4, 222 p. Cloos, Ernst, 1968, Experimental analysis of Gulf Coast fracture patterns: Am. Assoc. Petroleum Geologists Bull., v. 52, p. 420—444. Decker, R. W., 1965, Vertical ground displacements over the east rift of Kilauea Volcano, Hawaii: Am. Geophys. Union Trans., V. 46, p. 185. 1969, Land surface defamation related to volcanic activity in Hawaii: Am. Phil. Soc. Yearbook, 1969, p. 295—296. Decker, R. W., Hill, D. P., and Wright, T. L., 1966, Deformation measurements on Kilauea Volcano, Hawaii: Bull. Volcanol., v. 29, p. 721—732. Decker, R. W., and Wright, T. L., 1968, Deformation measurements on Mauna Loa Volcano, Hawaii: Bull. Volcanol., v. 32, p. 401. Dieterich, J.V., and Decker, R. W., 1975, Finite element modeling of surface deformation associated with volcanism: J our. Geophys. Res., v. 80, p. 4094—4102. Duffield, W. A., 1975, Structure and origin of the Koae fault system, Kilauea Volcano, Hawaii: US. Geol. Survey Prof. Paper 856, 12 p. Duffield, W. A., Jackson, D. B., and Swanson, D. A., 1976, Shallow, forceful intrusion of magma and related ground deformation at Kilauea Volcano, May 15—16, 1970: Bull. Volcanol. (in press). Duffield, W. A., and Nakamura, Kazuaki, 1973, A photogeologic method for determining the direction of horizontal dilation from patterns of en echelon fracturing: U.S. Geol. Survey Jour. Re- search, v. 1, p. 283—287. Eaton, J. P., 1962, Crustal structure and volcanism in Hawaii: Am. Geophys. Union Mon. 6, p. 13—29. Eaton, J. P., and Fraser, G. D., 1958a, Hawaiian Volcano Observa- tory Summary 9, 8 p. 1958b, Hawaiian Volcano Observatory Summary 10, 9 p. Ellis, William, 1827, Narrative of a tour of Hawaii, or Owhyhee, with remarks on the history, traditions, manners, customs, and language of the inhabitants of the Sandwich Islands: Honolulu, Advertiser Pub. Co., reprinted in 1963, 342 p. 38 Endo, E. T., 1971, Focal mechanism for the May 15—18, 1970, shallow Kilauea earthquake swarm: San Jose State College, San Jose, Calif, M.S. thesis, 165 p. Finch, R. H., 1925, The earthquakes at Kapoho, Island of Hawaii, April 1924: Seismol. Soc. America Bull., v. 15, p. 122—127. 1950, The December 1950 subsidence at Kilauea (Hawaii): Volcano Letter, no. 510, p. 1—3. Fiske, R. S., and Jackson, E. D., 1972, Orientation and growth of Hawaiian volcanic rifts: the effect of regional structure and gravitational stresses: Royal Soc. [London] Proc., v. 329, p. 299—— 326. Fiske, R. S., and Kinoshita, W. T., 1969, Inflation of Kilauea Volcano prior to its 1967—1968 eruption: Science, v. 165, p. 341—349. Fiske, R. S., and Koyanagi, R. Y., 1968, The December 1965 eruption of Kilauea Volcano, Hawaii: U.S. Geol. Survey Prof. Paper 607, 21 p. Gossett, F. R., 1959, Manual of geodetic triangulation: U.S. Coast and Geodetic Survey Spec. Pub. 247, 344 p. Hill, D. P., 1969, Crustal structure of the Island of Hawaii from seismic-refraction measurements: Seismol. Soc. America Bull., V. 59, p. 101—130. Jackson, D. B., Swanson, D. A., Koyanagi, R. Y., and Wright, T. L., 1975, The August and October 1968 east rift eruptions of Kilauea Volcano, Hawaii: U.S. Geol. Survey Prof. Paper 890, 33 p. Jaggar, T. A., 1924, Volcanic conditions in April (1924): Hawaiian Volcano Observatory Bull., V. 12, no. 4, p. 17—20. 1947, Origin and development of craters: Geol. Soc. America Mem. 21, 508 p. Karren, R. J., 1959, Report on leveling in eruptive areas, Hawaii, 1957—1958: U.S. Geol. Survey, 9 p. and appendix. Kinoshita, W. T., 1967, May 1963 earthquakes and deformation in the Koae fault zone, Kilauea Volcano, Hawaii, in Geological Survey research 1967: U.S. Geol. Survey Prof. Paper 575—C, p. Cl73~Cl76. Kinoshita, W. T., Koyanagi, R. Y., Wright, T. L., and Fiske, R. S., 1969, Kilauea Volcano: The 1967—1968 summit eruption: Sci- ence, v. 166, p. 459—468. Kinoshita, W. T., Krivoy, H. L., Mabey D. R., and Macdonald, R. R., 1963, Gravity survey of the island of Hawaii, in Geological Sur- vey research 1963: U.S. Geol. Survey Prof. Paper 475—C, p. 0114—0116. Kinoshita, W. T., Swanson, D. A., and Jackson, D. B., 1974, The measurement of crustal deformation related to volcanic activity at Kilauea Volcano, Hawaii, in Civetta, L., Gasparini, P., Luongo, G., and Rapolla, A., eds., Physical volcanology: Elsevier, Amsterdam, p. 87—115. Koyanagi, R. Y., Endo, E. T., and Okamura, A. T., 1975, Hawaiian Volcano Observatory Summary 61, p. 1. Koyanagi, R. Y., Krivoy, H. L., and Okamura, A. T., 1966, The 1962 Kaoiki, Hawaii, earthquake and its aftershocks: Seismol. Soc. America Bull., v. 56, p. 1317—1335. Koyanagi, R. Y., Swanson, D. A., and Endo, E. T., 1972, Distribution of earthquakes related to mobility of the south flank of Kilauea Volcano, Hawaii, in Geological Survey research 1972: U.S. Geol. Survey Prof. Paper 800—D, p. D89—D97. Koyanagi, R. Y., Unger, J. D., and Endo, E. T., 1973, Seismic evi- dence for magma intrusion in the eastern Koae fault system, Kilauea Volcano, Hawaii: Am. Geophys. Union Trans, v. 54, no. 11, p. 1216. Lloyd, C. R., 1964, Investigation of horizontal displacement of ground surface, Kilauea and Puna areas, Hawaii: U.S. Geol. Survey, 27 p. Macdonald, G. A., 1954, Activity of Hawaiian volcanoes during the years 1940—1950: Bull. Volcanol., v. 15, p. 119—179. DISPLACEMENT OF THE SOUTH FLANK OF KILAUEA VOLCANO 1955, Hawaiian volcanoes during 1952: U.S. Geol. Survey Bull. 1021—B, 108 p. 1956, The structure of Hawaiian volcanoes: K. Ned. Geol.- Mijnb. Genoot. Verh., Geol. Ser., v. 16, p. 274—295. Macdonald, G. A., and Abbott, A. T., 1970, Volcanoes in the sea: Univ. Hawaii Press, Honolulu, 441 p. Macdonald, G. A., and Eaton, J. P., 1957, Hawaiian volcanoes during 1954: U.S. Geol. Survey Bull. 1061—B, 72 p. 1964, Hawaiian volcanoes during 1955: U.S. Geol. Survey Bull. 1171, 170 p. Mogi, Kiyoo, 1958, Relations between the eruptions of various vol- canoes and the deformation of the ground surfaces around them: Tokyo Univ. Earthquake Research Inst. Bull., v. 36, p. 99—134. Moore, J. G., 1964, Giant submarine landslides on the Hawaiian Ridge, in Geological Survey research 1964: U.S. Geol. Survey Prof. Paper 501—D, p. D95—D98. 1970, Relationship between subsidence and volcanic load, Hawaii: Bull. Volcanol., v. 34, p. 562—576. 1971, Bathymetry and geology—East cape of the Island of Hawaii: U.S. Geol. Survey Misc. Geol. Inv. Map I—677, scale 1:62,500. Moore, J. G., and Fiske, R. S., 1969, Volcanic substructure inferred from dredge samples and ocean-bottom photographs, Hawaii: Geol. Soc. America Bull., v. 80, p. 1191—1202. Moore, J. G., and Koyanagi, R. Y., 1969, The October 1963 eruption of Kilauea Volcano, Hawaii: U.S. Geol. Survey Prof. Paper 614—0, p. 01—013. Moore, J. G., and Krivoy, H. L., 1964, The 1962 flank eruption of Kilauea Volcano and structure of the east rift zone: Jour. Geophys. Research, v. 69, p. 2033—2045. Nakamura, Kazuaki, 1970, En echelon features of Icelandic ground fissures: Acta Nat. Islandica, v. 2, no. 8, p. 1—15. Okamura, R. T., and Swanson, D. A., 1975, January—March 1971 leveling at Kilauea and tabulated comparisons with previous data: Hawaiian Volcano Observatory Summary 61, p. 7—42. Peck, D. L., Wright, T. L., and Moore, J. G., 1966, Crystallization of tholeiitic basalt in Alae lava lake, Hawaii: Bull. Volcanol., V. 29, p. 629—656. Peterson, D. W., 1967, Geologic map of the Kilauea Crater quad- rangle, Hawaii: U.S. Geol. Survey Geol. Quad. Map GQ—667, scale 1224,000. Powers, H. A., 1948, A chronology of the explosive eruptions of Kilauea: Pacific Sci., v. 2, p. 278—292. Richter, D. H., Ault, W. U., Eaton, J. P., and Moore, J. G., 1964, The 1961 eruption of Kilauea Volcano, Hawaii: U.S. Geol. Survey Prof. Paper 474—D, 34 p. Richter, D. H., Eaton, J. P., Murata, K. J., Ault, W. U., and Krivoy, H. L., 1970, Chronological narrative of the 1959—60 eruption of Kilauea Volcano, Hawaii: U.S. Geol. Survey Prof. Paper 537—E, p. E1—E7 3. Rubin, Meyer, and Suess, H. E., 1956, U.S. Geological Survey radiocarbon dates III: Science, v. 123, p. 442—448. Ryall, Alan, and Bennett, D. L., 1968, Crustal structure of southern Hawaii related to volcanic processes in the upper mantle: J our. Geophys. Research, v. 73, p. 4561—4582. Shaw, H. R., 1973, Mantle convection and volcanic periodicity in the Pacific; evidence from Hawaii: Geol. Soc. America Bull., v. 84, p. 1505—1526. Stearns, H. T., 1947, Geology and ground-water resources of the Island of Niihau, Hawaii: Hawaii Div. Hydrography Bull. 12, 38 p. Stearns, H. T., and Macdonald, G. A., 1946, Geology and ground- water resources of the Island of Hawaii: Hawaii Div. Hydrog- raphy Bull. 9, 363 p. Stearns, H. T., and Vaksvik, K. N., 1935, Geology and ground-water REFERENCES CITED resources of the Island of Oahu, Hawaii: Hawaii Div. Hydrog- raphy Bull. 1, 479 p. Swanson, D. A., 1972, Magma supply rate at Kilauea Volcano, 1952—1971: Science, v. 175, p. 169—170. Swanson, D. A., Duffield, W. A., Jackson, D. B., and Peterson, D. W., 1972, The complex filling of Alae Crater, Kilauea Volcano, Hawaii: Bull. Volcanol., v. 36, no. 1, p. 105—126. Swanson, D. A., Jackson, D. B., Duffield, W. A., and Peterson, D. W., 1971, Mauna Ulu eruption, Kilauea Volcano: Geotimes, v. 16, p. 12—16. Swanson, D. A., Jackson, D. B., Koyanagi, R. Y., and Wright, T. L., 1976, The February 1969 east rift eruption of Kilauea Volcano, Hawaii: US. Geol. Survey Prof. Paper 891, 30 p. Swanson, D. A., and Okamura, R. T., 1975, Trilateration network on Kilauea, fall 1970: Hawaiian Volcano Observatory Summary ———*1 39 60, p. 3—12. US. Geological Survey, 1966, Leveling: Part 2E in Topographic in- structions of the United States Geological Survey,‘63 p. Volcano Letter, 1949, nos. 503—505. Walker, G. W., 1969, Geologic map of the Kau Desert quadrangle, Hawaii: US. Geol. Survey Map GQ—827, scale 1124,000. Wilson, R. M., 1935, Ground surface movement at Kilauea Volcano, Hawaii: Hawaii Univ. Research Pub. 10, 56 p. Wingate, E. G., 1933, Puna Triangulation: Volcano Letter, no. 400. Wright, T. L., 1971, Chemistry of Kilauea and Mauna Loa lava in space and time: US. Geol. Survey Prof. Paper'735, 40 p. Wright, T. L., Kinoshita, W. T., and Peck, D. L., 1968, March 1965 eruption of Kilauea Volcano and the formation of Makaopuhi lava lake: Jour. Geophys. Research, v. 73, p. 3181—3205. {EGPO 691- 308—1976 iiiéflgEg 7 D AY 7 ”BMW” 9, (”i A Land Use and Land Cover Classification System for Use with Remote Sensor Data GEOLOGICAL SURVEY PROFESSIONAL PAPER 964 \ ;’\ A Land Use and Land Cover Classification System for Use with Remote Sensor Data By JAMES R. ANDERSON, ERNEST E. HARDY, JOHN T. ROACH, and RICHARD E. WITMER GEOLOGICAL SURVEY PROFESSIONAL PAPER 964 A revision of the land use classification system as presented in US. Geological Survey Circular 671 UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON : 1976 UNITED STATES DEPARTMENT OF THE INTERIOR THOMAS S. KLEPPE, Secretary GEOLOGICAL SURVEY V. E. McKelvey, Director Library of Congress Cataloging in Publication Data Main entry under title: A Land use and land cover classification system for use with remote sensor data. (U.S. Geological Survey professional paper ; 964) Revision of the ed. by J. R. Anderson, E. E. Hardy, and J. T. Roach published in 1972 under title: A land-use classification system for use With remote sensor data. Bibliography: p. Supt. of Docs. no.: I 19.16:964 1. Land—United States—Classification. 2. Remote sensing systems. I. Anderson, James Richard, 1919— II. Anderson, James Richard, 1919- A land-use classification system for use with remote sensor data. III. Series: United States. Geological Survey. Professional paper ; 964. HD111.L258 333.7’012 75—619350 For sale by the Superintendent of Documents, US. Government Printing Office Washington, DC. 20402 Stock Number 024-001-02809-3 FIGURE TABLE 2. 2“?” PPM!" CONTENTS Page Abstract _________________________________________________________________ Introduction ______________________________________________________________ 1 Need for standardization __________________________________________________ 2 Historical development of the classification system ___________________________ 3 Designing a classification system for use with remote sensing techniques ________ 4 Classification criteria ______________________________________________________ 5 Developing the classification system ___________________ _,_ ___________________ 7 Using the classification system ____________________________________________ 9 Definitions _______________________________________________________________ 10 Urban or Built-up Land ______________________________________________ 10 Agricultural Land ____________________________________________________ 13 Rangeland _________________________________, __________________________ 14 Forest Land ________________________________________________________ 16 Water _______________________________________________________________ 17 Wetland _____________________________________________________________ 1'7 Barren Land _________________________________________________________ 18 Tundra ______________________________________________________________ 210 Perennial Snow or Ice _____________________ ' ___________________________ 2 1 Map presentation _________________________________________________________ 22 Selected bibliography ____________________________ _.. _______________________ 27 ILLUSTRATIONS Map of a part of the Indianapolis, Ind.-Ill., 1:250,000 quadrangle, showing Level I land use and land cover _____________________________________________________________________________________ Map of a part of the Indianapolis, Ind-111., 1:250,000 quadrangle, showing Level II land useand land cover ______________________________________________________________________________________ Map of a part of the Maywood, Ind., 1124,000 quadrangle, showing Level II land use and land cover __ Map of a part of the Maywood, Ind., 1124,000 quadrangle, showing Level III land use and land cover __ _—___———"— TABLES Major uses of land, United States, 1969 ___________________________________________________________ Land use and land cover classification system for use with remote sensor data _______________________ Standard land use code—first level categories _____________________________________________________ U.S.G.S. Level I land use color code ____________________________________________________________ III 23 24 25 26 N) NCDWW A .LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA By JAMES R. ANDERSON, ERNEST E. HARDY, JOHN T. ROACH, and RICHARD E. WITMER ABSTRACT The framework of a national land use and land cover classification system is presented for use with remote sensor data. The classification system has been developed to meet the needs of Federal and State agencies for an up-to-date overview of land use and land cover throughout the country on a basis that is uniform in categorization at the more generalized first and second levels and that will be receptive to data from satellite and aircraft remote sensors. The pro- posed system uses the features of existing widely used classi- fication systems that are amenable to data derived from re- mote sensing sources. It is intentionally left open-ended so that Federal, regional, State, and local agencies can have flexibility in developing more detailed land use classifications at the third and fourth levels in order to meet their particular needs and at the same time remain compatible with each other and the national system. Revision of the land use classification system as presented in U.S. Geological Survey Circular 671 was undertaken in order to incorporate the re- sults of extensive testing and review of the categorization and definitions. INTRODUCTION A modern nation, as a modern business, must have adequate information on many complex interrelated aspects of its activities in order to make decisions. Land use is only one such aspect, but knowledge about land use and land cover has become increas- ingly important as the Nation plans to overcome the problems of haphazard, uncontrolled develop- ment, deteriorating environmental quality, loss of prime agricultural lands, destruction of important wetlands, and loss of fish and wildlife habitat. Land use data are needed in the analysis of environmental processes and problems that must be understood if living conditions and standards are to be improved or maintained at current levels. One of the prime prerequisites for better use of land is information on existing land use patterns and changes in land use through time. The U.S. Department of Agriculture (1972) reported that during the decade of the 1960’s, 730,000 acres (296,000 hectares) were urbanized each year, trans- portation land uses expanded by 130,000 acres (53,000 hectares) per year, and recreational area increased by about 1 million acres (409,000 hec- tares) per year. Knowledge of the present distribu- tion and area of such agricultural, recreational, and urban lands, as well as information on their chang- ing proportions, is needed by legislators, planners, and State and local governmental officials to deter- mine better land use policy, to project transporta- tion and utility demand, to identify future develop- ment pressure points and areas, and ’to implement effective plans for regional development. As Claw- son and Stewart (1965) have stated: In this dynamic situation, accurate, meaningful, current data on land use are essential. If public agencies and private organizations are to know What is happening, and are to make sound plans for their own future action, then reliable infor- mation is critical. The variety of land use and land cover data needs is exceedingly broad. Current land use and land cov- er data are needed for equalization of tax assess- ments in many States. Land use and land cover data also are needed by Federal, State, and local agencies for water-resource inventory, flood control, water- supply planning, and waste-water treatment. Many Federal agencies need current comprehensive inven- tories of existing activities on public lands combined with the existing and changing uses ‘of adjacent private lands to improve the management of public lands. Federal agencies also need land use data to assess the environmental impact resulting from the development of energy resources, to manage wildlife resources and minimize man-wildlife ecosystem conflicts, to make national summaries of land use patterns and changes for national policy formula- tion, and to prepare environmental impact state- ments and assess future impacts on environmental quality, *—~‘ 2 A LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA NEED FOR STANDARDIZATION For many years, agencies at the various govern- mental levels have been collecting data about land, but for the most part they have worked independent- ly and without coordination. Too often this has meant duplication of effort, or it has been found that data collected for a specific purpose were of little or no value for a similar purpose only a short time later. There are many different sources of information on existing land use and land cover and on changes that are occurring. Local planning agencies make use of detailed information generated during ground surveys involving enumeration and observation. In- terpretation of large-scale aerial photographs also has been used widely (Avery, 1968). In some cases, supplementary information is inferred on the basis of utility hookups, building permits, and similar in- formation. Major problems are present in the appli- cation and interpretation of the existing data. These include changes in definitions of categories and data- collection methods by source agencies, incomplete data coverage, varying data age, and employment of incompatible classification systems. In addition, it is nearly impossible to aggregate the available data because of the differing classification systems used. The demand for standardized land use and land cover data can only increase as we seek to assess and manage areas of critical concern for environ- mental control such as flood plains and wetlands, energy resource development and production areas, wildlife habitat, recreational lands, and areas such as maj or residential and industrial development sites. As the result of long concern about duplication and coordination among Federal, State, and local governments in the collection and handling of vari- ous types of data, the United States has already achieved reasonably effective, though not perfect, standardization in some instances, as evidenced by present programs in soil surveys, topographic map- ping, collection of weather information, and inven- tory of forest resources. Recent developments in data processing and remote sensing technology make the need for similar cooperation in land use inven- tories even more evident and more pressing. Devel- opment and acceptance of a system for classifying land use data obtained primarily by use of remote sensing techniques, but reasonably compatible with existing classification systems, are the urgently needed first steps. This is not the first time that use of remote sensors has been proposed to provide the primary data from which land use and land cover types and their bound- aries are interpreted. During the past 40 years several surveys, studies, and other projects have successfully demonstrated that remote sensor data are useful for land use and land cover inventory and mapping. These surveys have contributed to our con- fidence that land use and land cover surveys of larger areas are possible by the use of remote sensor data bases. In the mid-1940’s, Francis J. Marschner began mapping major land use associations for the entire United States, using aerial photographs taken dur- ing the late 1930’s and the early 1940’s. Marschner produced a set of State land use maps at the scale of 1 : 1,000,000 from mosaics of the aerial photographs and then compiled a map of major land uses at 1 :5,000,000 (Marschner, 1950) . More recently, the States of New York and Min- nesota have used remote sensor data for statewide land use mapping. New York’s LUNR (Land Use and Natural Resources) Program (New York State Office of Planning Coordination, 1969) employs com- puter storage of some 50 categories of land use infor- mation derived from hand-drafted maps compiled by interpreting 1967—1970 aerial photography. This information can be updated and manipulated to pro- vide numerical summaries and analyses and com- puter-generated maps (Hardy and Shelton, 1970). Aerial photographs taken in the spring of 1968 and 1969 at an altitude of about 40,000 ft (12,400 m) yielded the data incorporated into the nine categories of the Minnesota Land Use Map, a part of the Min- nesota Land Management Information System (0r- ning and Maki, 1972) . Thrower’s map (1970) of the Southwestern United States represents the first large-area inventory of land use employing satellite imagery. Imagery from several manned and unman- ned missions was used in deriving the general land use map published at a scale of 1:1,000,000. Remote sensing techniques, including the use of conventional aerial photography, can be used effec- tively to complement surveys based on ground ob- servation and enumeration, so the potential of a timely and accurate inventory of the current use of the Nation’s land resources now exists. At the same time, data processing techniques permit the storage of large quantities of detailed data that can be or- ganized in a variety of ways to meet specific needs. The patterns of resource use and resource demand are constantly changing. Fortunately, the capability to obtain data about land uses related to resource development is improving because of recent tech- nological improvements in remote sensing equip- ————",. HISTORICAL DEVELOPMENT OF THE CLASSIFICATION SYSTEM 3 ment, interpretation techniques, and data process- ing (National Academy of Sciences, 1970). HISTORICAL DEVELOPMENT OF THE CLASSIFICATION SYSTEM The needs of Federal agencies for a broad over- view of national land use and land cover patterns and trends and environmental values led to the for- mation of an Interagency Steering Committee on Land Use Information and Classification early in 1971. The work of the committee, composed of rep- resentatives from the Geological Survey of the US. Department of the Interior, the National Aero- nautics and Space Administration (NASA), the Soil Conservation Service of the US. Department of Agriculture, the Association of American Geograph- ers, and the International Geographical Union, has been supported by NASA and the Department of the Interior and coordinated by the US. Geological Survey (U.S.G.S.). The objective of the committee was the develop- ment of a national classification system that would be receptive to inputs of data from both convention- al sources and remote sensors on high-altitude air- craft and satellite platforms, and that would at the same time form the framework into which the cate- gories of more detailed land use studies by regional, State, and local agencies could be fitted and aggre- gated upward from Level IV toward Level I for more generalized smaller scale use at the national level. Several classification systems designed for or amenable to use with remote sensing techniques served as the basis for discussion at a Conference on Land Use Information and Classification in Wash- ington‘, DC, June 28—30, 1971. This conference was attended by more than 150 representatives of Fed- eral, State, and local government agencies, univer- sities, institutes, and private concerns. On the basis of these discussions, the Interagency Steering Com— mittee then proposed to develop and test a land use and land cover classification system that could be used with remote sensing and with minimal reliance on supplemental information at the more generalized first and second levels of categorization. The need for compatibility with the more generalized levels of land use and land cover categorization in classifica- tion systems currently in use was clearly recognized, especially those levels of the Standard Land Use Cod— ing Manual published by the US. Urban Renewal Administration and the Bureau of Public Roads (1965), the inventory of Major Uses of Land made every 5 years by the Economic Research Service of the US. Department of Agriculture (Frey, 1973), and the national inventory of soil and water conser- vation needs, initiated in 1956 and carried out (for the second time in 1966 by several agencies of the US. Departments of Agriculture and Interior (US. Department of Agriculture, 1971). Two land use classification systems initially pro— posed by James R. Anderson for conference use were designed to place major relianceon remote sensing, although supplementary sources of information were assumed to be available for the more elaborate of the two (Anderson, 1971). The classification system for the New York State Land Use and Natural Re- sources Inventory, developed mainly at the Center for Aerial Photographic Studies at Cornell Univer- sity, had been designed for use with aerial photogra- phy at 1 :24,000 scale, and although devised specifi- cally for New York State, it was adaptable for use elsewhere. To take advantage of the New York experience, Ernest E. Hardy and John T. Roach were invited to collaborate in preparing the definitive framework of the proposed classification. Definitions of land use categories used in New York were care- fully reviewed and were modified to make them ap- plicable to the country as a Whole. The resulting classification was presented in US. Geological Sur- vey Circular 671. Because of his past experience with the Commission on Geographic Applications of Remote Sensing of the Association of American Ge- ographers, Richard E. Witmer was invited to partici- pate with the others in this revision of the classifica- tion system. ' Attention was given mainly to the more general- ized first and second levels of categorization. Defini- tions for each of the categories on these two levels were subjected to selective testing and evaluation by the U.S.G.S., using darta obtained primarily from high—altitude flights as part of the research in con— nection with the U.S.G.S. Central Atlantic Regional Ecological Test Site (CARETS) Project (28,800 mi2 or 74,700 km‘-’), the Phoenix Pilot Project (31,500 mi‘-’ or 81,500 km2), and the land use mapping for the Ozarks Regional Commission (72,000 mi2 or 186,500 km?). The work of Pettinger and Poulton (1970) pro- vided valuable insight into the land use mosaic of the Southwestern United States. Some of the categoriza- tion for barren land and rangeland suggested by these researchers has been adopted in this land use and land cover classification system. * 4 A LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA DESIGNING A CLASSIFICATION SYSTEMv- FOR USE WITH REMOTE SENSING TECHNIQUES There is no one ideal classification of land use and land cover, and it is unlikely that one could ever be developed. There are different perspectives in the classification process, and the process itself tends to be subjective, even when an objective numerical ap- proach is used. There is, in fact, no logical reason to expect that one detailed inventory should be adequate for more than a short time, since land use and land cover patterns change in keeping with demands for natural resources. Each classification is made to suit the needs of the user, and few users will be satisfied with an inventory that does not meet most of their needs. In attempting to develop a classification sys- tem for use with remote sensing techniques that Will provide a framework to satisfy the needs of the majority of users, certain guidelines of criteria for evaluation must first be established. To begin with, there is considerable diversity of opinion about what constitutes land use, although present use of land is one of the characteristics that is widely recognized as significant for planning and management purposes. One concept that has much merit is that land use refers to, “man’s activities on land which are directly related to the land” (Claw- son and Stewart, 1965). Land cover, on the other hand, describes, “the vegetational and artificial con- structions covering the land surface” (Burley, 1961). ' The types of land use and land cover categoriza- tion developed in the classification system presented in this report can be related to systems for classify- ing land capability, vulnerability to certain manage- ment practices, and potential for any particular ac- tivity or land value, either intrinsic or speculative. Concepts concerning land“ cover and land use ac- tivity are closely related and in many cases have been used interchangeably. The purposes for which lands are being used commonly have associated types of cover, whether they be forest, agricultural, resi- dential, or industrial. Remote sensing image-form- ing devices do not record activity directly. The remote sensor acquires a response which is based on many characteristics of the land surface, including natural or artificial cover. The interpreter uses pat- terns, tones, textures, shapes, and site associations to derive information about land use activities from What is basically information about land cover. Some activities of man, however, cannot be direct- ly related to the type of land cover. Extensive recrea- tional activities covering large tracts of land are not particularly amenable to interpretation from remote sensor data. For example, hunting is a very common and pervasive recreational use of land, but hunting usually occurs on land that would be classified as some type of forest, range, or agricultural land either during ground survey or image interpretation. Consequently, supplemental information is needed to identify lands used for hunting. Supplemental in- formation such as land ownership maps also is neces- sary to determine the use of lands such as parks, game refuges, or water-conservation districts, which may have land uses coincident with administrative boundaries not usually discernable by inventory using remote sensor data. For these reasons, types of land use and land cover identifiable primarily from remote sensor data are used as the basis for organiz- ing this classification system. Agencies requiring more detailed land use information may need to employ more supplemental data. In almost any classification process, it is rare to find the clearly defined classes that one would like. In determining land cover, it would seem simple to draw the line between land and water until one con- siders such problems as seasonally wet areas, tidal flats, or marshes with various kinds of plant cover. Decisions that may seem arbitrary must be made at times, but if the descriptions of categories are com- plete and guidelines are explained, the inventory process can be repeated. The classification system must allow for the inclusion of all parts of the area under study and should also provide a unit of refer- ence for each land use and land cover type. The problem of inventorying and classifying multiple uses occurring on a single parcel of land will not be easily solved. Multiple uses may occur simultaneously, as in the instance of agricultural land or forest land used for recreational activities such as hunting or camping. Uses may also occur alternately, such as a major reservoir providing flood control during spring runoff and generating power during winter peak demand periods. This same reservoir may have sufficient water depth to be navigable by commercial shipping the year round and may additionally provide summer recreational opportunities. Obviously all of these activities would not be detectable on a single aerial photograph. How- ever, interpreters have occasionally related flood- control activities to drawdown easements around reservoirs detectable on imagery acquired during winter low-water levels. Similarly, major locks at water-control structures imply barge or ship traffic, and foaming tailraces indicate power generation. Pleasure-boat marinas, as well as the wakes of the CLASSIFICATION CRITERIA 5 boats themselves, can be detected on high-altitude photographs. Although each of these activities is detectable at some time using remote sensing, many other multiple-use situations cannot be interpreted with the same degree of success. The example of the reservoir does provide insight into another facet of the problem’s solution, however, and that is the pos- sibility and need for acquiring collateral data to aid in the understanding of a multiple-use situation. The vertical arrangement of many uses above and below the actual ground surface provides additional problems for the land use interpreter. Coal and other mineral deposits under croplands or forests, electrical transmission lines crossing pastures, ga— rages underground or on roofs of buildings, and sub- ways beneath urban areas all exemplify situations which must be resolved by individual users and com- pilers of land use data. The size of the minimum area which can be de- picted as being in any particular land use category depends partially on the scale and resolution of the original remote sensor data or other data source from which the land use is identified and interpreted. It also depends on the scale of‘data compilation as well as the final scale of the presentation of the land use information. In some cases, land uses can- not be identified with the level of accuracy approach- ing the size of the smallest unit mappable, while in others, specific land uses can be identified which are too small to be mapped. Farmsteads, for example, are usually not distinguished from other agricultural land uses when mapping at the more generalized levels of the classification. On the other hand, these farmsteads may well be interpretable but too small to be represented at the final format scale. Analogous situations may arise in the use of other categories. When maps are intended as the format for pre—. senting land use data, it is difiicult to represent any unit area smaller than 0.10 inch (2.54 mm) on a side. In addition, smaller areas cause legibility problems for the map reader. Users of computer-generated graphics are similarly constrained by the minimum size of the computer printout. CLASSIFICATION CRITERIA A land use and land cover classification system which can effectively employ orbital and high-alti- tude remote sensor data should meet the following criteria (Anderson, 1971) : 1. The minimum level of interpretation accuracy in the identification of land use and land cover categories from remote sensor data should be at least 85 percent. 2. The accuracy of interpretation for the several categories should be about equal. 3. Repeatable or repetitive results should be ob- tainable from one interpreter to another and from one time of sensing to another. 4. The classification system should be applicable over extensive areas. 5. The categorization should permit vegetation and other types of land cover to be used as sur- rogates for activity. 6. The classification system should be suitable for use with remote sensor data obtained at differ- ent times of the year. 7. Effective use of subcategories that can be ob- tained from ground surveys or from the use of larger scale or enhanced remote sensor data should be possible. Aggregation of categories must be possible. 9. Comparison with future land use data should be possible. 10. Multiple uses of land should be recognized when possible. 9° Some of these criteria should apply to land use and land cover classification in general, but some of the criteria apply primarily to land use and land cover data interpreted from remote sensor data. It is hoped that, at the more generalized first and second levels, an accuracy in interpretation can be attained that will make the land use and land cover data comparable in quality to those obtained in other ways. For land use and land cover data needed for planning and management purposes, the accuracy of interpretation at the generalized first and second levels is satisfactory when the interpreter makes the correct interpretation 85 to 90 percent of the time. For regulation of land use activities or for‘tax assess- ment purposes, for example, greater accuracy usual- ly will be required. Greater accuracy generally will be attained only at much higher cost. The accuracy of land use data obtained from remote sensor sources is comparable to that acquired by using enumeration techniques. For example, postenumeration surveys made by the U.S. Bureau of the Census revealed that 14 percent of all farms (but not necessarily 14 per- cent of the farmland) were not enumerated during the 1969 Census of Agriculture (Ingram and Pro- chaska, 1972) . In addition to perfecting new interpretation tech- niques and procedures for analysis, such as the vari- ous types of image enhancement and signature idenA tification, we can assume that the resolution capa- bility of the various remote sensing systems will also 6 A LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA improve. Resolution, or resolving power, of an imag- ing system refers to its ability to separate two objects some distance apart. In most land use appli- cations, we are most interested in the minimum size of an area which can be recognized as having an interpretable land use or land cover type. Obviously, such a minimumarea depends not only on the type and characteristics of the imaging system involved, but pragmatically also on the order of “generation” of the imagery, that is, how far the study image is removed in number of reproduction stages from the original record. The user should refer to the most recent information available in determining the reso- lution parameters of the system. The kind and amount of land use and land cover information that may be obtained from different sensors depend on the altitude and the resolution of each sensor. There is little likelihood that any one sensor or system will produce good data at all alti- tudes. It would be desirable to evaluate each source of remote sensing data and its application solely on the basis of the qualities and characteristics of the source. However, it is common practice to transfer the data to a base map, and no matter what the guidelines are, it is diflicult to use a base map with- out extracting some additional data from such maps. Topographic maps, road maps, and detailed city maps will generally contribute detail beyond the capabilities of the remote sensor data. The multilevel land use and land cover classifica- tion system described in this report has been devel- oped because different sensors will provide data at a range of resolutions dependent upon altitude and scale. In general, the following relations pertain, assuming a 6-inch focal length camera is used in obtaining aircraft imagery. Classification level Typical data characteristics I ______________ LANDSAT (formery ERTS) type of data. II _____________ High-altitude data at 40,000 ft (12,400 m) or above (less than 1:80,000 scale). III _____________ Medium-altitude data taken between 10,000 and 40,000 ft (3,100 and 12,400 m) (120,000 to 1:80,000 .scale) . IV ____________ Low-altitude data taken below 10,000 ft (3,100 in) (more than 120,000 scale). Although land use data obtained at any level of categorization certainly should not be restricted to any particular level of user groups nor to any par- ticular scale of presentation, information at Levels I and II would generally be of interest to users who desire data on a nationwide, interstate, or statewide basis. More detailed land use and land cover data such as those categorized at Levels III and IV usual- ly will be used more frequently by those who need and generate local information at the intrastate, re- gional, county, or municipal level. It is intended that these latter levels of categorization will be developed by the user groups themselves, so that their specific needs may be satisfied by the categories they intro- duce into the structure. Being able to aggregate more detailed categories into the categories at Level II being adopted by the U.S.G.S. is desirable if the classification system is to be useful. In general,Level II land use and land cover data interface quite effec- tively with point and line data available on the stand- ard U.S.G.S. topographic maps. This general relationship between the categoriza- tion level and the data source is not intended to restrict users to particular scales, either in the original data source from which the land use infor- mation is compiled or in the final map product or other graphic device. Level I land use information, for example, while efficiently and economically gath- ered over large areas by a LANDSAT type of satel- lite or from high-altitude imagery, could also be interpreted from conventional large-scale aircraft imagery or compiled by ground survey. This same information can be displayed at a Wide variety of scales ranging from a standard topographic map scale, such as 1 :24,000 or even larger, to the much smaller scale of the orbital imagery, such as 1:1,000,000. Similarly, several Level II categories (and, in some instances, Level III categories) have been interpreted from LANDSAT data. Presently, though, Level II categories are obtained more accur- ately from high-altitude photographs. Much Level III and Level IV land use and land cover data can also be obtained from high-altitude imagery. This level of categorization can also be presented at a wide range of scales. However, as the more detailed levels of categorization are used, more dependence necessarily must be placed on higher resolution re- mote sensor data and supplemental ground surveys. The principal remote sensor source for Level II data at the present time is high-altitude, color-infra- red photography. Scales smaller than 1:80,000 are characteristic of high-altitude photographs, but scales from 1:24,000 to 1:250,000 generally have been used for the final map products. The same photography which now is used to con— struct or update 1 : 24,000 topographic maps or ortho— photoquads at similar scales is a potential data source for inventorying land use and land cover. The orthophoto base, in particular, commonly can enable rapid interpretation of Levels I and II informa- tion at relatively low cost. The cost of acquiring more detailed levels of land use and land cover data DEVELOPING THE CLASSIFICATION SYSTEM. 7 might prohibit including such data on large-scale maps over extensive areas. Recent experiments (Stevens and others, 1974) with Levels I and II land use data referenced to 1 :24,000 topographic maps have been conducted by researchers of the Maps and Surveys Branch of the Tennessee Valley Authority in conjunction with the Marshall Space Flight Center and Oak Ridge Na- tional Laboratories. Quite satisfactory results have been obtained when interpreting land use from high- altitude photography. In areas of considerable ter- rain relief a stereoplotter was used to avoid scale problems. The categories proposed at Level II cannot all be interpreted with equal reliability. In parts of the United States, some categories may be extremely difficult to interpret from high-altitude aircraft imagery alone. Conventional aerial photography and sources of information other than remote sensor data may be needed for interpretation of especially complex areas. On the basis of research and testing carried out in the U.S.G.S. Geography Program’s Central Atlantic Regional Ecological Test Site (CARETS) Project, the Phoenix Pilot Project, and in land use mapping for the Ozarks Regional Com- mission (U.S. Geological Survey, 1973), it has been determined that the cost of using such supplemen- tary information can be held to reasonable levels. At Level III, which is beyond the scope of the present discussion, use of substantial amounts of supplemental information in addition to some re- motely sensed information at scales of 1:15,000 to 1:40,000 should be anticipated. Surprisingly de- tailed inventories may be undertaken, and by using both remotely sensed and supplemental information, most land use and land cover types, except those of very complex urban areas or of thoroughly hetero- geneous mixtures can be adequately located, meas- ured, and coded. Level IV would call for much more supplemental information and remotely sensed data at a much larger scale. DEVELOPING THE CLASSIFICATION SYSTEM In developing the classification system, every ef- fort has been made to provide as much compatibility as possible with other classification systems current- ly being used by the various Federal agencies in— volved in land use inventory and mapping. Special attention has been paid to the definitions of land use categories used by other agencies, to the extent that they are useful in categorizing data obtained from remote sensor sources. The definition of Urban or Built-up Land, for ex- ample, includes those uses similarly classified (Woo- ten and Anderson, 1957) by the U.S. Department of Agriculture, plus the built-up portions of major recreational sites, public installations, and other similar facilities. Agricultural land has been defined to include Cropland and Pasture; Orchards, Groves, Vineyards, Nurseries, and Ornamental Horticultural Areas; and Confined Feeding Operations as the prin- cipal components. Certain land uses such as pasture, however, cannot be separated consistently and ac- curately by using the remote sensor data sources appropriate to the more generalized levels of the classification. The totality of the category thus close- ly parallels the U.S. Department of Agriculture defi- nition of agricultural land. The primary definition of Forest Land employed for use with data acquired by remote sensors ap- proximates that used by the U.S. Forest Service (un- published manual), with the exception of those brush and shrub-form types such as Chaparral and mesquite, which are classed as forest land by the Forest Service because of their importance in water- shed control. Because of their spectral response, these generally are grouped with Rangeland types in classifications of vegetation interpretable from re- mote sensing imagery. The principal concept by which certain types of cover are included in the Rangeland category, and which separates rangeland from pasture land, is that rangeland has a natural climax plant cover of native grasses, forbs, and shrubs which is potential- ly useful as a grazing or forage resource (U.S. Con- gress, 1936; U.S. Department of Agriculture, 1962, 1971). Although these rangelands usually are not seeded, fertilized, drained, irrigated, or cultivated, if the forage cover is improved, it is managed pri- marily like native vegetation, and the forage re- source is regulated by varying the intensity and seasonality of grazing (Stoddard and Smith, 1955). Since the typical cropland practices mentioned just above are characteristics of some pasture lands, these pasture lands are similar in image signature to crop- land types. The definition of Wetland incorporates the major elements of the original U.S. Department of the Interior definition (Shaw and Fredine, 1956) as well as the combined efforts of the U.S.G.S. working group on wetlands definition. Table 1 presents a general summary of land use compiled every 5 years by the Economic Research 8 A LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA Service of the US. Department of Agriculture and supplemented from other sources. These statistics, ‘ which are available only for States, are provided by the various government agencies which compile in- formation on some categories of land use, several of which parallel the U.S.G.S. land use classification system. TABLE 1.—Major uses of land, United States, 196.9 1 Acres Hectares (mil- (mil- Per- lions) lions) cent Cropland _______________________ 472 191 20.9 Cropland used for crops ___- W'T—T— Cropland harvested ______ 286 116 ___ Crop failure ____________ 6 2 ___ Cultivated summer fallow- 41 17 ___ Soil improvement crops and idle cropland _____________ 51 21 __- Cropland used only for pasture 88 35 ___ Grassland pasture and range2| ___- 604 245 26.7_ Forest land _____________________ 723 293 31.9 Grazed _____________________ 198 80 ___ Not grazed _________________ 525 213 ___ Special uses 3 ___________________ 178 72 7.9__ Urban areas ________________ 35 14 ___ Transportation areas ________ 26 11 ___ Rural parks ________________ 49 19 ___ Wildlife refuges ____________ 32 13 ___ National defense, flood control, and industrial areas _______ 26 11 ___ State-owned institutions and miscellaneous other uses ___ 2 1 ___ Farmsteads, farm roads, and lanes ________________ 8 3 ___ Miscellaneous land‘ ______________ 287 116 12L 1Frey, H. T., 1973. Does not include area covered by water in streams more than % of a mile in width and. lakes, reservoirs, and so forth of more than 40 acres in size. 2Includes pasture that is to be included with cropland in the U.S.G.S. classification system. 3Except for urban and built-up areas and transportation uses, these special uses will be classified by dominant cover under the U.S.G.S. classi- fication system. 4 Tundra, glaciers, and icefields, marshes, open swamps, bare rock areas, deserts, beaches, and other miscellaneous land. The land use and land cover classification system presented in this report (table 2) includes only the more generalized first and second levels. The system satisfies the three major attributes of the classifica- tion process as outlined by Grigg (1965): (1) it gives names to categories by simply using accepted terminology; (2) it enables information to be trans- mitted; and (3) it allows inductive generalizations to be made. The classification system is capable of further refinement on the basis of more extended and varied use. At the more generalized levels it should meet the principal objective of providing a land use and land cover classification system for use in land use planning and management activities. Attainment of the more fundamental and long-range objective of providing a standardized system of land use and land cover classification for national and regional TABLE 2.——Land use and land cover classification system for use with remote sensor data Level I Level II 1 Urban or Built-up Land 11 Residential. 12 Commercial and Services. 13 Industrial. 14 Transportation, Communi- cations, and Utilities. 15 Industrial and Commercial Complexes. 16 Mixed Urban or Built-up Land. 17 Other Urban or Built-up Land. 2 Agricultural Land 21 Cropland and Pasture. 22 Orchards, Groves, Vine- yards, Nurseries, and Ornamental Horticultural Areas. 23 Confined Feeding Opera- .tions. 24 Other Agricultural Land. 3 Rangeland 31 Herbaceous Rangeland. 32 Shrub and Brush Range- land. 33 Mixed Rangeland. 4 Forest Land 41 Deciduous Forest Land. 42 Evergreen Forest Land. 43 Mixed Forest Land. 5 Water 51 Streams and Canals. 52 Lakes. 53 Reservoirs. 54 Bays and Estuaries. 6 Wetland 61 Forested Wetland. 62 Nonforested Wetland. 7 Barren Land 71 Dry Salt Flats. ‘ 72 Beaches. 73 Sandy Areas other than Beaches. 74 Bare Exposed Rock. 75 Strip Mines. Quarries, and Gravel Pits. 76 Transitional Areas. 77 Mixed Barren Land. 8 Tundra 81 Shrub and Brush Tundra. 82 Herbaceous Tundra. 83 Bare Ground Tundra. 84 Wet Tundra. 85 Mixed Tundra. 9 Perennial Snow or Ice 91 Perennial Snowfields. 92’ Glaciers. studies will depend on the improvement that should result from widespread use of the system. As further advances in technology are made, it may be necessary to modify the classification system for use with automatic data analysis. The LANDSAT and Skylab missions and the high-altitude aircraft program of the National Aeronautics and Space Ad- ministration have offered opportunities for nation- wide testing of the feasibility of using this classifica- tion system to obtain land use information on a uniform basis. The approach to land use and land cover classifi- cation embodied in the system described herein is “resource oriented,” in contrast, for example, with the “people orientation” of the “Standard Land Use USING THE CLASSIFICATION SYSTEM 9 Coding Manual,” developed by the US. Urban Re- newal Administration and the Bureau of Public Roads (1965). For the most part the Manual is derived from the “Standard Industrial Classification Code” established and published by the former Bu- reau of the Budget (U.S. Executive Oflice of the President, 1957). The people-oriented system of the “Standard Land Use Coding Manual” assigns seven of the nine gen- eralized first level categories to urban, transporta- tion, recreational, and related uses of land, which account for less than 5 percent of the total area of the United States (tables 1 and 3). Although there is an obvious need for an urban-oriented land use classification system, there is also a need for a resource-oriented classification system Whose pri- mary emphasis would be the remaining 95 percent of the United States land area. The U.S.G.S. classifica— tion system described in this report addresses that need, with eight of the nine Level I categories treat- ing land area of the United States that is not in urban or built-up areas. Six of the first level cate- gories in the standard land use code are retained under Urban or Built-up at Level II in the U.S.G.S. system. Even though the standard land use code and the U.S.G.S. classification differ considerably in their major emphases, a marked degree of compatibility between these two systems exists at the more gen- eralized levels and even at the more detailed levels. TABLE 3.——Stanolard land use code—first level categories1 1 Residential. 2 Manufacturing (9 second level categories included). 3 Manufacturing (6 second level categories included). 4. Transportation, communications, and utilities. 5. Trade. 6. Services. 7 Cultural, entertainment, and recreation. 8. Resource production and extraction. 9. Undeveloped land and water areas. 1 S tandard land use coding manual, 1965,12. 29. USING THE CLASSIFICATION SYSTEM The use of the same or similar terminology does not automatically guarantee that the land use data collected and coded according to two systems will be entirely compatible. The principal points of depar- ture between other classifications and the U.S.G.S. system originate because of the emphasis placed on remote sensing as the primary data source used in the U.S.G.S. classification system. Because of this emphasis, activity must be interpreted using land cover as the principal surrogate, in addition to the image interpreter’s customary references to pattern, geographic location, and so forth. This process neces- sarily precludes the possibility of information being generated which identifies ownership-management units such as farms or ranches or relating detached uses, included in a specific ownership complex, to the parent activity. For example, warehouses cannot be related to retail sales when the two occurrences are separated spatially. The actual cover and related uses are mapped in each case, rather than injecting inference into the inventory process. Inferences used for prediction could cause prob- lems for the land use interpreter where land use is clearly in transition, with neither the former use nor the future use actually being present. In most such cases, it is tempting to speculate on future use, but all that can actually be determined in such wide- ranging situations is that change is occurring. Large clear-cut areas in the southeastern forests, for ex- ample, are not always returned to forests and might assume any of a variety of future uses, such as a residential subdivision, an industrial site, an area of cropland, or a phosphate mine. The “sagebrush sub- division” of the Southwest may have all the potential earmarks of future settlement, such as carefully platted streets, and yet never experience any con- struction. Such cleared open areas should be identi- fied as “Transitional Areas.” Since Level II will probably be most appropriate for statewide and interstate regional land use and land cover compilation and mapping, and since Level II categories can be created by aggregating similar Level III categories, the Level II categorization may be considered to be the fulcrum of the classification system. The classification system may be entered at the particular level appropriate to the individual user, and the information generated may be added together with data generated by others to form an aggregate category at the next higher level. As an example, if a local planning group had devised a Level III classification of a particular group of land uses and had included sufficient definitional informa- tion of their land use categories, their data could be compiled into a larger inventory by a state or re- gional planning group compiling data by use of the Level II categories. Such data,.in turn, could serve as part of the data base for a national inventory. Seldom is it necessary to inventory land uses at the more detailed levels, even for local planning. Having greater detail does, however, provide flexi- bility in manipulating the data when several differ- ent purposes must be served. The cost of interpret- ing, coding, and recording land use data at the more detailed levels is necessarily greater than if the data were handled at more generalized levels. This extra cost reflects the increase in cost of remote sensor and 10 A LAND USE AND LAND COVER CLASSIFICAT collateral data acquired at larger scales, as well as the increase in interpretation costs. The U.S.G.S. classification system provides flexi- bility in developing categorization at the more de- tailed levels. Therefore, it is appropriate to illustrate the additive properties of the system and to provide examples for users wishing to develop more detailed} categorization. The several examples given below represent possible categorizations. Users should not consider themselves limited to categories such as these but should develop categories of utmost utility; to their particular needs. It should be emphasize that, whatever categories are used at the variou classification levels, special attention should be give to providing the potential users of the data with sufii‘ cient information so that they may either compile the data into more generalized levels or aggregate more detailed data into the existing classes. 3 One example of subcategorization of Residential Land as keyed to the standard land use code wouli be: Level I Level 11 Level III L. Urban or 11. Residential. 111. Single-family Units. Built-up 112. Multi—family Units 113. Group Quarters. 114. Residential Hotels. 115. Mobile Home Park ;. 116. Transient Lodging 5.. 117. Other. This particular breakdown of “Residential” em- ploys criteria of capacity, type, and permanency (f residence as the discriminating factors amorg classes. Criteria applied to other situations could pos- sibly include density of dwellings, tenancy, age of construction, and so forth. Obviously, such a Level III categorization would require use of supplemental information. Users desiring Level IV informatim could employ a variety of additional criteria in d s- criminating among land uses,but it can be seen that the element which allows aggregation and transfer between categories is the proper description of what is included in each individual category at whatever level the data are being classified. The Level II category, Cropland and Pasture, may be simply subdivided at Level III. Level II 21. Cropland and Pasture. Level III 211. Cropland. 212. Pasture. Some users may wish such additional criteria em- ployed at Level III as degree of activity or idleness or degree of improvement, while others may place such items in Levels IV or V. What may be a primary category for one user group may be of secondary importance to another. As stated by Clawson 1nd [ON SYSTEM FOR USE WITH REMOTE SENSOR DATA Stewart (1965), “One man’s miscellany is another man’s prime concern.” No one would consider pub- lishing a map of current land use of any part of the Western United States without having irrigated land as a major category. With the flexibility inher- ent in this classification system, an accommodation of this type of need can be made easily, provided that irrigated land is mapped or tabulated as a dis- crete unit which can be aggregated into the more general categories included in the framework of the classification. A possible restructuring which would accommodate the desire to present irrigated land as a major category would be: Irrigated agricultural land Cropland Pasture Orchards, Groves and so forth Nonirrigated agricultural land Cropland Pasture Orchards, Groves and so forth DEFINITIONS An attempt has been made to include sufficient detail in the definitions presented here to provide a general understanding of what is included in each category at Levels I and 11. Many of the uses de- scribed in detail will not be detectable on small-scale aerial photographs. However, the detail will aid in the interpretation process, and the additional infor- mation will be useful to those who have large-scale aerial photographs and other supplemental informa- tion available. 1. URBAN OR BUILT-UP LAND Urban or Built—up Land is comprised of areas of intensive use with much of the land covered by struc- tures. Included in this category are cities, towns, vil- lages, strip developments along highways, transpor- tation, power, and communications facilities, and areas such as those occupied by mills, shopping cen- ters, industrial and commercial complexes, and insti- tutions that may, in some instances, be isolated from urban areas. As development progresses, land having less inten- sive or nonconforming use may be located in the midst of Urban or Built—up areas and will generally be included in this category. Agricultural land, for- est, wetland, or water areas on the fringe of Urban or Built-up areas will not be included except where they are surrounded and dominated by urban devel- opment. The Urban or Built-up category takes prece- dence over others when the criteria for more than one category are met. For example, residential areas that have sufficient tree cover to meet Forest Land criteria will be placed in the Residential category. DEFINITIONS 11 _11. RESIDENTIAL Residential land uses range from high density, represented by the multiple-unit structures of urban cores, to low density, where houses are on lots of more than an acre, on the periphery of urban expan- sion. Linear residential developments along trans- portation routes extending outward from urban areas should be included as residential appendages to urban centers, but care must be taken to dis- tinguish them from commercial strips in the same locality. The residential strips generally have a uni- form size and spacing of structures, linear drive- ways, and lawn areas; the commercial strips are more likely to have buildings of different sizes and spacing, large driveways, and parking areas. Resi— dential development along shorelines is also linear and sometimes extends back only one residential parcel from the shoreline to the first road. Areas of sparse residential land use, such as farm- steads, will be included in categories to which they are related unless an appropriate compilation scale is being used to indicate such uses separately. Rural residential and recreational subdivisions, however, are included in this category, since the land is almost totally committed to residential use, even though it may have forest or range types of cover. In some places, the boundary will be clear where new housing developments abut against intensively used agricul- tural areas, but the boundary may be vague and diffi- cult to discern when residential development occurs in small isolated units over an area of mixed or less intensive uses. A careful evaluation of density and the overall relation of the area to the total urban complex must be made. Residential sections which are integral parts of other uses may be difiicult to identify. Housing situa— tions such as those existing on military bases, at col- leges and universities, living quarters for laborers near a work base, or lodging for employees of agri- cultural field operations or resorts thus would be placed within the Industrial, Agricultural, or Com- mercial and Services categories. 12. COMMERCIAL AND SERVICES Commercial areas are those used predominantly for the sale of products and services. They are often abutted by residential, agricultural, or other con- trasting uses which help define them. Components of the Commercial and Services category are urban cen- tral business districts; shopping centers, usually in suburban and outlying areas; commercial strip de- velopments along major highways and access routes to cities; j unkyards; resorts; and so forth. The main buildings, secondary structures, and areas support- ing the basic use are all included—office buildings, warehouses, driveways, sheds, parking lots, land- scaped areas, and waste disposal areas. Commercial areas may include some noncommer— cial uses too small to be separated out. Central busi- ness districts commonly include some institutions, such as churches and schools, and commercial strip developments may include some residential units. When these noncommercial uses exceed one-thirdof the total commercial area, the Mixed Urban or Built- up category should be used. There is no separate category for recreational land uses at Level II since most recreational activity is pervasive throughout many other land uses. Selected areas are predomi- nantly recreation oriented, and some of the more dis- tinctive occurrences such as drive-in theaters can. be identified on remote sensor imagery. Most recrea- tional activity, however, necessarily will be identified using supplemental information. Recreational facili- ties that form an integral part of an institution should be included in this category. There is usually a major Visible difference in the form of parking facilities, arrangements for traffic flow, and the gen- eral association of buildings and facilities. The in- tensively developed sections of recreational areas would be included in the Commercial and Services category, but extensive parts of golf courses, riding areas, ski areas, and so forth would be included in the Other Urban or Built-up category. Institutional land uses, such as the various educa- tional, religious, health, correctional, and military facilities are also components of this category. All buildings, grounds, and parking lots that compose the facility are included within the institutional unit, but areas not specifically related to the purpose of the institution should be placed in the appropriate category. Auxiliary land uses, particularly residen- tial, commercial and services, and other supporting land uses on a military base would be included in this category, but agricultural areas not specifically as- sociated with correctional, educational, or religious institutions are placed in the appropriate agricul- tural category. Small institutional units, as, for ex- ample, many churches and some secondary and ele- mentary schools, would be mappable only at large scales and will usually be included within another category, such as Residential. 13. INDUSTRIAL Industrial areas include a Wide array of land uses from light manufacturing to heavy manufacturing 12 A LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA plants. Identification of light industries—those fo- cused on design, assembly, finishing, processing, and packaging of products—can often be based on the type of building, parking, and shipping arrange- ments. Light industrial areas may be, but are not necessarily, directly in contact with urban areas; many are now found at airports or in relatively open country. Heavy industries use raw materials such as iron ore, timber, or coal. Included are steel mills, pulp and lumber mills, electric-power generating stations, oil refineries and tank farms, chemical plants, and brickmaking plants. Stockpiles of raw materials and waste-product disposal areas are usu- ally visible, along with transportation facilities capable of handling heavy materials. Surface structures associated with mining opera- tions are included in this category. Surface struc- tures and equipment may range from a minimum of a loading device and trucks to extended areas with access roads, processing facilities, stockpiles, storage sheds, and numerous vehicles. Spoil material and slag heaps usually are found within a short trucking dis- tance of the major mine areas and may be the key indicator of underground mining operations. Uni- form identification of all these diverse extractive uses is extremely difficult from remote sensor data alone. Areas of future reserves are included in the appropriate present-use category, such as Agricul- tural Land or Forest Land, regardless of the ex- pected future use. 14. TRANSPORTATION, COMMUNICATIONS, AND UTILITIES The land uses included in the Transportation, Communications, and Utilities category occur to some degree within all of the other Urban or Built— up categories and actually can be found within many other categories. Unless they can be mapped sepa- rately at whatever scale is being employed, they usually are considered an integral part of the land use within which they occur. For that reason, any statistical summary of the area of land uses in this category typically represents only a partial data set. Statistical area summaries of such land uses aggre— gated from Levels III and IV, though, would include more accurate area estimates. Major transportation routes and areas greatly influence other land uses, and many land use bound- aries are outlined by them. The types and extent of transportation facilities in a locality determine the degree of access and affect both the present and po- tential use of the area. Highways and railways are characterized by areas of activity connected in linear patterns. The high- ways include rights-of-way, areas used for inter- changes, and service and terminal facilities. Rail facilities include stations, parking lots, roundhouses, repair and switching yards, and related areas, as well as overland track and spur connections of suffi- cient width for delineation at mapping 'scale. Airports, seaports, and major lakeports are iso- lated areas of high utilization, usually with no well- defined intervening connections, although some ports are connected by canals. Airport facilities include the runways, intervening land, terminals, service buildings, navigation aids, fuel storage, parking lots, and a limited buffer zone. Terminal facilities general- ly include the associated freight and warehousing functions. Small airports (except those on rotated farmland), heliports, and land associated with sea- plane bases may be identified if mapping scale per— mits. Port areas include the docks, shipyards, dry- docks, locks, and waterway control structures. Communications and utilities areas such as those involved in processing, treatment, and transporta- tion of water, gas, oil, and electricity and areas used for airwave communications are also included in this category. Pumping stations, electric substations, and areas used for radio, radar, or television antennas are the major types. Small facilities, or those associ— ated with an industrial or commercial land use, are included within the larger category with which they are associated. Long-distance gas, oil, electric, tele- phone, water, or other transmission facilities rarely constitute the dominant use of the lands with which they are associated. 15. INDUSTRIAL AND COMMERCIAL COMPLEXES The Industrial and Commercial Complexes cate- gory includes those industrial and commercial land uses that typically occur together or in close func- tional proximity. Such areas commonly are labeled with terminology such as “Industrial Park,” but since functions such as warehousing, wholesaling, and occasionally retailing may be found in the same structures or nearby, the more inclusive category title has been adopted. Industrial and Commercial complexes have a defi- nite remote sensor image signature which allows their separation from other Urban or Built-up land uses. Because of their intentional development as dis- crete units of land use, they may border on a wide variety of other land use types, from Residential Land to Agricultural Land to Forest Land. If the separate functions included in the category are iden- DEFINITIONS l3 tified at Levels III or IV using supplemental data Or with ground survey, the land use researcher has the discretion of aggregating these functions into the appropriate Level II Urban or Built-up categories or retaining the unit as an Industrial and Commercial Complex. 16. MIXED URBAN OR BUILT-UP LAND The Mixed Urban or Built-up category is used for a mixture of Level II Urban or Built-up uses where individual uses cannot be separated at mapping scale. Where more than one-third intermixture of another use or uses occurs in a specific area, it is classified as Mixed Urban or Built-up Land. Where the inter- mixed land use or uses total less than one-third of the specific area, the category appropriate to the dominant land use is applied. This category typically includes developments along transportation routes and in cities, towns, and built-up areas where separate land uses cannot be mapped individually. Residential, Commercial, In- dustrial, and occasionally other land uses may be included. A mixture of industrial and commercial uses in Industrial and Commercial Complexes as de- fined in category 15 are not included in this category. Farmsteads intermixed with strip or cluster settle- ments will be included within the built-up land, but other agricultural land uses should be excluded. 17. OTHER URBAN OR BUILT-UP LAND Other Urban or Built-up Land typically consists of uses such as golf driving ranges, zoos, urban parks, cemeteries, waste dumps, water-control structures and spillways, the extensive parts of such uses as golf courses and ski areas, and undeveloped land within an urban setting. Open land may be in very intensive use but a use that does not require struc— tures, such as urban playgrounds, botanical gardens, or arboreta. The use of descriptions such as “idle land,” “vacant land,” or “open land” should be avoided in categorizing undeveloped lands within urban areas on the basis of the use of remote sensor data, since information generally is not available to the interpreter to make such a refinement in categorization. 2. AGRICULTURAL LAND Agricultural Land may be defined broadly as land used primarily for production of food and fiber. On high-altitude imagery, the chief indications of agri- cultural activity will be distinctive geometric field and road patterns on the landscape and the traces produced by livestock or mechanized equipment. However, pasture and other lands where such equip- ment is used infrequently may not show as well— defined shapes as other areas. These distinctive geo- metric patterns are also characteristic of Urban or Built-up Lands because of street layout and develop- ment by blocks. Distinguishing between Agricultural and Urban or Built-up Lands ordinarily should be possible on the basis of urban-activity indicators and the associated concentration of population. The num- ber of building complexes is smaller and the density of the road and highway network is much lower in Agricultural Land than in Urban or Built-up Land. Some urban land uses, such as parks and large ceme- teries, however, may be mistaken for Agricultural Land, especially when they occur on the periphery of the urban areas. The interface of Agricultural Land with other categories of land use may sometimes be a transition zone in which there is an intermixture of land uses at first and second levels of categorization. Where farming activities are limited by wetness, the exact boundary also may be difl‘icult to locate, and Agricul- tural Land may grade into Wetland. When the pro- duction of agricultural crops is not hindered by wet- land conditions, such cropland should be included in the Agricultural category. This latter stipulation also includes those cases in which agricultural crop production depends on wetland conditions, such as the flooding of ricefields or the development of cran- berry bogs. When lands produce economic commodi- ties as a function of their Wild state such as wild rice, cattails, or certain forest products 'commonly associated with wetland, however, they should be in- cluded in the Wetland category. Similarly, when wet- lands are drained for agricultural purposes, they should be included in the Agricultural Land cate- gory. When such drainage enterprises fall into dis- use and if wetland vegetation is reestablished, the land reverts to the Wetland category. The Level II categories of Agricultural Land are: Cropland and Pasture; Orchards, Groves, Vineyards, Nurseries, and Ornamental Horticultural Areas; Confined Feeding Operations; and Other Agricul- tural Land. 21. CROPLAND AND PASTURE The several components of Cropland and Pasture now used for agricultural statistics include: crop- land harvested, including bush fruits; cultivated summer-fallow and idle cropland; land on which crop failure occurs; cropland in soil-improvement grasses and legumes; cropland used only for pasture in rotation with crops; and pasture on land more or 14 A LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA less permanently used for that purpose. From imag- ery alone, it generally is not possible to make a dis- tinction between Cropland and Pasture with a high degree of accuracy and uniformity, let alone a dis- tinction among the various components of Cropland (Hardy, Belcher, and Phillips, 1971). Moreover, some of the components listed represent the condi- tion of the land at the end of the growing season and will not apply exactly to imagery taken at other times of the year. They will, however, be a guide to identification of Cropland and Pasture. Brushland in the Eastern States, typically used to some extent for pasturing cattle, is included in the Shrub-Brush- land Rangeland category since the grazing activity is usually not discernible on remote sensor imagery appropriate to Levels I and II. This activity possibly might be distinguished on low-altitude imagery. Such grazing activities generally occur on land where crop production or intensive'pasturing has ceased, for any of a variety of reasons, and which has grown up in brush. Such brushlands often are used for grazing, somewhat analogous to the extensive use of rangelands in the West. Certain factors vary throughout the United States, and this variability also must be recognized; field size depends on topography, soil types, sizes of farms, kinds of crops and pastures, capital invest- ment, labor availability, and other conditions. Irri~ gated land in the Western States is recognized easily in contrast to Rangeland, but in the Eastern States, irrigation by use of overhead sprinklers generally cannot be detected from imagery unless distinctive circular patterns are created. Drainage or water con- trol on land used for cropland and pasture also may create a recognizable pattern that may aid in identi- fication of the land use. In areas of quick-growing crops, a field may appear to be in nonagricultural use unless the temporary nature of the inactivity is recognized. 22. ORCHARDS, GROVES, VINEYARDS, NURSERIES, AND ORNAMENTAL HORTICULTURAL AREAS Orchards, groves, and vineyards produce the vari- ous fruit and nut crops. Nurseries and horticultural areas, which include floricultural and seed-and-sod areas and some greenhouses, are used perennially for those purposes. Tree nurseries which provide seed- lings for plantation forestry also are included, here. Many of these areas may be included in another cate- gory, generally Cropland and Pasture, when identifi- cation is made by use of small—scale imagery alone. Identification may be aided by recognition of the combination of soil qualities, topography, and local climatological factors needed for these operations: water bodies in close proximity which moderate the effects of short duration temperature fluctuations; site selection for air drainage on sloping land; and deep well-drained soils on slopes moderate enough to permit use of machinery. Isolated small orchards, such as the fruit trees on the family farm, usually are not recognizable on high-altitude imagery and are, therefore, not included. 23. CONFINED FEEDING OPERATIONS Confined Feeding Operations are large, specialized livestock production enterprises, chiefly beef cattle feedlots, dairy operations with confined feeding, and large poultry farms, but also including hog feedlots. These operations have large animal populations re- stricted to relatively small areas. The result is a con- centration of waste material that is an environmental concern. The waste-disposal problems justify a sepa- rate category for these relatively small areas. Con- fined Feeding Operations have a built-up appear- ance, chiefly composed of buildings, much fencing, access paths, and waste-disposal areas. Some are located near an urban area to take advantage of transportation facilities and proximity to process- ing plants. , Excluded are shipping corrals and other tempo- rary holding facilities. Such occurrences as thor- oughbred horse farms generally do not have the animal population densities which would place them in this category. 24. OTHER AGRICULTURAL LAND Other land uses typically associated with the first three categories of Agricultural Land are the princi- pal components of the Other Agricultural Land cate- gory. They include farmsteads, holding areas for livestock such as corrals, breeding and training fa— cilities on horse farms, farm lanes and roads, ditches and canals, small farm ponds, and similar uses. Such occurrences generally are quite small in area and often uninterpretable by use of high—altitude data. Even when they are interpretable from such data, it may not be feasible to mapthem at smaller presenta- tion scales, which generally results in their inclusion with adjacent agricultural use areas. This category should also be used for aggregating data for land uses derived at more detailed levels of classification. 3. RANGELAND Rangeland historically has been defined as land where the potential natural vegetation is predomi- nantly grasses, grasslike plants, forbs, or shrubs and DEFINITIONS 15 where natural herbivory was an important influence in its precivilization state. Management techniques which associate soil, water, and forage-vegetation resources are more suitable for rangeland manage- ment than are practices generally used in managing pastureland. Some rangelands have been or may be seeded to introduced or domesticated plant species. Most of the rangelands in the United States are in the western range, the area to the west of an irregu- lar north-south line that cuts through the Dakotas, Nebraska, Kansas, Oklahoma, and Texas. Range— lands also are found in certain places historically not included in the western range, such as the Flint Hills, the Southeastern States, and Alaska. The his- torical connotation of Rangeland is expanded in this classification to include those areas in the Eastern States which commonly are called brushlands. The Level 11 categories of Rangeland are: Herba- ceous Range, Shrub and Brush Rangeland, and Mixed Rangeland. 31. HERBACEOUS RANGELAND The Herbaceous Rangeland category encompasses lands dominated by naturally occurring grasses and forbs as well as those areas of actual rangeland which have been modified to include grasses and forbs as their principal cover, when the land is man- aged for rangeland purposes and not managed using practices typical of pastureland. It includes the tall grass (or true prairie), short grass, bunch grass or palouse grass, and desert grass regions. Respective— ly, these grass regions represent a sequence of de- clining amounts of available moisture. Most of the tall grass region has been plowed for agriculture and the remaining tall grass range is now in North Da— kota, Nebraska, southern Kansas and Oklahoma, and the Texas Coastal Plain. Short grass rangeland oc- curs in a strip about 300 miles (500 km) wide from the Texas Panhandle northward to the Dakotas where it widens to cover the western half of the Dakotas, the eastern three-fourths of Montana, and the eastern third of Wyoming. Bunch grass and desert grass are found in many locations, represent- ing transitional situations to desert shrub. Typical occurrences of grasslands include such species as the various bluestems (Andropogon) , grama grasses (Bouteloua) , wheatgrasses (Agropyron) , needle- grasses (Stipa), and fescues (Festuca). This category also includes the palmetto prairie areas of south-central Florida, which consist mainly of dense stands of medium length and tall grasses such as wiregrass (Aristida stricta) and saw pal- mettos (Seronoa ripens), interspersed occasional palms (Sabal palmetto), and shrubs (Shelford, 1963). Those palmetto prairie areas now in im- proved pasture would not be included in this cate- gory, nor would the herbaceous varieties of tundra vegetation. 32. SHRUB AND BRUSH RANGELAND The typical shrub occurrences are found in those arid and semiarid regions characterized by such xerophytic vegetative types with woody stems as big sagebrush (Artemisia tridentata), shadscale (Atri- plex confertifolia), greasewood (Sarcobatus vermi- culatus), or creosotebush (Lawea divam‘cata) and also by the typical desert succulent xerophytes, such as the various forms of Cactus (Kuchler, 1964). When bottom lands and moist flats are characterized by dense stands of typical wetland species such as mesquite (Prosopis), they are considered Wetland. Where highly alkaline soils are present, halophytes such as desert saltbush (Atriplex) may occur. The type, density, and association of these various species are useful as indicators of the local hydrologic and pedologic environments. Also included in this cate- gory are Chaparral, a dense mixture of broadleaf evergreen schlerophyll shrubs, and the occurrences of mountain mahogany (Cercocarpus ledifolius) and scrub oaks (Quercus). The eastern brushlands are typically former crop- lands or pasture lands (cleared from original forest land) which now have grown up in brush in transi- tion back to forest land to the extent that they are no longer identifiable as cropland or pasture from re- mote sensor imagery. Many of these brushlands are grazed in an extensive manner by livestock and pro- vide wildlife habitat. These areas usually remain as part of the farm enterprise, even though not being used at their former levels of intensity. Eastern brushland areas traditionally have not been included in the rangeland concept because of their original forested state prior to clearing for cropland or pas- ture and generally have been summarized statistical- ly with pastureland. Because they function now pri- marily as extensive grazing land, they are included here as part of the Rangeland category. After suffi- cient forest growth has occurred, they should be classified as either Deciduous, Evergreen, or Mixed Forest Land. Those occurrences of shrubs and brush which are part of the Tundra are not included under Rangeland. 3 3. MIXED RANGELAND When more than one-third intermixture of either herbaceous or shrub and brush rangeland species oc- curs in a specific area, it is classified as Mixed 16 A LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA Rangeland. Where the intermixed land use or uses total less than one-third of the specific area, the category appropriate to the dominant type of Range— land is applied. Mixtures of herbaceous and shrub or brush tundra plants are not considered Rangeland. 4. FOREST LAND Forest Lands have a tree-crown areal density (crown closure percentage) of 10 percent or more, are stocked with trees capable of producing timber or other wood products, and exert an influence on the climate or water regime. Forest Land generally can be identified rather easily on high-altitude imag- ery, although the boundary between it and other categories of land may be difficult to delineate precisely. Lands from which trees have been removed to less than 10 percent crown closure but which have not been developed for other uses also are included. For example, lands on which there are rotation cy- cles of clearcutting and blockplanting are part of Forest Land. On such lands, when trees reach mar- ketable size, which for pulpwood in the Southeastern United States may occur in 2 to 3 decades, there will be large areas that have little or no visible forest growth. The pattern can sometimes be identified by the presence of cutting operations in the midst of a large expanse of forest. Unless there is evidence of other use, such areas of little or no forest growth should be included in the Forest Land category. Forest land which is grazed extensively, as in the Southeastern States, would be included in this cate- gory because the dominant cover is forest and the dominant activities are forest related. Such activities could form the basis for Levels III or IV categoriza- tion. Lands that meet the requirements for Forest Land and also for an Urban or Built-up category should be placed in the latter category. The only exceptions in classifying Forest Land are those areas which would otherwise be classified as Wetland if not for the forest cover. Since the wet condition is of much interest to land managers and planning groups and is so important as an environmental surrogate and control, such lands are classified as Forested Wetland. Auxiliary concepts associated with Forest Land, such as wilderness reservation, water conservation, or ownership classification, are not detectable using remote sensor data. Such concepts may be used for creating categories at the more detailed levels when supplemental information is available. At Level II, Forest Land is divided into three categories: Deciduous, Evergreen, and Mixed. To differentiate these three categories effectively, se- quential data, or at least data acquired during the period when deciduous trees are bare, generally will be necessary. 41. DECIDUOUS FOREST LAND Deciduous Forest Land includes all forested areas having a predominance of trees that lose their leaves at the end of the frost-free season or at the begin- ning of a dry season. In most parts of the United States, these would be the hardwoods such as oak (Quercus), maple (Acer), or hickory (Ca/rya) and the “soft” hardwoods, such as aspen (Populus tremu- loides) (Shelford, 1963). Tropical hardwoods are included in the Evergreen Forest Land category. Deciduous forest types characteristic of Wetland, such as tupelo (Nyssa) or cottonwood (Populus deltoides) , also are not included in this category. 42. EVERGREEN FOREST LAND Evergreen Forest Land includes all forested areas in which the trees are predominantly those which remain green throughout the year. Both coniferous and broad-leaved evergreens are included in this category. In most areas, the coniferous evergreens predominate, but some of the forests of Hawaii are notable exceptions. The coniferous evergreens are commonly referred to or classified as softwoods. They include such eastern species as the longleaf pine (Pinus palustm's), slash pine (Pinus ellioti), shortleaf pine (Pinus echinata), loblolly pine (Pinus taeda), and other southern yellow pines; various spruces (Picea) and balsam fir (Abies balsamea); white pine (Pinus strobus), red pine (Pinus resino- sa), and jack pine (Pinus banksiana) ; and hemlock (Tsuga canadensz’s); and such western species as Douglas-fir (Pseudotsuga menziesii), redwood (Se- quoia sempervirens), ponderosa pine (Pinus manti- cola), Sitka spruce (Picea sitchensis), Engelmann spruce (Picea engelmanm') , western redcedar (Thu- ja plicata) , and western hemlock (Tsuga heterophyl— la) (Shelford, 1963). Evergreen species commonly associated with Wetland, such as tamarack (Lam'x lam‘cina) or black spruce (Picea marioma), are not included in this category (Kuchler, 1964). 43. MIXED FOREST LAND Mixed Forest Land includes all forested areas where both evergreen and deciduous trees are grow- ing and neither predominates. When more than one- third intermixture of either evergreen or deciduous species occurs in a specific area, it is classified as Mixed Forest Land. Where the intermixed land use DEFINITIONS 17 or uses total less than one-third of the specified area, the category appropriate to the dominant type of Forest Land is applied, whether Deciduous or Ever- green. 5. WATER The delineation‘ of water areas depends on the scale of data presentation and the scale and resolu— tion characteristics of the remote sensor data used for interpretation of land use and land cover. (Water as defined by the Bureau of the Census includes all areas within the land mass of the United States that persistently are water covered, provided that, if linear, they are at least 1/3 mile (200 m) wide and, if extended, cover at least 40 acres (16 hectares) .) For many purposes, agencies need information on the size and number of water bodies smaller than Bureau of the Census minimums. These frequently can be obtained from small-scale remote sensor data with considerable accuracy. 51. STREAMS AND CANALS The Streams and Canals category includes rivers, creeks, canals, and other linear water bodies. Where the water course is interrupted by a control struc- ture, the impounded area will be placed in the Reservoirs category. The boundary between streams and other bodies of water is the straight line across the mouth of the stream up to 1 nautical mile (1.85 km). Beyond that limit, the classification of the water body changes to the appropriate category, whether it be Lakes, Reservoirs, or Bays and Estuaries. These latter cate- gories are used only if the water body is considered to be “inland water ” and therefore included in the total area of the United, States. No category is ap- plied to waters classified as “other than inland water ” or offshore marine waters beyond the mouths of rivers '(U.S. Bureau of the Census, 1970). 52. LAKES Lakes are nonflowing, naturally enclosed bodies of water, including regulated natural lakes but ex— cluding reservoirs. Islands that are too small to delineate should be included in the water area. The delineation of a lake should be based on the areal extent of water at the time the remote sensor data are acquired. 53. RESERVOIRS Reservoirs are artificial impoundments of water used for irrigation, flood control, municipal water supplies, recreation, hydroelectric power generation, and so forth. Dams, levees, other water-control structures, or the excavation itself usually will be evident to aid in the identification, although the water-control structures themselves and spillways are included in the Other Urban or Built-up Land category. In most cases, reservoirs serve multiple purposes and may include all of the land use functions just mentioned. In certain cases like the Tennessee River, the entire length of the trunk stream is impounded. In such a situation, the stream exists as a stairstep series of impoundments with waterway, flood-con- trol, recreation, and power-generation functions but is still considered a reservoir, since the additional functions are the result of impoundment. 54-. BAYS AND ESTUARIES Bays and Estuaries are inlets or arms of the sea that extend inland. They are included in this system only when they are considered to be inland water and therefore are included within the total area of the United States. Those bay and estuarine water areas classified as “other than inland water” are not included within the total area of the United States. These “other than inland water” areas are adjacent to certain States and fall under their jurisdiction. They occur in primary bodies of water such as the Atlantic Ocean coastal waters, Chesapeake Bay, Delaware Bay, Long Island Sound, Gulf of Mexico, Pacific Ocean coastal waters, Puget Sound, the Straits of Georgia and Juan de Euca, Gulf of Alaska, Bering Sea, Arctic Ocean coastal waters, and the Great Lakes (US. Bureau of the Census, 1970). Only those bays and estuaries classified as inland water are included in this category. No category is applied to offshore waters beyond the limits of Bays and Estuaries. 6. WETLAND Wetlands are those areas where the water table is at, near, or above the land surface for a significant part of most years. The hydrologic regime is such that aquatic or hydrophytic vegetation usually is established, although alluvial and tidal flats may be nonvegetated. Wetlands frequently are associated with topographic lows, even in mountainous regions. Examples of wetlands include marshes, mudflats, and swamps situated on the shallow margins of bays, lakes, ponds, streams, and manmade impoundments such as reservoirs. They include wet meadows or perched bogs in high mountain valleys and season- ally wet or flooded basins, playas, or potholes with no surface-water outflow. Shallow water areas 18 A LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA where aquatic vegetation is submerged are classed as open water and are not included in the Wetland category. Extensive parts of some river flood plains qualify as Wetlands, as do regularly flooded irrigation over- flow areas. These do not include agricultural land where seasonal wetness or short-term flooding may provide an important co ponent of the total annual soil moisture necessary f r crop production. Areas in which soil wetness or flooding is so short-lived that no typical wetland vegetation is developed properly belong in other categories. Cultivated wetlands such as the flooded fields associated with rice production and developed cran- berry bogs are classified as Agricultural Land. Un- cultivated wetlands from which wild rice, cattails, or wood products, and so forth are harvested, or wetlands grazed by livestock, are retained in the Wetland category. Remote sensor data provide the primary source of land use and vegetative cover information for the more generalized levels of this classification system. Vegetation types and detectable surface water or soil moisture interpreted from such data provide the most appropriate means of identifying wetlands and wetland boundaries. Inasmuch as vegetation re- sponds to changes in moisture conditions, remote sensor data acquired over a period of time will allow the detection of fluctuations in wetland conditions. Ground surveys of soil types or the duration of flooding may provide supplemental information to be employed at the more detailed levels of classifica- tion. Wetland areas drained for any purpose belong to other land use and land cover categories such as Agricultural Land, Rangeland, Forest Land, or Urban or Built-up Land. When the drainage is dis- continued and such use ceases, classification may revert to Wetland. Wetlands managed for wildlife purposes may show short-term changes in land use as different management practices are used but are properly classified Wetland. Two separate boundaries are important with re- spect to wetland discrimination: the upper wetland boundary above which practically any category of land use or land cover may exist, and the boundary between wetland and open water beyond which the appropriate Water category should be employed. Forested Wetland and Nonforested Wetland are the Level 11 categories of Wetland. 61. FORESTED WETLAND Forested Wetlands are wetlands dominated by woody vegetation. Forested Wetland includes season- ally flooded bottomland hardwoods, mangrove swamps, shrub swamps, and wooded swamps in- cluding those around bogs. Because Forested Wet- lands can be detected and mapped by the use of seasonal (winter/summer) imagery, and because delineation of Forested Wetlands is needed for many environmental planning activities, they are sepa- rated from other categories of Forest Land. The following are examples of typical vegetation found in Forested Wetland. Wooded swamps and southern flood plains contain primarily cypress (Taxodium), tupelo (Nyssa), oaks (Qaercus), and red maple (Acer rubram). Mangroves (Avicenm'a and Rhizophora) are dominant in certain subtropi- cal Forested Wetland areas. Central and northern flood plains are dominated by cottonwoods (Pop- ulus), ash (Framinus), alder (Alnus), and willow (Salim). Flood plains of the Southwest may be domi- nated by mesquite (Prosom's), saltcedar (Tamar-ix), seepwillow (Baccham's), and arrowweed (Pluchea). Northern bogs typically contain tamarack or larch (Larisa), black spruce (Picea mariana), and heath shrubs (Ericaceae). Shrub swamp vegetation in- cludes alder (Alnus), willow (Salim), and button- bush (Cephalanthus occidentalis) . 62. NONFORESTED WETLAND Nonforested Wetlands are dominated by wetland herbaceous vegetation or are nonvegetated. These wetlands include tidal and nontidal fresh, brackish, and salt marshes and nonvegetated flats and also freshwater meadows, wet prairies, and open bogs. The following are examples of vegetation asso- ciated with Nonforested Wetland. Narrow-leaved emergents such as cordgrass (Spartina) and rush (Juncus) are dominant in coastal salt marshes. Both narrow-leaved emergents such as cattail (Typha), bulrush (Scirpus), sedges (Caress), sawgrass (Cladium) and other grasses (for example, Pam'- cum and Zizam'opsis miliacea), and broad-leaved emergents such as waterlily (Nuphar, Nymphea), pickerelweed (Poatedem‘a), arrow arum (Peltan- dra), arrowhead (Sagittaria), water hyacinth (Eichhomia crassipes), and alligatorweed (Altern- anthera philoxeroides) are typical of brackish to freshwater locations. Mosses (Sphagnum) and sedges (Caress) grow in wet meadows and bogs. 7. BARREN LAND Barren Land is land of limited ability to support life and in which less than one-third of the area has vegetation or other cover. In general, it is an area of thin soil, sand, or rocks. Vegetation, if present, is more widely spaced and scrubby than that in the DEFINITIONS 19 Shrub and Brush category of Rangeland. Unusual conditions, such as a heavy rainfall, occasionally result in growth of a short—lived, more luxuriant plant cover. Wet, nonvegetated barren lands are in- cluded in the Nonforested Wetland category. Land may appear barren because of man’s activ- ities. When it may reasonably be inferred from the data source that the land will be returned to its former use, it is not included in the Barren cate- gory but classified on the basis of its site and situ- ation. Agricultural land, for example, may be tem- porarily without vegetative cover because of crop- ping season or tillage practices. Similarly, industrial land may have waste and tailing dumps, and areas of intensively managed forest land may have clear- cut blocks evident. When neither the former nor the future use can be discerned and the area is obviously in a state of land use transition, it is considered to be Barren Land, in order to avoid inferential errors. Level 11 categories of Barren Land are: Dry Salt Flats, Beaches, Sandy Areas other than Beaches; Bare Exposed Rock; Strip Mines, Quarries, and Gravel Pits; Transitional Areas; and Mixed Barren Land. 71. DRY SALT FLATS Dry Salt Flats occurring on the flat-floored bot- toms of interior desert basins which do not qualify as Wetland are included in this category. On aerial photographs, Dry Salt Flats tend to appear white or light toned because of the high concentrations of salts at the surface as water has been evaporated, resulting in a higher albedo than other adjacent desert features. 72. BEACHES Beaches are the smooth sloping accumulations of sand and gravel along shorelines. The surface is stable inland, but the shoreward part is subject to erosion by wind and water and to deposition in pro- tected areas. 73. SANDY AREAS OTHER THAN BEACHES Sandy Areas other than Beaches are composed primarily of dunes—accumulations of sand trans- ported by the wind. Sand accumulations most com- monly are found in deserts although they also oc- cur on coastal plains, river flood plains, and deltas and in periglacial environments. When such sand accumulations are encountered in tundra areas, they are not included here but are placed in the Bare Ground Tundra category. 74. BARE EXPOSED ROCK The Bare Exposed Rock category includes areas of bedrock exposure, desert pavement, scarps, talus, slides, volcanic material, rock glaciers, and other accumulations of rock without vegetative cover, with the exception of such rock exposures occurring in tundra regions. 75. STRIP MINES, QUARRIES, AND GRAVEL PITS Those extractive mining activities that have sig- nificant surface expression are included in this cate- gory. Vegetative cover and overburden‘are removed to expose such deposits as coal, iron ore, limestone, and copper. Quarrying of building and decorative stone and recovery of sand and gravel deposits also result in large open surface pits. Current mining activity is not always distinguishable, and inactive, unreclaimed, and active strip mines, quarries, bor- row pits, and gravel pits are included in this cate- gory until other cover or use has been established, after which the land would be classified in accord- ance with the resulting use or cover. Unused pits or quarries that have been flooded, however, are placed in the appropriate Water category. 76. TRANSITIONAL AREAS The Transitional Areas category is intended for those areas which are in transition from one land use activity to another. They are characterized by the lack of any remote sensor information which would enable the land use interpreter to predict re- liably the future use or discern the past use. All that actually can be determined in these situations is that a transition is in progress, and inference about past or future use should be avoided. This transi- tional phase occurs when, for example, forest lands are cleared for agriculture, wetlands are drained for development, or when any type of land use ceases as areas become temporarily bare as con- struction is planned for such future uses as resi- dences, shopping centers, industrial sites, or subur- ban and rural residential subdivisions. Land being altered by filling, such as occurs in spoil dumps or sanitary landfills, also is indicative of this transi- tional phase. 77. MIXED BARREN LAND The Mixed Barren Land category is used when a mixture of Barren Land features occurs and the dominant land use occupies less than two-thirds of the area. Such a situation arises, for example, in a desert region where combinations of salt flats, sandy areas, bare rock, surface extraction, and transi— 20 A LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA tional activities could occur in close proximity and in areal extent too small for each to be included at mapping scale. Where more than one-third inter- mixture of another use or uses occurs in a specific area, it is classified as Mixed Barren Land. Where the intermixed land use or uses total less than one- third of the specific area, the category appropriate to the dominant type of Barren Land is applied. 8. TUNDRA Tundra is the term applied to the treeless regions beyond the limit of the boreal forest and above the altitudinal limit of trees in high mountain ranges. In the United States, tundra occurs primarily in Alaska, in several areas of the western high moun- tain ranges, and in small isolated locations in the higher mountains of New England and northern New York. The timber line which separates forest and tundra in alpine regions corresponds to an arctic transition zone in which trees increasingly are re- stricted to the most favorable sites. The vegetative cover of the tundra is low, dwarfed, and often forms a complete mat. These plant characteristics are in large part the result of adaptation to the physical environment—one of the most extreme on Earth, where temperatures may average above freezing only 1 or 2 months out of the year, where strong desiccating winds may occur, where great variation in solar energy received may exist, and where permafrost is encountered almost everywhere beneath the vegetative cover. The number of species in the tundra flora is rela- tively small compared with typical middle- and low- latitude flora, and this number of species decreases as the environment becomes increasingly severe with changes of latitude and altitude. The tundra vegetation consists primarily of grasses, sedges, small flowering herbs, low shrubs, lichens, and mosses. The vegetative cover‘is most luxuriant near the boreal forest, with the ground surface usually being completely covered. As the plant cover be- comes sparse, shrubs become fewer and more bare areas occur. Species diversity is lowest near the boundaries of permanent ice and snow areas, where only isolated patches of vegetation occur on the bare ground surface. , The vegetation of the tundra is closely associated with other environmental factors. Minor manmade disturbances, as well as microenvironmental changes over short distances, can have significant effects. Minor changes in available moisture or wind protec- tion, for example, can result in different plant asso- ciations. Similarly, man’s activity in the tundra may engender new drainage patterns with resultant changes in plant community or erosion character- istics (Price, 1972). The boundaries between Tundra, Perennial Snow or Ice, and Water are best determined by using images acquired in late summer. The Forest Land- Tundra boundary in the Arctic tends to be transi- tional over a wide area and characterized by either incursion of forests where site improvement occurs, as along the flood plains or river valleys, or by in- creasing environmental severity, as on exposed dry uplands. This Forest Land-Tundra boundary is much easier to delineate in alpine areas. The Barren Land- Tundra interface occurs where one or more of the environmental parameters necessary for vegetation growth is deficient and also would be determined best with late-summer imagers. Using the results of various investigations, Level II categories of Tundra based primarily on what is interpretable from remote sensor image signatures are: Shrub and Brush Tundra, Herbaceous Tundra, Bare Ground Tundra, Wet Tundra, and Mixed Tundra. 81. SHRUB AND BRUSH TUNDRA The Shrub and Brush Tundra category consists of the various woody shrubs and brushy thickets found in the tundra environment. These occur in dense-to-open evergreen and deciduous thickets, with the latter dominated by types such as the various birches (Betula), alders (Alnus), or willows (Salim), as well as many types of berry plants. Low evergreen shrub thickets are characterized by such dominant types as Empetrum and various members of the heath family, such as Cassiope, Vaccinium, and Ledum (Viereck and Little, 1972). 82. HERBACEOUS TUNDRA Herbaceous Tundra is composed of various sedges, grasses, forbs, lichens, and mosses, all of which lack woody stems. A wide variety of such herbaceous types may be found in close proximity on the tundra. Sites having sufficient moisture usually are covered with a thick mat of mosses together with sedges such as Carer and Eriophomm (cotton grass) in almost continuous and uniform tussocks, as well as other herbaceous forms such as types of bluegrass (Poa), buttercups (Rammculus), and lichens such as Cladom'a and Cetmria. Drier or more exposed sites usually trend toward a sparse moss-lichen mat. 83. BARE GROUND TUNDRA The Bare Ground Tundra category is intended for those tundra occurrences which are less than one- ————i . DEFINITIONS 21 third vegetated. It usually consists of sites visually dominated by considerable areas of exposed bare rock, sand, or gravel interspersed with low herbace- ous and shrubby plants. This type of tundra is in- dicative of the most severe environmental stress and usually occurs poleward of the areas supporting the more luxuriant herbaceous and shrub forms and on'higher mountain ridges. The various species of Dryas, such as white mountain-avens, are dominant in Arctic regions, as are the sandworts (Mimo- a/rtia) and mountainheaths (Phyllodoce). Bare Ground Tundra gradually merges with one or more of the Barren Land categories on its more severe margin. 84. WET TUNDRA Wet Tundra is usually found in areas having little topographic relief. Standing water is almost always present during months when temperatures average above the freezing level. Numerous shallow lakes are also common (Joint Federal-State Land Use Planning Commission for Alaska, 1973). Perma- frost is usually close to the surface, and various patterned ground features may be evident. Sedges (Carex) such as cotton grass are characteristically dominant, and a few shrubby plants may occur on adjacent drier sites. Rooted aquatic plants are also common. Wet Tundra is delineated best on imagery acquired in late summer. 85. MIXED TUNDRA The Mixed Tundra category is used for a mixture of the Level 11 Tundra occurrences where any particular type occupies less than two-thirds of the area of the mapping unit. Where more than one- third intermixture of another use or uses occurs in a specific area, it is classified as Mixed Tundra. Where the intermixed land cover categories total less than one-third of the specific area, the category ap- propriate to the dominant type of Tundra is applied. 9. PERENNIAL SNOW OR ICE Certain lands have a perennial cover of either snow or ice because of a combination of environ- mental factors which cause these features to sur- vive the summer melting season. In doing so, they persist as relatively permanent features on the land— scape and may be used as environmental surrogates. Snow, firn (coarse, compacted granularsnow) , or ice accumulation in these areas exceeds ablation, which is the combined loss of snow or ice mass by evapora- tion and melt-water runoff. Adjacent lands most commonly will be classed as Water, Wetland, Barren Land, or Tundra, with their common boundaries being distinguished most'readily on late summer imagery. . The terminology and nomenclature of any sub- division of Perennial Snow or Ice areas are always subject to considerable debate, but a Level II break- down into categories of Perennial Snowfields and Glaciers seems to be appropriate for use with remote sensor data. Such a subdivision is based on surface form and the presence or absence of features indi- cating glacial flow. In addition, these forms and flow features may be related to stage of develop- ment and certain periglacial or glacial processes. 91. PERENNIAL SNOWFIELDS Perennial Snowfields are accumulations of snow and firn that did not entirely melt during previous summers. Snowfields can be quite extensive and thus representative of a regional climate, or can be quite isolated and localized, when they are known by various terms, such as snowbanks. The regional snowline is controlled by general climatic conditions and closely parallels the regional 32°F (0°C) isotherm for the average temperature of the warmest summer month. The use of the term “line” is somewhat misleading, because the “snow- line” represents an irregular transitional boundary, which is determined at any single location by the combination of snowfall and ablation, variables which can change greatly within short distances be- cause of changes in local topography and slope orientation. Small isolated snowfields occurring in protected locations can develop into incipient or nivation cirques, which become gradually holloWed by the annual patterns of freezing and thawing, aided by downslope movement of rock material. They are circular to semicircular and often develop ridges of mass-wasted materials called protalus ramparts at their downslope margins. As Flint (1957) has pointed out, “Such cirques, of course, are not in themselves indication of , glaciation, they indicate merely a frost climate.” Snowfields can normally be distinguished from the following Glacier category by their relative lack of flow features. 92. GLACIERS Glacial ice originates from the compaction of snow into firn and finally to ice under the weight of sev- eral successive annual accumulations. Refrozen melt water usually contributes to the increasing density of the glacial ice mass. With .sufi‘icient thickness, .. i 22 A LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA weight, and bulk, flow begins, and all glaciers ex- hibit evidence of present or past motion in the form of moraines, crevasses, and so forth. Where the snowline of adjacent ice-free areas extends across the glacier, it is known as the firn limit, which represents the dividing line between the glacier’s two major zones, the zone of accumulation and the zone of ablation. While glaciers normally are recognized easily, certain glacial boundaries may be subject to misinterpretation, even by the experi- enced interpreter. Flow features upglacier from the firn limit typically are obscured by fresh snow, forc- ing the image interpreter to depend on secondary information such as valley shape or seek a more discriminating sensor. Similarly, morainal material may cover the terminus (or snout) of the glacier because of ablation, making boundary determination in that vicinity difficult. This latter problem occa- sionally is compounded by the presence of consider- able vegetation rooted in the insulating blanket of ablation moraine. Further subdivision of glacial occurrences, mainly on the basis of form and topographic position, would include: small drift glaciers (sometimes called Ural-type or cirque glaciers) ; valley glaciers (also called mountain or alpine glaciers) ; piedmont gla- ciers; and icecaps (or ice sheets). Other features have somewhat the surface form of true glaciers, such as “rock glaciers.” Since these are composed primarily of fragmented rock mate- rial together with interstitial ice, they are classified as Bare Exposed Rock. MAP PRESENTATION Figures 1 through 4 depict typical maps which have been produced using the U.S. Geological Sur- vey land use and land cover classification system. The land use and land cover maps have been pro- duced by conventional interpretation techniques and are typical examples of maps produced from high- altitude color-infrared photographs. In order to provide a systematic and uniform ap- proach to the presentation of land use and land cover information in map format, a scheme of color coding is employed (table 4). In this scheme, Level I land uses are color coded using a modified version of the World Land Use Survey (International Geo- graphical Union, 1952) color scheme. Level II land uses can be presented using the two-digit numeral appropriate to the land use category, such as “21,” which would signify Cropland and Pasture. The use of some type of system other than a further strati- TABLE 4.—U.S.G.S. Level I Land Use Color Code 1. Urban or Built-up Land__Red (Munsell 5R 6/12). 2. Agricultural Land _______ LighLBrown (Munsell 5YR 7 . 3. Rangeland ______________ LighLOrange (Munsell 10YR 4. Forest Land ____________ Green (Munsell 10GY 8/5). 5. Water __________________ Dark Blue (Munsell 10B 7/7). 6. Wetland ________________ Light/Blue (Munsell 7.5B 8.5 3 . 7. Barren Land ____________ Gray (Munsell N 8/0). 8. Tundra _________________ Grlee51;-Gray (Munsell 10G 8.5/ 9. Perennial Snow or Ice ___White.(Munsell N 10/0). fication by color is necessary at Level II since it would be a considerable problem to select 37 differ- ent colors which would be distinguishable at the size of the minimum mapping unit. A numerical sys- tem, with the number of digits equaling the level of categorization, forms a flexible classification sys- tem that permits continuation to Levels III and IV or beyond. In addition, retaining a discrete color code for each Level I land use or land cover category permits rapid visual integration of the areas char- acterized by that use or cover type. Even though a numerical system for the Level II land uses has been illustrated, such a system is not the only method of presenting Level II land use in- formation. What is proposed is the use of the modi- fied International Geographical Union World Land Use Survey color code at Level I. Alternatives to a numerical code at Level II could take the form of graphic symbols such as dots, stipples, cross-hatch- ing, swamp or marsh symbols, or any of the great variety of such items available to the cartographer. Such a method, together with the Level I color cod- ing, would allow the reader rapid visual orientation to each discrete Level II land use category but would impede statistical inventory of the area in- cluded in each land use and would be difficult to subdivide further into Level III categories. Another alternative for land use symbolization at Level II is the use of an alphabetical code for each category such as “Ur,” representing (Urban or Built-up) Residential Land, or “Ac,” for (Agricul- tural) Cropland and Pasture. Such a system has the merit of suggesting the logical name of each cate- gory but also impedes interpretation and enumer- ation at the more detailed levels because of increased complexity of the alphabetical code. In addition, the increase in length of the alphabetical code used for the more detailed levels will cause placement prob- lems as the minimum size of a mapping unit is ap- proached. MAP PRESENTATION 23 86°15' 86°10, 2 . 4 2 2 2 2 1 7 '39!° '31“ 39° 45 45! 1 0 EXPLANATION e 1 Urban or built-upland z 2 Agricultural land 3r 2 4 Forest land 12 9" _ 5 Water 7 Barren land I 2 J 5 2 a l ”(a 4‘ 3Q} If u hine 2 "" j a; :13 0f? 39° I 7 4 . E4 x 5 a 7 r- 4 E 2 2 N 5‘ HCE ‘ 1 g; M ‘ l i’ @W 1 {P 1 2 f 1 2 MILES E E 1,} an if ‘ , 1 2 K'LOMETRES j? a uh, dawn, $3 1 5" g ‘4 ’ ream , ,- 2. 86°15' 86°10’ FIGURE 1.——Level I land use and land cover in an enlarged part of the northeast quarter of the Indian- apolis, Indiana-Illinois, 1:250,000 quadrangle. Area outlined in center of map corresponds to May- wood area shown in figures 3 and 4. 24 A LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA 810 6° , ' ~—"- "W/dlfi"s§"l1’ {-313 I» I ._..__ 'Hfiisl‘dllll .h', ‘ '- L , ,J-T-mnm e. IllV’y “ ‘ ‘E’Q‘ ‘ 7 ,r ‘ g ' , '7 ”,flg;lfigm =" $ , a . i l My» 5 ”A“ n ‘u‘ 77 L r 9.3 i 11 I E ‘ , 12 39° ‘. V ~ , 45' 7.535%” ‘ 13 It”? 5‘ ‘ 7‘ 14 “LE 15 Lift: '3: 17 SI 21 hilly ‘ 22 3‘35— :” i?” 23 r ,I , 41 ‘ 51 53 _ 75 76 o o EXPLANATION Residential Commercial and services Industrial Transportation, communications, and utilities Industrial and commercial complexes Mixed urban or built-up land Other urban or built-up land Cropland and pasture Orchards, groves, vineyards, nurseries, and ornamental horticultural areas Confined feeding operations Deciduous forest land Streams and canals Reservoirs Strip mines, quarries, and gravel pits Transitional areas 1 2 MILES l 2 KILOM ETRES FIGURE 2.—Level II land use and land cover in an enlarged part of the northeast quarter of the Indianapolis, Indiana- Illinois, 1:250,000 quadrangle. Area outlined in center of map corresponds to Maywood area‘shown in figures 3 and 4. MAP PRESENTATION 86° u A VENUE ._. ox r?! .— 39° _ o 41' J - - 41! 15" ll 13 l. 15" E /4 E r :7 21 .- , 21 17 86°12’30" A 86°11'15” 11 12 13 14 17 21 22 24 41 42 43 51 52 53 62 75 25 EXPLANATION Residential Commercial and services Industrial Transportation, communications. and utilities Other urban or built-up land Cropland and pasture Orchards, groves, vineyards, nurseries, and ornamental horticultural areas Other agricultural land Deciduous forest land Evergreen forest land Mixed forest land Streams and canals Lakes Reservoirs Nonforested wetland Strip mines, quarries, and gravel pits 1/2 MILE .5 KILOMETRE FIGURE 3.—Level II land use and land cover in a part of the Maywood, Indiana, 1:24,000 quadrangle. Level III inter- pretations for the same area are shown in figure 4. 26 A LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA pit 532 EXPLANATION 111 Single family 122 Retail trade 131 Primary processing 132 Fabrication 134 Extraction facilities 141 Highways 144 Airports 145 Communications 147 Utilities 39° 39° 173 Waste dumps 42’ 42: 174 Urban undeveloped 30” 211 Cropland 212 Pastureland 224 Nurseries and floriculture 242 Farmsteads 412 10—30 percent crown cover, deciduous 413 30—70 percent crown cover, deciduous 414 >70 percent crown cover, deciduous 424 >70 percent crown cover, evergreen 432 10—30 percent crown cover, mixed 51 1 Streams H I, 521 Lakes x (3:532 -. i}; 532 Water-filled quarries 622 Mudflats 753 Sand and gravel pits (active) 39° _ _ o, 41:, A 2?, 15 i ”132 -1 .i 15” 0 V2 MlLE PEI—*4 _ am - .- . o .5 KlLOMETRE , - 211 211 212 17411 86°12’30” 86°11’15” FIGURE 4.—Level III land use and land cover in a part of the Maywood, Indiana, 1:24,000 quadrangle. Level II in- terpretations for the same area are shown in figure 3. SELECTED BIBLIOGRAPHY 27 SELECTED BIBLIOGRAPHY Anderson, James R., 1971, Land use classification schemes used in selected recent geographic applications of remote sensing: Photogramm.Eng., v. 37, no. 4, p. 379—387. Anderson, James R., Hardy, Ernest E., and Roach, John T., 1971, A land-use classification system for use with re mote-sensor data: U.S. Geol. Survey Circ. 671, 16 p., refs. Avery, T. Eugene, 1968, Interpretation of aerial photographs [2nd ed.]: Minneapolis, Burgess Pub. Co., 324 p. Barlowe, Raleigh, 1972, Land resource economics [2nd ed.]: Englewood Clifis, N.J., Prentice-Hall, Inc., 585 p. Burley, Terence M., 1961, Land use or land utilization?: Prof. Geographer, v. 13, no. 6, p. 18—20. Clawson, Marion, and Stewart, Charles L., 1965, Landuse in— formation. A critical survey of U.S. statistics including possibilities for greater uniformity: Baltimore, Md., The Johns Hopkins Press for Resources for the Future, Inc., 402 p. Colvocoresses, Alden P., 1971, Image resolution for ERTS, Skylab, and Gemini/Apollo: Photogramm. Eng., v. 38, no. 1, p. 33—36. Colvocoresses, Alden P., and McEwen, Robert B., 1973, Prog- ress in cartography, EROS program: Symposium on Significant Results Obtained from the Earth Resources Technology Satellite—1, Natl. Aeronautics and Space Admin. Pub. SP—327, p. 887—898. Ellefsen, R., Swain, P. H., and Wray, J. R., 1973, Urban land use mapping by machine processing of ERTS—l multispectral data: A San Francisco Bay area example: West Lafayette, Ind., Purdue Univ. Lab. for Applications of Remote Sensing Inf. Note 101573. Flint, R. F., 1957, Glacial and Pleistocene geology: New York, John Wiley and Sons, Inc., 553 p. Frey, H. Thomas, 1973, Major uses of land in the United States—summary for 1969: U.S. Dept. of Agriculture, Econ. ~Research Service, Agr. Econ. Rept. no. 247. Gleason, Henry A., and Cronquist, Arthur, 1964, The natural geography of plants: New York, Columbia‘Univ. Press, 420 p. Grigg, David, 1965, The logic of regional systems: Annals Assoc. Amer. Geographers, v. 55, no. 3, p. 465—491. Hardy, Ernest E., Belcher, Donald J ., and Phillips, Elmer S., 1971, Land use classification with simulated satellite pho- tography: U.S. Dept. of Agriculture, Econ. Research Service, Agr. Inf. Bull., 352 p. Hardy, Ernest E., and Shelton, Ronald L., 1970, Inventory- ing New York’s land use and natural resources: New York’s Food and Life Sciences, v. 3, no. 4, p. 4—7. Hawley, Arthur J., 1973, The present and future status of Eastern North Carolina wetlands: Chapel Hill, Univ. of North Carolina, Water Resources Res. Inst., Rept. no. 87. Ingram, J. J., and Prochaska, D. D., 1972‘, Measuring com- pleteness of coverage in the 1969 census of agriculture: Am. Stat. Assoc., Business and Econ. Sect, ann. mtg., Montreal 1972, Proc., p. 199—215. International Geographical Union, 1952, Report of the commit- tee on world land survey for the period 1949-1952: Worcester, England, 23 p. Joint Federal-State Land Use Planning Commission for Alaska, 1973, Major ecosystems of Alaska: Anchorage, Joint Federal-State Land Use Planning Comm. for Alaska, map, scale 1:2,500,000, incl. text. Kuchler, A. W., 1964, Potential natural vegetation of the con- terminous United States: Amer. Geog- Soc., Spec. Pub. no. 36, 116 p. Marschner, F. J ., 1950, Major land ’uses in the United States [map, scale 1:5,000,000]: U.S. Dept. of Agriculture, Agr. Research Service. National Academy of Sciences, 1970, Remote sensing with special reference to agriculture and forestry: Washing- ton, D.C., Natl. Acad. Sci., 423 p. New York State Office of Planning Coordination, 1969, Land use and natural resources inventory of New York State: Albany, New York State Office of Planning Coordination, 67 p. Oosting, Henry J., 1956, The study of plant communities [2nd ed.]: San Francisco W. H. Freeman 00., 440 p. Orning, George W., and Maki, Les, 1972, Land management information in northwest Minnesota: Minneapolis, Univ. of Minn. Center for Urban Studies, Minn. Land Manage- ment Inf. System Study, Rept. no. 1. Pettinger, L. R., and Poulton, C. E., 1970, The application of high altitude photography for vegetation resource in~ ventories in southeastern Arizona: Final Rept., Contract no. NAS 9—8577, Natl. Aeronautics and Space Admin., 147 p. Price, Larry W., 1972, The periglacial environment, perma- frost, and man: Washington, D.C., Assoc. of Amer. Geographers, Comm. on College Geography, Resource Paper No. 14, 88 p. Rosenberg, Paul, 1971, Resolution, detectability, and recog- nizability: Photogramm. Eng., v. 37, no. 12, p. 1255— 1258. Shaw, Samuel P., and Fredine, G. Gordon, 1956, Wetlands of the United States: U.S. Dept. of the Interior, Fish and Wildlife Service Circ. 39. Shelford, Victor E., 1963, The ecology of North America: Urbana, Univ. of Illinois Press, 810 p. Stevens, Alan R., Ogden, W. H., Wright, H. B., and Craven, C. W., 1974, Alternatives for land use/cover mapping in the Tennessee River watershed: Amer. Cong. on Survey- ing and Mapping, Amer. Soc. of Photogramm., ann. mtg., 34th, St. Louis, Mo., Mar. 10—15, 1974, p. 533—542. Stoddard, Lawrence A., and Smith, Arthur D., 1955, Range management [2nd ed.]: New York, McGraw-Hill Book Co., 433 p. Sweet, David C., and Wells, Terry L., 1973, Resource man- agement implications of ERTS—l data to Ohio: Sympo- sium on Significant Results Obtained from the Earth Re- sources Technology Satellitel, Natl. Aeronautics and Space Admin. Pub. SP—327, p. 1459—1466. Thrower, Norman J. W., 1970, Land use in the Southwestern United States from Gemini and Apollo imagery (map suppl. no. 12): Annals Assoc. Amer. Geographers, v. 60, no. 1. U.S. Bureau of the Census, 1970, Areas of the United States: U.S. Dept. of Commerce, Bureau of the Census, Area Measurement Rept. GE—20, no. 1. U.S. Congress, 1936, The Western Range: U.S. 74th Cong, 2d sess., Senate Doc. 199. 1973, The land use policy and planning assistance act: U.S. 93rd Cong., 1st sess., Senate Bill 268. U.S. Department of Agriculture, Conservation Needs In- ventory Committee, 1971, National inventory of soil and 28 A LAND USE AND LAND COVER CLASSIFICATION SYSTEM FOR USE WITH REMOTE SENSOR DATA water conservation needs, 1967: Statistical Bull. 461, 211 p. 1972, Farmland: Are we running out?: The Farm Index, v. XI no. 12, p. 8—10. U.S. Department of Agriculture, Soil Conservation Service, 1962, Classifying rangeland for conservation and plan- ning: U.S. Dept. of Agr. Handbook 235. [U.S.] Executive Office of the President, Bureau of the Bud- get, 1957, Standard industrial classification code: Wash- ington, D.C. U.S. Geological Survey, 1973, Geological Survey research 1973: U.S. Geol. Survey Prof. Paper 850, p. 255—258. U.S. Urban Renewal Administration, Housing and Home Fi- nance Agency, and Bureau of Public Roads, 1965, Stand- ard land use coding manual, a standard system for identifying and coding land use activities: Washington, D.C., 111 p. Viereck, Leslie A., and Little, Elbert L., Jr., 1972, Alaska trees and shrubs: U.S. Dept. of Agriculture, Forest Serv- ice Handbook 410, 265 p. Welch, Roy, 1973, Cartographic quality of ERTS—l image: Symposium on Significant Results Obtained from the Earth Resources Technology Satellite—1, Natl. Aeronau- tics and Space Admin. Pub. SP—327, p. 879—886. Wooten, Hugh H., and Anderson, James R., 957, Major uses of land in the United States—summer for 1954: U.S. Dept. of Agriculture, Agr. Research S rvice, Agr. Inf. Bull. 168. £331,339; The Relation of Geology to Stress Changes Caused by Underground Excavation in Crystalline Rocks at Idaho Springs, Colorado GEOLOGICAL SURVEY PROFESSIONAL PAPER 965 f. WW ”0? {512; S" 0‘40 ‘7’? J£1N 2. 7 1977 i. \ 4‘ " \ix":7_§13;rmcg \Wii/ \(‘i JAN 19 197? LIBRARY 0 JAN 4 1977 DOCUMENTS ommmrmw‘ ' The Relation of Geology to Stress Changes Caused by Underground Excavation in Crystalline Rocks at Idaho Springs, Colorado By FITZHUGH T. LEE, U.S. GEOLOGICAL SURVEY, JOHN F. ABEL, JR., COLORADO SCHOOL OF MINES, and THOMAS C. NICHOLS, JR., U.S. GEOLOGICAL SURVEY GEOLOGICAL SURVEY PROFESSIONAL PAPER 965 A study to understand the effects of geology and rock stresses on mining in complexly deformed anisotropic metasedimentary rocks using theoretical, laboratory, and field investigations UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON : 1976 UNITED STATES DEPARTMENT OF THE INTERIOR THOMAS S. KLEPPE, Secretary GEOLOGICAL SURVEY V. E. McKelvey, Director Library of Congress Cataloging in Publication Data Lee, Fitzhugh T. The relation of geology to stress changes caused by underground excavation in crystalline rocks at Idaho Springs, Colorado. (Geological Survey Professional Paper 965) Bibliography: p. 1. Rock mechanics. 2. Rocks, Crystalline. I. Abel, John F., joint author. II. Nichols, Thomas C., joint author. 111. Title: The relation of geology to stress changes caused by underground excavation. . . IV. Series: United States Geological Survey Professional Paper 965. TA706.L43 557.3'08s [622'.2] 76—16054 For sale by the Superintendent of Documents, US. Government Printing Office Washington, DC. 20402 Stock Number 024—001-02916—2 CONTENTS Page Page Metric-English equivalents ........................................................... IV Laboratory model-tunnel investigations.. 15 Abstract ................................................................... 1 Acrylic model .................... 17 Introduction .............................................................. 1 Concrete model. ..... 19 ACkm‘NIngmems ------ 2 Granite model ............................. 20 Instrumentation analyses """"""" . 2 Evaluation of model results ...................... 22 Stress concentration calculations .............................. ,, 3 Stress at b . f l. d 22 _ , , pro e in excess 0 average app 1e stress.. ........ Mathematical calculation of stress concentrations ........ 4 Stresses r d' th (1 . t 1 22 _ . p ece mg e a vancmg unne ........... Stress concentrations perpendicular to borehole 4 Discussion of lateral deformation ......................................... 24 axrs ....................................................................... Suess concentrations parallel to borehole axis ______ 5 Conclusions from model studies.. .................... 25 Summary Of mathematical calculations Of stress Field investigations ...................... 25 concentrations ...................................................... 7 , . . , Finite-element analyses of stress concentrations ............ 7 ROCk mechanics investigations at the Colorado SChOOI Axisy m metric analyses ......................................... 9 of Mines experimental mine .............................................. 25 Preliminary analyses.. 9 Scope of field studies ........................ 25 Probe analyses ...... 10 Geologic setting of the Idaho Springs area .......................... 26 Stress distribution .............................. . 10 In situ stress field ................................... 27 Summary of stress concentration factor . 11 Experimental room 1 ........... 31 Comparison of stress concentration factors Explanation of stress changes in room 1 ....................... 34 obtained by mathematical and finite— Stress changes ahead of the central crosscut. element methods ........................................... 13 extension ..................................................................... 35 Analysis of laboratory biaxial test ................... 14 Experimental room 2 ..................................................... 38 Plane-strain analyses ............................................... 14 . . . , . . . . . Discuss10n of stresses in the experimental mine .......................... 45 Preliminary plane-strain analysrs .................... 14 Analysis of overcoring ..................................... 15 References cited ............................................................................. 46 Requirement of tensile strength ...................... l6 ILLUSTRATIONS Page PLATE 1. Geologic map of part of Colorado School of Mines experimental mine .................................................................. . ............... In pocket FIGURE 1. Diagram of three-dimensional borehole probe, showing relation of sensing sphere to cylindrical epoxy probe and host rock ............................................................................................................................................................................................ 3 2. Cross sections of borehole probe showing dimensions and geometric relationships ..................................................................... V 3 3. Flow diagram of mathematical methods for calculation of stress concentration factor ................................................................. 4 4. Graph showing spherical stress concentration ................................................................................................................................ 5 5. Diagrams showing stress concentration for example conditions in the R direction ...................................................................... 6 6. Diagrams showing determinations of stress concentrations in the Z direction ............................................... 7 7. Graphs showing basic mesh for finite-element model ..................................................................................... 8 8. Schematic diagram of axisymmetric problem ......................................... 9 9. Schematic diagram of plane—strain problem .............................................................................................................................. 9 10. Graphs showing stress distribution from finite-element analysis along radial line cutting center of probe—steel sensor within epoxy mass under hydrostatic stress and under triaxial stress .................... 11 11. Graphs showing stress distribution from finite-element analysis along a radial line cutting the center of the probe for five analyses ......... 12 l2. Graph of stress concentration factors for spherical steel sensor surrounded by epoxy cylinder and host rock ............................. 14 13. Graphs showing stress distributions along a radial line cutting the center of the probe and along the X axis for the plane-strain analysis ................................................................................................................................................................. 15 l4. Graph showing plane-strain finite-element analysis of stress distribution in borehole probe and rock along R axis ................. 17 15. Photograph of acrylic model showing three-dimensional borehole probe grouted into central drill hole ................................... 18 16. Graphs showing changes of magnitude in principal stresses and maximum principal stress in acryl:c model compared to results of Galle and Wilhoit ..................................................................................................... 18 17. Equal-area diagram showing bearing and plunge of principal stress changes in acrylic model .................................................. 19 18. Photograph of concrete tunnel model ............................................................................................................................................ 19 III IV CONTENTS Page FIGURE 19. Graphs showing changes of magnitude in principal stresses and maximum principal stress in concrete model compared to the results of Galle and Wilhoit ............................ - ............................................................................................................... 20 20. Equal-area diagram showing bearing and plunge of principal stress changes in concrete model ...................... . 20 21. Photograph of granite tunnel model showing grouted sensor ....................................................................................................... 21 22. Graphs showing changes of magnitude in principal stresses and maximum principal stress in granite model compared to the results of Galle and Wilhoit ........................................................................................................................................... 21 23. Equal-area diagram showing bearing and plunge of principal stress changes in granite model .................................................. 21 24. Graph showing stress distribution adjacent to model tunnel in photoelastic material ................................................................. 23 25. Index map showing location of field site ..................................................................................................... 26 26. Plan of field site showing general arrangement of workings, excavation sequence, borehole probe locations, and major faults .......................................................................................................................................................................................... 28 27. Equal-area diagram showing principal in situ stress orientation determinations .............. 29 28. Equal-area diagram of foliation joints in study area ........................................................ 29 29. Equal-area diagram of 159 joints in study area ................................ . .......................... . 30 30. Generalized view showing relationship of hanging wall and footwall of fault I to foliation and to probes 5 and 8.. 31 31. Graph showing vertical normal stress in rock adjacent to probes 5 and 8 as excavation advanced .............................................. 31 32. Equal-area lower hemisphere diagram of poles of principalrstress changes, as determined from probe 8 ................................... 32 33. Comparative equal-area lower hemisphere diagrams of undifferentiated principal stress-change directions determined at 1.-. probes 5 and 8 ............................... 33 34. Map of room 1 in Colorado School of Mines experimental mine showmg strike of vertical sections .......................................... 33 35. Equal-area diagram showing orientation of in situ stresses determined in walls of fault I .......................................................... 34 36. Cross section AD containing probe 5 showing hypothetical lines of zero change in stress level after two stages of excavation. 35 37. Graph showing principal stress changes caused by advance of central crosscut ............. 36 38. Plan view of stress-change trajectories and stress-change vectors .......................................................................... 37 39. Generalized section showing relation of borehole probes, multiple-position borehole extensometers, and fault 2 ..................... 38 40. Graphs of principal stress changes in roof during excavation of room 2 .......................................................................... 39 41. Graphs of strain changes for multiple-position borehole extensometers in roof of room 2.. .. ....... 40 42. Equal-area diagrams showing principal stress changes of probes 24 and 25 before room 2 excavat10n.. 40 43. Diagrams showing generalized in situ stress conditions at several stages in the excavation of room 2... 42 44. Equal-area diagram showing stress-change orientations determined from borehole probes around room 2 ............................... 43 TABLES Page TABLE 1. Differences between theoretical stress concentrations perpendicular to and parallel to the borehole axis... ............... 7 2. Finite-element model properties ............................................................................................................... lO 3. Variations in radial and tangential stresses computed from finite element analyses .................... 10 4. Average stress concentration factors for axisymmetric finite-element solution .......................... 11 5. Finite-element model properties for biaxial test ................................................................................................... l4 6. Finite-element model properties for plane-strain analysis ................................................................................... l4 7. Average stresses determined ahead of tunnel face in acrylic model ................................................................................................ l8 8. Average stresses determined ahead of tunnel face in concrete model .............................................................................................. 19 9. Average stresses determined ahead of tunnel face in granite model .................................................................... 21 10. Summary of model material properties ............................................................................................................. 22 11. Change in lateral tensile stress in granite model... ............................................................................ 24 12. Calculated in situ stresses .............................................................................................. 29 13. Principal in situ stress clusters determrned in field study area ....................................................................................................... 30 METRIC-ENGLISH EQUIVALENTS Metric unit English equivalent Metric unit English equivalent Length Stress millimetre (mm) metre (m) kilometre (km) 0.03937 inch (in) . 3.28 fee; (h) kiloneWtons per .62 mile (mi) square meter (kN/mz) = 0.145 pound per square inch (lb/in?) ll II II THE RELATION OF GEOLOGY TO STRESS CHANGES CAUSED BY UNDERGROUND EXCAVATION IN CRYSTALLINE ROCKS AT IDAHO SPRINGS, COLORADO By FITZHUGH T. LEE, JOHN F. ABEL, JR.,l and THOMAS C. NICHOLS, JR. ABSTRACT A comprehensive rock-mechanics study, including theoretical, labo- ratory, and field investigations, was performed to better understand the influence of certain geologic features and excavation procedures on the stress changes and deformations induced by excavations in a faulted and altered crystalline rock mass. The geologic and excavation factors con- sidered included the structural geologic framework, the tectonic and residual components of the in situ stress field, and the sequence and orientation of excavations. The field results are based upon three- dimensional measurements of geologic features, stress changes, and the nature of the excavation. Part of the study consisted of developing a basis for interpreting stress data obtained with the US Geological Survey solid-inclusion borehole probe. For this purpose, extensive theoretical analysis and testing in models were required before field data from the probe could be ade- quately interpreted. These preliminary investigations established the theoretical basis for the probe and developed the confidence necessary for its use. No complete theoretical three-dimensional elastic analysis exists for the complex geometry and stress-strain relations of the borehole probe-host material system. However, two approximate methods were used to analyze the response of the probe to changes of stress. These involved the mathematical combination of two plane-strain elastic analyses of planes parallel to and perpendicular to the probe axes and the elastic evaluation of the probe using axisymmetric finite—element models of the probe. Results from the two approaches were in close agreement; average stress-concentration factors (SCF) were calculated for a range of rock and sensor properties. The SCF is necessary to convert strains measured on the spherical sensor of the probe to stress changes in the host-rock body. In the field study we recognized discrepancies between measured stresses and those predicted on the basis of simple gravity-loading and I elastic behavior. Geologic discontinuities, especially faults and foliation as well as excavation procedures, control the orientation and magnitude of stresses in granitic and metasedimentary rocks at the field site at Idaho . Springs, Colo. Stress changes, determined at several locations in the mine at an average depth of 107 m (350 ft), were not satisfactorily predicted by simple gravity loading through elastic analysis. The magnitudes of the in situ stresses, determined by the overcoring method, and of stress changes induced by excavating a crosscut and two rooms greatly exceeded the stresses predicted using elasticity relations and overburden loading; the average horizontal stress components and the average vertical stress component of the in situ stress field are three times and twice as large, respectively. Reasons for such discrepancies, in some instances, could be identified. If not near a fault, two of the in situ principal stresses are commonly parallel to the pervasive foliation and the other principal stress is commonly normal to foliation. The principal stress orientations vary approximately with the foliation attitude. Adjacent to faults, however, major and minor principal stress changes either were coincident with the 1 Colorado School of Mines. Golden, Colo. fault normal or were in the plane of the fault. Strain energy, naturally concentrated along faults, was further concentrated in rock adjacent to both faults and underground openings; vertical decompression of 7,000 lb/in2 was triggered in one place by excavation through a fault. Stresses? determined by overcoring were unequal in opposite walls of faults, differing by a factor of.2. The excavation process also contributed to changes in the magnitudes and directions of the principal stresses in the vicinity of the excavation. The direction of the greatest decompressive stress changed during excavation so that it remained perpendicular to the greatest room cross section. Stress changes occurred some 7 diameters ahead of the advancing crosscut driven in the jointed and closely foliated gneiss and gneissic granite. Instrumentation placed ahead of three model tunnels detected compressive stress changes 4 diameters in concrete, and 2 diameters in granite. These findings are far different from a theoretical elastic esti- mate of the onset of detectable stress change at l tunnel diameter ahead of the face. In addition, minor compressive stress peaks were detected about 6 diameters ahead of the crosscut, 2 diameters ahead of the face in acrylic, 1.25 diameters in concrete, and 1 diameter in granite. These subsidiary stress peaks are not explained by available theory. Such rock-mass behavior, if not anticipated, might result, especially in a complex (multiple-opening) underground operation, in damage of support systems owing to overloading. Laboratory and field data suggest that the stress field may have both residual (locked-in) and tectonic or gravity components. The release of residual strain energy is principally brought about by excavation through excavation-induced fault movements. Disturbances of the stress field caused by the intermittent excavation in the mine in part were time dependent, with stabilization requiring weeks to months. INTRODUCTION The laboratory and field investigations described in this report were started by the US. Geological Survey in July 1965 and completed in August 1971. The investigation was conducted as part of a continuing program in rock mechanics research whose main purpose is to examine rock behavior with respect to engineering problems associated with surface and underground excavation in rock. The literature contains very few studies that recognize or demonstrate the importance of three-dimensional in- formation in the analysis of rock deformation due to tunneling or mining. Because this problem is three- dimensional, an appropriate solution should logically be expressed in this form. In the past, the lack of adequate in- strumentation together with inadequate geologic data has 1 2 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO greatly restricted interpretations and conclusions of rock- mass behavior. The response of a rock mass to excavation is commonly analyzed by assuming: that the in situ stress field is gravi- tational, the rock mass is elastic, the excavation tech- niques are uniform, and the excavation is instantaneous. This approach cannot take into account a number of factors which could be important; namely, the geologic framework, tectonic and residual components of the in situ stress field, the variability of excavation techniques, and the sequential nature of excavation. The purpose of this investigation was to determine the importance of these factors to the response of a specific rock mass to the excavation of geometrically simple tunnels and rooms. The scope of the investigation included: (1) the devel- opment and application of a borehole probe to determine the initial in situ stress field and the stress changes induced by excavation, (2) a detailed investigation of the site geology, and (3) deformation analyses to discover short- comings of conventional analytical methods and the nature and importance of the effects of commonly neglected factors. The initial in situ stresses and the excavation-induced stress changes were monitored with the three-dimensional stress probe described by Nichols, Abel, and Lee (1968). Also, at some of the underground locations, other instru- ments were used. To provide an adequate basis for interpreting probe measurements, detailed theoretical and controlled labora- tory studies of the probe’s behavior were conducted. The theoretical studies consisted of finite-element analyses of probe geometry and boundary conditions, which were compared with other mathematical solutions. The labora- tory studies consisted of testing the probe under known applied stresses in three models of known physical proper- ties: acrylic, concrete, and Silver Plume Granite. Also, each probe was tested and was calibrated under hydro- static pressure. The results of these preliminary investiga- tions of the probe were further verified by driving (drilling) model tunnels into the three laboratory models. Stress changes were monitored during the course of model- tunnel driving. These stress distributions and magni- tudes were compared with theoretical elastic predictions and with field data available from other sources, and the probe-stress values were found to be reasonable and reliable. The field investigations described in this report were performed at the Colorado School of Mines experimental mine at Idaho Springs, Colo. The geologic framework—a faulted sequence of metasedimentary rocks intruded by granitic and pegmatitic rocks—is characterized by faults, foliation, and joints. These geologic structural features, often ignored in practice, largely controlled the behavior of the stress field and the deformation of the excavation. In addition to geologic mapping, the field investiga- tion included the installation of 17 borehole probes at various times in conjunction with the excavation of two rooms and a central connecting crosscut; 16 of these probes produced useful data. Seven were employed to obtain in situ stress-field information, and nine were used to moni- tor stress changes for periods of up to 4 years. These stress changes resulted from the excavation of the rooms and the central crosscut connecting them. The borehole extenso- meter measurements of other investigators at the field location contributed to our understanding of rock-mass response. ACKNOWLEDGMENTS The investigations benefited from the assistance of sev- eral individuals whose help is gratefully acknowledged. R. A. Farrow, US. Geological Survey, performed or super- vised the many laboratory deformation tests and assisted us in the model investigations. G. S. Erickson fabricated the borehole probes and assisted valuably in collecting readings from the several monitoring probes. We thank R. S. Culver and R. N. Cox, then at the Colorado School of Mines, who supervised the underground excavations and made available to us other rock-strain data from the experimental mine.. Cox also was responsible for the finite-element analyses of the borehole probe. J. J. Reed, Samuel Shaw, and Martin Forsmark, all of the Mining Engineering Department, Colorado School of Mines, generously made available appropriate sites in the experi- mental mine and provided necessary logistical services at the mine. Michael Haverland, US. Bureau of Reclama- tion, furnished in situ stress data from a borehole in the underground study area. We are indebted to D. J. Varnes, H. W. Olsen, A. T. F. Chen, and R. H. Moench for their thought-provoking and enlightening comments and criticism. INSTRUMENTATION ANALYSES The borehole probe (Nichols and others, 1968) mea- sures host-rock stress changes in three dimensions in terms of strains induced in the surface of the incapsulated spheri- cal metal sensor. The relation between measured strains and host-rock stresses varies with the geometry and material properties of the probe-rock system. The opera- tion of this complex system, not previously investigated, had to be explained before the desired information, stress in the rock, could be determined. With its incapsulated spherical sensor, the borehole probe acts as a welded cylindrical inclusion within a semi- infinite elastic rock mass (fig. 1). In addition to these geo- metric influences on strain measurements, the strain of the sensor is also influenced by differences in the elastic properties of the three materials. These two factors induce stress concentrations about the sphere and about the cylinder when the host body is stressed. (Stress concentra- tion is the ratio of the stress at any point to the applied stress.) The configuration and magnitude of these stress INSTRUMENTATION ANALYSES 3 Rosette 1 Dummy sphere (unrestrained) for thermal compensation in sensing circuit Host rock 1~in:-diameter metal spherical sensor Host rock Center:\ line Collar of 37in.—diameter borehole FIGURE l.—U.S. Geological Survey three-dimensional borehole probe. R, radial direction; Z, axial direction; 9, tangential direction. A, Relationship of spherical sensor to cylincrical epoxy probe and host rock. B, Location of strain-gage rosettes on spherical sensor. concentrations had to be determined in order to obtain the composite stress concentration factor (SCF) needed for the calculation of stresses in the host material. The borehole probe initially contained a steel sensor; subsequent probes utilized other more sensitive metals, including aluminum and brass. The test results are sufficiently general to apply to the other metal sensors. In this report the maximum principal stress is the most compressive (or least tensile) and compression is positive. This section describes the analytical bases developed to calculate host-rock stress changes from measurements of sensor strain changes. Three approaches were used: (a) adaptation of exact mathematical solutions for a cylinder and a sphere within a semi-infinite medium, (b) a finite-element representation of the probe-host-rock system, and (c) the analysis of stress changes in model tunnels. This laboratory study of model tunnels was also helpful in checking theoretically predicted probe behav- ior and in understanding the results of the field investi- gations. The following assumptions were required in order to proceed with these analyses: 1. The rock mass, epoxy grout, and sphere are elastic, homogeneous, and isotropic. 2. The stress field is hydrostatic, compressive, and equal to unity. In summary, the problem was to relate the strain in a semirigid spherical inclusion to the host-rock stress, where the spherical inclusion is enclosed within a cylinder of different material and the cylinder is bonded to and bounded by bedrock. STRESS CONCENTRATION CALCULATIONS In order to obtain stresses in the rock, the sensor-strain readings must be corrected for the stress-concentrating effects of geometry (sphere and cylinder) and of the differ- ent materials (rock, epoxy, metal). These effects interact simultaneously when the rock mass is subjected to a stress change. Further, the analysis must consider concentra- tions in both the R and Z directions (fig. 1). Changes in GR in the host rock are modified near and within the probe as follows: (1) the stresses in the rock mass near the probe are modified by the epoxy cylinder and the spherical sensor, and (2) at the same time, the stresses in the epoxy cylinder are modified by the presence of the spheri- cal sensor (fig. 2A). Goodier's (1933, p. 39) spherical influence zone- 4 sphere diae meters outside spherical sensor <— <— <— Metal sphere / Direction of load application Goodier's (1933, p. 39) influence zone- 4 diameters outside spherical Sensor I HD’l \ sin. —Epoxy cylinder—e borehole Metal . 8.5an sphere 8.5131] / i Z direction. \ B FIGURE 2.——Cross sections of borehole probe showing dimensions and geometric relationships A, R direction—Perpendicular to centerline of borehole probe. B, Z direction—Parallel to centerline of borehole probe. 4 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO Changes in a Z in the host rock are modified only by the epoxy cylinder (fig. 2B). In this direction there is suffi- cient epoxy (greater than 4 sphere diameters beyond the spherical sensor) so that there is no significant effect produced in the rock by the spherical sensor (Goodier, 1933, p. 39). Stress changes in the host rock acting at angles greater than 20.5° from the Z direction will have less than 4 sphere diameters of epoxy between the sensor and the epoxy-rock interface. Within an intermediate angular zone, from 20.5° to 90° to the Z direction, the overall stress concentration will lie between the R- and Z-direction analyses. The greater the thickness of epoxy between the spherical sensor and the epoxy-rock interface the more closely the sensor will follow the analysis in the Z direction. Conversely, the smaller the thickness of epoxy between the spherical sen- sor and the epoxy-rock interface the more closely the sensor will follow the analysis in the R direction. A flow diagram illustrating the mathematical calculations used to determine the SCF in the two directions is shown in figure 3. MATHEMATICAL CALCULATION OF STRESS CONCENTRATIONS STRESS CONCENTRATIONS PERPENDICULAR TO BOREHOLE AXIS (R DIRECTION) The stress-concentrating effects of the sphere and of the cylinder were calculated separately; these results were then superposed to obtain the composite SCF (fig. 3). The stress changes occurring in the rock and the epoxy cylinder at the boundary of a rock mass and in the epoxy cylinder can be calculated from the stress concentration STRESS CONCENTRATION FACTOR (SCF) DETERMINATION I l R direction Planar Z direction 7 l I I—Volumetric—| l Hemispheric Average influence 13.2 percent of spherical stress concentration rock to sensor (Srs) Times l 1.4 percent of spherical stress concentration rock to sensor (Srs) 6.6 percent of spherical stress concentration rock to sensor (Srs) Times Times 100 percent of cylindrical stress concentration rock to epoxy (Cre) 100 percent of cylindrical stress concentration rock to epoxy (Cre) 100 percent of cylindrical stress concentration rock to epoxy (Cre) 100 percent of cylindrical stress concentration rock to epoxy (Cm) Times l Times | l Times | Times 86.8 percent of spherical stress concentration epoxy to sensor (Ses) 100 percent of spherical stress concentration epoxy to sensor (Ses) 98.6 percent of spherical stress concentration epoxy to sensor (365) 93.4 percent of spherical stress concentration epoxy to sens‘or (Ses) Equals Equals Equals Equals SCF—Perpendicular SCF—Planar method—parallel SCF— Hemispheric method—parallel SCF—Average influence 0.132 srsx crex 0.868 SeS=SCFR C XS =SCF (1) re es Zplanar (2) method—— parallel 0.014 srsx crex 0.986 ses =SCF Z hemispheric (3) 0.066 SrSX CreX 0.934 535 =SCF . Z average influence (4) FIGURE 3.—Flow diagram of mathematical methods for calculation of stress concentration factor (SCF). INSTRUMENTATION ANALYSES 5 equation of N. J. Muskhelishvili (in Leeman, 1964, p. 55) . and are dependent on the relative stiffness and Poisson’s ratios of the rock and of the epoxy. This cylindrical stress concentration is actually a ratio: %’= (1~_V2) [______l_____ (v +l)+E/E’ (v’+1)(]r—2v’) 2 + E/E’(v’+l)+(v+l)(3—4v)] where a’=stress in cylindrical epoxy inclusion o=stress in rock mass (applied stress) ’=stiffness of epoxy (Young’s modulus) E=stiffness of rock mass (Young’s modulus) v’=Poisson’s ratio of epoxy v=Poisson’s ratio of rock mass. For example, the cylindrical stress concentration in an epoxy cylinder with E’=0.45‘><106 lb/in2 and v’=0.28 within a rock mass with E=4.5‘x 106 1b/in2 and v=0.25 is 0.26 or a’=0.26c. A spherical stress concentration is produced at the boundary and within a sphere surrounded by an infinite elastic body that is a function of the relative stiffness (E) and Poisson’s ratio (v) of the sphere and the host. This relationship has been determined mathematically by Goodier (1933) and Edwards (1951) and is shown in figure 4. For a steel sphere surrounded by an infinite body of elastic epoxy—that is, a cylinder greater than 4% diameters larger than the sensor diameter and with values of E50.451x106 lb/in2 and v=0.28, the spherical stress con- centration at the boundary and in the steel sphere is approximately 1.9, or a'=1.90, where a’=stress in spherical steel inclusion a=stress in infinite mass of epoxy i 100 | I 9 100 percent '— < D: 80— — E u" 66 8 percent 2 F ' Z — _ 0 NJ 60 U 0 U, tt u) Lu W “- _ 36.0 ercent I z 40 ° — 1— __ tn .1 < _ o 20 Center of cc sphere 1.4 0.5 % percent percent a. o l I i l w o 1 2 3 4 5 s 7 s DISTANCE FROM CENTER OF SPHERE, 1N SPHERE RADII FIGURE 4,—Spherical stress concentration. Percent of total concentra- tion at various diameters from center of sphere. (From Goodier, 1933, p. 41.) However, the epoxy cylinder extends only 1% sphere di- ameters beyond the center of the spherical sensor, so the epoxy is not an infinite body. According to information :in figure 4, the spherical stress concentration more than 1% sphere diameters from the center of the sphere (the rock mass-epoxy boundary) is approximately 13 percent. The remaining 87 percent of the spherical stress concentration (takes place between the rock-epoxy boundary and the (5) . sphere-epoxy boundary. Following the principle of superposition, the stress- concentrating effects of the spherical sensor may be com- bined with the stress-concentrating effects of the epoxy cylinder. For the example shown in figure 5A, the calculaa tion of the spherical stress concentration in the rock at the rock mass-epoxy boundary (SI,s ) in the R direction is 1+0.68(0.132)=1+0.09, where 1+0.68 is the maximum spherical stress concentration in the rock mass as determined graphically by Edwards’ method (1951, p. 27), and 0.132 is the percent spherical stress concentration more than 11/2 diameters from the center of the sphere (fig. 4). This stress is simultaneously acted upon by a cylindri- cal stress concentration in the rock mass due to the epoxy cylinder, C re, (l+0.09)(0.259)=0.282 (Leeman, 1964, p. 55). The epoxy between the rock and the steel sphere then produces another increment of stress concentration at the steel-epoxy boundary (Ses ). This in- crement is equal to l+spherical stress concentration from the epoxy cylinder to the spherical steel sensor times the remaining theoretical spherical stress concentration for a perfectly rigid sphere. This result is multiplied by the cylindrical stress concentration in the epoxy cylinder. [1+0.132(0.6.8)](0.259)[1+(1-0.132)(0.90)]=0.50=SCFR . (eq 1, fig. 3) Therefore, in this example the steel sphere should sense 0.50 lb/in2 for 1 1b/in2 stress change occurring in the R direction. Figure SB shows the steps in the stress concen- tration which have taken place. STRESS CONCENTRATIONS PARALLEL TO BOREHOLE AXIS (Z DIRECTION) We must now determine the stress concentrations in the host material in the Z direction and compare these with the stress concentrations in the R direction, and then we determine the required composite stress concentration factor (SCF). The stress acting in the Z direction (az) is influenced Applied stress radial to bore— hole (1—unit) ..____ __ __ K / ' 's influen \ / odlel’ Ce 2 / GO Che \ outside spher‘ \ / A dmmeters [cal sensor) \ / K \ Elastic rock host (E24.5,v=0.25) V \( Cylindrical epoxy probe (E=0.45, 1120.28) Spherical steel sensor (E= 30.5, v =O.285) FIGURE 5.—Stress concentrations for example conditions in the R direction. STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO STRESS CONCENTRATlON PERPENDICULAR TO BOREHOLE CENTERLINE 4.5 U) LL c 13 O Ell g 0: El lg- 3 Lu 3 *w O E ‘r ‘ r. A. “J 8 la 9 0 Z (3 ‘1: E L” o g 10 o __ O J \3 D: o «(Inn :44 m 3 u. LLI \: m (n l— g; 0 EC 0) \g “ l5 J 6' m g l= 2 _ <( cc l eEpoxy~rock 5 LU / interface I I n. |.09 L” (n 0: Lu I a. U) «Surface of sensor a—Center of sensor 1.0 1.2 1.4 A, Goodier’s (1933) spherical influence zone. B, Superposition of stress concentrations in rock mass. less by the spherical stress concentration within the rock mass than is the stress acting in the R direction (0R), as mentioned previously. Three different approximations for estimating “2 concentrations were made (fig. 6). In the first method, the adjacent rock was assumed to have no significant influence on oz because more than 4 sphere diameters of epoxy are present in that direction (fig. GB). Strain changes in the Z direction are transmitted, across the epoxy-rock interface by the bond between the rock and epoxy. This is referred to as the planar method of approximation because the stress is assumed to be transmitted in a plane in the Z direction and confined to the epoxy. This method does not consider any spherical stress concentration from the rock to the spherical sensor. In the second method, the cylinder of epoxy is replaced by a hemisphere whose volume equals half the volume of the epoxy cylinder (fig. 6C). This method assumes that there is a constant spherical stress concentration pro- duced by the spherical sensor on the rock regardless of the angle the stress change makes with the Z direction. The third method assumes that the spherical stress con- centration (rock to spherical sensor) in the rock has an average influence, or effectiveness, which ranges from 13.2 percent to 0.0 percent in the R direction (fig. 6D). For the planar method, the overall stress concentration is equal to the cylindrical rock-to-epoxy stress concentra- tion times the full spherical stress concentration of the epoxy acting upon the steel sphere. For the example, this stress concentration is equal to 0.259(1+0.90)=0.49=SCF2. (eq 2, fig. 3) The hemispheric method is nearly the same as the method of calculation of am, once the radius of the hemi- sphere has been calculated (fig. 6C ). Because of the finite geometry of the probe, only 1.4 percent of the total spheri- cal stress concentration of the rock on the steel sensor would occur beyond the calculated 6.35-cm (2.5 in.) sphere-radius hemisphere: ' l+(0.68)(0.014)=1.01. This concentration is also acted upon by the cylindrical stress concentration of the rock on the epoxy. l.01(0.259)=0.26. A further spherical concentration occurs at the epoxy- steel interface, as follows: 0.26[1+0.90(l—O.Ol4)]=[1+0.90(0.986)]0.26=0.49=SCFZ. (eq 3, fig. 3) The average influence method represents a compro- mise which should approximate the average conditions in the Z and R directions. This method follows the same cal- culation as for «IR except that the average stress concen- INSTRUMENTATION ANALYSES 7 4V2—sensor—diameter Rock mass influence zone x xxxxxxx X xx XXX //—| //'1 /’-| _. / —. / l / I / —’ /—“—’ l / .——> / Rocklmass / 0.0 percent / Rock mass 1 xx xx xxxxxxxxorxglgxxxxxm / __. l W . — 13.2 percent Borehole Z—* 1 Sensor Borehole Z_’l Sensor Borehole l probe 31.34“} probe _. Z") probe \ 2—) \ \ l Sensor W xx xxy‘x x—‘x xxx xlyx—‘xxxx/X \ XV‘xx xx‘x ‘x xxxl’Yx _. —. c_, 63.7—in.3 \ 1 \\ I _. \ ' l _, \_~. 1 \ \\J \\ .1 \\_1 KXXXXXXX xxxx xxx A B C D FIGURE 6.—Determinations of stress concentrations in the Z direction. Arrows indicate direction of stress application. Volumes are ex- plained in text (p. 5 ). A, Actual case. B, Planar method (assumes that diameter of borehole probe exceeds diameter of influence). tration effect is reduced from 13.2 percent to 6.6 percent. This results in an overall ”Z concentration of 0.50, the same as that calculated for (JR. The calculations for the average influence method proceed as follows: Spherical stress concentration (rock on steel sensing sphere) 1+0.68(0.066)=1.04, Cylindrical stress concentration (rock on epoxy cylinder) 1.04(0.259)=0.27. Spherical stress concentration (epoxy on steel sensing sphere) 0.27[1+0.90(0.934)]=0.50=scrz. (eq 4, fig. 3) SUMMARY OF MATHEMATICAL CALCULATIONS OF STRESS CONCENTRATIONS Table 1 gives calculated percentage differences between the stress magnitudes that should be sensed by the probe, perpendicular to and parallel to the probe axis, for various host-rock and sensor properties using the methods just described. The close correspondence between the mathematical analyses (table 1) indicated that a composite SCF value could be expressed by any of the equations (1—4) in figure 3. Equation 4 was used for the parallel concentration factor because it is closer to the actual geometry. An appro- priate SCF value is applied to all stress calculations using borehole-probe strain measurements. FINITE-ELEMENT ANALYSES OF STRESS CONCENTRATIONS The finite-element analytical method provided an in- dependent check on the SCF developed in the preceding section. The symmetry of construction of the borehole probe was appropriate for analysis by finite-element tech- niques. The principles of this method were outlined by Medearis (1974, p. 219-266). The US Geological Survey C, Hemispheric volume method (assumes that cylindrical volume of epoxy is replaced by equal-volume hemisphere). D, Average influence method (assumes that only average stress con— centration in rock mass is effective; that is. 13.2 percent of theoretical normal stress). TABLE 1.—Differences between theoretical stress concentrations per- pendicular to and parallel to the borehole axis [E=Young’s modulus (106 lb/in‘); v=Poisson's ratio; epoxy grout: E=0.45,‘v =0.28; steel sphere: E=30.5,fl=0.285; brass sphere: E=15.0, V‘=0.285; aluminum sphere: E=10.6, V‘=0.33. +, exceeds stress concentration perpendicular to borehole axis. -, less than stress concentration perpendicular to borehole axis] Percent difference in stress E13595 parallel to borehole axisl host rock Stress concentra- tions perpendicular ‘ Average to borehole axis Linear Hemispheric influence E v method method method Steel sphere 0.5 0.25 1.94 —4.4 —3.8 —2.0 1.0 .25 1.44 —4.0 —3.4 -1.9 1.5 .25 1.14 -3.7 —3.2 —1.6 3.0 .25 .70 —2.9 —2.3 —l.2 4.5 .25 .50 —1.8 -1.8 - .6 6.0 .25 .40 —l.5 —l.0 — .5 Brass sphere 0.5 0.25 1.90 -4.0 —3.5 —1.8 1.0 .25 1.40 —3.7 —3.3 —1.6 1.5 .25 1.11 —3.3 —2.9 —1.4 3.0 .25 .67 —1.6 —1.3 —.6 4.5 .25 .48 —.2 .0 +.2 6.0 .25 .38 +1.1 +1.1 +.8 Aluminum sphere 0.5 0.25 1.89 —3.7 —3.2 —l.4 1.0 .25 1.39 —3.4 —2.9 —l.4 1.5 .25 1.10 -2.7 —2.3 -1.3 3.0 .25 .66 .0 +.6 +.3 4.5 .25 .47 +2.1 +2.3 +1.6 6.0 .25 .37 +2.7 +2.7 +2.2 1Stress in axial direction of probe. contracted with R. M. Cox, then (1969) a graduate student in mining engineering at the Colorado School of Mines, to perform these analyses. The results agreed closely with the STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO WALLOFBOREH INNER SEGMENT OF MODEL BOREHOLE CENTERLINE DJ 2 3 CCI Lu I'l 2 Lu 0 Lu .4 o I LU E o [D 0.25 0.5 0.75 1.0 0 1 2 3 4 5 6 7 8 | | | | l | I l | I | I l l A B i FIGURE 7.——Basic mesh for finite-element model. The mesh contains 665 elements and 573 nodal points. A, Mesh for inner'segment B, Mesh for outer segment. Scales are sphere diameters from center of sphere. INSTRUMENTATION ANALYSES 9 Z Borehole w, Rock Epoxy probe , Metal sensor \0 % 0 1 2 3 4 5 CENTIMETRES 0 1 2 INCHES FIGURE 8,—Schematic diagram of axisymmetric problem. R, radial direction; Z, axial direction; 0, tangential direction. mathematically derived SCF. The following presentation is taken in part from the report by R. M. Cox (written commun., 1969). A finite-element model was constructed to facilitate the solution of a variety of problems concerned with stress dis— tributions in the probe with a minimum number of changes (fig. 7). Element size was determined by prox- imity to critical stress areas. The model, when appro- priately modified, could be used to study the effects of changes in rock properties, epoxy grout properties, and sensor properties on various loading conditions. Two types of analyses have been made using the finite- element method: three-dimensional axisymmetric analy- ses and plane-strain analyses. The axisymmetric analysis, a more general technique, is shown schematically in fig— ure 8. The exterior boundary of the rock mass is far enough from the probe (41/2-sensor diameters in the R direction) to minimize stress effects on this boundary due to the sphere and cylinder. The plane-strain analysis (fig. 9) permitted estimation of stress distributions in the probe by an alter- nate method. It also was used to analyze stress-relief by overcoring of the borehole probe. AXISYMMETRIC ANALYSES The axisymmetric method was used to study the effects of changes in rock properties, epoxy properties, and spherical sensor properties as well as various load condi- tions on sensor strains and indicated stresses. After two preliminary analyses were made to test the effectiveness of the axisymmetric method, six additional axisymmetric analyses were made. PRELIMINARY ANALYSES Two preliminary finite-element analyses (A and B) were devised to represent two theoretically solved cases of a sphere within a nearly infinite mass under two boundary- loading conditions (fig. 7). Analysis A was subjected to a simulated hydrostatic stress field and analysis B was sub- jected to a deviatoric stress field. The results were com- —->RX Metal sensor FIGURE 9.—Schematic diagram of plane-strain problem. Metal sensor simulated by cylinder of semi-infinite length. 10 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO pared with solutions of Goodier (1933, p. 40). Goodier found the radial stress concentration on such a sphere in a hydrostatic stress field to be 7" =3[(1— v >/<1+v>1. <6) where v =Poisson’s ratio of host material al=stress in steel sphere a”=hydrostatic stress It is apparent from this equation that the radial stress concentration depends on the Poisson’s ratio of the host material. For analysis A the epoxy has a Poisson’s ratio of 0.28. This value of Poisson’s ratio yields a radial stress con- centration of 1.69. Analysis A, with a hydrostatic stress of 1,000 lb/inz, gives a radial stress of 1,686 lb/in2 at the steel-epoxy bound- ary. The radial stress concentration is 1.69, which agrees with the theoretical solution within 0.2 percent. The tan- gential stress concentration is 1.70 at the locations of rosettes 1 and 2 (rosette locations are shown in fig. 18). The axial stress concentration (parallel to borehole centerline) for these rosettes is 1.64, which is within the probable 4- percent accuracy of the method. At the location of rosette 3, the radial and tangential stress concentrations are 1.77‘i 0.07. A plot of the stress distribution for analysis A is shown in figure 10A. The SCF in this case would be 1.72, with a maximum error of 5 percent in any one direction. The second preliminary axisymmetric analysis (analy- sis B) had an applied radial stress of 1,000 lb/in2 and an applied axial stress of 560 lb/inz. The results of this solu- tion are plotted in figure 103. The radial stress concentra- tion at the steel-epoxy boundary is 1.73, which falls close to the value of 1.69 derived from the mathematical solu- tion of Goodier (1933). The average axial stress concentra- tion is 1.63, with a maximum variation of 6 percent. Again, the axial stress concentration differs only slightly from the radial stress concentration. From these results we conclude that the finite-element solutions, which can be checked for accuracy by Com- paring them with other solutions, can be used to estimate the SCF. PROBE ANALYSES After verification of the effectiveness of the axisym- metric finite-element method, several analyses of this model were made to study: 1. Distribution of stress in the steel-epoxy probe. 2. Effects of changes in host-rock properties on the stress distribution in the probe. 3. Distribution of stress for boundary-loading conditions present in the laboratory biaxial overcore testing device. A list of the material properties and boundary-loading conditions used in these analyses is given in table 2. TABLE 2.—Finite-element model properties Directions of EPOXY boundary-loading Sphere cylinder Host rock conditions Analysis l[5(10“) 1: 1E00") V lE(10‘5) 1; Radial1 Axial Axisyrnmetlic analyses 1 30.5 0.285 0.45 0.28 0.5 0.20 1,000 Fixed. 2 30.5 .285 .45 .28 1.0 .20 1,000 Do. 3 30.5 .285 .45 .28 2.0 .20 1,000 Do. 4 30.5 .285 .45 .28 4.0 .20 1,000 Do. 5 30.5 .285 .45 .28 6.0 .20 1,000 Do. 1Units in pounds per square inch. Any finite-element solution is approximate. The preci- sion of the approximation is dependent on the fineness of the mesh. In the case of the axisymmetric model, the theo- retical tangential stress computed for rosettes l and 2 should be equal to the radial and tangential stresses computed for rosette 3 (fig. 1B). Table 3 shows the variations in these computed values. The indicated varia- tion among the analyses is approximately 4 percent. TABLE 3.— Variations in radial and tangential stresses computed from finite element analyses [Median maximum variation from mean computed stress =4.1 percent] Host rock Stresses computed at rosettesI l and 2 3 5 Maximum Analysis I500‘) 11 (Iangen- ( tangen- (radial) Meanl variation tial) tial) (percent) 1 0 5 0.20 1,832 1,976 1,885 1,898 4.1 2 1.0 .20 1,392 1,497 1,423 1,437 4.2 3 2.0 .20 936 940 934 937 .3 4 4 0 .20 566 607 577 583 4.1 5 6 0 .20 405 434 413 417 4.1 1Units in pounds per square inch. The SCF variation of about 4 percent indicated in table 3 could be reduced to less than 2 percent by using one factor for the tangential stress of rosettes 1 and 2 and radial stress of rosette 3, and another factor for the axial stress of rosettes 1 and 2 and the tangential stress of rosette 3. How- ever, such precision is greater than the precision (about 4 percent) of the finite-element method and an average value for the SCF is used. These SCF’s are listed in table 4 and plotted in figure 12B. Over the range of conditions tested, the analyses indi- cate that the stresses in the metal sensor of the probe vary inversely with the host-rock modulus and that the stress concentration in the host rock at the epoxy-rock contact varies directly with the host modulus. STRESS DISTRIBUTION Analyses 1—5 (table 2) were made with the ends fixed (no deformation) and with a radial stress of 1,000 lb/inz. Re- sults from these analyses were used to construct the stress INSTRUMENTATION ANALYSES 11 2000 I , l I Steell Epoxy ‘I‘ . 32:1 E = 0.45 (In/inwfin') 1500 — _ 1000 = 30.5 v = 0.285 ix E 500 2000 ‘ I Epoxy E = 0.45 (lb/in2 ”(0%) V = 0.28 1 500 COMPRESSIVE STRESS, IN POUNDS PER SQUARE INCH 1000 B I | I I | o \ I I | o 1 2 3 4 5 6 7 8 DISTANCE FROM CENTER OF PROBE, IN SENSOR—DIAMETERS EXPLANATION _+._+_+—— Tangential stress (0,.) —-—o—o——0— Radial stress (as) _—.—o—o— Axial stress (02‘) FIGURE 10.—Stress distribution from finite-element analysis along radial line cutting center of probe— steel sensor within epoxy mass. A, Hydrostatic stress of 1,000 lb/in2 (analysis A). B, Triaxial stress of 05:0; +1,000 lb/in2, 02:560 lb/in2 (analysis B). distributions along a radial line perpendicular to the axis of the borehole probe. The resulting stress distributions are shown in figure 11. Results of the mathematical stress concentration calculations (fig. 5B) are in good agree- ment with those from the finite-element analvsis (fig. 1 1D ). SUMMARY OF STRESS CONCENTRATION FACTORS The locations of the strain-gage rosettes on the probe. sensor are shown in figure 18. Because of the axial sym- metry of the model, rosettes l and 2 should register the same strain (or stress) under uniform radial stress. Stress TABLE 4,—Average stress concentration jactors (SCF’S) for axisym- metric finite—element solution [Factor 1 is an average of the parallel tangential stress on rosettes 1 and 2 and the parallel radial stress on rosette 3. Factor 2 is an average of the parallel axial stress on rosettes l and 2 and the parallel tangential stress on rosette 3] Host rock Average Maximum Factor Maximum Factor Maximum Analysis ___—_ factor difference difference difference 15(10‘5) " (percent) (percent) (percent) 1 0 5 0.20 1.90 6.5 1.85 1.6 1.95 5.7 2 1.0 .20 1.44 8.0 1.41 1.4 1.47 4.4 3 2.0 .20 .94 7.2 .94 1.1 .93 7.0 4 4.0 .20 .58 6.7 .57 1.7 .58 3.2 5 6 0 .20 .42 7.2 .41 1.2 .43 2.3 1Units in pounds per square inch. l2 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO 2000 M I 4 I I I I I I T I I I | I I Rock E = 0.5 (Ib/in2/Li1') A E In. 1500 V=0.2 1000 500 0 2000 I I I I SteeIL Epoxy _I_ Rock ‘I‘ T I - _ 1500* I E=1.0(Ib/in2/fl) B E In. V 0. 1000 500 — L41 COMPRESSIVE STRESS, IN POUNDS PER SQUARE INCH 2000 I I I Steel Epoxy .I‘ Rock I I E= 2.0 (lb/in2/p‘_m ) In. _ 1500—Ln g I\+ v=0.2 C o (‘I \ m o E = 0.45 I +\ II II in A‘ V =F O 28 I \+\+\ I +\+\+\ 1000 — I I I I | I I I _ ,+ 500 “{AL I I I \ J 0 I I | I I I I 0 1 2 3 4 5 6 7 8 DISTANCE FROM CENTER OF PROBE, IN SENSOR—DIAMETERS concentration factors determined from finite element However, an average stress concentration factor (SCF) analyses for different host-material properties are given in could be used for calculation of stress in the host material table 4. The radial and tangential stress concentrations are with a maximum error of 8 percent. A plot of this SCF ver- slightly less than the axial stress concentrations (fig. 12A). sus the host-rock modulus is shown in figure 123. INSTRUMENTATION ANALYSES l3 2°°° l l I l l Steel Epoxy Rock l l + 1500— i \+ D — ml \ 19 gl \ 8 all E= 0.45 l + II n - v=0.28 E=4. (Ib' 2 #‘n- +\+—————-+ 10°03: ;ll 1, =02 N" / in. ) + l | l 500m — 0 l l COMPRESSIVE STRESS, IN POUNDS PER SQUARE INCH 1500— — ID L0 (0 o “l W 0 ll || .— 10003 A 500-— .— 0 0 DISTANCE FROM CENTER OF PROBE, IN SENSOR—DlAMETE RS EXPLANATION —-+—-—+———+— Tangential stress (09) —o——o——0— Radial stress(UR) —o——o——-o— Axial stress (Oz) FIGURE 11 (above and facing page).—Stress distribution from finite-element analysis along a radial line cutting the center of the probe (analysis 1). A, Analysis 1; B, analysis 2; C, analysis 3; D, analysis 4; E, analysis 5. COMPARISON OF STRESS CONCENTRATION FACTORS OBTAINED BY MATHEMATICAL AND FINITE-ELEMENT METHODS The stress concentrations which were determined for the effect of the geometry on the borehole probe and the rock in which it is placed are given in figure 123. One curve was obtained from the theoretical spherical-cylindrical analy- sis, which is summarized in table 1, and the other curve is the average stress concentration factor (SCF) computed by the axisymmetric finite-element method, which is sum- marized in table 4. In view of the assumptions of elasticity and the approxi— mations needed for both techniques, the agreement is close, and use of the probe in monitoring three- dimensional stress changes in elastic material would appear justified. The stress monitored by the probe sensor is, however, sensitive to the modulus of the host material. For a steel sensor, the SCF ranges from 1.90 to 0.42 as the 14 2.5 . . . . a l o 1 ‘12 g 2.0 U EXPLANATION B \ . Axial SCF If \ Cl Tangential SCF 21.5 \ o RadialSCF 9 I— <( Cl: '— z 1.0 L” o z o \ o \\ \ U) 05 \‘N (n m A fig I l- (I) 0 0.5 1.0 2.0 3.0 4.0 5.0 6.0 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO EXPLANATION i 1.94 Theoretical SCF —-— 1.90 X Calculated \ —-—* lnterpolated 1 44 Finite-element axisymmetric SCF 1.44 ' 0 Calculated 1 14 lnterpolated 1.14 0.96 0.94 0.72 \ \ 0-70 \\\ 058 0.53 \ 0.42 K\\\ 0'55 0.50 ‘~‘~< B 0.40 . . . . I 0 0.5 1.0 2.0 3.0 4.0 5.0 6.0 MODULUS OF DEFORMATION OF HOST ROCK, IN POUNDS PER SQUARE INCH PER MICROINCHES PER INCH FIGURE l2.—Stress concentration factors held constant (0.2). (SCF’s) for spherical steel sensor surrounded by epoxy cylinder and host rock. Poisson’s ratio A, Axial, tangential, and radial stress concentration factors plotted against modulus of deformation of host rock. B, Averaged radial and tangential stress concentration factors change of l lb/in2 in the host rock of modulus 6x106 steel sensor; i.e., SCst 0.4} host-rock modulus increases from 0.5 ><106 to 6.0x 106 lb/ in2. Thus, knowledge of the modulus of deformation of the host rock is essential to accurately determine stress changes. ANALYSIS OF LABORATORY BIAXIAL TEST Another use of the axisymmetric finite-element method is in the evaluation of in situ stresses determined by over- coring the borehole probe. Overcoring relieves the rock surrounding the probe from the in situ stress field, causing a stress change in the spherical sensor. This change should allow the calculation of the in situ stress field present in the rock mass. The rock cylinder recovered from the bore- hole with its enclosed probe can be tested in a biaxial loading devlce in the laboratory. By virtue of these known applied loads, a check on the theoretically derived SCF can be obtained. The axisymmetric finite-element method is a means of checking laboratory biaxial core-loading re- sults. The results of this analysis (with free end) (analysis 6, fig. 13A; table 5), when compared with the results of analysis 4 (similar material properties but fixed end), show that the laboratory biaxial device produces the stress con- centration in the radial and tangential directions that would be expected in the field. However, the axial boundary conditions in the field and in the biaxial device are probably different. The axial stress is considerably less in the laboratory biaxial device because no axial stress is applied, whereas some axial component of stress would be anticipated. plotted against modulus of deformation of host rock. [Note A stress lb/in2 results in a stress change of approximately 0.4 lb/in2 on the TABLE 5.—Finite-element model properties for biaxial test (axisym‘ metric analysis) Directions of Epoxy Sphere cylinder Host rock boundary-loading . condItIons Analysts 1E00“) 12 1E00“) 11 lE005) 1; Radial1 Axial 6 30.5 0.285 0.45 0.28 4.0 0.25 1,000 Free 1Units in pounds per square inch. PLANE-STRAIN ANALYSES Plane-strain finite-element analyses were made pri- marily to test qualitatively the effectiveness of the probe in estimating in situ stresses. The method was first tested to verify that it would reproduce a theoretical plane-strain problem. PRELIMINARY PLANE—STRAIN ANALYSIS The results of the preliminary plane-strain analysis (an- alysis 7) are shown in figure 138 as a plot along a radial line perpendicular to the long axis of the probe (similar to fig. 1 l). The host-rock properties (table 6) were the same as TABLE 6,—Finite-element model properties for plane-strain analysis Analysis Sphere EPOXY clender Host rock XI Y1 1E00“) 1} ‘E(10‘) 1; 1E00“) 11 7 30.5 0.285 0.45 0.28 2.0 0.20 1,000 _1,000 1Units in pounds per square inch. INSTRUMENTATION ANALYSES 15 2000 I I I SteelL Epoxy _| Rock T I E= 0.45 | V: 0.28 | + " = 0'25 1500 —m g: l \+ ' N c 8 O" I \+ o H u I | E “1 “I l E | 0.1000— I S ' l o l | l I l | 500 STRESS CHANGE, IN POUNDS PER SQUARE INCH E= 4.0 (lb/inZ/Ifirl Tangential stress Radial stress I 1 In. A m. 3| E=0.45 \ v=0.2 8 oil v = 0 28 u ? ' Y-stress l | l “4 l I I 1000— 500 PER SQUARE INCH COMPRESSIVE STRESS, IN POUNDS S 0 W 2 Axial stress 3 l i 1 | 1 | | l x —250 m 1500 l l I— l l =¢ - I Steer? Epoxy T;\Rock Ex= 2_0 “Mini/"11$ ) + . 4. N‘+\+ +*+h + DISTANCE FROM CENTER OF PROBE, IN SENSOR-DIAMETERS FIGURE 13.—Stress distributions. A, Along a radial line cutting the center of the probe (analysis 6, biaxial test). B, Along the X axis for the plane-strain analysis (analysis 7). for analysis 3 of the axisymmetric analysis. The resulting stress concentration for this analysis, however, is only 0.8 compared to 0.94 for the similar axisymmetric case. This result was to be expected because the plane-strain analysis is a representation of a cylinder superposed on a cylindri- cal inclusion, and the magnitude of stress concentration is greater for a spherical inclusion than for a cylindrical in- clusion (Goodier, 1933). However, in general the stress distributions are similar to those of the axisymmetric analyses, and qualitative indications of the stress orienta- tions can be obtained using a plane-strain model. ANALYSIS OF OVERCORING The plane-strain finite-element method was utilized to analyze simulated stress-relief overcoring procedures in a step-by-step manner using both the US. Bureau of Mines deformation gage and the three-dimensional bore- hole probe. The U.S. Bureau of Mines theoretical biaxial plane-strain solution was used to check the finite-element method. This comparison was desirable because no exact theoretical three—dimensional solution exists for analyz- ing strain changes produced by overcoring the borehole probe. The analysis of this simulated operation proceeded as follows: 1. The model simulated a rock mass with the boundaries sufficiently removed from the eventual borehole location to minimize the boundary effects of the borehole. 2. The rock was then placed under a biaxial hydrostatic stress of 1,000 lb/in2; the borehole was then placed in the model and the resulting stress pattern and displacements were determined. 3. The appropriate gage was then installed and the in- strument was overcored. The resulting stress pattern and displacements were determined. The results of the finite-element analysis of the US. Bureau of Mines stress-relief procedure show a stress dis— tribution about the borehole prior to overcoring that is 16 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO identical with the distribution predicted by theoretical solutions. The stress in the rock cylinder (after over- coring) is zero, as predicted by elastic theory. The radial displacements of the borehole boundary were calculated to be 0.0017 cm (0.00067 in.) after over- coring. This gives a diameter change of 0.0034 cm (0.00134 in.) for the plane-strain finite-element solution. The theoretical diameter change for plane strain under hydrostatic stress derived from Obert and Duvall (1967, p. 413) is U=2pd(1-v2)/E (hydrostatic), (7) where U=the diameter change, d=the borehole diameter, v=Poisson’s ratio, E=Young’s modulus, and p=hydrostatic pressure. Using the same material properties and loads as in the finite-element solution, E=2 X we lb/inz, where v=0.20, d=l.5 in., and p=1,000 lb/in2, this equation yields a diameter change of 0.0037 cm (0.00144 in.). Thus, the theoretical and finite-element solutions agree within 6.9 percent. It is not known whether this difference is due to the size of the finite— element mesh or whether a difference actually exists. The results of a plane-strain finite-element analysis of the borehole-probe overcoring operation are shown in fig- ure 14. The group of curves B shows the stress distribu- tions around the drill hole before installation of the probe, which are the same stress distributions as those previously determined theoretically. The group of curvesA shows the stress distributions within the probe and the rock overcore after overcoring. The stresses in the probe are tensile, as would be expected from a relieved compressive stress field. The radial stress in the rock overcore is tensile, and the tan- gential stress is compressive. The stress concentration in the steel sensor is approxi- mately 0.60, which is lower than the stress concentration of 0.80 predicted in analysis 7, which has the same material properties and load conditions. This is probably due to the use of a plane-strain method of analysis which simulates a cylinder within a cylinder. The results can be applied only qualitatively to the actual three-dimensional problem of a sphere within a cylinder. REQUIREMENT OF TENSILE STRENGTH The plane-strain analysis of overcoring assumes ade- quate tensile strength in the rock mass around the bore- hole probe. In order to obtain stress-relief information, the rock within the overcore must be capable of transmitting a tensile stress. In the preceding example the rock (E=2.0 x106 lb/inz, v=0.2) was subjected to about 600 lb/in2 tension when a 1,000-lb/in2 equal biaxial stress was removed. Field experiments have demonstrated that when over- coring in low-modulus- and low-tensi1e~strength rocks in a compressive stress field tensile failures can occur in the rock near the rock-epoxy interface, prohibiting meaning- ful results. The level of the tensile stress generated in the rock by overcoring and by removal of a compressive stress is a function of the ratio of the rock modulus to the epoxy modulus. If the rock modulus is decreased and approaches the modulus of the epoxy grout (0.45><106 lb/in2), the tensile stress rises. Tuffaceous rock under an estimated 1,000-lb/in2 compression, with a modulus of between 0.7 ><106 and 0.9 x 106 lb/inz, failed in tension in the rock ad- jacent to the probe during overcoring. Granitic rock under 2,000-1b/in2 compression, but with an average modulus of 7><106 lb/inz, did not fail in tension during overcoring. The indicated beneficial effect of reducing the modulus of the epoxy grout for overcoring low-modulus rock will be studied. LABORATORY MODEL-TUNNEL INVESTIGATIONS The purpose of the laboratory-model studies was two- fold: to evaluate the studies of probe behavior previously described and to define the effects of tunneling in simple systems omitting body loads and geologic factors, such as faults and foliation. The laboratory model-tunnel results were compared with the findings from the field tunnel study. The materials used in the model-testing program were chosen to represent a range Of fabric properties, from an amorphous material (acrylic) to an artificial granular aggregate (concrete) to a natural granular aggregate (granite). Because the model tunnels were smooth-walled cylin- drical boreholes, they would resemble “mole”-drilled tunnels more closely than conventional “drill-and-blast” tunnels. The model tunnel in the acrylic was advanced by a twist drill; the concrete and granite model tunnels re- quired use of a diamond-core drill. The drillng caused a significant thermal disturbance in the acrylic model which necessitated a wait to verify stabilityafter each drill- ing increment. The loading was adjusted to maintain the same average stress in the models after each increment of tunnel advance. All model testing was done under room temperature and humidity conditions. The probes used to monitor stress changes in the model-tunnel studies differed somewhat from those used in the field. Rather than the 7 .6-cm (3 in.) borehole used in the field, we grouted the laboratory probes in a 5.1-cm (2 in.) borehole. Also, the laboratory probe was 15.2 cm (6 in.) long, while the field-borehole probe is 43.2 cm (17 in.) long. The LABORATORY MODEL-TUNNEL INVESTIGATIONS 17 I Steel —I Epoxv I Rock overcore I Rock 2000 I I I I I I I I E= 30.5 | E= 0.45 E: 2.0 (Ib/inz/ “II—:1 | EXPLANATION V = 0.285I V = 0-28 I V = 0-2 I Before overcore I o 0 stress (Ba) I I ' I D Rstress (BR) 1500 I . I After overcore _ I I o I o 0 stress (A0) I I . @Rstress (AR) I . I 1000 Ib/in2 biaxial I . I (radial) load 3 I I .. I g I I I . . ' 0 31000 — I I I Be . E g I - W U. I I 0: I - g I I . B. 8 I .. a I . I 0. I I I U) 500 — ' — 2 I I I 3 I I - 2 I I A I E I I I 9 (If If} I E 0 Ba and I B, I I o I I 1 2 O 3 4 5 6 o I o SENSOR—DIAMETERS FROM CENTER OF PROBE I 0 Al? I I A0 I o I .5 AR I R 0 ° ‘6’ I g —500 — and I o o I — I3 A9 0 I I o 0 O I I I a I I I .4000 I I I I I R FIGURE 14,—Plane-strain finite-element analysis of stress distribution in borehole probe and rock along R axis. Curves B show stress in rock around drill hole before installation of probe. Curves A show stress in probe and rock overcore after stress relief by overcoring. diameter of the spherical sensor is 2.5 cm (1 in.) for all applications. Stress changes were monitored for each increment of tunnel advance. Tunnel advance was accomplished by re- moving the model from the hydraulic press, drilling the necessary distance, replacing the model in the press, and successively loading and unloading the model. Normally, two cycles were sufficient to demonstrate the repeatability of stress-strain response at the probe location. It was not physically possible, although desirable, to advance the tunnels under load. Changes in stress, however, appear to be consistently related to changes in tunnel length which occurred between loadings. The tunneling methods had no measurable effect on stress changes near the moni- toring probe. ACRYLIC MODEL The acrylic model containing a centrally located probe (fig. 15) was tested under an average uniaxial stress of 1,950 lb/in2 prior to drilling the tunnel. The resulting strains measured by the sensor were equivalent to a compressive stress of 2,100 lb/in2 parallel to the vertical loading direc- tion, 1,000 lb/in2 in tension horizontal and approxi- mately perpendicular to the axis of the probe borehole, 18 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO _ ~‘ Borehole probe \f". FIGURE 15.—View of acrylic model showing three—dimensional bore- hole probe grouted into central drill hole. Strain gages (A—E) were used in previous testing (Nichols and others, 1968, p. C14). Tunnel was advanced from rear face. Model measures 48.3 cm (19 in.) high, 33.0 cm (13 in.) wide, and 30.5 cm (12 in.) deep. Arrows indicate loading direction. and 1,820 lb/in2 in tension horizontal and approximately parallel to the axis of the probe borehole. The 4.78-cm- (1.88 in.) diameter tunnel was advanced by drilling into the side of the acrylic model opposite the sen- sor implacement hole (in a direction arbitrarily chosen as PJ.45° E.) As the tunnel in the acrylic model was advanced in in- crements toward the probe, the probe sensed an increase in compression in the direction of loading. The principal stress changes in the plane normal to the direction of load- ing were very small. The directions of the principal stress changes remained essentially fixed (table 7; figs. 16A, 17). Stresses estimated from photoelastic studies by Galle and Wilhoit (1962), in a smaller but similarly loaded plas- tic block ahead of a 3.18-cm- (1.25 in.) diameter bore, are in good agreement with the stresses monitored by the probe for a 1.50-tunnel-diameter interval between the probe and the tunnel face (fig. 163). Differences existed, however, when the face was between 1.50 and 2.25 tunnel diameters from the monitoring probe. TABLE 7.—Average stresses determined ahead of tunnel face in acrylic model [Acrylic properties: E=0.327><106 lb/in2, v =0.39, tensile elastic limit=12,000 lb/inz. Probe properties: epoxy grout, E=l.55 X106 lb/in2. v=0.39; steel sensor, E=30.5x105 lb/in’,"’=0.285; SCF=2.73. Average stress applied=l,950 lb/in2 vertical. +, compressive stress; —,tensile stress] Dia _ Maximum principal Intermediate principal Minimum principal m stress stress stress eters from - . ._ sensor Magnt Bearing Plunge Magni Bearing Plunge Magm Bearing Plunge rude tude center (lb /in2) (degrees) (degrees) (lb /in2) (degrees) (degrees) (1b /in2) (degrees) (degrees) 4.47 +2,l00 N. 66 W 74 -l,000 S 39 E. 14 -l,820 N. 50 E. 7 3.93 +2,100 N. 65 W 73 -980 S 36 E. 15 -l,750 N. 52 E. 8 3.39 +2.170 N. 65 W 73 -l,040 S 38 E. 15 —l,820 N. 50 E. 8 2.85 +2.190 N. 66 W 73 -990 S 36 E. 15 -l.780 N. 52 E. 8 2.52 +2260 N. 65 W 73 —l,020 S 36 E. 15 —l,790 N. 52 E. 8 2.05 +2300 N. 63 W 73 -l,020 S 36 E. 15 —l,780 N. 52 E. 8 1.51 +2.210 N. 70 W 73 -1,040 5 35 E. 14 —1,950 N. 52 E. 10 1.24 +2,180 N. 70 W 72 -l,000 S 33 E. 14 -1.B60 N. 54 E. 10 .97 +2290 N. 69 W 72 —980 S 82 E. 14 4,860 N. 55 E. IO .70 +2520 N. 66 W 72 —l,000 S 30 E. 14 -l,800 N. 57 E. 10 .43 +2.570 N. 65 W 71 4,030 S 25 E. 15 -l.860 N. 62 E. 12 +500 +400 — In D z t: D -9+3oo — O I 3 n. U E z E g -.I.U 0 +200 _ Lu I U (D < <2: 3 O _ I ‘0 +100 0 a: a r u.I 0 D: 5 s .5 _100_ + Segsor I: o ‘— —200 1 ' ‘ A +3200 I I I I l +3000 _ I Elastic stress data from g +2800 - Galle and Wilhoit - Lu 0: < +2600 * / — D 8 / I +2400 _ Stress determined by probe / / _ 3 One standard / —\ deVIatIon / +2200 _.,——/ f +2000 I I I I I I | I 5°15“ 41/. 4 3y. 3 27. 2 11/2 1 y. TUNNEL DIAMETERS FROM CENTER OF SENSOR COMPRESSIVE STRESS, IN POUNDS FIGURE Iii—Changes of magnitude in principal stresses (A) and maxi- mum principal stress compared to results of Galle and Wilhoit (1962, p. 148, fig. 7b) (B) in acrylic model. LABORATORY MODEL-TUNNEL INVESTIGATIONS 19 NORTH principal Maximum principal\* stress stress 63 . Direction of loading Intermediate principal\ stress xX" x FIGURE 17,—Bearing and plunge of principal stress changes in acrylic model (lower hemisphere equal-area projection). Specific findings from the acrylic model study requir- ing further explanation are: 1. The stress before tunneling is 7.7 percent above the average applied uniaxial stress, 2,100 lb/in2 instead of 1,950 lb/inz. 2. The tensile stresses present in the model with respect to the loading (maximum principal) stress. 3. The approximate 9_-percent peak in the maximum principal stress concentration that occurred at 1.75 tunnel diameters ahead of the advancing model tunnel. These findings will be evaluated after the presentation of the results obtained from the concrete and granite models. CONCRETE MODEL A second model was constructed of high-strength con- crete (fig. 18). The concrete model with its grouted sensor was initially tested under an average vertical uniaxial stress of approximately 1,3101b/in2 after the 4.78-cm- (1.88 in.) diameter tunnel had been driven 4.44 cm (1.75 in.). The resulting average strains measured by the sensor were equivalent to approximately 1,420 lb/in2 in compression in the vertical loading direction, 500 lb/in2 in tension nor- mal to both the vertical loading direction and to the axis of the model tunnel, and 1501b/in2 in tension parallel to the horizontal axis of the model tunnel. The same incremental tunnel-advancing procedure was used with the concrete as had been used for the acrylic tunnel-model experiment. The direction of tunnel advance was arbitrarily denoted east. The stress magni- tudes and directions sensed by the probe in the concrete FIGURE 18.—Concrete tunnel model. Block has been sawed parallel to borehole axis, exposing strain sensor. Tunnel was advanced from left to right. Arrows indicate loading direction. Scale in inches. model after each advance of the tunnel are presented in table 8 and in figures 19A and 20. The results from the concrete model in the direction of applied uniaxial loading (fig. 193) agree rather closely with the photoelastic data in the region less than 1 tunnel diameter ahead of the advancing tunnel face. The stress— strain response of the concrete model as detected by the probe was remarkably similar to that of the acrylic model, as follows: 1. The stress before tunneling is 8.4 percent greater than the applied uniaxial stress—1,420 lb/in2 rather than 1,3101b/in2. 2. The sizable variation between the two tensile princ- ipal stresses normal to and parallel to the tunnel axis and (or) the axis of the probe hole that occurred in the acrylic model was also present in the concrete. TABLE 8.—Ave1age stresses determined ahead of tunnel face in concrete model [Concrete properties: E=3.5x10‘ lb/in’, v'=0.26, compressive strength=7,580 lb/inz. Probe properties: epoxy grout, E=0.45x105 1b/in2,v=0.28; brass sensor, E=15.0x1061b/in2, u=0.285; SCF=0.614. Average stress applied = 1,310 lb/in2 vertical +, compressive stress; —, tensile stress] Diam- Maximum principal Intermediate principal Minimum principal eters stress ‘_ > stress stress from '. ‘. sensor Mtifitl Bearing Plunge Mia? Bearing Plunge Malgm Bearing Plunge center (1b /in2) (degrees) (degrees) (lb/inz) (degrees) (degrees) (lb /in2) (degrees) (degrees) 2.66 H.420 S. 31 W. 82 -140 N. 5 W. 7 -510 N. 86 E. 5 2.00 +l,580 S. 30 W. 82 —140 N. 5 W. 7 —540 N. 86 E. 5 1.45 +1,570 S. 30 W. 82 —l70 N. 7 W. 6 —610 N. 84 E. 5 1.18 H.640 S. 32 W. 83 -100 N. 5 W. 6 —580 N. 85 E. 4 .91 +1,610 S. 34 W. 83 -70 N. 6 W. 5 4320 N. 85 E. 5 .64 +1,670 S. 36 W. 82 —l70 N. 6 W. 6 —710 N. 84 E. 5 .37 +1,870 S. 42 W. 83 -220 N. 7 W. 4 —740 N. 83 E. 5 20 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO +500 | +400 n” + to o o | Compression + M o o I +100 — Sensor STRESS CHANGE, IN POUNDS PER SQUARE INCH L O O | Decompression l M o o | —300 +2200 I , Elastic stress data from Galle and Wilhoit _ (1962, p. 148, fig. 7b) +2000 +1800 —‘ Stress determined / / biz/\pmbe>< One standard /// \"‘ +1600 — +fl400 — ,a’ | I I I I +12003 2y; 2 1y, 1 y. o TUNNEL DIAMETERS FROM CENTER OF SENSOR COMPRESSIVE STRESS, IN POUNDS PER SQUARE INCH FIGURE 19.—Changes of magnitude in principal stresses (A) and maxi- mum principal stress compared to the results of Galle and Wilhoit (1962, p. 148, fig. 7b) (B) in concrete model. 3. A minor, but significant, compressive stress concen- tration in the loading direction was sensed by the probe. In the concrete model this was between approximately 1 and 2.25 tunnel diameters ahead of the advancing tunnel, whereas in the acrylic model the concentration occurred between 1.5 and 2.5 tunnel diameters ahead. A maximum principal stress concentration peak of 9 percent occurred approximately 1.25 tunnel diameters in front of the advancing tunnel in the concrete model (fig. 19B). This compares to a similar peak at 2 tunnel di- ameters in the acrylic model (fig. 16B). 4. The minimum principal stress showed a sudden in— crease when the tunnel was about 1.75 tunnel diam- eters from the center of the probe. In the concrete model, however, the minimum principal stress did not return to its previous low level, as it had in the nearly elastic acrylic model. NORTH 1M 1 ‘ \ Intermediate principal stress Minimum Direction principal Stress 1 BOREHOLE of loading \: CENTERLINE . \ . . . MaXImum prInCIpaI stress FIGURE 20.—Bearing and plunge of principal stress changes in concrete model (lower hemisphere equal- area projection). GRANITE MODEL The third tunnel model (fig. 21) consisted of a block of Silver Plume Granite 30.48 cm (12 in.) high, 43.18 cm (17 in.) wide, and 43.18 cm (17 in.) deep. Concrete platens, 10.16 cm (4 in.) high, 33.02 cm (13 in.) wide, and 30.48 cm (12 in.) deep, were cemented with epoxy to the sides of the granite, extending its height to 50.80 cm (20 in.). The con- cre te platens were cut from the concrete model, for which E=3.3x106 lb/inz, v =0.26. The properties of the granite as determined from specimens removed from the block after testing are similar to those of the concrete; namely, E=4.5><106 lb/in2, v =0.23. The effect of the lower stiffness (E) and higher Poisson’s ratio of the concrete platens is to increase the lateral tensile strain in the granite at the granite-concrete interface by approximately 11.7 ,uin./in., or 53 lb/in2 per 1,000 lb/in2 of applied uniaxial (vertical) stress—this in addition to a measured lateral tensile stress of 295 lb/in2 per 1,000 lb/in2 applied vertical compressive stress. Any lateral tension due to the mismatch of model and platen materials appears to have been restricted to the immediate vicinity of the platen-rock interface because the probe did not detect such an increase in lateral tensile stress. Therefore, the probe was influenced only by those LABORATORY MODEL-TUNNEL INVESTIGATIONS 21 FIGURE 2l.—Granite tunnel model showing grouted sensor. A 5.24-cm- (21/15: in.) diameter tunnel was advanced from the rear side. Arrows indicate loading direction. Scale in inches. lateral extensional strains resulting from the absence ot lateral boundary loads. The granite model with its grouted sensor was initially tested under an average vertical uniaxial stress of 1,155 lb / in2 after the 5.26-cm- (2.07 in.) diameter tunnel had been driven 5.08 cm (2 in.) in a direction arbitrarily designated north. The resulting average strains measured by the sensor were equivalent to 1,380 lb/in2 in compression in the vertical loading direction, 90 lb/in2 in tension 40° clockwise from the tunnel axis in the horizontal plane nor- mal to the vertical loading direction, and 690 lb/in2 in ten- sion 50° counterclockwise from the tunnel axis in the hori- zontal plane. An incremental tunnel-advancing procedure was used, as in the previous models. The changes in principal stress magnitudes and directions during tunnel advance in the granite model are presented in table 9. Figure 22A pre- sents the principal stress changes and figure 23 the princi- pal stress orientations. The stress changes near the tunnel face in the granite TABLE 9.——Avemge stresses determined ahead of tunnel face in granite model [Granite properties: E=4£5>x106 lb/in”, V=0.23. Probe properties: epoxy grout, E=0.45 X106 lb/ini, u=0.28; steel sensor, E=30.5)Ol_ 110 ZO—-|O|T1:U-U an IN 0=o° PLANE (PARALLEL TO LOADING DIRECTION) EXPLANATION 0'5 = Radial stress =Tunne| diameter Contours in Ib/in 2 Compression (—) FIGURE 24.—Stress distribution adjacent to model tunnel in photo- elastic material modified from Galle and Wilhoit, 1962, p. 148). for the acrylic model, in figure 198 for the concrete model, and in figure 228 for the granite model suggest that these subsidiary higher stressed zones owe their existence to a real stress concentration and not to instrument error. Our field evidence agrees apprOximately with an onset of stress increase at about 2— 5 tunnel diameters ahead of a model- tunnel face. The model used by Galle and Wilhoit (1962) may have been of too limited an extent to show this stress peak because their models extended only 2.25 diameters ahead of their model-tunnel face (fig. 24). These tests suggest that model geometry and dimensions strongly in- fluence elastic solutions. The anomalous vertical compressive stress concentra- tions which were measured from 2 to 5 diameters ahead of the model tunnels did not continuously increase to the predicted stress of approximately 1.50 times the average before-tunneling stress. A decrease in vertical compressive stress occurs prior to the final stress rise ahead of the tunnel face. In order to explain this phenomenon, one must con- sider that a decrease in confinement in the direction of the advancing tunnel would produce a decrease in stress (re- duction of restraint) in the direction of tunnel advance. The increase in maximum principal compressive stress in the direction of loading results in small additional increases in the intermediate and minimum tensile stresses acting perpendicular to the loading direction (figs. 16A, 19A, 22A). A small increase in tensile stress in the direction of the approaching tunnel in the acrylic model accompanying a small decrease in the maximum principal compressive stress occurs between 1 and 2 tunnel diameters ahead of the tunnel (table 7; figs. 16A, 17). The tensile stress perpen- dicular to the direction of the approaching tunnel decreases slightly. In the concrete model a small increase in the lateral ten— sile stresses occurs in the direction of the approaching tun- nel, accompanied by a decrease perpendicular to the ap- proaching tunnel, within the zone of decreased maxi- mum principal compressive stress from 0.75 to 1.25 tun- nel diameters ahead of the advancing tunnel (table 8; figs. 19A, 20). These results'are similar to those obtained from the acrylic model. In the granite model the minimum principal stress in- creases in tension between 3.50 and 1.50 tunnel diameters ahead of the tunnel face, whereas the intermediate princi- pal stress remains slightly compressive (table 9; figs. 22A, 23). The average lateral tensile stress ((63+02)/ 2) increases throughout the zone ahead of the advancing tunnel. The granite model differs from the other two models in that the intermediate and minimum principal stress directions are not oriented perpendicular to and parallel to the tunnel axis. The granite was anisotropic in the horizontal plane, having a specimen modulus of 4.89x106 lb/in2 and a Poisson’s ratio of 0.210 in the intermediate principal stress direction and a specimen modulus of4.85><106 lb/in2 and a Poisson’s ratio of 0.244 in the minimum principal stress direction. The influence of the anisotropy in Poisson’s ratiocould account for a difference in horizontal stress of approximately 80 lb/in2 between the intermediate and minimum principal stresses in the granite. The average actual difference was approximately 590 lb/inz, or about seven times larger than can reasonably by explained by elastic anisotropy. No Obvious explanation exists for the magnitude Of the nonsymmetric lateral stress in the granite model. The directions of the intermediate and minimum principal stresses are, however, in agreement with those calculated from the elastic anisotropy, with the stiffer member carrying the greater stress. The average horizontal-to-vertical stress ratio deter- mined for the granite model agrees reasonably well with the plane-strain predicted stress ratio (table 11). The ratio 24 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO TABLE 11.—Change in lateral tensile stress in granite model [Mean difference=2.9 percent; standard deviation=7.5 percent: 95- percent confidence interval=2.9 to 8.7 percent] Diameters ~ . . from Measured Tallo Predicted ratioI Percent sensor AVg’qior /overt. Avg. o‘hoir /avert difference? 3.53 0.275 0.295 —6.8 2.57 .282 .295 -3.4 2.09 .286 .295 -3.0 1.61 .300 .295 +1.7 1.37 .306 .295 +3.7 l.l3 .307 .295 +4.1 .89 .308 .295 +4.4 .65 .335 .295 +13.6 .41 .331 .295 +122 1Based on absence of restraint, plane~strain elastic relations, and physical specimen tests. 2 Measured ratio — predicted ratio _——-—— x 100. Predicted ratio of average horizontal stress ((az'+a'3)/ 2) to average vertical stress increased with the increased longitudinal stress resulting from tunnel advance. This implies a change in Poisson’s ratio with either the change in confinement or increasing uniaxial stress. During testing of granite speci- mens, an increase in Poisson’s ratio from 0.20 to 0.24 was measured from 0 to 10,000 lb/inz. No change in Poisson’s ratio could, however, be detected over the small stress range measured in the granite model (1,370—l,560 lb/inz). The release of confinement, induced by the approach- ing tunnel, appeared to be the dominant factor in the behavior of the acrylic model. Poisson’s strain was ap- parently more important in the concrete model, but release in confinement was present. The response of the granite appeared to be related more to anisotropic variations in the granite than to release of restraint related to tunneling. In all three models, the maximum tensile stress developed in the material most capable of carrying it: the stiffer epoxy grout in the case of the acrylic model, the stiffer concrete in the case of the concrete model, and the direction of lower Poisson’s ratio and greater stiffness of rock in the case of the granite model. Indicated anomalous stresses cannot be ascribed to instrument error. But an anomalous condition exists that cannot be supported by available elastic theory or by measurements. DISCUSSION OF LATERAL DEFORMATION The before-tunneling load-induced horizontal tensile stresses in the acrylic model were approximately 1,8201b/ in2 parallel to the direction of the emplacement borehole and approximately 1,000 lb/in2 normal to the emplace- ment borehole. The induced horizontal tensile stresses in the concrete block 2.50 tunnel diameters ahead of the sensor were approximately 150 lb/in2 parallel to the direc- tion of the emplacement hole and approximately 500 lb/ in2 normal to the borehole. The induced horizontal ten- sile stresses in the granite block, in the zone from 1.50 to 3.50 tunnel diameters ahead of the sensor, were approxi- mately 90 lb/in2 42° clockwise from the axis of the em- placement hole and approximately 700 lb/in2 49° coun- terclockwise from the axis of the emplacement hole. The existence of internal lateral tensile stresses in a uni- axial test specimen has been observed or has been postu- lated by other investigators. Conway(1963, p.131) used photoelastic freezing techniques to observe high lateral tensile stresses in uniaxial tests conducted on CIBA epoxy, with both low—friction end conditions and high-friction end conditions. With the low-friction end condition, Con- way observed a nearly uniform radial tensile stress across all longitudinal sections of the specimen, with a stress dis- continuity at the external boundary. With the high- friction end condition, he again observed a nearly uni- form radial tensile stress across all longitudinal sections to within one-tenth of the specimen length from the ends. Conway stated (p. 131) that these tensile stress com- ponents directly contradict elastic theory and that (p. 149) they are approximately equal to one-half the axial princi- pal stress. Brown and Trollope (1967, p. 234), using the principle of superposition, postulated that there are effective stresses (stresses induced perpendicular to the uniaxial loading direction) within a loaded body which are related by the expressions: (T; =(Tx_ v,(0'y+a' z); (8) 0-3, :oy-V'er +oz), (9) and 02 :az‘vlfirx‘l'a'y), (10) where v’ is a parameter dependent upon the structure of the material, a; , a3? , and 0'," are internal effective stresses, and ox, 0’3), and U; are boundary stresses. From these ex« pressions it may be seen that internal tensile horizontal effective stresses are present in a uniaxial test. Judging from our experimental results and those of Conway (1963), the concept of effective stresses seems to be real and in disagreement with elastic theory. If we use equations 8, 9, or 10 to solve for effective internal stress components, then we must determine the value of v’ and its physical significance. Brown and Trollope (1967, p. 234) stated that v’ for ideal linear materials may be equal to Poisson’s ratio (v), but our data do not confirm this. The acrylic model and Conway’s epoxy model probably are as close to ideal linear materials as any real material can be. The concrete and granite models probably are anisotropic with local stress concentrating structures (pebbles and grains), but they have a nearly linear response. Our data, however, indicate that v’ more nearly approximates v/(l—v), and Conway’s data also indicate that the same relationship may exist. FIELD INVESTIGATIONS 25 The restraining stress necessary to prevent elastic tens- ile strains from developing in such uniaxially loaded iso- tropic elastic bodies is equal to v/ (1— v) times the applied uniaxial stress (Obert and Duvall, 1967, p. 474). In the case of the acrylic model, the restraining compressive stress necessary to prevent that model from deforming laterally should be 0.39 140.39 Similarly, the restraining lateral, or horizontal, stress in the concrete model should have been about 500 lb/in2 in tension and uniform in all directions: < 0‘26 >(1,4201b/in2)=5001b/in2. 1'—0.26 >(2,100 lb/in2)=1,340 lb/inz. The granite model would be subject to a restraining stress of 395 lb/in2 in all lateral directions: < 023 )(1,3201b/in2)=395 lb/in2. 1—»0.23 The average lateral stress components measured in the acrylic-epoxy model, the granite model, and the concrete model prior to tunneling were, respectively, “41,410, —395, and +325 lb/in2 (all in tension). The absolute values of the effective lateral tensile stress components in a uni- axial test, therefore, approximately agree with the com- pressive lateral stresses required for perfect lateral confinement. Our test results are similar to Conway’s (1963), and indi- cate a sharp stress discontinuity at the boundary of the model. This is necessary to account for the difference between a’ and a. In addition to the effects of material structure, 1” may also depend on the geometry and the degree of confine- ment afforded by the body, and possibly on the intrinsic characteristics of the body which determine the nature of the elastic constants v and E. The variation of the elastic constants may well determine the value of v’. The average lateral stress, 1,410 lb/in2 in tension, measured in the acrylic-epoxy model prior to the start of model tunneling is greatly exceeded by the tensile strength of the acrylic (12,000 lb/in2). The tensile strength of the high-compressive-strength concrete was estimated to exceed 800 lb/in2 (Portland Cement Assoc., 1952, p. 5—6). Tensile tests performed on concrete cored from this model indicateda tensile strength of 485 lb/inz, considerably greater than the average lateral tensile stress of 3251b/in2 measured in the anisotropic concrete model before tun- neling. The average lateral tensile stress measured by the probe in the granite prior to tunneling, 395 lb/inz, was much less than thé 880-lb/in2 tensile strength determined by physical specimen testing. CONCLUSIONS FROM MODEL STUDIES The model studies increased our confidence in the bore- hole probe and justified its further use. These results showed that the borehole probe could accurately deter- mine changes in stress magnitude and direction. Both compressive and tensile changes were correctly observed. The reproducibility of stress determinations was shown over a range of known applied stresses within three mater- ials of contrasting physical properties. The influence of boundary conditions was clearly detected by the borehole probe. These studies indicated the areas of application and limitation of elastic theory to predict stress distribution related to the tunnel excavation in rock. The model-tunnel studies revealed two surprising find- ings: first, that stress changes occur considerably farther ahead of an advancing tunnel face than theoretically expected, and second, an anomalous stress concentration effect preceding (one or more tunnel diameters ahead of) the anticipated stress buildup near the tunnel face. These tunnel effects were useful in the interpretation of the field-tunneling studies which are reported in the fol- lowing section. FIELD INVESTIGATIONS ROCK MECHANICS INVESTIGATIONS AT THE COLORADO SCHOOL OF MINES EXPERIMENTAL MINE The analysis of stress changes related to underground excavation requires a knowledge of the in situ stress field and the significant geologic properties of the rock mass. These significant geologic properties primarily include rock-mass structural discontinuities and compositional features. Knowledge of these properties is necessary because stresses commonly have geologic controls (Bielen- stein and Eisbacher, 1969; Lee and others, 1969). The de- formation of a rock mass is never strictly homogeneous, in part because geologic materials are, as a rule, heterogen- eous and discontinuous. The scale of significant geologic properties may range in size from individual mineral grains to mountain chains to tectonic provinces. There- fore, a satisfactory understanding of the stress field in one part of a mine usually requires knowledge of the geologic history and geologic features over a larger area. SCOPE OF FIELD STUDIES The field investigations were carried out in the experi- mental mine of the Colorado School of Mines, located at Idaho Springs, Colo., 45 km (28 mi) west of Denver in the Front Range of the Rocky Mountains (fig. 25). R. S. Culver of the Colorado School of Mines helped to coordinate the initial US. Geological Survey instrumentation program 26 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO 106‘ 105° l WELD ,2 . _ 6, <§> BOULDER GRANDQ Q: T ’ —— — ADAMS ‘27 ,9 GILPIN I J {i/ Efp‘Eiirrr‘wgntal = {a DENVER Springs CLEAR CREEK 39’45’ — Georgetown ARAPAHOE M | l JEFFERSON if ,/ | ’/ DOUGLAS i / 1 I 20 MILES 0 10 20 KILOMETRE-S FIGURE 25.—Location of Colorado School of Mines experimental mine at Idaho Springs, Colo. with the excavation schedule and with rock mechanics studies of other groups in the mine. This coordination provided more information from several types of instru- ments than otherwise would have been available. These re- sults will be mentioned, where pertinent, in the following sections. The purpose of our field investigations was to deter- mine what geologic properties cause deviations from gravitational stresses and, in turn, the relation of stress changes produced by tunneling to specific significant geo- logic properties. GEOLOGIC SETTING OF THE IDAHO SPRINGS AREA The general geology of the Idaho Springs area has been described comprehensively by several investigators, in- cluding Spurr and Garrey (1908), Bastin and Hill (1917), and Lovering and Goddard (1950). Various features of the Precambrian bedrock have been treated in recent, more de- tailed studies; structural geology and rock units have been reported by Moench (1964), joint patterns by Harrison and Moench (1961), wallrock alteration by Tooker (1963), general economic geology by Moench and Drake (1966a), and uranium deposits by Sims and Sheridan (1964). Moench, Harrison, and Sims (1962) and Tweto and Sims (1963) gave comprehensive summaries of igneous re- lationships, tectonic activity, and structural geology. The Precambrian bedrock in the Idaho Springs area comprises a generally conformable series of folded meta- sedimentary gneisses and metaigneous and igneous rocks (Harrison and Moench, 1961). Most of the gneissic rocks are thought to represent clastic sedimentary rocks that have been deformed, recrystallized, and partly reconstitu- ted at considerable depth and at high temperatures. Most investigators believe that the ubiquitous foliation repre- sents bedding planes. Three varieties of granitic rocks have been recognized in the Idaho Springs area. These rocks were intruded into the metasedimentary gneisses during Precambrian time and include granodiorite, quartz diorite, both evidently related to the Boulder Creek batholith, and biotite-muscovite granite, related to the Silver Plume Granite. The granitic rocks form many small bodies that are generally concor- dant but may be locally discordant. Bodies of granodiorite and quartz diorite have gneissic border zones that are parallel to the foliation and lineation in the gneissic country rocks, suggesting that these rocks were deformed and metamorphosed after their emplacement. The biotite- muscovite granite is rather massive and was emplaced after the principal gneiss-forming metamorphic event. The Precambrian geologic history of the Idaho Springs area has been summarized by Harrison and Moench (1961, p. B2). The major events recognized by these authors are as follows: 1. Precambrian sediments were deeply buried and reconstituted into high-grade gneisses. 2. The foliated metasedimentary rocks were plastically deformed into major folds with north-northeast-trending axes. The deformation was accompanied by the intrusion of granodiorite, and then minor amounts of quartz diorite and associated hornblendite. 3. Biotite-muscovite granite was intruded near the end of the period of plastic folding. 4. Uplift and erosion of several thousand feet of cover occurred. 5. The Precambrian rocks were deformed locally. Where deformed, the more massive rocks were crushed and granulated; the more foliated gneissic rocks were formed into small terrace, mono- clinal, or chevron folds; also some foliated metasedimentary rocks were cataclastically deformed. The generally deformed condition of most rocks in cen- tral Colorado is testimony that this area has been a region of crustal activity throughout much of geologic time. The gneisses show strong foliation, drag folds, mineral aline- ments, and slickensides, as well as high-grade metamor- phic mineral assemblages. These rocks belong to the sillimanite-almandine-orthoclase subfacies of the al- mandine-amphibolite facies of Turner and ‘Verhoogen (1960, p. 550). The same authors stated (p. 553) that rocks such as these were metamorphosed in an environment having a temperature range of 550°—750°C and pressures of between 60,000 and 120,000 lb/in2. During the Tertiary Laramide orogeny, a sequence of porphyry dikes and irregular plutons was intruded into the Precambrian metamorphic and igneous terrane in the Idaho Springs area. These rocks constitute part of a belt of porphyries that extends northeastward across the Front Range and, together with Tertiary mineral deposits, con- stitutes the Front Range mineral belt (Moench and Drake, 1966a, p. 25). The porphyries were intruded along preex- FIELD INVESTIGATIONS 27 isting planes of weakness in the Precambrian rocks: along joints, foliation surfaces, contacts, faults, and axial planes of folds (Tweto and Sims, 1963). Most of the dikes trend northeastward, but some strike northwest and a few short ones strike east. Faulting occurred on a grand scale during Tertiary time. Early Miocene to Pliocene faulting displaced mountain versus valley blocks as much as 12,200 m (40,000 ft) vertically (Scott, 1973). Some of the youngest (vein) deposits have been offset in places by movement along faults or fractures (Lovering and Goddard, 1950, p. 170-171), indicating that Laramide stresses were signifi- cantly affecting the rock long after the major Laramide igneous (porphyry) intrusions. The bedrock in the Idaho Springs area indicates a com- plex geologic history, and the stress field associated with the bedrock of this area has changed in response to geologic events. The regional geologic events have pro- duced the existing geologic and stress conditions present at the field site. These conditions significantly influenced the behavior of the excavations made in the experimental mine. Field experiments by the US. Geological Survey in the Colorado School of Mines experimental mine began in December 1966, when two three-dimensional borehole probes were installed (fig. 26). This date also marked the first field test of the borehole probe. At later dates, 12 addi- tional probes were installed at several locations in the mine. Five instruments were strain-relief overcored in order to determine the in situ stress field. Nine probes were used to monitor time-dependent changes of rock stress related to an active excavation. These were monitored pe - riodically for as long as 4 years. The methods used to in- stall the three-dimensional borehole probe were described by Lee, Nichols, and Abel (1969). The experimental excavation involved two large rooms (1 and 2) and a central access crosscut between drifts B Left and C Left. Plate 1 is a map of rooms 1 and 2 showing the geologic structures and the locations of the probes. Figure 26 shows the mining sequence in each room and in the central crosscut. The overburden in this part of the mine is about 107 m (350 ft). Geologic mapping was done throughout the period of . field investigations. This procedure allowed us to record rock-mass information as new excavations were made near previously mapped locations. The detailed mine maps of Moench and Drake (1966b) provided helpful background information on the major geologic structures in the ex- perimental mine. In the vicinity of the two rooms, most of the rock is bio- tite or hornblende gneiss and gneissic granite with lesser pegmatitic granite and amphibolite. Geologic structural features include two major through-going subparallel faults, three major joint sets, and a conspicuous gneissic foliation (pl. 1; fig. 29). The faults trend east-northeast with a shallow southeast dip and intersect nearly at right angles with the steeply dipping east-northeast-trending foliation. Fault 1 (the southern fault) intersects room 1; fault 2 (the northern fault) intersects the extended central crosscut (pl. 1). Fault 1 has as much as 0.03 m (0.1 ft) of associated gouge and as much as 0.6 m (2 ft) of sheared, mineralized rock on both the hanging wall and the foot- wall. Fault 2 has 0.02—0.12 m (0.05—0.4 ft) of gouge and as much as 3.7 m (12 ft) of sheared and altered rock—— mainly confined to the hanging wall. Both faults are ex— posed in other openings, as much as 30.5 m (100 ft) away, where they have similar attitudes. One of the major joint sets parallels the generally steeply northwest-dipping, east-northeast-trending foliation usually; another major joint set strikes northwest and dips steeply to the north- east; and a third set dips steeply and strikes northeast approximately perpendicular to the excavation direction. Mapping of joints in the study area included a visual assessment of the amount of alteration associated with the joint as well as significance (rank) and average spacing of joints. This information was helpful in understanding the complex deformation behavior of the rock mass during and after excavation. For example, where there was strong decompression (A05) coincident with the orientation of a joint set the infOrmation recorded on the geologic map ad- jacent to each joint symbol (pl. 1) was examined. If gouge, or intense fracturing were associated with the joint sur- faces and if the j oints were closely spaced, we would expect the rock to be more compressible perpendicular to such joints than parallel to them. This directional anisotropy, if marked, should be detectable from analysis of the three- dimensional stress changes. Indeed, the rock mass seldom responded uniformly to excavation-induced stress changes. Differences in stiffness, controlled by the proper- ties of joints, were important, particularly in the behavior of the rock mass around room 2. IN SITU STRESS FIELD Five determinations of the in situ stress field were made in the study area by overcoring probes B-3, 17,18, 21, and 22 at various locations (pl. 1). The three-dimensional stress field was determined by stress—relief overcoring at five sites, three approximately 4.6 m (15 ft) northwest of room 2 and two approximately 7.6 m (25 ft) west of room 1. Figure 27 and table 12 show the stresses determined. The composite results of these in situ principal stress determinations show a consistent clustering of orienta- tions (fig. 27). These clusters of preferred principal stress orientations each contain five principal stress deter- minations. Cluster ”A contains three maximum and two intermediate principal stress determinations; cluster aB contains two maximum, two intermediate, and one minimum principal stress determinations; and cluster "C contains four minimum and one intermediate principal stress determinations. Figure 28 suggests that there may be 28 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO . Z Probes B~2 (overcored Feb. 27, 1968) 18 (oyercored Mar. 20, 1968) Probe 21 (overcored Feb. 26, 1968) Probe 22 (overeored Mar. 19, 1968) Probe 8—5 (Apr. 30, 1968) Probe 25 at 1.2 m (4 ft) Apr. 1, 1968 Probe 24 at 5.5 "1118 ft)ADr.1,1968 Probe 19 at 1.2m (4 ft)Apr.1,1968 Probe 15 at 5.5 rn (1B ft)ADr-1,1968 Probe9(July 28,1967) 10/26 Apr. 4, 1968 presplit June 18, 1968 Dresplit USBR strain relief overcore borehole Mar. 1968 Probe 10 (July 28,1967) Apr. 4, 1968 presplit > Probe 8 0.3m or 1 ft 7/15 below floor (Dec. 9, 1966) EXPLANATION 7/11 / fl— _—?— 4/13 35 5/11 Fault, showing dip—Dashed where approximately located; queried where inferred a 4/27 0 w. o 5- 75 Vertical hole in roof Probe 17 (overcored Mar,12,1968) NI 35° w., 2° 6 4%: ’59 Borehole, showing strike and dip— 6’. BlacR symbol indicates probe Plan at1.4m(4.5 m location Probe B—3 (overeored (0 Probe 5 ) _ above invert . M 5,1968 ct.19,1966 Excavation sequence (8/11 , August 11, etc). 3' l 9” 1—3 excavated in 1966 4—253 excavated in 1967 26-35 excavated in 1968 USBR O 10 20 30 FEET U.S. Bureau of Reclamation 0 2 4 6 SMETRES FIGURE 26.—Plan of field site at Colorado School of Mines experimental mine, showing general arrangement of workings, excavation sequence, borehole probe locations, and major faults. FIELD INVESTIGATIONS 29 TABLE 12.—Calculated in situ stresses Probe 8-?) Probe 17 Probe 18 Probe 21 Probe 22 Magni- Bearing Plunge Magm- Bearing Plunge Magm- Bearing Plunge Magm' Bearing Plunge Magm- Bearing Plunge tude d d 5) tude d ) (de ees) (de rees) (de tees) tude (degrees) (degrees) tude (degrees) (degrees) (lb/in?) ( agrees) ( egree (lb/in?) ( egrees gr (lb/mg) g g (lb/inz) (lb/in? , Principal stresses:1 ”1"" .. 620 N. 68 E. 16 1,805 N. 47 E. 20 1,415 S. 8 E. 38 1,440 S. 74 E. 53 1,020 N. 15 E. 28 0’2 515 S. 17 E. .17 1,225 S. 16 E. 51 930 N. 66 E. 20 1,045 N. 0 E. 10 770 S. 73 W. 53 0'3} ...................... 420 N. 62 W. 66 690 N. 56 W. 32 870 N. 46 W. 45 510 S. 86 W. 35 480 S. 62 E. 22 Near-horizontal stress compon- ents along probe axes: 0"), .................. 540 S. 54 E. 10 1,115 N. 15 W. 0 960 N. 63 E. 0 1,015 N 75 E. 10 715 N. 72 E. 10 0'," .................. 585 N. 36 E. 2 1,480 N. 75 E. 2 1,190 S. 27 E. 2 1,010 N. 18 W. 2 810 N. 18 W. 2 Near-vertical stress compon- ent along probe axis: ”x .................. 430 N. 66W. 80 1,125 S. 75 W. 88 1,065 N. 27 W. 88 970 S 62W 80 745 S 62W 80 Horizontal stress components with respect to central r cut axis: Ct?“ 111 ............ 555 N. 36 W. 0 1,085 N. 36 W. 0 1,140 N. 36 W. 0 1,085 N. 36 W. 0 675 N. 36 W. 0 a?“ e. 600 N. 54 E. 0 1,680 N. 54 E. 0 995 N. 54 E. o 795 N. 54 E. o 810 N. 54 E. o perpendicular Vertical stress component: 0‘ . ............ 435 90 1,145 90 1,080 90 1,115 90 785 90 Vertlcal= 'All stresses compressive. NORTH Probe Symbol @-Principal stress clusters as I '7 E :5___ Enclose 95 ’5 9 TB percent 2‘ o C _____ confidence 22 0 intervals FIGURE 27.—Principal in situ stress orientation deter- minations (lower hemisphere equal-area projection). a significant relationship of stress to foliation. The 95- percent confidence interval for the orientations of cluster ”B covers the full variation of the poles of the foliation present in the study area. The average orientations of the other two principal stress clusters, "A and ”C, lie on and close to the trace of the average foliation plane (fig. 28). NORTH 0‘71: ,/ \\ \\ eeee FIGURE 28.—Foliation joints in study area. Lower hemi- sphere equal-area projection; contoured on 3, 6, and 9 poles per l-percent area; 51 poles. a' A,a'B, and ”-c are centers of principal stress clusters shown in fig- ure 27. The average maximum principal stress cluster (aA) is oriented approximately perpendicular to a joint set present in room 2 (SW. quadrant, fig. 29). The apparent relationship of the present stress-field or- ientation to much older geologic features, the foliation and jointing, is indicative of the importance of previous geologic events to the orientation of the principal stresses. 30 NORTH q‘ . - PE RPENDICULARS PE R'PENDICU LA 85 OTC FAU LT 1 FIGURE 29.-—Equal-area diagram of 159 joints in study area. Lower hemisphere projection, contoured on 2 poles per l-percent area; 159 poles. Also shown are stress-change orientations around room 2. The foliation is the most persistent geologic structural feature in the study area (pl. 1; fig. 28). Faults l and 2 appear to exert” considerable local influ- ence on the stress-field orientation. Figures 27 and 29 show that the average intermediate principal stress direction lies close to the planes of faults l and 2. Also, the principal stress orientations determined at probe 17 (located 3.7 m (12 ft) in the footwall northwest of fault I) were rotated into or toward the trace of fault I and toward and perpen- dicular to fault 1 relative to the principal stresses deter- mined at probe B-3 in the hanging wall of fault I. The variations in the magnitudes and orientations of the principal stresses included in the individual clusters may be related to variations in stiffness between foliation layers and to joints perpendicular to the foliation. An accuracy of about 3 percent for the average of each cluster of stress orientations is indicated by the independent check for orthogonality shown in figure 27. The influence of anisotropic changes in modulus across foliated lithologic contacts ‘or across and along faults and STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO joints apparently caused large local variations in the back- ground stress field. A nearby stress-field determination was used, in preference to the average condition, for analy- sis of rock behavior throughout this report whenever available. A nearby stress-field determination should be more accurate near the site of determination. Because of the complexity of the geologic environment and the small number of in situ stress determinations, extrapolation of a local stress-field determination, particularly stress magni- tudes, over more than a few feet can be grossly misleading (table 12). When a nearby stress-field determination was not available, the average in situ stress field was used for analysis. The U.S. Bureau of Reclamation made nine biaxial plane-strain overcore stress-relief measurements in one borehole with an instrument similar to the U.S. Bureau of Mines borehole deformation gage at the location shown in figure 26. The average results in a vertical plane oriented N. 80° W. were 850 lb/in2 vertical compressive stress, 680 lb/in2 horizontal compressive stress acting N. 80° W., and a shearing stress of +170 lb/in2 acting in the plane. The comparable average stresses calculated from the com- posite stress-cluster data (fig. 27) are 780 lb/in2 vertical compressive stress, 890 lb/in2 horizontal compressive stress acting N. 80° W., and a shearing stress of +180 lb/in2 acting in the plane. The stress values of the U.S. Bureau of Reclamation lie within the variations of the principal stress clusters (table 13). The magnitudes and orientations of the principal stresses and principal stress clusters cannot be explained on the basis of the 106.7 m (350 ft) of overburden present. Overburden can account for only approximately 385 lb/in2 of vertical stress and 115 lb/in2 of horizontal re- straining stress. Nor were we convinced that active tec— tonic boundary forces produced these stresses. In order to determine whether residual stresses (Varnes, 1970; Varnes and Lee, 1972) might provide an explanation for the observed stress magnitudes and orientations, several laboratory tests were performed. Six-inch-diameter core samples of gneiss were taken from the Idaho Springs Formation in the study area, placed in a temperature-humidity-controlled room, strain gaged, and overcored. Relieved residual extensional TABLE 13,—Principal in situ stress clusters determined in field study area Bearing Plunge Magnitude‘ . . (degrees) (degrees) (lb/in?) Princrpal Stress 95-percent 95-percent 951mm“: cluster confidence confidence confidence interval interval interval Average Average Average From To From To From 1 o A N. 3 E. N. 5.8 E. N. 72.2 E. 19 11.4 26.6 1,085 583 1,587 B S. 35 E. S. 2.0 E. S. 68.0 E. 36 17.1 57.9 1,015 463 1,567 C N. 73 W. N. 43.0 W. S. 77.0 W 44 23.9 64.1 650 437 863 lAll stresses are compressive. FIELD INVESTIGATIONS v 31 strains measured by six overcored strain gages stabilized from + l to +554 ,u.in./ in, 1,050 hours after overcoring. The average residual strain measured was 175 yin/in. The indicated equivalent stored compressive stress prior to release ranged from 8 to 4,350 lb/inz. Thus, sufficient residual strains are present in some rocks of the Idaho Springs Formation to account for the stress magnitudes measured by in situ stress-relief over- coring and to provide the driving stresses indicated by the magnitudes of decompression during long term stress monitoring, which are discussed in the following sections. The range of magnitudes of the residual strains would favor an in situ stress field containing residual compon- ents in addition to those induced from other sources. EXPERIMENTAL ROOM 1 Prior to excavation of room 1, two three-dimensional borehole probes, 5 and 8, were installed (fig. 26). Probe 5 was in the southwest wall of the future room, approxi- mately 0.9 m (3 ft) above floor level, and probe 8 was in the northeast wall. Both installations were completed before section 2 of the room excavation began. Figure 26 shows the sequence and date of the sections removed. The room, primarily excavated for rock-bolt performance studies, was to have been 7.3 by 12.2 m (24 by 40ft). Access was pro- vided to the room from B—Left drift by a smaller central crosscut. Additional strain information was available from single- and multiple-position extensometers and from instrumented rock bolts that had been installed in the roof of the room by R. S. Culver, Colorado'School of Mines, as excavation proceeded. Some of these extensom— eters were positioned across fault I; thus, displacements along the fault were well documented (Culver, 1967). The relationships of fault I and of the foliation to Foliation Probe 5 Probe 8 ‘“('30 ft (18 ml“J FIGURE 30.—Generalized view showing relationship of hanging wall and footwall of fault I (stippled) to foliation and to probes 5 and 8. stresses determined at probes 5 and 8 are shown in figure 30, both instruments being approximately 3.7 m (12 ft) (normal distance) from the fault plane, probe 5 in the hanging wall and probe 8 in the footwall. Excavation pro- ceeded northwesterly, beginning in the hanging wall of fault I. When the excavation of section 3 (fig. 26) inter- sected the fault surface, measured displacements of as much as 0.01 cm (0.004 in.) occurred across the fault (Cul- ver, 1967). Vertical decompression (that is, a tensile stress change) followed on footwall probe 8, and at the same time hanging wall probe 5 experienced moderate increase of compression. Figure 31 shows the vertical components of time-dependent stress changes determined by data moni- tored on probes 5 and 8. The beginning of a large decom- pression on probe 8 (fig. 31) coincided with measured dis- placements of as much as 0.5 cm (0.2 in.) across the fault (Culver, 1967), which occurred after the last round (14) was excavated from the room. In the footwall at probe 8, a ver- tical decompression of approximately 3,500 lb/in2 took place during a period of a little more than a month, indi- cating continuing displacement along the fault plane accompanied by stress readjustments in the rock mass. During this time, there was no excavation activity, but the loosening of slabs of rock from the roof and wall above probe 8 became serious and caused the excavation plan to be abandoned. Remedial bolting and subsequent renewed excavation of the central crosscut in July (fig. 31) resulted in further adjustments in the stress field around probe 8. Load-carrying ability of the rock was effectively increased by the rock reinforcement, as shown by the 3,200-lb/in2 increase in compression in the rock. The renewed excava- 2000 I I I I I I I I Wall slabbing and maximum _ movement on fault 1 1000 — Renewal of central drift excavation +Compresswe O | Remedial rock bolting lntersected fault 2 ~1000 — 72000 — I w o o o l Room 2 excavation _ <—.—> Decompresswe 1 rs O O O I {In D o o I No excavation CHANGE IN VERTICAL NORMAL STRESS, IN POUNDS PER SQUARE INCH —6000 — EXCAVATION SEQUENCE 77000 — (FIG. 26) _ /\ 25 25 30 3132 33 16 1320 22 / 26 27 29 34 35 . ...././I. . d 13 3 4 5 57 89 1611214 15 171921/2I3 2 _8000 I . I... .. I..... I .. II...” I I I I DEC FEB APR JUNE AUG oer DEC FEB APR JUNE AUG OCT 1966 1967 1968 FIGURE 3l.—-Vertical normal stress in rock adjacent to probes 5 and 8 as excavation advanced. For clarity, only data at inflection points have been used. 32 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO tion activity and rock bolting were apparently too far away from probe 5 to greatly affect the stress conditions in its vicinity; however, a slow decompression continued, prob- ably caused by continued adjustment along fault 1. When the central crosscut penetrated fault 2 in July and August 1967, another large time-dependent decompression was produced in the footwall around probe 8 (fig. 31), and a smaller decompression was produced on probe 5. In late July 1967, probes 9 and 10 were installed approx- imately 15.8 m (52 ft) ahead of the face of the central cross- cut (fig. 26). These probes, which are discussed in detail later, were subjected to increased compression during the period of fault 2—related decompression of probes 5 and 8 (fig. 31). Fault 2 was partially exposed in the lowerrright- hand corner of section 16 and was fully exposed around the entire section of the central crosscut with the removal of section 17. Apparently, the intersection of fault 2 caused a time-dependent transfer of compression stress from the rock in the vicinity of probes 5 and 8 to the rock in the vicinity of probes 9 and 10. Such time dependence can be reasonably explained by slow adjustments of individual joint blocks responding to changing stress conditions along and adjacent to excavation-activated faults. When viewed in three dimensions, the stress-change orientations reveal even more strikingly their relation to geologic features and excavation activity. Figure 32 shows the traces of the plane of fault I and the average foliation plane in the same vicinity, the poles of the fault, and per- pendiculars to the foliation planes. The other poles in the figure represent 10 principal stress changes at probe 8 ob- tained from daily measurements over a 2-week period be- ginning at the time excavation intersected fault 1. The atti- tudes of the three principal stress changes are shown by the indicated symbols. The data show that the intermediate principal stress changes consistently are oriented nearly horizontally in a plane parallel to the fault and that they bear to the southwest, nearly at right angles to the dip of the fault and foliation planes. Both the maximum and the minimum principal stress changes are oriented either approximately in the plane of the fault or approximately perpendicular to the fault. The poles of the maximum and minimum stress changes exchange positions from time to time, as shown by their alternations in readings 4 through 10. Later measurements (fig. 33) showed that the poles shown in the north half of the diagram migrated with time to form a maximum perpendicular to the fault. For the 2-week period, all poles in the southeast sector (on or near the fault plane) represent the greatest absolute stress changes—changes of either maximum compression or maximum decompression. This fact indicates that, as excavation in the footwall continued, the rocks bordering the fault were expending strain energy—rebounding—in a series of contractions and dilations having an almost daily periodicity. This behavior is attributed to the sudden removal of natural rock support across fault I by move- NORTH +1 Perpendicular to plane of fault 1 Borehole axis Perpendicular to average foliation plane EXPLANATION . + a Minimum principal stress change,a’3 Intermediate principal stress changemz Maximum principal stress change, :71 FIGURE 32,—Equal-area lower hemisphere diagram of poles of principal stress changes, as determined from probe 8. Shown are 10 readings taken over a 2-week period, be- ginning when mining intersected fault 1 (fig. 26). ment along fault 1. Movement on the fault appears to have allowed stored (potential) energy to be gradually expended, as indicated by the overall decompression in the rock around probe 8 during this 2-week period (fig. 32). Figure 33 shows equal-area diagrams for the directions of all 219 principal stress changes (A01, A112, and A03 not differentiated) measured by probes 8 and 5 during room 1 excavation. Contours represent percent of inter— cepts in a l-percent area. Also shown in the equal-area plots are planes representing the attitude of fault 1 and the average attitude of foliation. Orientations of the three principal stress changes during the period are concen- trated within small intercept areas. However, the individual orientations of A01, A02, and A03 shifted from time to time from one intercept area to another, but at any one time one principal stress-change intercept is present within each area of concentration. The plots for both probes show very strong concentra- tions that are related to geologic features or to excavation direction and excavation procedures. Probe 8 in the foot- wall shows a very strong concentration of stress-change directions in the southeast quadrant, on the trace of the fault plane. A second concentration in the northwest quadrant lies near the trace of the average foliation plane and includes the direction perpendicular to the fault. The third concentration coincides with the pole of a major joint set (not shown in the diagram). The plots of stress- change directions for probe 5 in the hanging wall are dif- FIELD INVESTIGATIONS PROBE 5 Vertical section 30 inclined plane striking parallel to direction of excavation and dip- ping in the zones of greatest stress~change locations Perpendicular to fault 1 Average foliation plane intersection of fault plane and inclined plane parallel to excavation PROBE 8 NORTH Inclined plane striking parallel to direction of excavation and dip- ping in the zones of greatest stress—change locations Perpendicular . to fault 1 Average foliation plane Intersection ‘of fault plane and inclined plane parallel to excavation FIGURE 33.—Comparative equal-area lower hemisphere diagrams of ,un- differentiated principal stress-change directions (Ala-1, A02, A03) determined at probes 5 and 8. Probe 5; 219 poles, Jan. 5 to June 6, 1967; probe 8; 225 poles, Dec. 12, 19§6, to June 6, 1967. Contours represent percent of stress-change directions per l—percent area. See figure 34 for location of lines of vertical sections AD, BD, and CD. ferent from those of probe 8 in that only one concentra- tion is strongly related to geologic features. There is a con- centration in the southeast quadrant that lies across the trace of the fault plane, but it is weaker and does not have the same orientation as the similar concentration for probe 8. The strong stress-change clusters for probe 5 in the 33 northeast and southwest quadrants are not clearly asso- ciated with any geologic features, but they are alined very .closely with the average strike of the vertical section through probe 5 that corresponds to the longest dimen- sion of the room during that particular stage in the exca- vation sequence. Figure 34 shows the average strike (AD) of the longest vertical section through the room and the strike CD and BD of the vertical sections, which show the > limits of migration of this plane during excavation. These same sections, CD and BD, can be seen to limit the boun- daries of the strong concentration of stress changes (fig. 33). The weak concentration seen in the northwest quad- rant is not clearly associated with any known geologic or excavation feature, although it approximates the direc- tion perpendicular to the fault. The data from probe 8 (fig. 33) show that within the footwall of fault I the directions of principal stress changes, which are nearly all decompressive, aline closely with geologic structures. Without exception, the major and minor principal stress changes for any long-term, large-magnitude change are either perpendicular to the intersected fault or lie within that fault plane. The same long-term changes are not so strongly related to the folia- Average strike of vertical section through probe 5 that was the long- B est average dimension of the room during the excavation EXPLANATION 35 +— Fault, showing clip 0 1O 20 FEET 0 2 4 METRES FIGURE 34,—Map of room 1 in Colorado School of Mines experimental mine showing strike of vertical sections AD, BD, and CD. 34 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO NORTH tion. The fact that the foliation and the nearby fault planes are practically at right angles tends to obscure the role played by foliation. At probe 8, the directions of stress changes appear to be controlled by geologic structure, and directional in- fluences of excavation procedures are obscured. However, the equal-area plot indicates that the direction of excava- tion may have some influence because two of the largest stress-change clusters lie on the trace of a plane that strikes parallel to the general direction of excavation. One of these clusters lies nearly at the intersection of this plane and the fault plane. Another plane, X, whose intercept on the lower hemisphere coincides with a great circle cutting through the two smallest clusters, has no apparent rela- tion to the geologic structure or to the excavation. The data for probe 5 indicate that room geometry exerts more influence on the changing stress field in the hang- ing wall than in the footwall of fault 1. The majority of stress changes are compressive, and the largest cluster lies in the vertical plane containing the longest average dimension of the room during that particular stage in the excavation sequence. Also, similar to probe 8, two of the largest stress-change clusters for probe 5 occur on the trace of a plane parallel to the direction of excavation, and one of the clusters occurs at the intersection of traces of the excavation plane and the fault plane. Excavation direc- tions appear to exert a significant influence on the orien- tation of stress-change directions. An explanation other than one dependent on a gravity- load elastic analysis is needed to provide a reasonable understanding of the changing stress field. The natural (in situ) stress-field information (obtained after room 1 was excavated) was helpful in this respect. The directions of the measured maximum principal in situ stresses in both the hanging wall(probevB-3) and the footwall (probe 17 ) are alined approximately parallel to the strike of fault 1 (table 12 and fig. 35). The maximum principal stress in the footwall is nearly three times the subparallel maximum principal stress in the hanging wall. The direction of the intermediate principal stresses ‘in both the hanging wall and the footwall is approxi- mately down the dip of fault 1. The intermediate prin- cipal stress in the footwall is approximately 2% times the subparallel intermediate stress in the hanging wall. The minimum principal stresses in both the hanging wall and the footwall are roughly normal to the plane of fault I. The minimum principal stress in the footwall is 1% times the subparallel minimum principal stress in the hanging wall. These differences in magnitude occur over a dis- tance of only 7.6 m (25 ft) across fault 1, and, assuming that the rock at these locations was unaffected by the excava- tion, these stresses represent the preexcavation stress field. During faulting, more energy may have been released in the hanging wall than in the footwall, owing to less con- fining pressure on the hanging wall than on the footwall. U \ PERPENDICULARS ‘E T 1 - 2/ ‘73 Q9\ Tp FAUL 1805\Ig/I/ a 0691,) P‘+/ ,1 690 lb/inz \5‘0"\ 0‘ / 9c, / 620 lb/in 9 QA o 9/ o / (13 0 \ 42o lb/inz EXPLANATION Probe Symbol Location B 3 G) Hanging wall 17 El Footwall FIGURE 35.—-Orientation of in situ stresses determined in walls of fault I. Equal-area lower hemisphere projection. The vertical normal stress component is 435 lb/in2 in the hanging wall, a figure that corresponds approximately to the stress produced by the estimated 107 m (350 ft) of over- burden. However, the vertical normal stress component in the footwall was 1,1451b/in2, almost three times that in the hanging wall, too high to be explained by the same 107 m (350 ft) of overburden. As the result of the principal stress differences across fault I, rather large shear stresses must have been present within the rock mass surrounding fault I. When downdip confinement along fault I was reduced during the excavation of room 1, the shear strength of the rock material in the fault was exceeded. Movement occurred in the fractured rock zone around fault I, and further enlargement of room 1 had to be stopped. EXPLANATION OF STRESS CHANGES IN ROOM 1 In the initial stages of excavation, the access drift was ad- vanced toward fault I and the natural high level of stress within the footwall was further concentrated. When the advancing opening penetrated the fault displacement occurred and stresses in the footwall were partially relieved; but the stresses in the hanging wall continued to build up. As can be seen in figure 36, when the room was excavated beyond the fault (first row of excavation) the hanging wall apparently began to act as a support mem- ber for the roof load (commonly called arch load), as evidenced by increased compressive stress in probe 5. The effective ovaloid room cross section intersecting probe 5 and parallel to the longest ceiling support distance should contain the highest magnitude tangential stresses. The FIELD INVESTIGATIONS 35 major principal stress-change directions occur in or near this plane, as demonstrated in figure 33. If an ovaloid section represented by a hypothetical line of zero change in stress level, as shown in figure 36, is used to represent the longest effective section of the excavated room, a tan- gential stress concentration at probe 5 would be approxi- mately three times the vertical normal stress of the theo- retical stress field (Obert and others, 1960, p. 12). The hypothetical zero stress-change line was approximately located using extensometer data obtained by Culver (1967). The vertical normal stress, as determined from overcoring in the hanging wall near probe 5, was 435 lb/in2 (table 12) as compared to approximately 900 lb/in2 increase in vertical normal stress at probe 5 caused by exca- vation during the first 5 months. At the end of this period the total vertical stress on probe 5 was approximately 1,335 lb/in2 or slightly greater than three times the original ver- tical stress, which agrees with the stress concentration pre- dicted by the ovaloid opening analysis. None of the principal stress-change directions shown in figure 33 are vertical, but all are shallow plunging. This information agrees with the finding that the plunge of the major principal stress direction of the natural stress field was 16° in the hanging wall and 20° in the footwall. The low-angle, high-magnitude tangential stress at the curved corner of the ovaloid opening is, therefore, further Hypothetical line of zero change in stress level, after completion of excavation, based on data from multiple»position extensometers MA 1, MB 2, and MC 3 (Culver, 1967, p.45). Compression above, decompression below ll ll Hypothetical line of zero ll H H . change in stress level II after first row of exca— H ll 12.2 m vation was completed. \ ll (40 ft) FAULT1 4 Compression above, ll decompression below /l/ H Northwest extension of room (projected) FIGURE 36.—Cross section AD (fig. 34) containing probe 5, constructed parallel to the longest average dimension of room 1 during most of the excavation, showing hypothetical lines of zero change in stress level after two stages of excavation. increased by the relatively high horizontal stress com- ponent of the in situ stress field. Continued fault move- ment and decreased shear resistance along the fault caused the roofload and, hence, the line of zero ’stress level to pro- gressively shift upward and northward, allowing the foot- wall support section in which probe 8 is located to be relieved of its load (fig. 36). However, probe 5 continued to show a net increase in compressive stresses related to the arching roofload. Decompressive stress changes then began, probably as a result of fault movement that caused a redistribution of the roofload over an even wider span extending beyond probe 5. Probe 8 showed large decom- pressive stress components parallel to and perpendicular to the fault surface, indicating a decrease of compressive stress in these directions. Apparently the stresses in the footwall were decreasing because the footwall near probe 8 was no longer supporting the roofload and because the confinement of this highly prestressed rock had been removed. The wall, relieved of its load, was deflecting into the room. Decompression continued on probe 5 during July, Au— gust, September, and October 1967 (fig. 32), apparently resulting from the advancing excavation and continued disturbance of the existing fault and joint system. STRESS CHANGES AHEAD OF THE CENTRAL CROSSCUT EXTENSION Most field measurements of rock response axially be- hind an advancing face differ from the anticipated elastic response. For example, in the Straight Creek pilot bore in Colorado (Abel, 1967, p. 16) rock-mass deformations took place over a zone 6—60 tunnel diameters behind the ad- vancing face. Field measurements generally agree, however, with the extent of the radial influence zone around a tunnel as predicted elastically and photoelas- tically (for example, see Abel, 1967, p. 19). The distance ahead of the tunnel face in which the stress distribution is affected by the advancing tunnel is essen- tially unknown. To our knowledge, no measurements of deformation, strain, or stress have been reported for the rock ahead of a tunnel. There are obvious physical diffi- culties in instrumenting this region. A number of European investigators, however, have made convergence measurements ahead of advancing longwall mining faces. A longwall face, normally on the order of 183-914 In (600—3,000 ft) long and 0.9—2.7 In (3—9 ft) high, approximates the geometry of a tunnel in the plane perpendicular to the longwall face. The divergence of the reported results is informative in anticipating the zone of influence in the rock ahead of a tunnel where a strain sensor could conceivably detect stress changes associated with the tunnel excavation. Various investiga- tors at the International Conference on Strata Control in Paris, France, in 1960 reported on their findings on the 36 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO extent of the zone of influence ahead of longwall mining faces as follows: Potts (1961), (>15—20 ft). Denkhaus and Hill (1961, p. 247-248), Quartzite, South Africa: Onset >6.1 m (>20 ft), peak 4.6—6.1 m (15—20 ft). Leeman (1961, p. 308), Quartzite, South Africa: Onset at 7.6 m (25 ft), peak <3.7 m (<12 ft). de Reeper and Bruens (1961, p. 328), Coal, Belgium: Onset at 4.6 m (15 ft) (approximately five thick- nesses of seam ahead of face). Carter (1961, p. 471), Coal, Great Britain: Onset <66 m (<216 ft), peak <14 m (<46 ft). Spackeler (1961), Coal, East Germany: Peak <5.5 In (<18 ft) (probably <3.7 m (<12 ft)). Henshaw (1961), Coal, Great Britain: Onset >14 m (>46 ft). Kolar Schist, India: Onset >4.6—6.1 m The wide variation in these field measurements, show- ing onset of stress from 5 to 25 seam thicknesses ahead of the face, and peak stress from 0 to 8 seam thicknesses, departs markedly from the theoretical prediction of the onset of such stress changes 1 diameter ahead of the tunnel face. Following the termination of excavation of room 1, as the result of bad roof and wall conditions associated with fault 1, the central crosscut was extended on the same N. 36° W. bearing until it intersected C—‘Left drift (pl 1; fig. 26). Probes 9 and 10 were installed ahead of the advancing tunnel and were monitored starting on July 28, 1967, fol- lowing excavation of section 16 (fig. 26). The results of this field sutdy of the stress changes occurring ahead of an advancing tunnel were compared with the laboratory model-tunnel studies. The stress-change histories for probes 9 and 10 from August 4 until they were destroyed by advance of the cen- tral crosscut are presented in figure 37. They demonstrate an initial increase in compressive stress followed by a large decrease in compressive stress. The magnitudes of the stress changes measured by probe 10 were consistently less than those measured by probe 9. The onset of stress changes associated with the advanc- ing tunnel took place as early as or before the probes were installed because stress changes were recorded upon in- stallation of the probes. This change occurred more than 7 diameters ahead of the advancing tunnel face (figs. 27, 37). Elastic analysis predicts the onset of stress change at l tunnel diameter ahead of the tunnel face. In the elastic acrylic model the onset occurred more than 4 tunnel di- ameters ahead of the model tunnel,in the heterogeneous isotropic concrete model it occurred at more than 2% tun- nel diameters ahead, and in the anisotropic homogeneous granite model it occurred at more than 2 tunnel diameters ahead. The results of studies of longwall behavior men- tioned above indicate that the onset of stress influences +4000 I 1 C .9 a w §+2ooo— 0x8 Au, PROBE 9 g {3\ RAM I0 9 I:\W o o— e ._‘ ‘° a 3 5 . ‘3. A“ 5 g 5 . m ‘2 3’ 10 °‘ 01 93 LB 3 DC ~ < “ P. " ~ 0 3 <1 c—2ooo~ V. as g u; g " m 3.9 g7 2 . s .4 H. LO 1 m 3 < 8’ e % a N. 9 < LO (0 a: 4.. CC u—4ooo— 52 ‘0 3 E: E _ <®\ tn 8 5 ° °\ 03 . \. - Z ~6000— 3’ \o 3 < o a , g _8000 I I I | I I I C “1.99000 I I I . 74> 6(1) 54> 4 34> 24> 15> o R TUNNEL DIAMETERS FROM TUNNEL FACE TO SENSO FIGURE 37,—Principal stress changes caused by advance of central cross- cut. Date of excavation is shown. related to longwall mining could be detected from 5 to more than 25 seam thicknesses ahead of the advancing face. A compressive stress peak was measured by probes 9 and 10 between approximately 6 and 6’/2 tunnel diameters ahead of the advancing tunnel face (fig. 37). No such stress peak is predicted by elastic theory. The absolute stress of this peak is not known, because probes 9 and 10 sensed stress changes relative to an initial reading without refer- ence to a known premining in situ stress field. A similar compressive stress peak was observed in each of the tunnel models, at 2 tunnel diameters for the acrylic model, at 11A tunnel diameters for the concrete model, and at 1 tunnel diameter for the granite model. The results from the long- wall measurements indicate a peak of from less than four seam thicknesses to somewhat more than eight seam thick- nesses ahead of the advancing longwall face. The results from probes 9 and 10 of our field study ap- proximately agree with the available field data but dis- agree in magnitude with the laboratory results and dis- agree with elastic theory. The orientation of the average maximum principal background stress cluster (0' A) in the study area is flat lying and trends northeast-southwest, and the orientation of the mean intermediate principal background stress is gently plunging and trends northwest-southeast. Because FIELD INVESTIGATIONS 37 PROBE 9 \ PROBE 10 5(1) 4(1) Central crosscut 14> 0 CENTIMETRES 30 24> 14> DIAMETERS () FROM PROBES 9 AND 10 TO CROSSCUT FACE AND DISTANCE (ft) FROM FACE EXPLANATION Trajectories of principal stress —><—— Maximum Intermediate Decompression vector Compression vector VECTOR SCALE (Ib/inz) 0 5,000 15,000 FIGURE 38.—Plan view of stress-change trajectories and stress-change vectors. Follow vectors and trajectory lines from left to right across diagram to simulate the effect of approaching excavation. the maximum and intermediate stress changes near probes 9 and 10 were also nearly flat lying, a plan view contain- ing these near-horizontal principal stress-change vectors could be constructed (fig. 38) which would show the stress response as the crosscut advanced. This diagram shows stress changes with respect to distance and to crosscut di - ameters ahead of the face. The locations of the stress- changeyectors are related to their distance from the face of the crosscut after each increment of advance and there- fore do not represent a typical plan illustration. The pattern of the stress-change trajectories suggests: ( l) the stress orientations are influenced by foliation and fault attitudes, (2) the onset of tunnel-excavation-induced stress response is at least 7 diameters ahead of the tunnel, (3) the orientation of the minor asymmetry of stress trajectories ahead of the advancing crosscut is grossly similar to the orientation of in situ stresses, which suggests an inter- action between stress changes induced by mining and the mean maximum principal background stress cluster (0A), and (4) the tensile stress changes progressively increase in all directions as the central crosscut approaches the probes. The departure of the stress change versus distance rela- tionship from elastic predictions, laboratory model measurements, and photoelastic measurements suggests 38 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO that rock-mass properties controlled the rock—mass response. The central crosscut was driven normal to the strike of the major geologic weakness in the rock mass, the foliation. This rock-mass weakness could allow exten- sional strain relief parallel to the crosscut axis and normal to the plane of foliation. Apparently the rock expanded preferentially perpendicular to these discontinuities, toward the crosscut face. The influence of fault I on the nearby stress field, as pre- viously discussed, directed our attention to the possible similar influence of fault 2 on probes 9 and 10. The rela- tive locations of fault 2 and probes 9 and 10 are shown in figure 26. Initial compression, followed by large decom— pression, occurred on the lower stressed (footwall) side of fault 2 (fig. 37) as the tunnel advanced, a behavior similar to that of fault 1 (fig. 31). The advance of the excavation toward probes 9 and 10 provided a means for the stress to be transferred from the higher stressed hanging wall to the lower stressed footwall of fault 2. This decompressive stress release dominated the response of probes 9 and 10. The overall effect of crosscut advance was to decompress the rock ahead of and to the side of the crosscut. EXPERIMENTAL ROOM 2 Room 2 was excavated after the central crosscut had been driven to the intersection with C-Left drift. The exca- vation of room 2 was accomplished by widening the cen- tral crosscut to 7.3 by 18.3 m (24 by 60 ft). The sequence of excavation is shown in figure 26. R. M. Cox of the C010- rado School of Mines designed the room and supervised its excavation for rock-bolt performance studies (Cox, 1971). Prior to excavation, probes 15, 19, 24, and 25 were in- stalled in the roof of the central crosscut at the locations shown in figure 26. Probes l5 and 24 were installed 5.5 m (18 ft) above the ceiling, whereas probes l9 and 25 were only 1.2 m (4 ft) above the ceiling. The location of probe B-5, also shown in figure 26, was in the northwest wall, about 5.5 m (18 ft) from the central crosscut. Multiple-position borehole extensometers (MPBX’s) were installed by Cox (1971) at locations shown in figure 39. These MPBX’s were placed so that they extended 12.2 m (40 ft) above the roof of the central corsscut. Figure 39 shows the relationships of probes l5, 19, 24, and 25 and MPBX’s 1, 2, and 3 to fault 2, all being located in the foot- wall. The roof probes were spaced from 7.6 to 14.2 m (25—465 ft) from the fault surface, probe 15 being the closest and probe 25 the farthest. It is doubtful that any of the MPBX installations intersect the plane of fault 2 unless the dip of the fault above the roof decreases to less than the 45° dip measured in the central crosscut. The 12.2-m (40 ft) anchor of MPBX—l would be located very close to the fault plane and probably is in locally sheared rock adjacent to the fault. Probe B-5, which lies northwest (to the right)of the area of the cross section shown in figure 39, is approxi— mately 18.9 m (62 ft) from the surface of fault 2. 12.2%“ 12.24“ 12.2—fm <40 m (40 ft) (40 t) anchor anchor MPBX-1 MPBX—2 MPBX—3 FIGURE 39.—Generalized section showing relation of borehole probes, multiple-position borehole extensometers (MPBX’s), and fault 2. In order to show true dip of fault, horizontal distances are exaggerated. The excavation of room 2 began in April 1968 and pro— ceeded essentially by excavating four 1.2- by 2.4-m (4- by 8— ft) slabs the length of the 15.2-m (50 ft) room, two on each side of the central crosscut (fig. 26). The room was success— fully completed in September 1968. Other smaller excava- tions were made as needed to trim the ends (transitions) of the room. The extensometers and stress probes were read by Cox (1971) before and after excavation of each slab. The data used in the following discussions were compiled from the borehole stress probes and the MPBX installations before, during, and after each slab was excavated from the room, during the period April 4, 1968, through January 1, 1969. Figure 40 shows principal stress changes with time for probes 15,19, 24, and 25 above room 2 and probe B-5 to the west of room 2. Figure 41 is a similar plot of the total strains that occurred between anchors of each MPBX in- stallation. Prior to May 26, 1968, some small excavations were made in the transition zone on the right side and the first slab was presplit. During this period the rock near deep probe 24 compressed and the rock near shallow probe 25 decompressed (fig. 40). The equal-area plot (fig. 42A) from deep probe 24 over the five intervals prior to May 26 shows that the principal stress-change directions (A01,A0'2, A03) cluster in three small groups that are related to the orientation of the in situ stress field, the excavation direc- tion, and fault 2. The orientation of these clusters is very close to that of the average principal stresses and to the principal stresses determined from overcoring nearby probe 18. Two stress concentration clusters lie in or close to the vertical plane striking in the direction of the central crosscut axis and normal to the foliation. One of these clusters (fig. 42A) appears to be related to the intersection of the vertical plane and the trace of the plane of fault 2, and the other cluster is nearly perpendicular to the fault. The third cluster (consisting only of Aaz) is nearly hori— FIELD INVESTIGATIONS First slab Second slab Third slab Fourth slab (27)‘ (29) (34) (35) I I I I O 01 02 #0 —400 03 Stress changes at depth of 5.5 m (18 ft) (probe 15), 7.6 m (25 ft) from fault 2 -800 +400 I O U E LU cc Stress changes at depth of 1.2 m (4 ft) ‘3‘ (probe 19), 10.7 m (35 ft) from fault 2 d -400 l | m I m n. (n 3 Stress changes at depth of 5.5 m (18 ft) :0) +800 (probe 24), 11.1 m (36.5 ft) from fault 2 n. E uJ 0 E +400 I U (n U) Lu 0': .— V) 0 O —400 —800 ‘ Stress changes at depth of 1.2 m (4 ft) ‘ (probe 25), 14.2 m (46.5 ft) from fault 2 —1 20° APR. I MAY | JUNE I JULY I AUG. I SEPT. I OCT. | Nov. I DEC. 1968 D FIGURE 40,—Principal stress changes in roof during excavation of room 2. ‘, see figure 26 for location of slabs. +, compression; — tension. , 39 40 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO l l l | | Strain changes _ for MPBX71 Second Third slab Fourth slab slab (29) (34) +250 +200 First slab (27)’ +1 00 +300 +200 + .1 O O STRAIN (10'6) +400 Strain changes _ for MPBX—2 —100 i +200 l Strain changes for MPBX— 3 MAY | JUNE l JULY l AUG.l SEPT. 1968 EXPLANATION Strain changes for borehole intervals, measured from collar of borehole m ft 6 0— 1.2 0— 4 D 1.2— 3.0 4—10 <> 1.2- 6.1 4—20 0 1242.2 4—40 0 3.0— 6.1 10—20 A 6.1—12.2 20—40 FIGURE AIL—Strain changes for multiple-position borehole extenso- meters in roof of room 2. *, see figure 26 for location Of slabs. +, compression; =, tension. Strains are based on anchor displacements. NORTH PERPENDICULARS TO FAULT 2 . u __ O ’9 j . 01(10851b/in2) 2‘ Probe 24 (5.5 m or 18 ft depth) A01 9 TIME lNTERVALS A02 ' Pmbe 24 1 Apr. 5, 1968 “a '3 2 Apr. 23, 1968 F 0 Average principal 3 Apr. 3, 1968 to Apr. 30, 1968 stresses 4 May 7, 1968 A ' _ 5 May 26,1968 Probe 18 pnnc1pal stresses N O RTH 4903 ERPENDICULARS 01 TO FAULT2 ”no 0 A 2 9 P22 0 (71(10851b/in2) \/ /\ / (,2 P~ fi4/ \ 4’ TRACES OF l/\\\§> »/ . \. FAULT 274 P22 \092 1 900 \,>'\’ 4 B Probe 25 (1.2 m or 411 depth) Au. 01 TIME INTERVALS Au; 0 Probe 25 1 Apr. 11, 1968 A03 5’ 2 Apr. 15,1968 3 0 Average principal 3 Apr. 5. 1968 to Apr. 23, 1968 stresses 4 May 7,1968 <7 A . _ 5 May 26, 1968 Probe 22 princupal St re 5885 FIGURE 42.—Principal stress changes of probes 24 (A) and 25 (B) before room 2 excavation. FIELD INVESTIGATIONS 41 zontal and approximately perpendicular to the crosscut axis and parallel to the strike of fault 2. The data show an exchange of positions of A0,, A02, and A03 with time, indi- cating alternation of stress axes. We observed a similar behavior when fault I was penetrated during the excava- tion of room 1 (figs. 32, 33). The data from probes 24 and 25 suggest that there was movement along fault 2 prior to May 26 that disturbed the stress field. A similar plot (fig. 423) of stress-change data before room widening from shallow probe 25, which was more remote from fault 2 than probe 24, shows the principal stress-change directions also clustering in three small areas. These changes, unlike the stress-change directions at probe 24, do not appear to be related to either the plane of fault 2 or the direction of the excavation. Figure 42 also has plotted on it the average principal stress directions and magnitudes of the in situ stress field and the orientations determined by the overcoring of probes 18 and 22 (fig. 28). The correspondence of the average in situ principal stress directions with the principal stress-change directions at probe 25 suggests that the orientation and magnitude of in Situ stresses at this location are probably the controlling factors on stress changes taking place before major room 2 excavation. - During this time, MPBX—3 (fig. 41C) showed vertical compressive strains from 0 to 1.2 in (0—4 ft) above the roof, low-magnitude tensile strains from 1.2 m to 6.1 m (4—20 ft), and nearly zero strain from 6.1 to 12.2 m (20—40 ft). MPBX’s l and 2 (figs. 41A, 413) showed similar strains during this interval. MPBX—l, the closest instrument to fault 2, showed compressive strain from 0 to 1.2 In (0—4 ft), and slight tensile strain from 1.2 to 12.2 m (4—40 ft). MPBX—2, adjacent to probes l5 and 19, (figs. 40A, 408) showed compressive strain from 0 to 1.2 In (0—4 ft), tensile strain from 1.2 to 3.0 m (4—‘10 ft), and nearly zero strain from 3.0 to 6.1 m (10—20 ft) and from 6.1 to 12.2 m (20—40 ft). These data suggest that the initial transition rounds and presplits on the southeast side of room 2 triggered additional adjustment along fault 2, which in turn distrubed the stress field in the footwall of fault 2. The in situ stress data for probes 18, 21, and 22, located 94—113 m (31—37 ft) west-northwest of probe 25 (table 12), show that the horizontal stresses are similar in magnitude to the vertical stresses, and the vertical stresses are about two to three times greater than can be explained by the overburden load. The principal stress directions deter- mined from probes l8 and 22 are included in figure 42. The changes that took place during the excavation of the northeast side of the transition zone can be explained by considering figure 43, a vertical section\ along the central crosscut. Figures 43A and 43B are a sketch of the probable orientation of maximum and minimum subsid- iary principal stress trajectories before fault movement and of the subsequent mining-induced principal stress changes measured by probes 24, 25, 15, and 19. Figure 43C shows a sketch suggesting the effect on the stress field of movement along fault 2 and the probable stress-relieved zones and stress-increased zones in the footwall. The fault movement apparently decreased the compressive stress acting parallel to the crosscut axis in approximately the first 1.2 m (4 ft) of roof rock in the footwall. This roof- load had to be shifted upward along the fault until these stresses could be carried by the still-confined (stiffer) parts of fault 2. Probe 24, closer to the fault than probe 25 and in a sec- tion of the footwall not relieved by fault movement, has superposed principal stress changes that are oriented per- pendicular to and parallel to the fault plane. These stress changes correspond approximately to the in situ stress directions (fig. 44). The increased loads in the vicinity of probe 24 consisted of an initial substantial stress increase normal to the fault plane, followed in a month by a simi- lar increase parallel to the fault plane. The rock adjacent to probe 25, on the other hand, showed principal stress decreases which were nearly hori- zontal and vertical (fig. 42B). This probe, at a depth of 1.2 m (4ft) in the ceiling, is in a thin zone of rock that was par- tially destressed by fault movement and excavation. The components of stress changes (figs. 42B, 43B, 43C) repre- sent decreases in the magnitudes of the in situ stresses in directions of decreasing confinement. Using relations shown in 1960 by Obert, Duvall, and Merrill (p. 12, fig. 5c), and approximation of the stresses in the vicinity of probe 25 prior to the fault movement was made. Excavation of the central drift in rock having approximately equal magnitude in situ horizontal and vertical stress com- ponents (table 12) probably moderately increased these stress magnitudes. As we explain in the following section, the theoretical tangential stress (at) at the top of the drift is about two times the vertical (cry) or the horizontal (oh) stress. The maximum and minimum stress-change direc- tions (fig. 42A) were rotated perpendicular and parallel, respectively, to the axis of the crosscut. This behavior sug- ges ts that there was a loss of confinement of the rock and a release of stress that was greatest approximately parallel to the central crosscut axis ( a 3 of in situ stress field). The pre- split blasting on April 4, 1968, and the transition round blasting on April 23 and 27, 1968, may have triggered a stress release along fault 2. This stress release, accom- panied by a confinement loss, could have caused the stress reorientation. Maximum confinement was then in a direction intermediate between.the normal to the crosscut axis and ”A of the in situ stress field. Intermediate con- finement was approximately vertical. The stress-change. directions (fig. 42B) are closely alined to corresponding directions of in situ stress confinement; that is, the least de- compression,Atr 1’ is in the direction of maximum con- finement, the intermediate decompression,Acz, is in the STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO 42 Central crosscut 2 m 0 O r m t n e m .n e D. X E Central crosscut Transition zone Transition zone Approximate line of /’ zero stress ch ange Partially relieved\ stress zone / /// e Approximate Iin Central crosscut of zero stress Experimental room 2 Transition zone Transition zone fi FIELD INVESTIGATIONS 43 4 FIGURE 43 (facing page).—Generalized in situ stress conditions at several stages in the excavation of room 2. Excavation triggered fault move— ment, which in turn caused changes in the stress field. No stress scale is implied. A. Approximate orientation ofmaximum and minimum principal stress trajectories before excavation of room 2 and before move- ment along fault 2. B. Inferred orientation of maximum and minimum principal stress trajectories after movement along fault 2 and after excavation of room 2. C. Stress zones resulting from deformations and stress field shown in B, above. direction of intermediate confinement, and maximum de- compression,Aq3, is in the direction of minimum confine- men t. The vertical compressive strain increase measured be- tween 0 and 1.2 m (0 and 4 ft) at all MPBX installations prior to the excavation of room 2 (fig. 41) also can be explained by reverse movement on fault 2. Such move- ment would produce a decrease in compression in the roof rock parallel to the crosscut axis near the fault. Decom- pression was facilitated by the unrestrained roof at the boundary of the central crosscut (fig. 43). At this boundary the horizontal compressive stress acting parallel to the crosscut axis would be reduced. The vertical compressive strain (Poisson’s effect from the horizontal decom- pression) would be greatest where restraint was least, in the near~roof rock. The low-magnitude vertical tensile strains that occurred above 1.2 m (4 ft) at all MPBX installations were probably caused by minor downward adjustments of the rock above the 1.2-m- (4 ft) thick near-roof partially stress-relieved zone. Major slabbing for room enlargement started on May 26, 1968, after the preliminary transition rounds were excavated. The four main slabs were taken at about monthly intervals, and the excavation of the room was finished by September 19 of the same year. Data presented in figure 40 show that after the first slab was taken in room 2, and until the room was finished, the stress levels on all the probes generally became less com- pressive with time. After each slab was taken, sudden stress changes occurred that recovered rapidly. During the same interval, the MPBX data presented in figure 41 show that the 0- to 1.2-m (0—4 ft) intervals either were stable or under- went tensile strain, the l.2- to 6.1-m (4—20 ft) intervals underwent tensile strain, and the 6.1- to 12.2-m (20—40 ft) intervals were nearly. stable. It would appear that the rock mass up to 6.1 m (20 ft) above the central crosscut was decompressing as the room was widened. Deeper than 6.1 m (20 ft), negligible change occurred in the rock mass as the result of the excavation. EXPLANATION Probe Symbol Distance from room 2 Distance from into the roof (wall‘) fault 2 ft m ft m 15 0 18 5.5 25 7.6 19 ® 4 1.2 35 10.7 24 0 18 5.5 36.5 11.1 25 E! 4 1.2 46.5 14.2 BS 0 16 4.9 ‘ 62 18.9 FIGURE 44.—Stress-change orientations determined from bore- hole probes around room 2: equal-area lower hemisphere pro- jection showing joints, fault 2, and excavation orientation. Contour interval of joints is 2 poles per l-percent area; 159 poles. Rock-mass response, reflecting prolonged structural controls and high stress levels, such as we have just de- scribed, is probably most likely in massive rock rather than severely fractured rock. Indeed, room 2, the largest room in the mine, was excavated in some of the most competent rock in the study area (pl. 1) and required no artificial support. An elastic analogy helps to illustrate our interpreta- tions. The excavation of room 2 by enlarging the central crosscut can be compared to the plane-strain approxima- tion of widening a nearly circular (ovaloidal) opening within a semi-infinite plate having a width-to-height ratio of slightly less than unity (W0 /Ho =0.8), to a rectangular opening with rounded corners having a width-to-height ratio of 3 (W0 /Ho=3). Again using the relationships developed by Obert, Duvall, and Merrill (1960, p. 9, 10, 12, 14), we can compare the initial central crosscut to an ovaloid opening in a hydrostatic stress field having a hori- zontal stress component nearly equal to the vertical com- ponent; thus, o-h/ o-v ~1=M (table 12). The tangential i 44 STRESS CHANGES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO stress concentration at the center top surface of the nearly circular (ovaloid) opening would be approximately twice the horizontal or vertical stress component; that is, a, N 20h N 203,. In widening the opening to a rectangular opening with W0 /Ho =3, the tangential stress on the center top surface would then be about three-quarters of the hori- zontal or vertical stress; that is, a, @340}, z 31101,. The tan- gential stress at the center of the roof of the rectangular opening was then less than half what it was before the Opening was widened. Thus, the elastic response of the tangential stress in the roof of the central drift should have been to decompress as the widening of room 2 proceeded. The decompressive stress changes should and did reach a maximum within the completely relieved zone, as shown in figure 43, decreasing upward and becoming negligible at approximately 6.1 m (20 ft) above the roof of the room in the undisturbed in situ stress field. This decompression above the ceiling indicates that the roofload or abutment load caused by lateral excavation was being transferred upward and was arching progressively higher above the excavation. Probes 15 and 24, approximately 4.3 m (14 ft) above the initial completely relieved zone, partially reflect this progressive upward transfer of load. Probe 24, which had a preroom-excavation monitoring period, showed an initial increase in compression before room excavation. Both instruments showed a net decompression in all direc- tions during room excavation, followed by stabilization. Probes 19 and 25, on the other hand, being within the 1.2- m (4 ft) relieved zone, should have undergone only minor decompression during room widening. The data for probe 25 indicate that this did happen. The net change was slightly compressive after excavation ceased. The zone probably was not completely relieved in all directions by the initial fault movement, and some components of de- compression were observed in directions of decreased con- finement. Probe 19, however, showed a progressive net de- compression in all directions throughout the period of measurement. Perhaps continued fault adjustments caused localized stress relief of rock during excavation. The concentration centers of the principal stress-change directions for probes 15, 19, 24, and 25 are shown in figure 29. Superposed on the diagram are major geologic fea— tures and excavation orientation that appear to exert varying degrees of control upon the stress field. The geologic features include the orientation of fault 2 and contours of poles of major joint sets. The most prominent joint set is that of foliation joints which are oriented nearly perpendicular to the fault plane, a common relation in this part of the mine. The stress changes that occurred at probes l5 and 19 appear to be strongly influenced by fault 2 in that the principal stress-change directions (Aal,A02,Aag) either lie within the fault plane or nearly coincide with the perpen- dicular to the fault. The stress changes occurring at probes 24 and 25 may have been more strongly influenced by the room orientation and joint sets than by the fault. The principal stress-change directions aline closely with these features. The A01 direction for probes 15, 19, and 25 and the A02 direction for probe 24 are nearly perpendicular to the room axis or in the same direction as the widening of the crosscut. One of the major joint sets is nearly perpendicu- lar to this direction and parallel to the room axis. The least decompression occurs normal to the room axis for three of the four probes in the roof and only inter— mediate decompression for the fourth probe; it is inferred that the greatest constraint to deformation (confinement) is perpendicular to the room axis. The joint set parallel to the room axis (fig. 29) does not significantly affect con- finement across these joints, probably owing to their tight- ness and lack of alteration. Although the room excavation proceeded perpendicular to the crosscut axis, the room was always longer parallel to the crosscut axis. Maximum con- finement was horizontal and perpendicular to the room' axis. Accordingly, minimum decompression in the rock in the roof of the room occurred as anticipated by elastic theory, perpendicular to the room axis. The long axis of the room should be parallel to the direction of least confinement of the roof rock. For probe 25, in the lower partially relieved zone, this appears to be true. In addition, the greater compressibility of the rock perpendicular to the foliation and prominent foliation joints acts to decrease confinement along the room axis. However, probe 24, which is 4.3 rn (14 ft) above the relieved zone, was still compressed by the arched roofload. The fact that more rock has been excavated in the direction of the room axis than in other directions may explain why the roofload at depth is greatest in this direction. As the room was widened the compressive roofload was transferred higher into the rock and a strong constraint was thereby maintained in the direction of the room axis. A minimum decompression along the crosscut axis was observed at deep probe 24 after postexcavation stability was achieved. The maximum decompression direction at deep probe 24 and the intermediate decompression direction at shallow probe 25 are nearly coincident. They dip steeply parallel to the room (crosscut) axis and normal to a strong joint set (fig. 29). The deep probe 24 had the least confine- ment in the direction of decreasing overburden to the ground surface and apparently confinement was reduced by the strong joint set, which is approximately parallel to the ground surface. At probe 25 the intermediate confine- ment was apparently controlled by the same joint set. In summary, the compressive stress concentration in the rock mass in the roof of the room at probe 24 migrated up- ward and southerly—a direction that coincides with the minimum overburden thickness, with the joint set nor— mal, and with the room axis. The resulting decompres- sion was maximum in this direction. On the other hand, probe 25, in rock that was partially relieved prior to room excavation, was not so much affected by the upward DISCUSSION OF STRESSES IN THE EXPERIMENTAL MINE 45 migration of the roofload as it was affected by reduced con- finement in the direction of the room axis. Foliation joints had the greatest influence on decreasing restraint in this zone. Fault 2 had a great influence upon the principal stress- change directions of probes 19 and 15. Deep probe 15, which is in the unrelieved zone, underwent maximum decompression normal to the fault, accompanied by mini- mum and intermediate decompression in directions paral- lel to the fault surface. The minimum decompression direction was nearly normal to a major joint set and to the room axis. It would appear that as the roofload migrated upward during room excavation fault 2 provided the least constraint, whereas the short room dimension provided the maximum constraint. Other joints were not effective in reducing constraint perpendicular to the room axis. The intermediate constraint was parallel to the fault sur- face but does not appear to be related to other geologic or excavation factors. Probe 19, in the shallow, partially relieved zone, be- haved in a manner similar to deep probe 15 above it in that the principal stress-change directions were nearly parallel to or perpenducular to the surface of fault 2. As seen in fig- ure 29, A01, the least decompression, was parallel to the sur- face of fault 2 and perpendicular to a major joint set and the room axis. The intermediate decompression, A02, was nearly perpendicular to fault 2, and the maximum de- compression, Aoa, which lay in the fault plane, was approximately perpendicular to the foliation joints and approximately parallel to the room axis. The constraint near shallow probes 19 and 25 in the re- lieved zone was similar, except that probe 19 was influ- enced more by fault 2 than was probe 25. Both the maxi- mum and minimum stress-change directions sensed by probe 19 are parallel to fault 2. The intermediate stress- change direction sensed by probe 19 lay closer to the per— pendicular to fault 2 than the intermediate stress-change direction sensed by probe 25, which was more influenced by foliation joints. DISCUSSION OF STRESSES IN THE EXPERIMENTAL MINE Harrison and Moench (1961) and Moench and Drake (1966a) have distinguished both Laramide and Precam- brian regional joint systems in the Idaho Springs area. Moench and Drake ( 1966a, p. 43) suggested, on the basis of data on faults, joints, and veins, that the stress field was compressive and was oriented east-northeast in early Ter- tiary time. Later in Tertiary time, these authors postu- lated that the region was subjected to tensional stress also oriented east-northeast. R. H. Moench (written commun., 1974) found that the youngest fault movements (postmineralization) were noted on faults that trend north and northwest, appropriate for a, oriented about N. 12°W. and as about N. 78°E. (both presumably horizontal). Our measurements throw little light on these generalities. In fact, the five in situ stress determinations were marginally adequate to describe the stress field in one section of a mine in the Idaho Springs district. If one considers the effects of residual stresses in addition to tectonic and gravitational forces and the complex tectonic history, as well as variations in rock composition, it is perhaps reasonable to expect variation, rather than consistency, in stress magnitudes and orientations. Boundary stresses could change with depth and, in the case of most near-surface measurements in mountainous regions, they would be influenced by topography. We have, therefore, resisted the temptation to apply the data gained in this investigation to areas remote from the field site. We believe that to do so, without intermediate measurement locations, would oversimplify, perhaps incorrectly, regional structural relations. We can inter- pret the rock behavior encountered in the experimental mine with greater confidence. The unequal compressive stresses measured in oppos- ing walls of fault I are difficult to explain by either over- burden or tectonic loading; likewise, the high-magnitude vertical decompressive stresses associated with the excava- tion of room 1 do not fit these conventional loading con- ceptions. It is unlikely that such large stress concentra- tions (differences) could be due to the geometry of the opening or to contrasting material properties. The magni- tudes of vertical stresses approximate those of horizontal stresses at a given location and, at the shallow depth of the mine, are probably too large to be produced by either a gravitational or a tectonic boundary load. The experi- mental mine is situated well above the nearby surface drainage to the south and east, which would reduce the magnitudes of transmitted boundary loads. In addition to the overburden load, and possibly tec- tonic stress components, the rock mass at the field site contains releasable residual (”locked-in”) stresses that are in turn related to past geologic processes. The presence of unequal compressive stresses in the walls of faults suggests to us that ancient fault movement preferentially relieved stresses, or concentrated more compression in one wall than in another. The effect would be to store more strain energy in the less relieved block while releasing part of the strain energy in the adjacent block. Rock near several probe locations showed early stress changes alined with foliation and later changes alined with fault orientation. Drilling and blasting apparently disturbed the fault equilibrium, triggering a release of residual stress. The re- sulting stress change imposed a new orientation on the stress field at several locations. Stress changes can also be activated by overcoring. The strain energy released can be large, as shown by the 4,350- lb/in2 compressive stress released by coring a rock sample after it was removed from the mine. Residual stresses of 46 STRESS CHAN GES CAUSED BY EXCAVATION, IDAHO SPRINGS, COLORADO these magnitudes, locked in the rock fabric during past episodes of deeper burial and more active tectonism, can account for the magnitude of the stress changes measured in the vicinity of the excavations. Stress Changes occurred as much as seven tunnel diam- eters ahead of the face, suggesting that releases of resi- dual strain energy may act for considerable distances along faults and foliation. Decompressions, such as were monitored by probes 5 and 8 in the walls of room 1, the roof of room 2, and ahead of the advancing crosscut, may be reasonably explained by the release of residual strain energy. An expected and observed concentration of compression at shallow depths in the walls was caused by load transfer due to removal of rock. Then, as excavation removed confinement, the walls of the excavations progressively deflected and deteri- orated. The result was a large docompression. The laboratory model-tunnel studies were only partly successful in predicting the stress changes encountered at the field location. The overriding influence of faults, joints, and foliation was not predictable from our simple unit-block models. The presence and behavior of residual stresses were not anticipated, and such a complex natural stress field would be extremely difficult or impossible to accurately simulate in a model. The models did, however, anticipate the underground findings of stress changes as much as two diameters ahead of the advancing tunnel face, but much less than the actual distance. Continued effort will be required to improve our know- ledge of geologic structures and associated stresses. A better understanding should come as investigators are able to distinguish overburden, boundary-tectonic, and residual components of the in situ stress field and then establish their relative importance. REFERENCES CITED Abel, J. F., Jr., 1967, Tunnel mechanics: Colorado School Mines Quart, v. 62, no. 2, 88 p. Bastin, E. S., and Hill, J. M., 1917, Economic geology of Gilpin County and adjacent parts of Clear Creek and Boulder Counties, Colorado: U.S. Geol. Survey Prof. Paper 94, 379 p. Bielenstein, H. U., and Eisbacher, G. H., 1969, Tectonic interpretation of elastic-strain-recovery measurements at Elliot Lake, Ontario: Ottawa, Dept. Energy, Mines and Resources, Mines Br. Research Rept. R 210, 64 p. Brown, E. T., and Trollope, D. H., 1967, The failure of linear brittle materials under effective tensile stress, in Rock mechanics and engineering geology: Vienna, Springer—Verlag, v. 5, no. 4, p. 219—241. Carter, W. H. N ., 1961, A review of strata control experience in long- wall working in Great Britain, in Internat. Conf. Strata Control, 3d, Paris 1960, Rev. de l’Industrie Minerale: p. 471—482. Conway, J. C., 1963, An investigation of the stress distribution in a circular cylinder under static compressive load for varying boundary conditions: Pennsylvania State Univ. M. S. thesis, 201 p. Cox, R. M., Jr., 1971, Rock behavior during experimental room excava- tion in Idaho Springs gneiss: Colorado School Mines D. Sci. thesis T—1283, 215 p. Culver, R. S., 1967, Evaluation studies on rock bolts and rock mechan- ics instruments, phase 1: Omaha Dist. Corps Engineers Tech. Rept. 4, 74 p. Denkhaus, H. G., and Hill, F. G., 1961, The conditions of the ground around excavations in hard rock at great depth, in Internat. Conf. Strata Control, 3d, Paris 1960, Rev. de l’Industrie Minerale: p. 243-251. Edwards, R. H., 1951, Stress concentrations around spheroidal inclu- sions and cavities: Jour. Appl. Mechanics, v. 73, p. 19—30. Galle, E. M., and Wilhoit, J. G., Jr., 1962, Stresses around a well bore due to internal pressure and unequal principal geostatic stresses: Soc. Petroleum Engineers Jour., v. 2, no. 2, p. 145—155. Goodier, J. N., 1933, Concentration of stress around spherical and cylindrical inclusions and flaws: Jour. Appl. Mechanics, Am. Soc. Mech. Engineers Trans., v. 55, p. 39—44. , Harrison, J. E., and Moench, R. H., 1961, Joints in Precambrian rocks, Central City-Idaho Springs area, Colorado: U.S. Geol. Survey Prof. Paper 374—8, 14 p. Henshaw, H., 1961, Discussion of Carter, W. H. N., A review of strata control experience in longwall working in Great Britain, in Internat. Conf. Strata Control, 3d, Paris 1960, Rev. de l’Industrie Minerale: p. 484. Lee, F. T., Nichols, T. C., Jr., and Abel, J. F., Jr., 1969, Some rela- tions between stress, geological structure, and underground ex- cavation in a metamorphic rock mass west of Denver, Colorado, in Geological Survey research 1969: U.S. Geol. Survey Prof. Paper 650-C, p. C127—Cl32. Leeman, E. R., 1961, Measurement of stress in abutments at depth, in Internat. Conf. Strata Control, 3d, Paris 1960, Rev. de l’Industrie Minerale: p. 301—311. 1964, The measurement of stress in rock; I, The principles of rock stress measurement; II, Borehole rock stress measuring instru- ments; III, The results of some rock stress investigations: South African Inst. Mining and Metallurgy Jour., v. 65, p. 45—114, 254-284. Lovering, T. S., and Goddard, E. N., 1950, Geology and ore deposits of the Front Range, Colorado: U.S. Geol. Survey Prof. Paper 223, 319 p. [1951]. Medearis, Kenneth, 1974, Numerical-computer methods for engineers and physical scientists: Fort Collins, Colo., KMA Research, 275 p. Moench, R. H., 1964, Geology of Precambrian rocks, Idaho Springs district, Colorado: U.S. Geol. Survey Bull. 1182—A, 70 p. [1965]. Moench, R. H., and Drake, A. A., Jr., 1966a, Economic geology of the Idaho Springs district, Clear Creek and Gilpin Counties, Colorado: U.S. Geol. Survey Bull. 1208, 91 p. 1966b, Mines and prospects, Idaho Springs district, Clear Creek and Gilpin Counties, Colorado—Descriptions and maps: U.S. Geol. Survey open-file report, 214 p. Moench, R. H., Harrison, J. E., and Sims, P. K., 1962, Precambrian folding in the Idaho Springs-Central City area, Front Range, Colorado: Geol. Soc. America Bull., v. 73, no. 1, p. 35—58. Nichols, T. C., Jr., Abel, J. F., Jr., and Lee, F. T., 1968, A solid- inclusion borehole probe to determine three-dimensional stress changes at a point in a rock mass: U.S. Geol. Survey Bull. l258—C, 28 p. Obert, Leonard, and Duvall, W. 1., 1967, Rock mechanics and the design of structures in rock: New York, John Wiley 8c Sons, Inc., 650 p. Obert, Leonard, Duvall, W. I., and Merrill, R. H., 1960, Design of underground openings in competent rock: U.S. Bur. Mines Bull. 587, 36 p. Portland Cement Association, 1952, Design and control of concrete mixtures [10th ed.]: Portland Cement Assoc., 68 p. Potts, E. L. J., 1961, Discussion of Denkhaus, H. G., and Hill, F. G., The conditions of the ground around excavations in hard rock at great depth, in Internat. Conf. Strata Control, 3d Paris 1960, Rev. de l'Industrie Minerale: p. 252. ——'———' REFERENCES CITED 47 Reeper, F. J. M. de, and Bruens, F. P., 1961, Measuring the loads on roadway supports by means of strain gauges, in Internat. Conf. Strata Control, 3d, Paris 1960, Rev. de l’Industrie Minerale: p. 321—337. Scott, G. R., 1973, Tertiary surfaces and depsoits of the Southern Rockies and their recognition [abs]: Geol. Soc. America Abs. with Programs, v. 5, no. 6, p. 510. Sims, P. K., and Sheridan, D. M., 1964, Geology of uranium deposits in the Front Range, Colorado, with sections by R. U. King, P. B. Moore, D. H. Richter, and J. D. Schlottmann: U.S. Geol. Survey Bull. 1159, 116 p. Spackeler, G., 1961, Discussion of Carter, W. H. N., A review of strata control experience in longwall working in Great Britain, in Internat, Conf. Strata Control, 3d, Paris 1960, Rev. de l’Industrie‘ Minerale: p. 483. Spurr, J. E., and Garrey, G. H., 1908, Economic geology of the George- town quadrangle (together with the Empire district), Colorado, with general geology, by S. H. Ball: U.S. Geol. Survey Prof. Paper 63, 422 p. ’ Tooker, E. W., 1963, Altered wallrocks in the central part of the Front Range mineral belt, Gilpin and Clear Creek Counties, Colorado: U.S. Geol. Survey Prof. Paper 439, 102 p. [1964]. Turner, F. J., and Verhoogen, John, 1960, Igneous and metamorphic petrology [2d ed.]: New York, McGraw—Hill Book Co., Inc., 694p. Tweto, Ogden, and Sims, P. K., 1963, Precambrian ancestry of the Colorado mineral belt: Geol. Soc. America Bu11., v. 74, no. 8, p. 991—1014. Vames, D. J., 1970, Model for simulation of residual stress in rock, in Symposium on rock mechanics, 11th, Proc.: New York, Soc. Mining Engineers, AIMMPE, p. 415—426. Varnes, D. J., and Lee, F. T., 1972, Hypothesis of mobilization of residual stress in rock: Geol. Soc. America Bull, v. 83, no. 9, p. 2863—2865. 9 U.S. GOVERNMENT PRINTING OFFICE 1976-777034/21 UNITED STATES DEPARTMENT OF THE INTERIOR PROFESSIONAL PAPER 965 GEOLOGICAL SURVEY PLATE 1 Medium, to coarse— grained banded biotite-horn ~ blende gneiss Blocky but competent \ @015? 70 ®o.1o 55 © )§;« 0:30 \ ® ®;\ Generally massive granite gneiss; 0030 47/59 migmatitic zones ’6, 5 .. Altered and fractured zone (gouge 690 ,1 ° and crushed rock). Most of south 5g 9/ J / so * 9 70 0 wall altered 2—3 ft in from wall )fio 63'6“) @013 and in roof to centerline of C— Left (D 0' QO 0 drift {’0‘ _' o Shear zone parallel to foliation; as © +60 75 . 75 0. / ®~ much as 6 in. gouge; extends to V0 @ 0; Miami tunnel o 50 / ‘ A dao 0: _ ®° y‘ 4 c; Masswe medium- to M @02 (-9 coarseegralned_ 45 60 pegmatitic gneiss )et o0 ‘ 0 ®\» Fine-grained banded . . . . Q \ hornblende bio— SIIICIerd altered zore parallel to 90 (D 0.5 o tite gneiss; blocky; 0.27 to 0.6»ft joint blocks joints and foiation Massive granitic to biotitic gneiss Gneiss as above but closer spaced joints\ Probes Bv2 (overcored) Very massive granite gneiss; \ 18 (overcored) 0 Probe 21 ‘ \ 3‘5 75 vercored \ / 90 Probe 22 ® 0&0 Biotite-hornblende 65 s o gneiss ® 04 0 \i® (D 0.5 o 0‘9\ 3,5/Probe 25 at 4 ft Closely banded (0.5—1 in.) t0 mas— Probe 24 at 18 ft sive hornblende gneiss; calcite o 60 veinlets 'e\ (l, )3 o‘ 70 @ ° «So \ $50.2 / @070 No 65 Q} 9 25 o @ Re 75 . \O/ 75 o 0.7. / €93 5,) Probe 19 at 4 ft Zone of minor shearing parallel to Pegmatite; biotite lenses o/F’robe 15 at 18 ft foliation '0‘) ® 0.6 \ 50 \ 55 4—... \0'3% Massive pegmatite parallel to © o' 55 > foliation 65\ 6 Probe 9 (D 12 50 Banded quartz-plagioclaseebiotite/ ® 0- 0 July 28, 1967 gneiss; pegmatite and sulfide veins / Massive / pegmatite Biotite gneiss and migmatite Probe 10 July 28,1967 Intensely sheared zone 6 in. to 2 ft wide; ~20 percent gouge / Moderately to intensely sheared ——/ Massive hornblende plagioclase— biotite gneiss and pegmatite gneiss;.minor :anding;lpegma- ® 01 tlte veins para 6 to o latlon / f/Well-foliated biotite / f / gneiss and interlay— / / ered amphibolite / '1. // ,f / f \3 1/ / Massive biotite gneiss p.0/ / 55 T/ m/ / J/ // ® 65 Q80 0 Massive granite gneiss 5 .1 / /_/_, 1 65 '9\Q47630\65® 09 O “we“ ///~ r / \®o.57 @100 a5 / / // Massive gneissic granite 60 _/J 65 03K“ 0 @03 o Banded biotite gneiss grades to . . ®//‘: . . Mainly granite; r granite gneiss moderately sheared [J70 55 (99-1+ Altered zone; intensely sheared; fflx Pegrnatite > 20 percent gouge // H (3 0 3 O o 65 05M //9 ”5 - EXPLANATION O / f / air 45 // / (9 60 x o / / SYMBOLS JOINT INFORMATION 03 N // / ® 0 65 "- /" (9 75 @331 5,5”\® 0 6 o + ’ 0 Granite gneiss‘ . _ _ t 75 to ‘ _ , Stoke and d1p of follat1on . . . / / \ . /‘6® c/L7O Pesmatlte Strlngers S1gn1f1cance Average spacmg Alteration ® 0.1 90 Pegmame 70 in feet 4y\50 0 55 gneiss 09‘ V» /// ‘ ~If . _ Q) Major + Gouge or intense Oé>®o (“2X @2 ()9 Finegrained biotite-hornblende 57’ v /// Stnke and dip OfJOIIltS fracturing Q? \3 o 's ’00 69’ gneiss; amphibolite lenses (:99 \N‘x’» // ® Intermediate . ® 65 o Pegmatite we" (0".165 // 40 0 Minor gouge and Medium— to coarseegralnid biotite, ® 0.2 o\ O ' gags e39“ / / I __ ® Minor f , t . quartz-plagioclase gneiss; minor 6;.6 75 75 360 Pegmatitic 099 ((30 / _ . “C urlng alteration; includes mmerous 0‘70 ® 01’ if/Vo zone “9’ 0“" // / Strlke and dlp 0f faults stringers of biotite—hlrnblende 2’. \k , ® 75 0W0 03 (5‘0 609‘ / , Fresh unaltered neiss . . . e‘ , ® 045 , o\t.\>9ve 9,9 / . A N V i m K r surface 9 0.1 75 (/5 a {\er f;/ Amphibolite; interlayered m ~ :1 ; NA» Pegmafite Qify/ biotite gneiss Zone of shearing N or alteration Probe 1 9 Drillhole with borehole probe (9 Vertical hole in roof Coarse~grained granite ® Probe 17 (overeo red) Gran itic gneiss and Iron—stained, ° NE' pegmatite veins blocky granite Massive gneissic Minor water from roof I. granite Medium»grained biotite gneiss, well (Do 6 . . . . o foliated; pegmatite stringers usur :5\ / ally parallel to foliation; wet; dis- seminated pyrite veins parallel to foliation 25 Blocky / granite Migmatite with pegmatite stringers Minor waterin @3030 10 o 10 2o 30 40 FEET '°°f and “"5 Massive l . ' l l ! l ‘ ‘ r J I l I I l l Coarse-grainedfeld» ‘5 granite 5 O 5 10 METRES Probe 5/ sparerlch peg- 80 Oct. 19, 1966 matite paral|e|®0~3 O to foliation Probe B»3 (overcored) Biotite schist, gneiss; minor granite; moi derately altered GEOLOGIC MAP OF PART OF COLORADO SCHOOL OF MINES EXPERIMENTAL MINE, IDAHO SPRINGS, COLORADO Plan view—4 5 ft above invert Coarseegrained feldsparerich/ . . pegmatite in gneiss Geology by F. T. Lee and T. C. Nichols, Jr., 1967—69 Cr U.S. GOVERNMENT PRINTING OFFICE I976—777-034/2l ? BAY The Geologic Retrieval and Synopsis Program (GRASP) GEOLOGICAL SURVEY PROFESSIONAL PAPER 966 DOCUMENTS DEFI‘RR'E‘MENT NOV 6 19375 {$23.1 IJNN’E:1::‘.:":Y m: taxingszrgm , The Geologic Retrieval and Synopsis Program (GRASP) By ROGER W. BOWEN and JOSEPH MOSES BOTBOL GEOLOGICAL SURVEY PROFESSIONAL PAPER 966 A portable data-retrieval system requiring minimal user training UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON : 1975 UNITED STATES DEPARTMENT OF THE INTERIOR GEOLOGICAL SURVEY V. E. McKelvey, Director Library of Congress Cataloging in Publication Data Bowen, Roger W. The geologic retrieval and synopsis program (GRASP) (Geological Survey professional paper; 966) Supt. of Docs. no.: I 19.161966 1. Information storage and retrieval systems--Geology. I. Botbol, Joseph Moses, joint author. II. Title. III. Series: United States. Geological Survey. Professional paper; 966. QE48.8.B68 029 '.9 '55 75-619314 For sale by the Superintendent of Documents, US. Government Printing Office Washington, DC. 20402 Stock Number 024-001-027-19-4 Y CON TENTS Page Abstract ________________________________________ 1 GRASP software specifications—Continued Introduction _____________________________________ 1 Subroutine name: COLPNT ___________________ _ Purpose and scope ___________________________ 1 Function name: COMP _______________________ General system description ________________________ 1 Subroutine name: CONDS ____________________ Design philosophy ____________________________ 1 Subroutine name: CONDTN __________________ Machine portability _______________________ 1 Subroutine name: DECOMP ___________________ Data-base independence ___________________ 2 Subroutine name: DEFINE ___________________ User community _________________________ 2 Subroutine name: DEFLST ___________________ Time sharing- ____________________________ 2 Subroutine name: DUMPIT ___________________ Present utilization ____________________________ 2 Function name: EVAL _______________________ COLFIL ________________________________ 2 Subroutine name: FDRIVE ___________________ MANFIL _______________________________ 2: Subroutine name: FILES _____________________ RASS ___________________________________ 3 Subroutine name: FIND ______________________ Future plans _________________________________ 3 Subroutine name: FINDGP ___________________ Detailed system description ________________________ 3 Subroutine name: FIT ———————————————————————— Data—file structures ___________________________ 3 Subroutine name: FTNC ______________________ Mask file ________________________________ 3 Subroutine name: GETPUT ___________________ Definitions file ___________________________ 4 Subroutine name: HELP _____________________ Dictionary file ___________________________ 4 Function name: ICONV ______________________ Multiple-choice file _______________________ 4 Subroutine name: IFILE ______________________ Numeric Master file ________________________ 5 Subroutine name: INIT _______________________ Data compression _____________________ 5 Subroutine name: KEYBRD ___________________ Machine dependencies _________________________ 6 Subroutine name: LENGTH __________________ Internal structure and functions _______________ 6 Subroutine name: LIST ______________________ Processing user input _________________________ 6» Subroutine name: LOGEXP ___________________ Conditional expreSSions ___________________ 6 Subroutine name: MEAN ____________________ Logical expressions _______________________ 6 Subroutine name: NAME _____________________ Number lists _____________________________ 8 Subroutine name: OBEY _____________________ Name lists _______________________________ 8 Subroutine name: OFILE _____________________ Arithmetic expressions ____________________ 8 SUbTOUtine name: OPREP ____________________ Searches ________________________________ 8 Subroutine name: PACK _____________________ Output __________________________________ 9 Subroutine name: PARSE ____________________ Data-base implementation _________________________ 9 :ubrilltine nameég’éz-E‘IRSE ———————————————————— unc ion name: ______________________ 33:6“. GRA.SPd """""""""""""""""""""" 10 Subroutine name: PREVAL ___________________ e rences ate -------------------------- 12 Subroutine name: QUIT ______________________ GRASP software specifications _____________________ 13 Subroutine name: RELEXP ___________________ Module name: DRIVER ______________________ 13 Subroutine name: RETRVE __________________ Subroutine name: ACCESS ___________________ 15 Subroutine name: RLIST _____________________ Subroutine name: BDEF _____________________ 17 Subroutine name: ROWPN T __________________ Subroutine name: BFIND ____________________ 19 Subroutine name: SCAN _____________________ Subroutine name: BINIT _____________________ 20 Subroutine name: START ____________________ Subroutine name: BINTYP ___________________ 21 Function name: UNCODE ____________________ Subroutine name: BLIST _____________________ 22 Subroutine name: VLIST _____________________ ILLUSTRATIONS Page FIGURE 1. Example of Mask-file and Dictionary-file arrangement ____________ 4 2. Example of Numeric Master file arrangement prior to compression __ 5 3. Pictorial summary of GRASP subroutine interrelationships _________ 7 III Page 23 25 27 28 29 31 32 33 35 37 38 40 41 43 45 46 47 48 49 50 51 52 53 56 58 59 61 62 63 64 68 71 72 73 76 81 83 84 85 86 A THE GEOLOGIC RETRIEVAL AND SYNOPSIS PROGRAM (GRASP) By ROGER W. BOWEN and JOSEPH MosEs BOTBOL ABSTRACT The Geologic Retrieval and Synopsis Program (GRASP) was designed and written to specifically accommodate inter- active access to earth-science data banks. GRASP is portable, easy to use, and data-base independent. Data banks accessed by GRASP must be partitioned and reformatted into five files which make up both data bank and pointers to parts of the bank. Machine dependencies include FORTRAN I/O unit numbers, direct-(random) access input, in-core read/write, and “prompting.” GRASP isolates these dependencies to FORTRAN subroutines designed to serve these functions specifically. GRASP is manipulated by 11 user commands which select, describe, access, retrieve, summarize, and display data. INTRODUCTION The US. Geological Survey presently has the re- sponsibility of developing and maintaining resource- data banks. Initially, many storage and retrieval systems were critically reviewed for their data-bank characteristics, ease of use, flexibility, portability, and applicability to Survey activity. The authors con- cluded that no available system was Wholly adequate for the needs of the Survey data banks. Generally, the observed systems were difficult to use, machine bound, or were oriented toward one type of data (for example, text oriented). The only logical al- ternative was to design, develop, and implement a geologic-data storage and retrieval system to be used primarily by geologists. PURPOSE AND SCOPE The Geologic Retrieval and Synopsis Program (GRASP) was written to provide a means of in- teractive access to geologic data stored in a time- sharing computer. GRASP can be implemented on any time-sharing computer system that has a FORTRAN IV com- piler. Data bases accessed by GRASP must con- tain fixed field data in alphameric, alphanumeric, and/or numeric modes. GENERAL SYSTEM DESCRIPTION To obtain a broad overview of the GRASP sys- tem, consideration should first be given to the philosophy governing the design of the system, use of the present system, and future system plans. This overview provides the “framework” and the anticipated “operational environment” which are necessary for the development of any system. Both the present utilization and future system plans show the correctness of the parameters and tech- niques used as well as the original assumptions re- garding “how,f’ “by whom,” and “Where” the sys- tem will be used. DESIGN PHILOSOPHY Three vital questions that must be answered prior to the implementation of any system are: (1) On what computer (s) will this system be used? (2) What are the characteristics of the data to be proc- essed? (3) Who will use the system? As is the case in many computer-system applications, these ques- tions originally had no definite answers. A 3-month effort was necessary to establish criteria that would govern GRASP design. MACHINE PORTABILITY To serve as broad a spectrum of the scientific community as possible, a system should be as port- able as possible. Because of differences in computer design, no system can be used on all computers With- out some modification. ANSI FORTRAN IV is uni- versally accepted as a standard programming lan- guage. It may be used on the vast majority of pres- ent-day computers that have the capacity for imple- menting compiler-level languages in a time-sharing mode of operation. For this reason, all the process- ing subroutines in GRASP are written in ANSI FORTRAN IV (machine-dependent features iso- lated to facilitate implementation). In this way, GRASP can be installed on virtually any modern 1 2 GEOLOGIC RETRIEVAL AND SYNOPSIS PROGRAM (GRASP) time-sharing computer. By designing GRASP to be portable, a much wider spectrum of the scientific community can be served by the system. Data need not be transferred from their resident banks in order to be accessed. GRASP could be used on most computers in order to access data files wherever they may reside. This machine independence elimi- nates the need for tedious data transformations to one system configuration, where the aggregating data eventually flood to the point of uselessness the peripheral storage of a central machine-dependent data bank. The authors believe that a common ac— cessing system for data residing in different com- puters is preferable to an accessing system that can be used for data resident in only one computer. DATA-BASE INDEPENDENCE GRASP is designed to operate using any data base that can be represented in conventional matrix form. In matrix form, the records (that is, items to be described) are the rows, and the attributes of each record are the columns. The real structure of the data base can be thought of as the titles and arrangement of the columns of a data matrix where the rows are merely instances or occurrences de— scribed by the columns. For example, a geochemical data base would have a matrix representation in which the columns might represent chemical anal- yses of various elements, and each row would rep— resent one sample. GRASP can function on any data matrix. The only requirement is that the vari- ables (or columns) be defined ahead of time in terms of their types (that is, alpha or numeric), and, where necessary, dictionaries of legitimate a1- phameric entries must be provided for alphameric variables. Thus, because of the matrix orientation of GRASP, any fixed field data base can be ac- commodated by the system. USER COMMUNITY GRASP is a retrieval system having its own rules and command language for operation. In other words, to use GRASP, the user does not need to be familiar with FORTRAN or any other computer language. The GRASP command language is de- signed to provide users with the ability to ask ques- tions of a data base and retain all items that an- swer “true” to the questions. The control language used to ask the questions (discussed in the section on “Use of GRASP” in this report) allows “retrieve only” data access to any GRASP user and does not require prior user knowledge of computer languages or system functions. Thus, GRASP can be imple- mented for a wide variety of users. TIME SHARING Inasmuch as GRASP is portable, data—base in- dependent, and serves a wide variety of users, GRASP should be implemented in a computational mode that has the most readily available user ac- cess, namely, time sharing. In its simplest form, time sharing allows any user to communicate with the computer from a terminal near a telephone. The entire design of GRASP is based on the premise that the user community will converse directly with a computer (via a terminal) in order to access, re- trieve, manipulate, summarize, and display data. This mode of computation provides the “instant” response necessary for timely decisionmaking, and also allows access by the user from the environment in which the computer response is of most value, that is, the laboratory, field, office, or conference room. PRESENT UTILIZATION In 1975, the GRASP system was being used to ac- cess data from six totally different data banks: (1) oil- and gaSapool characteristics of Colorado, (2) mineral deposits of the world, (3) geochemical ex- ploration data from the United States, (4) coal re- sources of the United States (prototypic data bank), (5) index of US. geologic map coverage (prototypic system), and (6) geothermal data bank (in Pisa, Italy). The first three of the above systems were implemented directly by the authors, and no attempt was made to redesign any of the original data-bank structures. COLFIL This file contains as many as 390_characteristics for each of 800 oil and (or) gas pools in Colorado. This file served as the original model for GRASP design and ultimately will contain 60,000 records. MANFIL The mineral deposits of the world (MANFIL) were the second file implemented using GRASP. It is a computerized batchdprocessing-oriented file con- taining geologic, production, and reserves data from about 4,000 nonferrous metal deposits throughout the world. Each record represents one deposit, and contains as many as 250 variables. Although GRASP was designed using the oil- and gas—pool file as a model, implementation of the world-mineral-de— posits file showed the flexibility of GRASP with re- spect to its data—base independence. DETAILED SYSTEM DESCRIPTION 3 RASS The RASS (Rock Analysis Storage System) file is -a batch-oriented geochemical data bank contain- ing limited geologic descriptions and comprehensive geochemical analyses of all samples processed by the laboratories of the US. Geological Survey. This file contains a unique type of numeric data called “quali- fied values.” Because of the upper and lower de— tectability limits of analytic devices, elements whose presence is known but whose content is out-side the analytical range of a device are sometimes reported at a given analytical cutoff value, accompanied by a ‘ letter indicating whether the content is less than, greater than, or in interference with another ele- ment. Typical qualified entries would appear as L5000, G1000, or H100, where .L signifies a content less than the attached value, G signifies a content greater than the attached value, and H signifies analytical interference at a concentration of the at- tached value. Because many of the RASS data were accompanied by alpha qualifiers, GRASP was modi- fied to accept and process this type of data in addi- tion to the conventional numeric- and alphamveric- data types. All the above files are implemented in a retrieve mode only, and graphics have not yet been added. Input to the files is done by people who are respon- sible for data entry and does not fall in the domain of the user. FUTURE PLANS Currently, the development of GRASP is primar- ily oriented toward implementation of techniques for interactive graphics storage and retrieval. Three problem areas are presently being researched: auto— matic recognition of features on scanned input docu- ments, annotation methods, and resolution of “in- tersecting feature” problems. Present research ef- forts are directed toward feature recognition and subsequent computational extraction of simple boundary vectors from scanned digital maps and photographs. In addition to recognition of features, methods are also being developed for annotation of both graphics-data entry and presentation of graphics data on output. One of the major anticipated technical and philo- sophical problems is concerned with the graphical resolution of intersecting features. Techniques are being developed that should resolve these problems for any particular data set. All the GRASP graphic-s output is being designed primarily for interactive graphics cathode ray tube (CRT) representation. This is in keeping with GRASP’s original “totally interactive” design philo- sophy. DETAILED SYSTEM DESCRIPTION GRASP is designed as a highly modular, hier- archically structured set of sub-routines (see section “GRASP Software Specifications”). Each subrou- tine performs a fixed task. The higher level sub- routines are primarily concerned with the flow of control required to execute a user command. The lower level subroutines are primarily concerned with extremely independent and specific tasks (such as “get a record,” “access a dictionary,” and “accept user input,” and so on). All information related to a specific data base is obtained from various files associated with that data base. Structuring the sys- tem in this way leads to a high degree of functional isolation. These design characteristics simplify the development, documentation, maintenance, growth, and inevitable change inherent in a system that sup— ports a variety of data bases on a wide spectrum of computer main frames. The section on GRASP soft- ware specifications is intended for use by those fa— miliar with FORTRAN language. DATA-FILE STRUCTURES Upon initial execution, the GRASP system reads an “index” file which contains the names of data bases available for access. Each record of the index file corresponds to a data base and contains the names of the files associated with that data base and a 40-character description of the data base. Each data base is composed of five files which con- tain the actual data, information on the structure of records, names which will be used to refer to par- ticular items within records, descriptive information on the names themselves, and a grouping into cate- gories of information. These files are called Mask, Definitions, Dictionary, Multiple-choice, and Nu- meric Master files. MASK FILE The Mask file contains the item names, item types (integer, real, character string, multiple choice, and qualified real), and pointers to the first entry in the Dictionary file for each character-type item. This file is read once and rewound when a data base is selected via the FILE command. An example of Mask file arrangement is shown in figure 1. GEOLOGIC RETRIEVAL AND SYNOPSIS PROGRAM (GRASP) Conceptual noncomputerized dictionaries for three character type variables: Variable No. 1 Variable N0. 2 Variable No.3 In the computer, the Dictionary File is arranged as follows: Dictionary File Continent Country Province (State) Item No. Pointer to Name next item North America USA California 1 2 North America South America Canada Virginia 2 3 South America Europe Mexico British Columbia 3 0 Europe Argentina Quebec . 4 5 USA Bram] Cordova Chile 5 6 Canada Germany 6 7 Mexico . fl J 7 8 Argentina In the computer, the MASK File is arranged as follows: 8 9 Brazfl 9 10 Chile Mask File 10 0 Germany Variable Name Variable Type Starting Position in 11 12 California Dictlonary File Continent Character 1 12 13 Virginia Country Character 4 13 14 British Columbia Province (State) Character 1 11 h 14 15 Quebec Production Numeric, real * 15 0 Cordova *Note: This variable is numeric, and does not require a pointer to the Dictionary File (“0" indicates end of list for particular variable) FIGURE 1.—Examp1e of Mask-file and Dictionary-file arrangement. DEFINITIONS FILE The Definitions file contains the following infor- mation : 1. 2. PS?" The number of categories in the file. ”The maximum number of (computer) words in a category name. The category names. For each category the following information is recorded: (a) category number. (b) number of lines used to describe this cate- gory. (c) maximum length (in computer words) of a description in this category. (d) number of variables appearing in this category. In some cases this will be dif- ferent from item b (the number of lines for description). (e) indices of the variables appearing in this category. (f) the variable names, types, and descrip- tions for this category. DICTIONARY FILE The Dictionary file contains all character-string values Which are assumed by character-type items. Each record contains a pointer to the record con- taining the next value, followed by the current value. The last value assumed by a character-type item is indicated by a pointer value .of zero (the record containing the first character-string value for a character-type item is pointed to by a value in the Mask file). The Dictionary file is designed as a ran- dom-access file whose values form a linked list. Fig- ure 1 shows an example of the Dictionary-file ar- rangement. MULTIPLE-CHOICE FILE The Multiple-choice file contains the acronyms and acronym meanings for the values assumed by mul- DETAILED SYSTEM DESCRIPTION 5 tiple-choice items. Each record of this file is com- posed of an item number indicating the multiple- choice item, the number of possible values this item assumes, the maximum length of an acronym value description, and a list of acronyms (which are double words) and their descriptions. NUMERIC MASTER FILE The Numeric Master file is composed of the rec- ords for a data base in a compressed form. Values for integer-type items are stored as integers. Values for fioating—point— (or real-) type items are stored as real numbers. Values for character-type items are stored as integer pointers to the entry number in the Dictionary file. Values for multiple—choice-type items are stored as integers containing a binary en- coding that represents the value set. (For example, if the second and fifth bit of the word are “on,” the value assumed is the second and fifth acronym value.) Each record of the compressed Master file is variable length in form and corresponds to an ex- panded 400-word record. Expansion of the com- pressed record is performed by subroutine GET— PUT. Figure 2 shows an example of the Numeric Master file prior to compression. DATA COMPRESSION The compression technique used is a form of blank suppression. The words of the compressed record are one of the following four types: A. Integer value. B. Real value. C. Integer blank count. D. Integer word count. The first word of all records is of type D (above) and is used to give the length of the record. Subse- quent words may be types A, B, and C. For types A and C, the last two bits give the type of the next word. The value of the word is obtained by divid- ing {by 4. The type of the next word is obtained via the remainder modulo 4, where the numbers 1—3 correspond to types A, B, and C. Type-A words are used for numeric integers, pointers to entries in character dictionaries, and binary encodings of mul- tiple-choice—type items. Type-C words are used to count the number of consecutive blanks to be in- serted in the expanded record. Type-B words are used for floating-point numeric values. The type of the next word is contained in the last 2 bits of the whole (integer) part of the words. For example, consider a data word having a value of 49.723. The Given the following two successive noncomputerized records to be entered into the Numeric Storage File: Continent* Record 1 Record 2 Continent ................................. North America South America Country .................................... United States Argentina Production ________________________________ 39281.6 49298.7 Identification No _______________________ ' 38 39 Province (or State) ____________________ Virginia Cordova Prior to compression, the computerized Numeric Storage File is arranged as follows: Country* Production ID No. Province* etc. 1 1 39281.6 38 2 2 4 49298.7 39 5 *See figure 1—for dictionary codification of continent, country, and province FIGURE 2.—Example of Numeric-Master-file arrangement prior to compression. 6 GEOLOGIC RETRIEVAL AND SYNOPSIS PROGRAM (GRASP) floating-point value would be 12.723 (12=49/4), and the type of the next word would be 49— (4X 12) = 1 or A. MACHINE DEPENDENCIES Although most of the GRASP-system code is written in ANSI FORTRAN IV, certain isolated functions must be tailored to the particular FOR- TRAN compiler on any given machine. These func- tions deal with the dynamic association of data set names and FORTRAN I/O unit numbers, direct- (random-) access input, the method of accommodat— ing “prompting,” and internal (in-core) transfer (writes) using format control. For the dynamic association of data-set names and FORTRAN unit numbers, the routines IFILE, OFILE and DEFINE are used. Details of these routines can be found in the section “GRASP Soft- ware Specifications.” Direct—access input is used to access the Diction- ary file (in subroutine ACCESS). The FORTRAN unit number, an integer expression giving the record number, and an input list are supplied in the READ statement. This form of direct-access input is com- patible with most FORTRAN compilers which sup- port direct-access input. Systems not having direct- access capabilities can be accommodated by modify- ing the logic of this subroutine. This modification involves positioning of a sequential file to the ap- propriate record prior to execution of the READ statement. For systems which do not accept the “prompt” op- tion in the READ statement, user “prompting” can be accomplished by using WRITE statements im- mediately preceding (in time) user input. The “prompt” message is contained in a FORMAT state— ment, along with a character which inhibits the generation of the normal carriage-return/line-feed usually associated with output directed to a time- share terminal. If a particular system does not have this capability, the “prompting” message will ap- pear on a separate line immediately preceding the user input. The internal transfer of data under format con— trol is accomplished via the ENCODE statement. The ENCODE statement is used in subroutine COLPNT to construct a line of output. The only other use of ENCODE is in subroutine PACK which is used to convert characters from unpacked to packed form. Most non-IBM FORTRANS support this statement in one form or another. In the case of IBM FORTRAN, a routine must be provided that allows internal data transfer under format control. INTERNAL STRUCTURE AND FUNCTIONS GRASP is designed to accept a “command” (or directive) from the user. Once the command has been recognized, the appropriate subroutine is exe- cuted. This subroutine will, in most cases, call other subroutines in order to accomplish its intended task. In some cases subroutine calls are nested to a depth of six. Figure 3 gives a pictorial summary of the calling hierarchy for subroutines which are in GRASP. This figure will be useful in implementing or modifying the GRASP system. PROCESSING USER INPUT All user input to GRASP is passed to the system in unpacked character form. At the highest level are single words used to execute a GRASP “com- mand.” In this case, the characters are packed, and the result is compared to the list of available com- mands. After a command has been issued, supple- mentary user input is usually required. This sup- plementary input must then be “parsed” (that is, converted) into a form more meaningful to the GRASP system. This parsed form is entirely nu- meric in nature. The numbers themselves may rep- resent values, integer encodings, or pointers. Sup— plementary input falls into five independent areas: conditional expressions, logical expressions, number lists, name lists, and arithmetic expressions. CONDITIONAL EXPRESSIONS A conditional expression is an attribute name, fol- lowed by a relation, followed by a value. The attri- bute name is identified using the binary-search technique. The relation is identified by a sequential table lookup. The value is converted to correspond in type with the attribute name referenced. This may result in a pointer to a character entry in the Dictionary file, a binary encoding of an acronym value in a record of the Multiple-choice file, or sim- ply a numeric value. Each conditional expression entered is associated with a letter (A—Z). LOGICAL EXPRESSIONS Logical expressions are composed of letters refer- ring to conditional expressions, the grouping sym- bols used to control the order of evaluation, and the logical operators .AND. (*), .O‘R. (+), .NOT. (—). For ease of evaluation logical expressions are con- verted to reverse—Polish form. This is a parenthesis- free form which permits rapid evaluation using a push-down stack technique. For a detailed descrip- tion of the conversion to and evaluation of reverse- DETAILED SYSTEM DESCRIPTION SUBROUTINE NAMES ACCESS BDEF BFIND BINIT BINTYP BLIST COLPNT COMP DECOMP DEFINE DEFLST DRIVER EVAL FDRIVE FIND FINDGP GETPUT ICONV SUBROUTINE NAMES IFILE INIT KEYBRD LENGTH OBEY OPREP PACK PAUSE PREVAL RELEXP RLIST ROWPNT SCAN UNCODE FIGURE 3.—Pictorial summary of GRASP subroutine interrelationships. The arrows point to subroutines that are called. 8 GEOLOGIC RETRIEVAL AND SYNOPSIS PROGRAM (GRASP) Polish form expressions, the reader is referred to Lee (1967, p. 162—180). NUMBER LISTS Number lists are composed of one or more in- teger numbers or number ranges (for example, 1, 2—7, 10). Each pair of elements in a number list must be separated by a comma. Individual elements are generated from the unpacked character form by constructing each number, one digit at a time, us- ing a sequential lookup on each character. Number ranges are generated by filling in the interior num- bers from the pair of extremes. NAME LISTS Name lists are composed of one or more names. When the name list contains a single element such as a file name, the packed form is obtained and as— sociated with the appropriate FORTRAN unit num- ber. If the name list is one or more attribute names, each pair of which is separated by a comma, each element is packed and looked up using the binary- search technique. ARITHMETIC EXPRESSIONS Arithmetic expressions may be entered in the place of single attribute names as supplementary input to the LIST command. These arithmetic ex- pressions may be composed of constants, the group- ing symbols (), attribute names, the arithmetic operator-s +, *, /, —, and the functions square (SQR), square root (SQRT), log base 10 (LOG) and power of 10 (EXP). The arithmetic expression is converted on input to reverse-Polish form. Eval- uation is done on output. If any of the attributes in the expression has no value for a given record, the expression is not evaluated. All conversion to re- verse-Polish form is done using transition-matrix parsing. Bauer and Samelson (1960) give a discus- sion of this technique. SEARCHES The two general types of GRASP searches are ex- ternal file and internal table. External-file searches are made on the Dictionary file, Multiple-choice file, and Numeric Master file. The Dictionary file is searched in two ways. The first way is as an indexed sequential file. When a con- dition relating an attribute name to a character- string value is entered, the record number of the first entry of the dictionary for that attribute is obtained from the “unnamed” common area. That record is read (directly), giving the entry value and a pointer to the record containing the next entry. The entry value is compared with the character- string value, and, if not equal, the record containing the next entry is read. This process continues until an entry is found that matches the character—string value, or until all entries of the dictionary have been read. The latter condition is detected by a next- record-pointer of zero. The second way of searching the Dictionary file is as a direct- (random-) access file. To display the value of a character-type at- tribute, its pointer is obtained from the current record of the selected Numeric Master file 1 and its value is obtained by a direct—access read on the Dictionary file. Multiple-choice acronym values are obtained when a condition is entered involving a multiple-choice— type attribute and when the value(s) of a multiple- choice-type attribute is to be displayed. Each record of the Multiple-choice file contains all the acronym values for a particular multiple-choice attribute. In all cases, the Multiple-choice file is read in direct- (random-) access mode using a pointer from the “unnamed” common area. After the correct record has been obtained, the attribute values are available in a tabular (array) form. The Numeric Master file is searched in a purely se- quential fashion. This search involves the application of a “question” to each record of the file where the “answer” can only be “yes” or “no.” If the answer is “yes,” the record is written on an output file. The question is posed by previously entering conditions and relating them by a logical expression. Internal table searches are made on attribute names and single characters. All lookups on attri- bute names are done using the “binary search” tech- nique on a sorted list. The list of names are read and sorted by execution of the FILE command. The interval that possibly contains the desired name is repeatedly halved until it is of length one. At that point, the position of the name is known, or the name is not in the list. Single characters are looked up sequentially when the list of possibilities is short, as in the case of digits in a number. A “hash code” technique is applied for longer lists such as alphabetic letters. This technique involves the initial storage of possi- bilities in a position dictated by a function applied to the value itself. This is done in a table whose size is greater than the number of possibilities. If 1 The selected Numeric Master file most probably will be some retrieved subset of the true Numeric Master file. DATA-BASE IMPLEMENTATION 9 the position is already occupied, an additional func- tion is applied to the value until an unoccupied position is found. Once the possibilities have been stored in this manner, the lookup of an arbitrary character is accomplished by applying the same pro- cedure. If an empty position is detected during look- up, the character is not in the list. Bell and Kaman (1970) give a more detailed description of the tech- nique. OUTPUT A data-retrieval system designed for interactive use should provide the user with information re- garding use of the system, the structure and con- tent of a particular data base, and the capability of displaying selective attribute values for records of some partition of a data base. These capabilities have been incorporated in GRASP and are individ- ually discussed in the following paragraphs. Information regarding use of the GRASP system is provided in two ways. First, all user response is preceded by system-generated “prompts” which indicate the type of response desired. Secondly, a command (HELP) has been implemented that gives the user a brief description of each command rec- ognized by the system. Information regarding the structure and content of a particular data base is obtain-able via the NAMES and FUNCTION commands. The NAMES command allows the user to determine attribute names (acronyms) and corresponding data types. A brief description is provided for each attribute name printed. After the selected attribute names have been printed, the user may examine the set of possible values assumed by character~and multiple-choice-type attributes. For numeric-type attributes, the user may obtain arithmetic means and ranges by selecting the MEAN function after issuing the FUNCTION command. A partition of a data base is created when a retrieval has been made using the CONDITIONS, LOGIC, and SEARCH commands. The displaying of selective attribute values for records of this partition is accomplished by using the LIST or DUMP com- mands. The DUMP command permits the user to print all values present for attributes in a selected set of categories. The values are printed one to a line with the corresponding attribute name. The LIST command permits the selection of specific attributes or arithmetic expressions containing attribute names for printing. The printing is select- ably formatted into columns or rows. For columnar output, the user may create a separate data set which could be used by other programs at a later time. DATA-BASE IMPLEMENTATION In the previous section on file description, it was noted that the various files were integral to and necessary for GRASP to function. There are ap- proximately as many methods of data collection as there are data bases, and it is not the intention of the writers to dictate data-base structures or meth- ods of data collection. However, the following sug- gestions will facilitate the construction of the files necessary for GRASP implementation. The structure of any GRASP data base must be such that it can be manifest in a tabular fashion. The table representing a data base is composed of columns that are attributes and rows that are items described by these attributes. For purposes of this report, the word “record” will be used in reference to row-s. Before any files can be constructed, a com- prehensive list of names of attributes (or column headings) must be compiled. Keep in mind that this arrangement of attributes will describe every rec- ord in the data base, and that although provision is made for all attributes, no record need contain data on every attribute. For each attribute that assumes a character-string value, a dictionary is compiled whose entries are the values assumed by that attribute. These dictionaries are used to create the Dictionary file. Once the Dictionary file is con- structed, the record number (that is, pointer) of the first entry (that is, value) for each character- type attribute is known. By using this information and the previously compiled list of names of attri- butes, the Mask file can be created. Next, all attri- butes should be grouped into categories of related information. This grouping provides the informa- tion necessary for the construction of the Defini- tions file. For multiple-choice-type attributes one simply needs to assign and to delineate value ac- ronyms for each attribute in a record. Each group of value acronyms forms a set, and the collection of sets forms the Multiple—choice file. Finally, the Numeric Master file is constructed, one record at a time. The individual record is con- structed by assigning values for each attribute in the order of its occurrence in the Mask file. For integer or real attributes, the value is inserted direct- ly. For character-type attributes, the entry number of the value in the appropriate dictionary is inserted. 10 GEOLOGIC RETRIEVAL AND SYNOPSIS PROGRAM (GRASP) For multiple-choice-type attributes, the binary en- coded word that describes its value is inserted. The record is then compressed as described in the sec- tion “Data File Structures.” USE OF GRASP From the viewpoint of a user, GRASP is a mech- anism for obtaining information from a data bank in a very simplistic and rigid manner. The “lang- uage” which is used to “direct” GRASP is composed of 11 “commands.” These commands can logically be divided into four groups. GROUP 1 (FILES, NAMES) is used to: A. Select or change the data base of interest. B. Obtain information regarding the nomenclature and content of the selected data base. GROUP 2 (LIST, DUMP, FUNCTION) is used to: A. Examine a selected set of records that is called a file. B. Perform selected computations of numeric attri- butes in a file. GROUP 3 (CONDITIONS, LOGIC, SEARCH) is used to perform a retrieval (SEARCH) on the data bank based on given criteria (CONDI- TIONS) which are combined via a logical ex- pression (LOGIC), a shorthand method of indi- cating which records of the data bank are to be retrieved. GROUP 4 (HELP, REVIEW, QUIT) is used to: A. Obtain brief information about the commands that GRASP will accept. B. Obtain information regarding the history and status of the current session with GRASP. C. Terminate the current session with GRASP. All commands except HELP and REVIEW will ask for some type of response. Each response enter- ed must end by striking the “cr” (RETURN) key. If a typing error or incorrect response is given, the system asks for another response. If at any point the system seems to be idle it is a good practice to strike the cr key. Certain commands (SEARCH, LIST, DUMP, FUNCTION) require an input file name. Entering a blank name in response to prompts generated by these commands (that is, cr only) results in the selection of the current Numeric Master file (as specified in the most recent FILE command). The LIST and DUMP commands also ask for the number of lines per page. This causes the system to pause after each printing of this number of lines, awaiting a response from the user. The user may then make a hard copy and (or) clear the screen if using a CRT terminal. Also, the user may terminate the printing altogether. At each pause, the user should enter a nonblank character followed by a or if it is desired to abort the rest of the printout; otherwise, only a cr will continue printing. The method of calling the GRASP system into execution will vary, depending on what compu- ter is used. At the beginning of execution, the GRASP system will print out the names and descrip- tions of the data bases available. The data-base name corresponds to the name of the Numeric Master file. Assume, for purposes of explanation, that a data consisting of oil and gas pools in the State of Colorado is available and named COLFIL. Following is a discussion of each command: FILES—This command is used to select a data base and may be issued at any time during a session. The individual-attribute names for any one data base will not be recognized by GRASP until this command has been issued. The user must enter a data—base name when the system asks for it. NAMES—This command is used to list the ac- ronyms which will be used to identify individual attributes within a record (that is, pool) and their meaning. First, 17 categories are printed. Then the system asks the user to enter a list of numbers cor- responding to the categories of interest. The list should be composed of individual numbers or num- ber ranges (such as 2—5), each of which must be separated by a comma. The list must be terminated by the cr key: for example, 1, 2—5, 9 cr and 1—4, 10, 11 cr. The system then lists each acronym and its meaning for all the categories of inter- est. After each category is complete, the sys- tem pauses. At this point the user must enter or to continue, or enter any letter (or digit) followed by or to stop. After all categories have been completed, the system asks if the user would like to see the possible values of any character-type or multiple-choice—type items. The user must then enter Y or N followed by or to indicate his decision. If the user enters N, the system will ask the user to enter his next command. If the user enters Y, the system asks for the names (acronyms) of the attri- butes of interest. The names are prompted and are given one per line followed by a or. After each name is given, the system skips to the next line and prints a numeral. To end the list (a maximum of 10 names may appear), enter cr (with no name). The system then starts printing the attribute names and possi- ble values, pausing after each name is complete. A pause also occurs after 30 lines of print. At each pause, enter or to continue or any letter (or digit) USE OF GRASP 11 ' followed by cr to stop. After this process is com- ‘ pleted, the system then asks if the user would like to see any more possible values. Again, enter Y for . yes or N for no. LIST.—This command is used to output selected attribute values (or expressions) from a selected file. Output may be to an interactive terminal or to . a specified data set which could be processed at some later time by other programs. The system first asks _‘ the user for the input—file name and the number of lines per page. The user is next asked to enter C for column printing or R for row printing. If column output is selected, the user is asked if he wants out- ? put to be to a disk data set in character form. If so, Tthe system will ask for a data-set name. Column output prints the selected acronyms as headings and their respective values below. Each column is com- posed of 8 character positions in a field of 10. One line of column output corresponds to one record. Row output consists of lines, each of which contain (an acronym and its corresponding value. If the value for a selected attribute is missing, the attri- ‘bute name is not printed. Records are separated by a line of asterisks. Before output proceeds, the sys- tem asks for the names of attributes or expressions ,which are desired. This is done by prompting with :index numbers. Expressions may optionally be preceded by some name. Five intrinsic functions are available: square root (SQRT), square (SQR), log base 10 (LOG), power of ten (TEN), and absolute value (ABS). Expressions may involve these intrinsic functions, attribute names, numeric constants, the arithmetic operators (+, —, *, /), and the grouping symbols ( ). The following is an example of a list to be output: 1. POOL 2. FIELD 3. DEPTH 4. LOG (DEPTH) 5. WELLAV= CRUAN69/ (NUMPOOL- TOTPROD) 6. iIn the above example, GRASP has prompted with {the index numbers 1—6. Note that the list is termi- nated by a blank entry. , DUMP.—This command is very similar to the LIST command having row printing specified. In- ,stead of asking for a list of names, the system asks for a list of category numbers. It then prints (in row fashion) the attribute name and value for each attribute present in the selected categories of the specified file. FUNCTION .——This command performs functions on a file. Currently, the only functions available are the arithmetic mean (MEAN) and a linear least- square fit (FIT) of two attributes. The system asks for the name of the input file. Next, the user is asked for the name of the function and names of the argu- ments. The argument names are the acronyms for attributes within a record; as many as five may be given. For instance, if MEAN DEPTH, TOTPROD, CRUCM70 or were entered, the range, mean, root mean square, sum, and sum of squares for DEPTH, TOTPROD, and CRUCM70 would be computed and printed. If FIT DEPTH, TOTPROD or were en- tered, the system would respond with the slope, intercept, and correlation coefficient. Values for all attributes in a record must be present for that rec- ord to be included in a computation. CONDITIONS.—This command is used to enter a set of retrieval criteria. Each criterion must be given in the form acronym relation value, where acronym is an attribute name (such as COUNTY, CRUCM69, POOL), where relation is EQ, NE, GT, LT, LE, GE, or BE, and where value is a number or a series of letters (such as ADAMS, 19342, MIS- SISSIPPIAN). The above relations have the follow- ing meanings: EQ—equal to. NE—not equal to. GT—greater than. LT—less than. LE—less than or equal to. 'GE—greater than or equal to. BE—«between (numerically, inclusive). The system precedes each condition with a letter prompt (up to 26 may be entered), which will be used in the logic expression that combines the con- ditions. Entering cr by itself terminates the list of conditions. Following is an example of a set of con- ditions: A. COUNTY EQ BACA B. DEPTH BE 5000,6000 C. TOTPROD GE 10 D. LITHOL NE DOLOMITE E. COUNTY EQ ADAMS F. POOL NE MISSISSIPIAN G. In the above example, the system provided the let- ters A through G as prompts. LOGIC.—This command is used to enter a logical or connective expression which combines the pre- 12 GEOLOGIC RETRIEVAL AND SYNOPSIS PROGRAM (GRASP) Viously entered conditions to form the retrieval cri- terion. The logical expression may be composed of the logical connectives (operators), the letters cor- responding to the criteria entered via the CONDI— TIONS command, and the grouping symbols ( ). The logical connectives are AND, inclusive OR, and NOT (written .AND., .OR., .NOT.). Note that they are each bracketed by periods. Provision has also been made to use * (for AND), + (for OR), — (for NOT). Assume that the example conditions given in the preceding CONDITIONS command section had been entered. If the user wanted to retrieve the pools in Baca County that had a depth of 5,000—6,000 feet, the logic expression would be A .AND. B cr. If all the pools in Adams and Baca Counties except those of Mississippian age having dolomite lithology were desired, the logic expression would be (A .OR. E) .AND. (D .AND. F) or. Note that the last pair of parentheses is not really needed. The ANDs will be applied before the CBS. NOTs are applied before ANDs and ORs. Hence, the first set (A. OR. E) is necessary so that the E is connected to A instead of to D. If one wanted to retrieve all pools with at least 10 producing wells having a depth greater than 6,000 feet or less than 5,000 feet, the logic expression would be .NOT. B .AND. C cr. If one wanted to retrieve all pools having less than 10 producing wells in the same range as above, one could use .NOT. (B .OR. C) or for a logic expression. This expression, in words, says “if the pool has a depth of 5,000—6,000 feet, or if it has 10 or more pro- ducing wells, I don’t want it.” SEARCH .—After the system has been given the conditions and connecting logic that compose the question to be asked of some file, an actual search of the data bank can be made. This is done with the SEARCH command. The system will ask for the name of the file to be searched (input file) and the name to call the file of records found (output file). After the search has been made, the system types ‘ the number of records searched and the number of records found. The capability of entering both input and output file names allows the user to perform “nested” searches. This means searches of files that were the result of previous searches. Frequently this is the most economical way of performing multiple or complex retrievals. For instance, sup— pose one wanted information on several sets of pools, all of which were in one county. One would first create an output file that contained all the pools in that county and then use that file as the input file for subsequent searches. HELP.—This command is used to print a list of the possible commands and a brief description of their functions. REVIEW.——-This command provides a review of the conditions and logic which are currently in effect. The names of input and output files for the last 10 retrievals are also printed. This command is used to refresh one’s memory on what was done recently during the current session. QUIT.—This command is used to exit from the GRASP system. A list of the files created during this session is printed, and the user is permitted to save them for future use. Abnormal session inter- rupts and terminations will cause GRASP to cease functioning. However, all files created during the active session are either saved or not saved, accord- ing to the abnormal termination rules of the partic- ular computing system. On abnormal termination, GRASP neither saves nor deletes files. REFERENCES CITED Bauer, F. L., and Samelson, K., 1960, Sequential formula translation: Assoc. Computing Machinery Commun., v. 2, no. 2, p. 76—83. Bell, J. R., and Kaman, C. H., 1970, The linear quotient hash code: Assoc. Computing Machinery Commun., v. 13, no. 11, p. 675—677. Lee, J. A. N., 1967, The anatomy of a. compiler: New York, Reinhold Publishing Corp., 275 p. GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) 13 GRASP SOFTWARE SPECIFICATIONS MODULE NAME: DRIVER Purpose: DRIVER is used primarily as a switching/calling mechanism. User commands are accepted and decoded. Con- trol is then passed to the routine designed to process the given command. This process continues until the user “quits.” Subroutines called: START, KEYBRD, CONDTN, LOGEXP, RETRVE, FTNC, FILES, CONDS, HELP, DUMPIT, NAME, LIST, QUIT, PACK. Common data referenced: None Called by: None Error checking and reporting: The command entered by the user is checked against the list of legal commands. If a command is not recognized, it is echoed back to the user GRASP SOURCE terminal with a message suggesting use of the HELP com- mand. Program logic: 1. Initialization is performed by zeroing counters and call- ing START. 2. An unpacked character string is accepted from the user via subroutine KEYBRD. 3. A four—character command is formed by packing the above string into COMMAND. 4. COMMAND is then compared with the list (WORDS) of acceptable command words (NAMES). When a match is found, control is transferred to the appropriate subroutine via a computed GO TO. 5. Steps 2 through 4 are repeated until an end-of—file (EOF) condition is sensed on the terminal or until the QUIT com- mand is executed. PROGRAM INTEGER WORDS(11) yCUMANDyNAMEPlem yRCODE1261,1VAL(26|yPOLISHl3010000001 l yIMAGE(51yIFILESlZOhOFILESlZO) 0000002 DATA NORDS/‘COND' ,‘LOGI' ,‘SEAR' 1'L IST' .‘F ILE' 9 'QUIT'. 'NAME'. 'HELP'0000003 ly'REVl','DUMP'.'FUNC'/,IBLNK/' '/ 0000004 NFILES=O 0000005 LPS=O 0000006 CALL START 0000007 110 TYPE 270 0000008 COMAND=IBLNK 0000009 CALL KEYBRD(&2609IMAGE.41 0000310 CALL PACK(IMAGE.COMAND,4,41 0000011 00 120 1=lyll 0000012 IF (COMAND.EQ.HORDS(I)) GO TO 130 0000013 120 CONTINUE 0000014 TYPE 290, COMAND 0000015 00 TO 110 0000016 130 00 T0 l1401150!160,240,1901260v23012101180722011701q 1 0000017 140 CALL CDNDTNl8110¢NAMEPToRCODE'IVAL,NREXP1 0000018 00 T0 110 0000019 150 CALL LOGEXP(81109POLISH9LPSpNREXP3 0000020 00 TO 110 0000021 160 CALL RETRVE(8150,&110,IFILES,UFILES,NF1LES,PULlSHyLPSyNAMEPT, 0000022 1 RCODE.IVAL,NREXP) 0000023 00 TO 110 0000024 170 CALL FTNClCllO) 0000025 00 T0 110 0000026 190 CALL FlLESlfillO) 0000027 00 T0 110 0000028 180 CALL CUNDS(NREXP,LPS) 0000029 1F (NFILES.GT.01 GO TO 200 0000030 TYPE 300 0000031 00 T0 110 ' 0000032 200 TYPE 310v (lFILESlIloOFILES(llyl=1oNFILESl 0000033 00 T0 110 0000034 210 CALL HELP(WORDS) 0000035 GO TO 110 0000036 220 CALL DUMPIT 0000037 00 TO 110 0000038 14 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) 230 CALL NAME(8110) GO TO 110 24C CALL LIST(8110) GO TO 110 260 CALL QUIT(OFILESpNFlLES) STOP 270 FORMAT (/l' ENTER COMMAND: "$) 290 FORMAT (leASy' ILLEGAL COMMAND. ENTER HELP IF YOU WISH TO SEE'I 1' THE LEGAL CUMMANDS.'/) 300 FORMAT (‘ NO FILES HAVE BEEN USED AT THIS TIME.') 310 FORMAT (' INPUT: OUTPUT:'/(2X9A5y2X,A5/)) END 0000039 0000040 0000041 0000042 0000043 0000044 0000045 0000046 0000047 0000048 0000049 0000050 GRASP SOFTWARE SPECIFICATIONS 15 SUBROUTIN E NAME: ACCESS Purpose: ACCESS looks up character-string values in dic— tionary files. In order to minimize disk accesses, five previous values are saved for as many as 100 distinct character-type items. Calling sequence: CALL ACCESS(NUMD,IVAL,TANK, NWORDSJSWTCH) Arguments : N UMD—Item number of the character-type variable whose values are to be accessed. IVAL—Direct-acoess key under which value is stored. TAN K—Contains the character value accessed. NWORDS—The number of words in TANK. ISWTCH—Switch to control which of the following four functions are desired: 1. Initialization for dictionary lookup. 2. Lookup a random item. 3. Return the direct-access key of the first item in this 4. Return the indicated (by IVAL) entry and the KEY for the next entry (that is, reset IVAL) . Subroutines called: None Common data referenced: IDPT in /DACOMM/ Called by: BDEF, COLPNT, DUMPIT, PNTER, ROWPNT Error checking and‘ reporting: N one Program Logic: The logic is divided into four sections relat- ing to values of. ISWTCI-L 1. If ISWTCHzl, initialize saved pointer arrays (USED, LASTDX) and set character variable counter NCVAR to zero. 2. If ISWTCH=2, see if the value has been stored in BUF- FER. If so, return it; otherwise access it on FORTRAN unit 21 and save its value (TANK), index (IVAL) and the item number (N UMD). 3. If ISWTCH=3, return the direct—access key for the first entry of the dictionary pointed to by NUMD. 4. If ISWTCH:4, access the entry pointed to by IVAL and reset IVAL to the key for the next entry in this dictionary. dictionary. G R A S P S O U R C E P R O G R A M SUBROUTINE ACCESS‘NUMDyIVALyTANKoNWURDSIISWTCH) 0000051 COMMON IDACOMM/ NVolDPT 0000052 INTEGER TANK(I)pBUFFER(59100v5)yUSED(100)leDEXllOO), 0000053 I LASTDXI57100)yIDPT(500l 0000054 DATA IBLNKvNDICT/' '121/ 0000055 00 T0 (5y15,lOOql50),ISHTCH 0000056 5 NCVAR=O 0000057 DO 10 J=19100 0000058 USED(J)=O 0000059 00 10 I=1y5 0000060 10 LASTDXlIoJ)=-999999 0000061 GO TO 320 0000062 15 IF(NCVAR.E0.0) GO TO 30 0000063 DO 20 KK=lyNCVAR 0000064 IF(INDEX(KK).EQ.NUMD) GO TO 40 0000065 20 CONTINUE 0000066 30 NCVAR=NCVAR+1 0000067 IF(NCVAR.GT.100) NCVAR=100 0000068 KK=NCVAR 0000069 INDEX(KK)=NUMD 0000070 40 IF!USED(KK’.EQ.0) GO TO 240 0000071 DO 50 K=lv5 0000072 IF(IVAL.EQ.LASTDX(K,KK)) GO TO 60 0000073 50 CONTINUE 0000074 GO TO 240 0000075 60 NHORDS=5 0000076 00 70 I=19NWORDS 0000077 70 TANK(I)=BUFFER(19KK’K) 0000078 GO TO 320 0000079 100 IVAL=IDPT(NUMD) 0000080 GO TO 320 0000081 150 READ(NDICT'IVAL) NP,NHORDS,(TANK(IlyI=loNHORDS) 0000082 . IVAL=NP 0000083 GO TO 320 0000084 16 2:. ,s U 260 270 320 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) ISTART=IDPT(NUMD) READ(NDICT'ISTART+IVAL'1) NP,NNORDS:(TANK(I)’I=loNWORDS) USED(KK)=MDD(USED(KK),5}+1 NUSED=USED(KK) DO 260 I=195 BUFFER(I.KK.NUSED)=IBLNK NHORD=MINO(NHOROS.5) DO 270 I=19NH0R0 BUFFER(viKyNUSED)=TANK(I) LASTOX(NUSEDyKK)=IVAL RETURN END 0000085 0000086 0000087 0000088 0000089 0000090 0000091 0000092 0000093 0000094 0000095 0000096 GRASP SOFTWARE SPECIFICATIONS 17 SUBROUTIN E NAME: BDEF Purpose: BDEF provides access to the character and multiple- choice dictionaries. Calling sequence: CALL BDEF (&n) Argument: n—Statement number (in caller) to which a branch is made if an EOF is sensed by KEYBRD. Subroutines called: KEYBRD, VLIST, ACCESS, PAUSE, BINTYP, IFILE Common data referenced: FNAMES, WHICH in /FILNAM/ ITYPE in blank common Called by: NAME Error checking and reporting: The user response to a “yes/ no” question is checked. If illegal, a message is typed, and the user is prompted for another response. If an item is selected that is not a multiple-choice or character-type item, a message is typed, and the user is requested to reenter a list of item names. GRASP SOURCE Program logic: The user is prompted to determine if a list of multiple-choice or character-type values is desired. If the response (obtained from KEYBRD) is “N”, a branch is made to the end of the routine. If the response is “Y,” a list of item names is obtained by a call to VLIST. If an EOF is sensed, the nonstandard return (from VLIST) exits via the nonstandard return of BDEF. The data type is deter- mined for each name (TAGS) returned by VLIST. If the type is not multiple choice or character, that message is typed, and the next element of TAGS is considered. If the type is multiple choice, a call to BINTYP is used to ob- tain the permissible values) for printing. If the type is char- acter, calls to ACCESS are made to obtain the possible values. When the second argument of ACCESS is returned as zero, all possible values have been referenced. A pro- grammed pause is generated after each 30 lines of print and after each item in TAGS is processed. Just prior to re- turn (standard or nonstandard), unit 22 (the binary or multiple-choice file) is rewound, and the unit number is reassociated with the current file name via a call to IFILE. PROGRAM SUBRDUTINE BDEF(*) 0000097 COMMON NAMES:ITYPE:IPTS,IPAD 0000098 COMMON IFILNAM/ FNAMES.HHICH9PAD 0000099 DOUBLE PRECISION LABEL(25)9NAMES(500)yTAGS(20) 0000100 INTEGER IPTS(500)'FNAMESIZIIywHICHyPADI4) 0000101 INTEGER BINLIZOI,YES.N0,REPLY,BITEM115v251yITYPEI50019TANKl25) 0000102 EQUIVALENCE (BITEMllplIoTANKllll 0000103 DATA YESyNOyNBIN/‘Y'u'N'oZZ/ 0000104 10 TYPE 120 0000105 20 TYPE 130 0000106 CALL KEYBRDIEIlOoREPLYyl) 0000107 IF (REPLY.EQ.ND) GO TO 100 0000108 IF (REPLY.EQ.YES) GO TO 30 0000109 TYPE 150 0000110 GO TO 20 0000111 30 CALL VL15T151101TAGSQBINLQNUM1 0000112 DO 90 N=11NUM 0000113 INDEX=BINLINI 0000114 IF (ITYPElINDEXl-BI 60,40170 0000115 40 TYPE 1909 TAGSlNI 0000116 CALL ACCESS(INDEX.K1TANK,NDMo3l 0000117 00 50 J=1,10000 0000118 CALL ACCESSIINDEXvaTANKyMy4) 0000119 TYPE 160. (TANK(II,I=1,M) 0000120 IFIK.EQ.O) GO TO 90 0000121 IF (MOD(J1301.NE.0) GO TO 50 0000122 CALL PAUSEIEIOO) 0000123 50 CONTINUE 0000124 60 TYPE 1701 TAGSINI 0000125 GO TO 90 0000126 70 IFIITYPE(INDEX).NE.4) GO TO 60 0000127 CALL BINTYPIINDEX,LABEL,BITEM,K9M1 0000128 TYPE 1901 TAGS(NI 0000129 00 80 J=1yM 0000130 80 TYPE 1801 LABELIJIMBITEMIIthI=1vK1 0000131 18 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) 90 CALL PAUSE(8100) GO TO 10 REHIND NBIN CALL IFILE1NBIN,FNAMES(16+HHICHD) 100 110 120 FORMAT (' WOULD YOU LIKE TO SEE THE POSSIBLE VALUES ' 1 CHOICE'l' OR CHARACTER TYPE TTEMS?') 130 150 160 170 180 190 RETURN REHIND NBIN CALL IFILE(NBIN,FNAMES(16+WHICH)1 RETURN FORMAT FORMAT FORMAT FORMAT FORMAT FORMAT END 1 1' (ENTER Y FOR YES , N FOR NO): ',$) (' YOUR REPLY WAS NOT UNDERSTOOD.') (10X’12A5) (leABy‘IS NOT A MULTIPLE CHOICE OR CHARACTER TYPE ITEM.') (11X,A7.'-',15A41 ‘1XyA8"=') 0000132 0000133 0000134 0000135 0000136 0000137 0000138 0000139 v'OF MULTIPLE0000140 0000141 0000142 0000143 0000144 0000145 0000146 0000147 0000148 GRASP SOFTWARE SUBROUTINE NAME: BFIND Purpose: BFIND is used to look up a double—word-item name in a list of double-word-item names having a given sorted order. Calling sequence: IN DEX,K) Arguments: n—Statement number (in caller) to which a branch is made if the name KEY is not in the list of names KEYS. KEY—The double-word item to be looked up. IPOST—The position in KEYS of the item KEY. CALL BFIND(&n,KEY,IPOST,KEYS, SPECIFICATIONS 19 KEYS—The list of double-word names that will be used to look up KEY. INDEX——Gives the indices of the sorted form of KEYS. K—The number of elements in KEYS. Subroutines called: None Common data, referenced: None Called by: DECOMP, PARSE, RELEXP, VLIST Error checking and reporting : None Program logic: The standard binary-search technique is used, which repeatedly halves the interval of search on a sorted list. If the interval of search goes negative, the element is not in the list and the nonstandard return is taken. G R A S P S 0 U R C E P R D G R A M SUBROUTINE BFIND(#.KEY,IPOSTyKEYSqINDEXyK) 0000149 DOUBLE PRECISION KEY,KEYS(1) 0000150 DIMENSION INDEX(1) 0000151 L1=1 0000152 L2=K 0000153 70 IF (L2.LT.L1) RETURN 1 0000154 J=(L1+L2)/Z 0000155 I=INDEX(J) 0000156 [F (DABS(KEY).GT.DABS(KEYS(Ill) GO TO 90 0000157 IF (DABS(KEY).LT.DABS(KEYS(I)l1 GO TO 80 0000158 GD T0 100 0000159 80 L2=J°1 0000160 GD TD 70 0000161 90 L1=J+1 0000162 GO TO 70 0000163 100 IPOST=I 0000164 RETURN 0000165 END 0000166 20 GEOLOGIC RETRIEVAL AND SYN‘POSIS PROGRAM (GRASP) SUBROUTINE NAME: BINIT of KEYS. M—The number of words in the KEYS and INDEX arrays. Subroutines called: None Common data referenced: None Purpose: BINIT is used to sort a list of double-precision words (NAMES, in this instance) into ascending order and re- turn an array of indices giving the sorted order of the elements in terms of the unsorted order. Called by: FILES Calling sequence: CALL BINIT (KEYS,INDEX,M) Error checking and reporting: None Arguments: Program logic: An in-place sort is performed using the stand- KEYS—The list of double-precision words to be sorted. and “Shell” technique. The original order of KEYS is over- INDEX—An array of indicies giving the unsorted order written, and the sorted order is returned in INDEX. GRASP SOURCE PROGRAM SUBROUTINE BINITIKEYS.INDEX,M) 0000167 DOUBLE PRECISION KEYSII),KTEMP 0000168 INTEGER INDEXII) 0000169 00 10 J=1,M 0000170 10 INDEX(J1=J 0000171 M0=M 0000172 20 1F (MO.LE.11 GO TO 60 0000173 J=4 0000174 IF (MD.GT.15) J=8 0000175 MO=2*(M0/J)+1 0000176 KD=M-MD 0000177 JO=1 0000178 30 I=JO 0000179 40 J=I+MO 0000180 IF (DABSIKEYSI1)).LE.DABS(KEYS(J))1 GO TO 50 0000181 KTEMP=KEYS(I) 0000182 KEYSIID=KEYSIJ1 0000183 KEYSIJ1=KTEMP 0000184 ITEMP=INDEX111 0000185 INDEXII)=INDEX(J) 0000186 INDEXIJI=ITEMP 0000187 [=1-M0 0000188 IF (1-1) 50940v40 0000189 50 JO=JO+1 0000190 1F (JO-K0) 30,30,20 0000191 60 RETURN 0000192 END 0000193 GRASP SOFTWARE SPECIFICATIONS 2‘1 SUBROUTINE NAME: BINTYP Purpose: BINTYP reads the values (LABEL, BITEM) which can be assumed by the multiple-choice-type item whose number is NUMD. Calling sequence: CALL BINTYP,(NUMD,LABEL,BITEM, LNG,NUM) Arguments: NUMD—Integer specifying a multiple-choice-type item. LABEL—Set of values (double word) that may be assumed by this item. BITEM—Set of descriptions corresponding to elements of LABEL. GRASP SOURCE SUBROUTINE BINTYPlNUMDyLABEL,BlTEMyLNGvNUM) COMMON IFILNAM/ FNAMESvWHICHgPAD DOUBLE PRECISION LABELlU INTEGER FNAMESIZI)yWHICHoPADUthBITEMl15' 25) DATA NBIN/ZZ/ 300 READ(NBIN.END=3I0) NAME'NUMyLNGy A (LABELlJ),(BITEM(IyJ) 91:19 LNGhJ=1.NUM) IFiNUMD-NAME) 310 REHIND NBIN CALL IFILElNBIN,FNAMES(16+WHICH¥D GO TO 300 320 RETURN END 31003209300 LNG—Maximum length of a description. N UM—The number of descriptions. Subroutines called: IFILE Common data, referenced: FNAMES, WHICH in /FILNAM/ Called by: BDEF, COLPNT, DUMPIT, PNTER, ROWPNT Error checking and reporting: None Program logic: Records of the multiple—choice (binary) file are read sequentially until the correct record is obtained. If the file was positioned past the desired record at call time, an EOF is sensed, the file is rewound, and the cur- rent multiple-choice file name is reassociated with unit 22. PROGRAM 0000194 0000195 0000196 0000197 0000198 0000199 0000200 0000201 0000202 0000203 0000204 0000205 0000206 22 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) SUBROUTINE NAME: BLIST Subroutines called: None Purpose: BLIST returns a list of numbers giving the bit posi- Common data referenced: None tions of the “ones” in a binary word. Called by: COLPNT, DUMPIT, ROWPNT Calling sequence: CALL BLIST(LIST,NUML,ICODE) Error checking and reporting: None Arguments: ' . _ LIST—Array of integers giving the bit positions in ICODE 1370ng logic: ICODE 15 moved to IDFJM" IDUM. 1s suc- which are “ones,” counting right boleft' cess1vely d1v1ded by 2, and the least Significant bit 1s ac- NUML—The number of items in LIST_ cessed by the MOD function. If the least significant bit is ICODE—The binary word to be examined by BLIST. “one,” the position counter is added to LIST. GRASP SOURCE PROGRAM SUBRUUTINE BLIST(LIST1NUML¢ICODE) 0000207 lNTEGER LIST(1) 0000208 NUML=0 0000209 IDUM=1CODE 0000210 00 10 [=1,25 0000211 IF (IDUM.EQ.0) GO TO 20 0000212 IF (M00llDUM,2).EQ-01 GO TO 10 0000213 NUML=NUML+1 0000214 LISTlNUML1=I 0000215 10 IDUM=IDUMI2 0000216 20 RETURN 0000217 END 0000218 GRASP SOFTWARE SPECIFICATIONS 23 SUBROUTIN E NAME: COLPNT Purpose: COLPNT outputs the values of as many as 20 se- lected items or expressions. Output is columnar and is di- rected to the terminal or to a disk data set. Calling sequence: CALL COLPNT(&n,NPAGE) Arguments: n—Statement (in calling routine) to which a branch is made if the nonstandard return is taken from VLIST (KEYBRD senses EOF). NPAGE—Number of lines between pauses per page of terminal output. Subroutines called: KEYBRD, OFILE, VLIST, ACCESS, GETPUT, PAUSE, EVAL, UNCODE, BINTYP, BLIST, PACK Common data referenced: POLISH, ICODE, LPS in/EXPRNS/ ITYPE in blank common Called by: LIST Error checking and reporting: None Program logic: 1. The user is asked if he would like the output to go to disk. His reply is returned by KEYBRD. If affirmative a logical flag is set, and he is prompted for a data-set name. This name is then associated with unit 24 via a call to OFILE. 2. A call to VLIST returns the item names (or expression pointers) that. are selected. 3. A call to ACCESS is made to initialize the lookup of character dictionary values. 4. Each record of the selected file is then obtained via GET- PUT, and a line (or record) of output is constructed. For numeric data, a format is constructed to maximize the num- ber of significant digits displayed, and the constructed line is printed. G R A S P S O U R C E P R O G R A M SUBROUTINE COLPNT(*.NPAGE) 0000219 COMMON NAMES,ITYPE.IPTS,IDlM 0000220 COMMON lEXPRNS/ POLISHleODEyLPS 0000221 DIMENSION ITYPE(500!,BITEM(15o251leEMS(20)leECl500inPTSl500)r 0000222 1REC15001, NAMES(500), TANKlZS)’ LABEL(251' LlSTlZS), POLISHi 0000223 215,819 ICODE(1518), LPS(8)1 IQUALib) 0000224 DOUBLE PRECISION DBLNK,AREA,LINE(20)1NAMESyLABEL9VNAME5120)y 0000225 1 FMT(3)1FMTS(8) 0000226 INTEGER BLANK,TANK,YES 0000227 LOGICAL BLNKvTTY 0000228 EQUIVALENCE (RECQIREC19‘IVALQVAL)y‘TANKYLIST,{(BLANK'IQUAL’ 0000229 DATA FMTyFMTS/'( '1' 'o'pAl) ’y'F8.6 '9'F8.5 ', 0000230 1 'F8.4 'y'F8.3 'y'F8.2 'y'F8.1 ','F8.0 'y'lPE8.1'/, 0000231 2 DBLNK,YES,lQUAL/' '1'Y',‘ ','G','H','L','N','T'/ 0000232 TYPE 100 0000233 CALL KEYBRD(829091911 0000234 TTY=I.NE.YES 0000235 IFlTTY) GO TO 115 0000236 NPAGE=10000000 0000237 TYPE 105 0000238 CALL KEYBRD1829091TEMSy5) 0000239 I=BLANK 0000240 CALL PACK(ITEMSy195,5l 0000241 CALL OFlLE124yll 0000242 115 KOUNT=O 0000243 CALL VLIST(&290.VNAMES,ITEMstUM) 0000244 IF (NUM.EQ.0) GO TO 280 0000245 IF(TTY) TYPE 300v (VNAMES‘I’91=17NUM1 0000246 CALL ACCESS(II,lVAL'TANK'LKyI) 0000247 120 CALL GETPUT(527011REC91) 0000248 KOUNT=KOUNT+1 0000249 IF (KOUNT.LE.NPAGE) GO TO 130 0000250 KOUNT=O 0000251 CALL PAUSEifiZYO) 0000252 TYPE 300v (VNAMESlIlyI=19NUM) 0000253 24 130 140 150 160 170 180 190 200 205 210 250 260 65 270 280 290 100 105 215 300 310 311 320 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) DO 260 JJ=11NUM AREA=DBLNK [I=ITEMS(JJ) IF (II.GT.0) GO TO 140 II=-II VAL=EVAL(IREC,ICODE(19111;POLISH(1yII)oLPS(II)yBLNK) IF(BLNK) GO T0 260 GO TO 160 IVAL=IREC(II) IF (IVAL.EQ.BLANK) GO TO 260 KIND=ITYPE(IIT GO TO (1509160v21012501205)v KIND ENCODE(8.320:AREA) IVAL GO TO 260 IF (VAL.EQ.0.) GO TO 180 A=ALOGIOIABS(VAL)) IF (A.GE.5.) GO TO 190 IF (A.LE.-4.) GO TO 190 IF (A.LE.0.1 GO TO 170 LK=IFIX(A)+2 GO TO 200 LK=1 GO TO 200 LK=2 GO TO 200 LK=8 FMT(2)=FMTS(LK) IF(KIND.NE.5) 10:1 ENCDDE‘9,FMT.AREA1 VAL,IQUAL(IQ) GO TO 260 VAL=UNCODE(VAL9IQ) GO TO 160 CALL ACCESS(IIgIVALyTANKyLKyZI ENCODE(812159AREA) (TANK(I)11=1oLKT GO TO 260 CALL BINTYPIII'LABELyBITEMoKoM! CALL BLIST(LIST,NUMSvIVAL) AREA=LABELILIST(1)D LINE(JJ)=AREA IF(TTY) GO TO 65 HRITE(241310) (LINE(JJ):JJ=19NUM) GO TO 120 TYPE 311.(LINE(JJ)9JJ=1vNUM) GO TO 120 CONTINUE IFI.NOT.TTY) REHINO 24 RETURN RETURN 1 FORMAT(' WOULD YOU LIKE OUTPUT TO BE TO DISK? (Y OR NI: FORMAT(' ENTER NAME OF DISK DATA SET TO BE CREATED: ',$) FORMATIA5gA31 FORMAT (ll/1X28A10) FORMAT (ZOAIO) FORMAT(1X,8A10) FORMAT (18! END 0000254 0000255 0000256 0000257 0000258 0000259 0000260 0000261 0000262 0000263 0000264 0000265 0000266 0000267 0000268 0000269 0000270 0000271 0000272 0000273 0000274 0000275 0000276 0000277 0000278 0000279 0000280 0000281 0000282 0000283 0000284 0000285 0000286 0000287 0000288 0000289 0000290 0000291 0000292 0000293 0000294 0000295 0000296 0000297 0000298 0000299 0000300 0000301 0000302 0000303 0000304 0000305 0000306 0000307 0000308 0000309 GRASP SOFTWARE SPECIFICATIONS FUNCTION NAME: COMP Purpose: COMP evaluates a relational expression. The rela- tional operators may be (in FORTRAN notation) EQ, LT, GT, LE, GE, NE, or BE (denoting between). The evalua- tion .is performed on integer, real, or binary—(that is, bit) type data. Calling sequence: HIT : COMP(IVAR,IVAL,VAR,VAL, ICODE,ISWTCH), where HIT is of type LOGICAL. Arguments: IVAR, IVAL—Integer-type arguments to be compared if ISWTCH is 1. VAR, VAIr—Real-type arguments to be compared ISWTCH is 2. ICODE—Encoding of comparison to be made. Assumes values 1 through 7, respectively, indicating the relations EQ, LT, GT, LE, GE, NE, BE. ISWTCH—Indicates the type of arguments for the com- parison (integer, real, or binary). if GRASP SOURCE LOGICAL FUNCTION COMP‘lVARyIVAL1VAR'VAL91CDDE’ISWTCH) COMMON /BTHN/ IVALSvNBE DIMENSION IVALSIZo 10): LOGICAL COMPAR, NONBLK VALS( 2,10) EQUIVALENCE (lVALSl 191) sVALSllvll ) , DATA BLANKI‘ '/ NONBLK=IVAR.NE. lBLNK.AND. IVAL.NE.IBLNK lFllSwTCH—Z) 5.859165 5 GO TO (10.2093094095056097OJv 10 COMPAR=IVAR.EQ-IVAL GO TO 160 COMPAR=IVAR.LT.IVAL.AND.NONBLK GO TO 160 COMPAR=IVAR.GT.IVAL.AND.NONBLK GO TO 160 COMPAR=IVAR.LE.IVAL.AND.NONBLK GO TO 160 COMPAR=IVAR.GE.IVAL.AND.NONBLK GO TO 160 COMPAR=IVAP.NE.IVAL GO TO 160 20 30 40 50 60 70 GO TO 160 85 90 COMPAR=VAR.EQ.VAL GO TO 160 COMPAR=VAR.LT.VAL.AND.NONBLK GO TO 160 COMPAR=VAR.GT.VAL.AND.NONBLK GO TO 160 COMPAR=VAR.LE.VAL.AND.NCNBLK GO TO 160 COMPAR=VAR.GE.VAL.AND.NONBLK GO TO 160 COMPAR=VAR.NE.VAL GO TO 160 100 110 120 130 140 25 Subroutines called: None Common data referenced: IVALS in /BTWN/ Called by: RETRVE Error checking and reporting: If ISWTCH its 3 (that is, binary-type arguments), ICODE is tested for 1 or 6 (EQ or NE). If the test fails, a message is typed and COMP returns FALSE as a value. Program logic: 1. The logical variable NONBLK is set to indicate nonblank operands. 2. If ISWTCH is less than 3, the two operands are compared according to ICODE. If ICODE is 7 (indicating the between operator), the operands are obtained from IVALS in /BTWN/ by use of IVAL as a pointer to the appropriate set. 3. If ISWTCH is 3, a bit (binary) comparison is made by ICODE COMPAR=I VAR.GE. lVALSll yIVALl.AND.IVAP.LE. lVALSlZv IVAL ) .AND.NDNBLK NONBLK=VAR. NE.BLANK.AND.VAL.NE.BLANK GO T3 (9091009110912011309140p150)v examining the IVAL’th bit in IVAR. PROGRAM 0000310 0000311 0000312 0000313 0000314 0000315 0000310 0000317 0000318 0000319 0000320 0000321 0000322 0000323 0000324 0000325 0000320 0000327 0000328 0000329 0000330 0000331 0000332 0000333 0000334 0000335 0000330 0000337 0000338 0000339 0000340 0000341 0000342 0000343 0000344 0000345 0000340 (IyVI), (BLANKylaLNK) ICODE 26 150 160 165 170 190 180 200 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) VI=VAL C0MPAR=VAR.GE.VALS(191).AND.VAR.LE.VALS(2,1).AND.NONBLK COMP=CCMPAR GO TO 180 IF(NDNBLK) GO TO 170 IF(ICUDE.EQ.1) GO TO 10 IF(ICDDE.EQ.6) GO TO 60 TYPE 200 COMP=.FALSE. GO TO 180 IDIGIT=IVAR/2**(IVAL-1T IF (ICODE.NE.1T GO TO 190 COMP=MOD(IDIGIT,2).EQ.1 GO TO 180 IF (ICUDE.NE.6) TYPE 200 COMP=MOD(IDIGIT.2).NE.1 RETURN FORMAT (52H RELATION MUST BE END .EQ./.NE. FDR BINARY TYPE VARIABLE) 0000347 0000348 0000349 0000350 0000351 0000352 0000353 0000356 0000355 0000356 0000357 0000358 0000359 0000360 0000361 0000362 0000363 0000364 0000365 GRASP SOFTWARE SPECIFICATIONS 27 SUBROUTINE NAME: CONDS Called by: DRIVER Purpose: CONDS types out the last set of conditions and logic E7707 checking and reporting: None enteped. Program logic: Calling sequence: CALL CONDS(NCONDS’LPS) 1. If any conditions have been entered (that is, NCOND >0), Arguments: they are typed out. NCONDS—The number of conditions currently entered. . . . _ LPS—TheT length of the current logical expression. 2- If a logo expresswn has been entered (that IS, LPS >0) 7 1t Common data referenced: EXPR, LOGIC in / IN PUT/ is typed out. GRASP SOURCE PROGRAM SUBROUTINE C0NDSLNCDND.LPST 0000366 COMMDN IINPUT/ EXPR.LOCIC 0000367 DIMENSION LABELS(26) 0000368 DOUBLE PRECISION EXPRL4.26T.LOGIC(8) 0000369 DATA LABELS/'A',‘B','C','D'.'E','F','G'.‘H',‘I'.‘J','K',‘L'.'M','N0000370 1".0','P".Q','R','S','T','U','V',IH"'X"'Y.,'Z'/ 0000371 TYPE 40 0000372 [F (NCOND.EQ.O) 00 TO 20 0000373 00 10 J=1.NC0ND 0000374 10 TYPE 50. LABELSLJ).(EXPRLI.JT.I=1.4) 0000375 00 TO 30 0000376 20 TYPE 60 0000377 30 IF (LPS.NE.O) TYPE 70. LOGIC 0000378 RETURN 0000379 40 FORMAT (//) 0000380 50 FORMAT (4x.A1.'. '.4A10T 0000301 60 FORMAT (' NO CONDITIONS HAVE BEEN ENTERED YET.‘) 0000382 70 FORMAT (' LOGIC STATEMENT IS: -.8A10T 0000383 END 0000384 28 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) SUBROUTINE NAME: CONDTN Purpose: This subroutine is used to control the entry of “conditions” (see RELEXP). Conditions are entered via KEYBRD, decoded via RELEXP, and saved (in char- acter form) in the common area named INPUT. Calling sequence: CALL CONDTN(&n,VARS,CODES,VALS, N COND) Arguments: n—Statement number (in caller) to which a branch will be made on a nonstandard return. VARS—Array of pointers to items referenced in the set of entered conditions. ' CODES—Array of integers giving the individual relational operators in the entered conditions. VALS—Array of values associated with the conditions en- tered. N COND—Counter giving the number of conditions entered. Subroutines called: KEYBRD, RELEXP, PACK Common data referenced : GRASP SOURCE SUBROUTINE CONDTNH‘y VARSvCODESyVALS vNCOND) COMMON IBTHN/ IAVALS’NBE COMMON [INPUT/ EXPRvLOGIC DOUBLE PRECISION EXPR(4926),AREA(419LOGIC(8) INTEGER lMAGE(80)yVARS(1),VALS(11.CODES(1hIVALS(21101'PR0MPT(261 LOGICAL ERR NBE in / BTWN/ EXPR in /INPUT/ Called by: DRIVER Error checking and reporting: An error flag returned by RELEXP is tested. If set, a request to retype the condition is issued. Program logic: 1. A letter (A—Z) is printed as a prompt. 2. The user response (a condition) is obtained via KEYBRD and passed to RELEXP. 3. If no errors have been detected by RELEXP, the user re- sponse is packed into AREA and then moved (first 40 char- acters) into the INPUT common block. 4. Steps 1—3 are repeated and incrementally counted until an all—blank response is entered. When this occurs, execu- tion resumes at statement 40, which sets the number of conditions entered. Control returns to the caller. 5. The nonstandard return is taken if an EOF is sensed from the terminal. PROGRAM 0000385 0000386 0000387 0000388 0000389 0000390 DATA PROMPT/'A."'B.'y'C.'.'D.'.'E.‘,'F.'y'Go','H-'y'1.','J.','K.’0000391 1y'L.‘,'M.'o'N.'y'O.'y'P.','Q.'.'R.' 2'Y.','Z.'/ NBE=0 J=1 10 TYPE 50' PROMPTLH CALL KEYBR018459IMAGE,80) CALL RELEXP(&40y IMAGEHJARSLI )gCODES(JhVALS(J1.ERR) 1F (ERRl GO TO 30 CALL PACKIIMAGEyAREAv40v40) DO 20 I=114 20 EXPR(1yJ)=AREA111 J=J+1 GO TO 10 30 TYPE 60 GO TO 10 40 NCOND=J-1 RETURN 45 RETURN 1 50 FORMAT (1X1A3y3) 60 FORMAT (' RE-TYPE CONDITION') END 'IS.I,IT.| ,‘U.','V.',‘H.’,'X.'y0000392 0000393 0000394 0000395 0000396 0000397 0000398 0000399 0000400 0000401 0000402 0000403 0000404 0000405 0000406 0000407 0000408 0000409 0000410 0000411 0000412 GRASP SOFTWARE SUBROUTINE NAME: DECOMP Purpose: DECOMP extracts a list of item names and a cor— responding list of item numbers from an unpacked character string. Callng sequence.- LIST,N) Arguments: l—Statement (in caller) which will be branched to if an invalid item name is detected. m—Statement (in caller) which will be branched to if the input string contains no item names. IMAGE—Unpacked input-character string. NLIST—List of item names (offset by one from LIST). LIST—A function name followed by a list of item numbers. N—Total number of items in LIST. Subroutines called: BFIND, PACK Common data referenced: NAMES, IPTS, IDIM in blank common Called by: FTNC Error checking and reporting: BFIND takes. the nonstandard return if a name is not found. This causes an error message to be typed and a new input to be requested. If a comma CALL DECOMP (&l,&m,IMAGE,NLIST, GRASP SOURCE SPECIFICATIONS 29 is detected before the list of item names begins, a message is typed and new input is requested. Program logic: The input-string IMAGE is scanned, a char- acter at a time, via a transition matrix. The list of names is created and the list of item numbers is obtained via calls to BFIN D. The following transition matrix is used: IMAT (4, 3) blank comma nonblank purpose 1 f(0)/1 error f(1)/2 start function name 2 f(2)/3 error f(0)/2 find end of function name 3 f(0)/3 f(0)/3 f(3)/4 start item name A f(2)/3 f(2)/3 f(O)/4 find end of item name where the f(i) are: f(0)——No operation. f (1)—Mark first character. f (2)—Mark last character. Pack and find index of item name (that is, find item number). f (3)—Increment list item counter and mark first char- acter of new list item. The entire input stream is scanned, and control is returned to caller. Refer to program logic section of LOGEXP for a more complete discussion of transition-matrix parsing. PROGRAM SUBROUTINE DECOMP(*1*9IMAGEpNLISToLISTle 0000413 COMMUN NAMESVITYPEpKPTSpIDIM 0000414 DIMENSION lTYPE(500lv IMAGE‘ I). LISTiliq IPTS‘500), IMATl493l 0000415 DOUBLE PRECISION NAMESl500)yNAMEyDBLNKvNLIST(5) 0000416 EQUIVALENCE (lNAMEyNAME) 0000417 DATA DBLNKgIBLNKyICOMMAI' 'y' '1'y'/ 0000413 DATA IMAT/le3o312312*4013923112'213414/ 0000419 N=l 0000420 IRUN=1 0000421 00 90 1:1180 . 0000422 [F (IMAGEilloEQ-ICOMMA) GO TO 10 0000423 IF (IMAGE(I).NE.IBLNK) GO TO 20 0000424 IVAL=IMATllROWyli 0000425 GO TO 30 0000426 10 IVAL=IMATiIPUH92) 0000427 GO TO 30 0000428 20 IVAL=IMAT(IRONy3l 0000429 30 IRON=MODiIVAL910) 0000430 JOB=IVAL/10+l 0000431 GO TO (90,50,70v409120)9 JOB 0000432 40 N=N+1 0000433 50 IC=I 0000434 GO TO 90 0000435 70 LC=I”1 0000436 NAME=DBLNK 0000437 CALL PACK‘IMAGE(IC’vNAMEyLC-IC+198l 0000438 IF (NoGTol’ GO TO 80 0000439 LIST(1l=INAME 0000440 GO TO 90 0000441 80 CALL BFINDlCIOOyNAMEyLISTlNipNAME$1IPTSpIDIM) 0000442 NLISTiN-ll=NAME 0000443 30 90 100 110 120 140 150 CONTINUE GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) IF (N.EQ.1) RETURN 2 RETURN TYPE 1409 RETURN 1 TYPE 1501 GO TO 110 FORMAT (1X1A89'IS AN INVALID NAME. FORMAT (- END NAME [MAGE(I) PUNCTUATION ERROR CAUSED BY ',A1y'. RE-ENTER LINE.’) RE-ENTER LINE.') 0000444 0000445 0000446 0000447 0000448 0000449 0000450 0000451 0000452 0000453 GEOLOGIC RETRIEVAL AND SUBROUTINE NAME: DEFINE Purpose: DEFINE is used to define the structure and name of a direct-access disk-data set having fixed-length records. Individual records may then be directly accessed by specify- ing the record number. Calling sequence: CALL DEFINE (U,S,V,F,PJ,PG) Arguments: U—The FORTRAN unit number expressed as an integer. S—The size of the records Within the file, expressed as an integer. For formatted records, S gives the number of characters per record. For unformatted records, S gives the number of words per record. V—The associated integer variable. The record number which would be accessed next if I/O were to continue sequentially is returned as an integer in the associated variable after each random read or write. SYNPOSIS PROGRAM (GRASP) 31 F~The name of the file which will be accessed when an I/ 0 statement references U (the above unit number). PJ—The project number in octal of the disk area in which the file resides. PG—The programmer number in octal of the disk area in which the file resides. Subroutines called : . None Common data referenced: None Called by: FILES Error checking and reporting: None Program logic: This is a DEC 1070, TOPS—10 system resi- dent routine. It provides the capabilities referred to in the Purpose section above. If the GRASP system is to be im- plemented on some other main frame, a comparable routine must be written or acquired. No listing is shown here. 32 GRASP SOFTWARE SPECIFICATIONS SUBROUTINE NAME: DEFLST Purpose: DEFLST outputs the category names to the user and allows him to indicate which categories are of interest. Calling sequence: CALL DEFLST(&m,&n,CAT,NUMC,MC, LIST) Arguments: m—Statement (in calling routine) that will be branched to if no category numbers are given when asked for. n—Statement (in calling routine) that will be branched to if an EOF is sensed in KEYBRD. CAT—Contains the category names as read from unit 20 (the “definitions file”) . NUMC—Number of categories selected by the user. MC—Maximum length in words of a category name. LIST—The category numbers selected. Subroutines called: IFILE, KEYBRD, LENGTH, RLIST Common data referenced: FNAMES, WHICH in /FILNAM/ Called by: NAME, DUMPIT GRASP SOURCE Error checking and reporting: All user response is checked for validity. If errors are detected, the response is requested again. Program logic: 1. A call to IFILE associates the “definitions” file name with FORTRAN unit 20, and the category names are read from this file. 2. The user is asked if he is interested in all categories. His response is checked against “Y” or “N.” If invalid, an error message is typed, and he is asked to respond again. 3. If the user’s response was “Y,” LIST is set to all the category numbers and control is returned to the calling routine. 4. If the response was “N,” the user is asked to enter a list of category numbers of interest. 5. His response, contained in IMAGE, is passed to RLIST to generate the values of LIST. PROGRAM SUBROUTINE DEFLST!*9*9CAT,NUMC,MC.LISTI 0000454 COMMON IFILNAM/ FNAMES,WHICH.PAD 0000455 INTEGER CATI811 IyLIST(1)vIMAGE(30l,FNAMES(21).HHICHyPADI4) 0000456 DATA IYES,NO/'Y'y'N‘/ 0000457 CALL IFILE(20,FNAMES(8+HHICHTl 0000458 READ (20) NCAT,MC,I(CAT(I,J).I=1yMClyJ=1,NCATI 0000459 25 TYPE 30 0000460 CALL KEYBRD‘81001191’ 0000461 IFII.EQ.IYESI GO TO 40 0000462 IFII.EQ.NO) GO TO 5 0000463 TYPE 35 0000464 GO TO 25 0000465 40 NUMC=NCAT 0000466 00 45 I=1.NUMC 0000467 45 LISTIII=I 0000468 00 TO 85 0000469 5 TYPE 110 0000470 DO 10 J=11NCAT 0000471 CALL LENGTH(CATII¢J)9MC,MCL) 0000472 10 TYPE 120v J9ICATI19leI=19MCLl 0000473 TYPE 130 0000474 20 CALL KEYBRDISIOOyIMAGEy30) 0000475 CALL RLISTI8209IMAGE,LIST'NUMC,NCAT) 0000476 IF (NUMC.EQ.0) GO TO 90 0000477 85 RETURN 0000478 90 RETURN 1 0000479 100 RETURN 2 0000480 30 FORMATI' SHALL ALL CATEGORIES BE CONSIDERED? (YES OR NO): '93) 0000481 35 FORMATI' YOUR REPLY HAS NOT UNDERSTOOD.'I 0000482 110 FORMAT (' EACH RECORD HAS BEEN DIVIDED INTO THE FOLLOWING 'o'GENER0000483 lAL CATEGORIES='/8X.'CAT. # CAT. NAME'ISX,‘ --------------- ') 0000484 120 FORMAT (10X112y4Xp9A5) 0000485 130 FORMAT (' ENTER A LIST OF ASCENDING NUMBERS MATCHING 'y'YOUR CATE00000486 IORIES OF INTEREST'/' (IE. 1,3,5 OR 2-5)” 0000487 END 0000488 GRASP SOFTWARE SPECIFICATIONS 33 SUBROUTINE NAME: DUMPIT Purpose: DUMPIT outputs to the terminal those values for all items present in a set of user-selected categories. The values are obtained from a user-selected file. Calling sequence: CALL DUMPIT Arguments: None Subroutines called: OPREP, DEFLST, FINDGP, ACCESS, GETPUT, PAUSE, LENGTH, UNCODE, BINTYP, BLIST Common data referenced: NFILE in /IOUNIT/ NAMES, ITYPE in blank common Called by: DRIVER Error checking and reporting: None Program logic: ‘ 1. Page size (NPAGE), input file name, and file unit are set up by a call to OPREP. 2. A call to DEFLST is made to determine categories to be dumped. 3. Calls to FIN DGP are made to determine pointers (KLIST) for those items in the selected categories. As DEFLST and FINDGP used FORTRAN unit 20, the “definitions” file for the current data base, the unit is rewound. GRASP SOURCE SUBROUTINE DUMPIT COMMON NAMES:ITYPE,IPTS,IPAD COMMON IIOUNIT/ NFILEoIOF DIMENSION ITYPE(500)9 l. REClSOO), TANKiZSh LOGICAL PNT HDG'NEHCATvHIT EQUIVALENCE (REC!IREC)9(IVAL0VAL)9 (TANK'LISle ( IBLNK, IQUAL) O, IGI"IHI’ILI'INI ’ITI/ DATA IQUALI' KOUNT=O CALL OPREP(8200982059NPAGE) CALL DEFLST (8200, 82009CAT,NUMC 1MC1KLIST) NTOT=0 DO 20 K=l oNUMC CALL FINDGP( EZOOyKLISHK) vi vJoNoviREC) READ (20! DO 10 I=lvNG NTOT=NTOT+1 10 SELECTlNTOT)=2*IREC(Il SELECTlNTOT-NG+1l=SELECTiNTOT-NG+1)+I 20 CONTINUE RENIND 20 CALL ACCESS([[01VAL9TANKpr1) 40 CALL GETPUTlCZOOvIRECvll HIT=OFALSEI KNT=O DU 190 JJ=10NTUT [I=SELECT(JJ)/2 NEWCAT=SELECT(JJloNE-2*II 1F (NEHCAT) KNT=KNT+1 PNTHDG=PNTHDG.OR.NEWCAT BITEM‘15925lv LABELl25ly DOUBLE PRECISION LABELyNAMESly NAMESlSOOl INTEGER TANKyCATl8y17lyKLlSTll7lySELECT(500)plPTSlSOOl 4. A call to ACCESS is made to initialize character dictionary lookups. 5. Each record in the input file is obtained by GETPUT, the selected items are tested for nonbllank characters, and their value is output. 6. The output algorithm is basically as follows: (a) Determine item type and switch to appropriate code section via a computed GOTO. (b) If type is integer, print under an I format. (c) If type is real, print under a G format. (d) If type is character, obtain string value by a call to ACCESS and print under an A format. (e) If type is multiple choice, obtain possible values by a call to BINTYP, and select the actual subset via a call to BLIST. Print this subset under an A format. (f) If type is qualified real, obtain value and qualifier via a call to UNCODE. Print under a G and A format. (g) After each line is printed, increment and test KOUNT against page size. If KOUNT is greater than page size, call PAUSE for a programmed pause and re- initialize KOUN T to zero. (h) After each record has been processed, print a line of asterisks as a record separator. PROGRAM 0000489 0000490 0000491 0000492 0000493 0000494 0000495 0000496 0000497 0000498 0000499 0000500 0000501 0000502 0000503 0000504 0000505 0000500 0000507 0000508 0000509 0000510 0000511 0000512 0000513 0000514 0000515 0000510 0000511 0000518 0000519 0000520 lREClSOO) 0 LlSTlZSl IQUAlelv 34 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) NAMESI=NAMESIIII 0000521 IVAL=IRECIIII 0000522 IF IIVAL.EQ.IBLNK) GO TO 190 00C0523 H1T=.TRUE. 0000524 KOUNT=KOUNT+1 0000525 IF (KOUNT.LE.NPAGE) GO TO 50 0000526 KOUNT=O 0000527 CALL PAUSEIEZOOT 0000528 50 KIND=ITYPEIIIT 0000529 IF(.NOT.PNTHDGI GO TO 55 0000530 KL=KLISTIKNT) 0000531 CALL LENGTHICAT(1vKL)yMCyMCL) 0000532 TYPE 210,(CATII.KLI,T=1,MCL) 0000533 PNTHDG=.FALSE. 0000534 55 GO TO I60o70v801170g160I.KIND 0000535 60 TYPE 230, NAMESI.IVAL 0000536 00 TO 190 0000537 70 TYPE 240, NAMESIyVAL 0000538 GO TO 190 0000539 80 CALL ACCESSIII.IVAL,TANK1Jy2) 0000540 TYPE 2501 NAMESIyITANKIII1I=1yJI 0000541 00 T0 190 0000542 160 VAL=UNCODEIVALyIQ) 0000543 TYPE 2401NAMESIoVALyIQUALIIQ) 0000544 GO TO 190 0000545 170 CALL BINTYP(IIvLABELvBTTEMvaM) 0000546 KCUNT=KOUNT+I 0000547 TYPE 2509 NAMES! 0000548 CALL BLISTILIST,NUMS.IVAL) 0000549 00 180 I=19NUMS 0000550 J=LISTIIT 0000551 180 TYPE 260, LABELIJI9IBITEMIL9J)7L=1,K) 0000552 190 CONTINUE 0000553 IF (HIT) TYPE 220 0000554 GO TO 40 0000555 200 REHIND NFILE 0000556 RENTND 20 0000557 205 RETURN 0000558 210 FORMAT (‘ CATEGORY: '98A5) 0000559 220 FORMAT (1X93I8H********T) 0000560 230 FORMAT (2X9A8.1X,I9) 0000561 240 FORMAT (2X:A891X11PGIZ.5:A1) 0000562 250 FORMAT (ZXVA891XOIZA5/11X112A5) 0000563 260 FORMAT (5X.A8.15A4) 0000564 END 0000565 GRASP SOFTWARE SPECIFICATIONS 35 FUNCTION NAME: EVAL Called by: COLPNT, ROWPNT . . Error checking and reporting : Purpose: By us1ng a partlcular set of values as operands, 1 D' . . b tt ted EVAL evaluates a previously parsed Reverse-Polish-form ' 1VLs10n y zero a emp ' arithmetic expression. 2. Log of a nonpositive value attempted. Calling sequence: VAL: EVAL(VALUES,TYPE,POLISH, 3. Square root of a negative value attempted. LBLNK) Program logic: A push-down stack technique is used to evalu- Argnments: ate the Reverse-Polish string contained in TYPE and POL- VALUES—Set of operand values. ISH. TYPE is scanned, an element at a time, pushing TYPE, POLISH—Arrays containing the encoded Reverse— operand values down on the stack until an operator is Polish form of the expression to be evaluated. The encod- sensed. Either the top or topmost two stack elements are ing is as follows: Let ITY be the I’th element of TYPE. then used as operands resulting in a new topmost-stack If ITY:0, the I’th element of POLISH is a numeric element which is the resulting value of the operator. Unary constant. If ITY>O, ITY is an index to the array operators/functions (absolute value, ABS; square root, VALUES. If ITY<0, ITY corresponds to an arithmetic SQRT; logarithm, L0G; square, SQR; ten exponent, TEN: operator or function. minus, —) operate on the top stack element. Binary opera- I—Gives the length of the arrays TYPE and POLISH. tors (—1—, —, *, / ) operate on the topmost-two stack elements. BLNK—Logical variable set to TRUE if any operand with After all elements of TYPE have been processed, the stack a blank value is sensed. should have one value in it. This value, the result, is re- Subroutines called: UN CODE turned. If a blank operand value is detected, the flag BLNK Common data referenced: ITYPE in blank common is turned on and zero is returned. GRASP SOURCE PROGRAM FUNCTION EVALIVALUESyTYPEyPOLISHoI,BLNK) 0000566 COMMON NAMES.ITYPE,IPTS,IPAD 0000567 DIMENSION NAMES(500)¢ VALUESIII, POLISH(lly STACK(411. TYPEIll. IT0000568 IYPEISOOIvIPTSISOOI 0000569 DOUBLE PRECISION NAMES 0000570 INTEGER TOP97YPEyVALUES 0000571 LOGICAL BLNK 0000572 EQUIVALENCE (VALyIVAL) 0000573 DATA IBLNKI' '/ 0000574 BLNK=.FALSE. 0000575 TOP=O 0000576 00 190 J=1'I 0000577 INDEX=TYPE(J) 0000578 IF (INDEX) 40130.10 0000579 10 TOP=TOP+1 0000580 IVAL=VALUESIINDEXl 0000581 IF (IVAL.EQ.IBLNK) GO TO 195 ' 0000582 IF IITYPEIINDEX).EQ.2) GO TO 15 0000583 IFIITYPE(INDEX).NE.5) GO TO 20 0000584 VAL=UNCODEIVALyIQI 0000585 15 STACKITOP1=VAL 0000586 GO TO 190 0000587 20 STACK(TOP)=IVAL 0000588 GO TO 190 0000589 30 TOP=TOP+1 0000590 STACKITOPl=POLISHIJ1 0000591 GO TO 190 0000592 40 IF (INDEX.GE.-4l GO TO 130 0000593 INDEX=INDEX+11 0000594 GO TO (50.60180’100y11091201v INDEX 0000595 50 STACK(TDP)=ABS(STACK(TOP)) 0000596 GO TO 190 0000597 36 60 70 80 90 100 110 120 130 140 150 160 170 180 190 195 200 210 220 230 240 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) TV=STACKITOPI IF (TV.GE.0.) GO TO 70 TYPE 230 GO TO 200 STACK(TOP)=SQRT(TV) GO TO 190 TV=STACK(TUP1 IF (TV.GT.O.) GO TO 90 TYPE 220 GO TO 200 STACK(TOP§=ALOGIO(TV) GO TO 190 STACK(TOP)=STACK(TOPT**2 GO TO 190 STACK(TOP)=10.**STACK(TOP) GO TO 190 STACK(TOP)=-STACK(TOPT GO TO 190 INDEX=INDEX+5 VT=STACK(TOP) TOP=TOP-1 GO TO (150,140.180,17019 INDEX STACKITOP)=STACK(TOP)*VT GO TO 190 IF (VT.NE.0.0) GO TO 160 TYPE 240 GO TO 200 STACK(TOP1=STACK(TOP)/VT GO TO 190 STACK(TOP)=STACK(TOP)+VT GO TO 190 STACKITOP)=STACK(TOP)°VT CONTINUE IF (TOP .NE. 1) GO TO 200 EVAL=STACK(1) GO TO 210 BLNK=.TRUE. EVAL=0.0 RETURN FORMAT (' ATTEMPTED TO TAKE LOG OF A ZERO OR NEG. VALUE-'1 FORMAT (' ATTEMPTED TO TAKE SORT FORMAT (' END DIVIDE BY ZEPO ATTEMPTED.') OF A NEGATIVE VALUEo'T 0000598 0000599 0000600 0000601 0000602 0000603 0000604 0000605 0000606 0000607 0000608 0000609 0000610 0000611 0000612 0000613 0000614 0000615 0000616 0000617 0000618 0000619 0000620 0000621 0000622 0000623 0000624 0000625 0000626 0000627 0000628 0000629 0000630 0000631 0000632 0000633 0000634 0000635 0000636 0000637 0000638 0000639 0000640 GRASP SOFTWARE SPECIFICATIONS 37 SUBROUTINE NAME: FDRIVE cating phase 1, 2, or 3. The phases are initialization, body, Purpose: FDRIVE provides for the single-pass computation and postprocessing'. of all implemented mathematical or statistical functions. Us- Subroutines called: MEAN, FIT ing this routine to make all calls to the subroutines cor- responding to implemented functions simplifies the addition of new functions. Common data referenced: None Called by: FTNC Calling sequence: CALL FDRIVE (ISWTCH) E7707“ checking and reporting: None Arguments ,- Program logic: This routine merely makes calls to the func- ISWTCH—Integer code passed to called subroutines indi- tions selected by the user via computed GOTO. GRASP SOURCE PROGRAM SUBROUTINE FDRIVE(ISNTCH) 0000641 COMMON [FTNCOM/ TAGS,IREC,ARGS,NARGSyIFTN,NFTN 0000642 DOUBLE PRECISION TAGS‘5y5) 0000643 INTEGER ARGSlva),NARCS‘S)11FTN(S)9IREC(500l 0000644 00 30 J=lyNFTN 0000645 I=IFTN(J) 0000646 GO TO (10,20,30,30,30,30), I 0000647 10 CALL MEAN(J,ISHTCHD 0000648 GO TO 30 0000649 20 CALL FIT(J,ISHTCH) 0000650 30 CONTINUE 0000651 RETURN 0000652 END 0000653 38 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) SUBROUTINE NAME: FILES Purpose: FILES prompts for and accepts a data-base name. After the name has been provided and verified, the Mask file of data-base characteristics associated with that name is read. Some preliminary processing is done on these char- acteristics. Calling sequence: CALL FILES (&n) Argument: n—Statement (in caller) to which a branch is made if the nonstandard return from KEYBRD is taken (namely, an EOF is sensed). Subroutines called: KEYBRD, IFILE, BINIT, DEFINE, PACK Common data referenced: NAMES, ITYPE, IPTS, IDIM in blank common. MASTER, MASK, DFILE, BFILE, NUMF, NUMI, IDIMS in /FILNAM/. I, IDPT in /DACOMM/. Called by: DRIVER GRASP SOURCE Error checking and reporting: The data-base name entered by the user is compared with the list (MASTER) of those available. If the data-base name is not recognized, an error message is typed. Program Logic: 1. A data-base name is prompted for and accepted via KEY- BRD, and is then packed into FILEID and compared with the list (MASTER) of available names. If not found, an error message and the list of available names is typed fol- lowed by a prompt for another name. 2. Once the data base has been established, its corresponding Mask file is read to fill the blank common area. 3. The item names from MASK are sorted via a call to BINIT and the array of pointers (IPTS) to the sorted NAMES is returned from BINIT. 4. NAMES is restored to its unsorted form and DEFINE is called to associate the name of the direct-access character dictionary with unit 21. IFILE is then called to associate the name of the multiple-choice file with unit 22. PROGRAM SUBROUTINE FILES(*I 0000654 COMMON NAMES,ITYPE,IPTS.IDIM 0000655 COMMON IFILNAM/ MASTER.MASK'DEFTN’DFILEvBFILEyNUMFQNUMIQIDIMS 0000656 COMMON /DACOMM/ IyIDPT 0000657 DIMENSION ITYPEISOOI. IPTSISOO), IDPTISCOI, IDIMSI4) 0000658 DOUBLE PRECISION NAMES(500).VNAMESI500) 0000659 INTEGER MASTERI4I,MASKI4IvDEFTN(4IyDFILE(4)yBFILEI4I’FILEID 0000660 DATA IZGyI21o122/20921922/ 0000661 1 TYPE 11 0000662 CALL KEYBRDI8999,ITYPE,5) 0000663 CALL PACKIITYPE'FILEIDQSvSI 0000664 00 3 J=11NUMF 0000665 IFIMASTERIJI.EQ.FILEID) GO TO 5 0000666 3 CONTINUE 0000667 TYPE 41FILE101‘MASTERIIIII=1yNUMFI 0000668 GO TO 1 0000669 5 NUMI=J 0000670 CALL IFILEIIZO.MASK(NUMI)) 0000671 IDIM=IDIMSINUMI) 0000672 READ(120) IITYPEIIIyI=lplDIMIyIIDPTIIIvI=19IDIMI, 0000673 I (NAMESIIIyI=1.IDIM) 0000674 REWIND 120 0000675 DO 10 I=1yIDIM 0060676 10 VNAMESIII=NAMESIII 0000677 CALL BINITINAMES;IPTS.IDIM) 0000678 DO 20 I=lyIDIM 0000679 20 NAMES(I)=VNAMES(I) 0000680 CALL DEFINEIIZle7vaDFILE(NUMI)."4129“176) 0000681 CALL IFILEIIZZvBFILEINUMIII 0000682 RETURN 0000683 999 RETURN 1 0000684 4 FORMAT(1X,A6,'NOT AN AVAILABLE DATA BASE NAME. '1 0000685 1 'USE ONE OF THE FOLLOWING:'/(1X'A5)I 0000686 11 FORMAT(' ENTER DATA BASE NAME: ’93) 0000687 END 0000688 GRASP SOFTWARE SPECIFICATIONS 39 SUBROUTINE NAME: FIND Purpose: FIND is used to look up a word in a “hash coded” table and return a code associated with its position. The “Linear quotient” technique, as described by Bell and Kaman (1970), is used. Calling sequence: CALL FIND (&n,ISYMBL,KODE,CHARS, CODES,M) Arguments: n—Statement (in caller) to which a branch is taken if ISYMBL is not in CHARS. ISYMBL—«Word to be looked up. KODE—Integer from the position in CODES corresponding to the position of ISYMBL in CHARS. CHARS—“Hash coded” table of words to be looked up. GRASP SOURCE SUBROUTINE FIND(*9ISYMBLvKODEvCHARSvCDDESoM’ INTEGER CHARSllloCODES(l)gFILLER DATA FILLER/'VOID'I L=IABS(ISYMBL) J=L/M I=L-M*J IF (MODlJyM).EQ.0) J=1 SC ICHAR=CHARS(I+1) IF (ICHAR.EQ.ISYMBL) GO TO 60 [F (ICHAR.EQ.FILLER) RETURN 1 [=MOD(I+J9Mi GO TO 50 60 KO0E=CODESll+1) RETURN END CODE S—Table corresponding to CHARS giving the original position of the word to be looked up. M—Table size (a prime number) for CHARS and CODES. Subroutines called: None Common data referenced: None Called by: LOGEXP, PARSE Error checking and reporting: If ISYMBL is not in CHARS, the nonstandard return is taken. Program logic: An initial location I and displacement J is determined from the internal machine representation of the word stored in ISYMBL. Initial and successive probes to CHARS are made until an empty location is probed. If ISYMBL is not in CHARS, the nonstandard return is taken. If ISYMBL is found in CHARS, the corresponding element of CODES is returned in KODE. PROGRAM 0000689 0000690 0000691 0000692 0000693 0000694 0000695 0000696 0000697 0000698 0000699 0000700 0000701 0000702 0000703 40 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) SUBROUTINE NAME: FINDGP Purpose: FIN DGP positions the file associated with FORTRAN unit 20 to a particular record paid associated with a category. Data concerning that category are returned to the caller. Calling sequence: CALL FINDGP(&n,KNUM,NUM,MAXL, NG,GROUP) Arguments: n—Statement number (in caller) to which control is passed if an EOF is sensed on unit 210. KNUM—The category number to which the file will be be positioned. NUM—The number of descriptions in this category. MAXL—The maximum length (in words) for a description. NG—Number of items referred to in this category. GROUP—List of item pointers associated with this category. GRASP SOURCE SUBROUTINE FINDGP!*pKNUM.NUM,MAXL,NG.GROUP1 COMMON IFILNAM/ FNAMESyHHICHpPAD INTEGER GROUP”1yFNAMESiZHyHHICHyPAD(41 DATA NDEF /20/ 30 READ (NDEF,END=901 KK’NUM'MAXLYNGO(GROUP(I,11:19NG) 1F (KK—KNUM) 40180170 40 J=KNUM-KK-1 50 READiNDEFl IF (J.LT.11 GO TO 30 DO 60 I=lpJ READlNDEF) 60 READ(NDEF) GO TO 30 70 REHIND NDEF CALL IFILEiNDEF,FNAMES(8*HH1CH)l J=KNUM GO TO 50 80 RETURN 90 RETURN 1 END Subroutines called: IFILE Common data referenced: FNAMES, WHICH in/FILNAM/ Called by: NAME, DUMPIT Error checking and reporting: None Program logic: 1. The next record on unit 20 is read, giving a category number KK and values for the last four arguments. 2. KK is tested against KNUM. (a) If KKKNUM, rewind unit 20 and reassociate it with the correct name via a call to IFILE having FNAMES and WHICH as arguments. Next, reposi— tion the file to the number pair of interest. PROGRAM 0000704 0000705 0000706 0000707 0000708 0000709 0000710 0300711 0000712 0000713 0000714 0000715 0000716 0000717 0000718 0000719 0000720 0000721 0000722 0000723 GRASP SOFTWARE SPECIFICATIONS 41 SUBROUTINE NAME: FIT Purpose: FIT is used to provide a least-square linear fit be- tween two items within a selected file. Calling sequence: CALL FIT (J,ISWTCH) Arguments: J—Pointer used to retrieve argument values from the com- mon area /FTNCOM/. ISWTCH—Switch indicating which of three parts (initial- ization, body, postprocessing) of the code is to be executed. Subroutines called: UNCODE Common data referenced: ITYPE in blank common TAGS, IREC, ARGS, NARGS, in /FTNCOM/ Called by: FDRIVE Error checking and reporting: If two arguments are not given, an appropriate error message is typed and return is immedi- ate. If the computation would yield an infinite slope, that message is typed. Program logic: The value of ISWTCH determines which of three sections of the code is executed. If ISWTCH=1,, the number of arguments is checked, and various sums are set GRASP SOURCE SUBROUT INE FIT( J. ISWTCH) COMMON NAMES: ITYPEylPTS. [PAD COMMON 1 SUMXS. SUMYSvaFN'Vl yVZ INTEGER IPTS( 500) INTEGER ARGSl6y5lyNARGS(5lyIFTN(5)'ITYPEi500lpIREC(500) LOGICAL ERR EQUIVALENCE IIVALI,VAL1I. DATA IBLNKI' ., IFIIszCH—ZI 2.4.6 2 ERR=NARGS(J).NE.2 [F (ERR) GO TO 20 N=O sunx=o.ooo SUMY=0.0DO sumxv=o.ooo sumxs=o.ooo sumvs=o.ooo GO T0 30 4 IF (ERR) GO T0 30 IVAL1=IRECIARGSI2.J)) IF (IVAL1.EQ.IBLNK) GO TO 30 IVAI2=IRECIARG$I3.JII IF IIVAL2.Eo.IBLNKI GO TO 30 V1=IVAL1 v2=IVAL2 1F IIIYPEIARGSI2.JII.EQ.2I V1=VAL1 IF IIIVPEIARGSI3,JII.EQ.2I V2=VAL2 IFiITYPEiARGSiZoJ) ).EQ.5) Vl=UNCODEiVALl 910) IF( ITYPEiARGSi3yJ) ).E0.5) V2=UNCODE(VAL2:IQ) N=N+1 SUM X= SUM X+V l IFTNCOM/ TAGS,IREC,ARGS,NARGS,IFTNyNFTN DOUBLE PRECISION NAMES‘SOOhTAGS(595)9SUMX’SUMY,SUMXY, (IVALZoVALZ) to zero. If ISWTCH=2, the error flag is tested. If not set, the appropriate values of the arguments are tested for non- blank. If nonblank, they are added to the appropriate sums. If ISWTCH:3-, the slope, intercept, and correlation coeffi- cient are calculated (if possible), using the sums previously determined. They are then printed out using the appropriate item names. All summations and least-square determinations are done using double-precision arithmetic to minimize the round-off effects introduced by performing the computation using only one pass on the data. Assuming the function FIT X,Y had been issued, the calcu- lations are performed using the following formula: DzN-EXg—EX-EX Bl: (N-EXY—EX-EYHD Bo:(EY—Bl-2X)/__N‘ C:D~Bi/VD-(N-2Y’—W where: Bo=intercept, Bi=slope, C—_-correlation coefficient, and N =number of nonblank X, Y points. PROGRAM 0000724 0000725 0000726 0000727 0000728 0000729 0000730 0000731 0000732 0000733 0000734 0000735 0000736 0000737 0000738 0000739 0000740 0000741 0000742 0000743 0000744 0000745 0000746 0000747 0000748 0000749 0000750 0000751 0000752 0000753 0000754 0000755 0000756 42 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) SUMXS=SUMXS+V1*V1 SUMXY=SUMXY+V1*V2 SUMY=SUMY+V2 SUMYS=SUMYS+V2*V2 GO TO 30 6 IF (ERR) GO TO 30 FN=N D=FN*SUMXS-SUMX*SUMX IF (0.50.0.000) GO TO 10 Bl=(FN*SUMXY-SUMX*SUMY)ID BO=ISUMY-BI*SUMX)/FN 0000757 0000758 0000759 0000760 0000761 0000762 0000763 0000764 0000765 0000766 0000767 CC=(FN*SUMXY-SUMX*SUMY)IDSQRT((FN*SUMXS-SUMX*SUMX)*(FN*SUMYS-SUMY*0000768 lSUMY)) TYPE 40, N,TAGS(1:J)oTAGS(2vJ)vBloBO,CC GO TO 30 10 TYPE 50, N GO TO 30 20 TYPE 60 30 RETURN 0000769 0000770 0000771 0000772 0000773 0000774 0000775 40 FORMAT (/lxsl5" POINTS USED TO FIT 'gABo'TO '1A8/. SLOPE='91PE12.0000776 159' INTERCEPT=',&12.5,' CORR. CDEFF.=',E12.5) 50 FORMAT (' UNABLE TO CALCULATE FIT HITH'.IS,' POINTS') 60 FORMAT (' THE FIT FUNCTION MUST HAVE 2 ARGUMENTS') END 0000777 0000778 0000779 0000780 GRASP SOFTWARE SPECIFICATIONS 43 SUBROUTINE NAME: FTNC Purpose: This routine acts as a driver for the processing of the FUNCTION command. It accepts (via KEYBRD) the function names and arguments, sets up system-required input-file information, and is then used to supply input records to the routines that actually calculate the requested functions. Calling sequence: CALL FTNC(&n) Argument: n—Statement (in caller) to which a branch is made if the nonstandard return (EOF) from KEYBRD is taken. Subroutines called: KEYBRD, OBEY, DECOMP, FDRIVE, GETPUT, PACK Common data referenced: FNAMES, WHICH in /FILNAM/ TAGS, IREC, ARGS, NARGS, IFTN, NFTN in /FTNCOM/ NFILE in /IOUNIT/ Called by: DRIVER Error checking and reporting: Function names entered by GRASP SOURCE SUBROUTINE FTNCH‘) COMMON IFILNAM/ FNAMES,HH[CH.PAO COMMON /FTNCOM/ TAGS. IREC:ARGS.NARGS, IFTN,NFTN COMMON /IOUNlT/ NFILEylPAD DOUBLE PRECISION TAGSlSyS) INTEGER DFAULT, FILE,DBLNK,FNAMES(21)9NHICH,PAD(4),IFTNK 53 ’ lEQUATE(4)yFTNSl51.ARGS(615)cIMAGE(80)1NARGS(5hIREC(50019 Z PROMPT(5) LOGICAL ANY vGOOD DATA DBLNKoEQUATEI' EQUIVALENCE (EQUATE(4)1FILET TYPE 100 DFAULT=FNAMES(HHICH) CALL KEYBRD(&909IMAGE95) CALL PACK(IMAGE1FILE!5,5) IF (FILE.EQ.DBLNK) FILE=DFAULT CALL OBEYl885yEQUATEi4l TYPE 1209 (FTNS([)91=111MPLTOl TYPE 130 10 NFTN=1 20 TYPE 150. PROMPTiNFTNi CALL KEYBRD(&90.IMAGE:80) CALL DECOMPUIZO .830 ,IMAGEgTAGS(11NFTN)'ARGS(1vNFTNhNARG) NARGSKNFTN)=NARG-l NFTN=NFTN+1 IF (NFTN-lMPLTD—ll 30 NFTN=NFTN~1 IF (NFTN.E0.0) RETURN ANY=.FALSE. DO 60 l=1vNFTN GOOO=.TRUE. lNAME=ARGS(lyI) DO 40 J=1,1MPLTD IF (iNAME.EQ.FTNS(Jll 20930.30 GO TO 50 I,OEQUAO,ITE IC'III'I 1/ DATA lMPLTDyFTNS/Zy'MEAN‘y'FIT',3*' '/ DATA PROMPT/'lo'y'zo'9'3o'9'4o'9'5.'/ the user are checked against a list of those available. If an invalid function name is entered, that message is typed. Program logic: 1. A prompted input-file name is obtained via KEYBRD, packed via a call to PACK, and associated with the FOR- TRAN input unit number via OBEY. 2‘. A list of the implemented function names is typed along with a request to enter the names of desired functions and their corresponding arguments. 3. As each function name is entered (via KEYBRD), it and its argument names are identified via DECOMP. 4. After the names have been entered and identified, a call to FDRIVE using an argument of 1 is issued to perform initialization. 5. Each record of the selected file is obtained via GETPUT and processed via a call to FDRIVE using an argument of 2. 6. Finally, a call to FDRIVE using an argument of 3 is made to accomplish any wrap-up processing associated with the selected function, and the input file is rewound. PROGRAM 0000781 0000782 0000783 0000784 0000785 0000786 0000787 0000788 0000789 0000790 0000791 0000792 0000793 0000794 0000795 0000796 0000797 0000798 0000799 0000800 0000801 0000802 0000803 0000804 0000805 0000806 0000807 0000808 0000809 0000810 0000811 0000812 0000813 0000814 0000815 0000816 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) 4O CONTINUE 0000817 TYPE 1601 INAME 0000818 J=6 0000819 GOOD=.FALSE. 0000820 50 IFTNIII=J 0000821 60 ANY=ANY.OR.GOOD 0000822 IF (.NOT.ANY) GO TO 10 0000823 CALL FDRIVEII) 0000824 70 CALL GETPUT(880.IREC:1I 0000825 CALL FDRIVE(2I 0000826 GO TO 70 0000827 80 CALL FDRIVE(3) 0000828 REHIND NFILE 0000829 RETURN 0000830 90 RETURN 1 0000831 100 FORMAT (' ENTER NAME OF FILE: '.$) 0000832 120 FORMAT (' FUNCTIONS AVAILABLE AT THIS TIME ARE:'/1X,5A81 0000833 130 FORMAT (' ENTER FUNCTION NAMES AND CORRESPONDING ARGUMENTS. ') 0000834 150 FORMAT (IX.A3,$) 0000835 160 FORMAT (1X:A5y' IS NOT AN AVAILABLE FUNCTION AND HAS ','BEEN IGNOR0000836 IED.'I 0000837 END 0000838 GRASP SOFTWARE SPECIFICATIONS SUBROUTINE NAME: GETPUT Purpose: GETPUT is used to md and unpack a record from the current input file, or to write the last-packed record obtained from the current input file on the current output file. Calling sequence: CALL GETPUT(&n,IREC,ISWTCH) Arguments: n—Statement (in caller) to which a branch is made if an EOF is sensed on the current input file. IREC—Contains the expanded record from the current input file. ISWTCH—Switch indicating whether record is to be read or written. Subroutines called: None Common data referenced: IN, IOUT in /IOUNIT/ Called by: COLPNT, DUMPIT, FTNC, RETRVE, ROWPNT Error checking and reporting: None Program logic: 1. If ISWTCH: 1, the next input record is read into TANK GRASP SOURCE 45 on the unit referenced by IN in /IOUNIT/. If an EOF is sensed, the nonstandard return is taken. If ISWTCH>1, the last—read input record is output on the unit referenced by IOUT in /IOUNIT/. 2.. The first word of TANK is assumed to be of type INTEGER and corresponds to the first word of IREC. The last 2 bits contain the type of the next word in TANK. TYPE values are: TYPE Indicates 1 Next word is an integer value. 2 Next word is a real value. 3 Next word is a blank counter. The value part of types 1 and 3 is in the leading bits (that is, 2—bit truncation). 3. If the word is of type REAL, it may be visualized as being composed of an integer and a fraction. The type for the next word is in the last 2 bits of the integer part and the associated real value is obtained by adding the fractional part to the leading bits of the integer part. PROGRAM SUBROUTINE GETPUT(*vIRECoISHTCH) 0000839 COMMON IIOUNIT/ INqIOUT 0000840 INTEGER IRECT l T,lTANK(lSO)yBLANK,TYPE 0000841 REAL TANKllSO) 0000842 EQUIVALENCE (FRACT11VALlo (ITANK(1I,TANK(1)) 0000843 DATA BLANKI' ‘/ 0000844 IF‘ISHTCH-Z) 5.70170 0000845 5 READ (IN,END=110) NUM,(TANK(I)1[=1vNUM) 0000846 TYPE=1 0000847 IPT=l 0000848 DO 50 I=19NUM 0000849 IFTTYPE.LT.0) TYPE=eTYPE 0000850 lF (TYPE-2T 10,20,30 0000851 10 IVAL=ITANK(I)/4 0000852 TYPE=ITANK(I)-4*IVAL 0000853 GO TO 25 0000854 20 TYPE=TANK(I) 0000855 FRACT=TANKTIT¢TYPE 0000856 J=TYPE/4 0000857 TYPE=TYPE-4*J 0000858 FRACT=J+FRACT 0000859 25 IRECIIPT)=IVAL 0000860 IPT=IPT+1 0000861 GO TO 50 0000862 30 J=lTANK(I)/4 0000863 TYPE=ITANK(I)—4*J 0000864 00 40 K=1,J 0000865 IREC(IPT)=BLANK 0000866 40 IPT=IPT+1 0000867 50 CONTINUE 0000868 60 RETURN 0000869 70 WRITETIOUTTNUMyTTANKil)vI=11NUM) 0000870 GO TO 60 0000871 110 RETURN 1 0000872 END 0000873 46 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) SUBROUTINE NAME: HELP Subroutines called: None . Common data referenced: None Purpose: HELP types the names and descriptions of GRASP Called by: DRIVER commands available to the user. Error checking and reporting: None Calling sequence: CALL HELP (WORDS) Program logic: The text associated with each of the commands Argument: is initialized in DATA statements. Each command name (in WORDS—An array containing the command names that WORDS) is typed with its corresponding description in may be issued by the user. TEXT. GRASP SOURCE PROGRAM SUBROUTINE HELP(WORDS) 0000874 INTEGER NORDSil) 0000875 DOUBLE PRECISION TEXTIIlvllinSGll11),M$GZ(II)7MSG3(11T1 0000876 I MSG4‘IIIQMSGSI11’1MSGéi11’9MSGT(11I1MSGSI11’7M569I11.1 0000877 2 MSGIOillIvMSGIITIII 0000878 EQUIVALENCE (MSGI(1)9TEXT(191I)vIMSGZ(IIQTEXT(ly2II1 0000879 I (MSGBTI),TEXT(1,3)IvTMSG4IIIyTEXTI194)),(MSGS(ITyTEXTIIySDI, 0000880 2 (MSGbTI),TEXTIlyb)i.(MSG7(1),TEXT(1o7))yTMSG8(1IvTEXTI1o8))9 0000881 3 (MSGQ‘l)!TEXT(199II.(MSGIO(l’oTEXTIlQlO)I)‘MSGII(1I7TEXT(1111II 0000882 DATA MSGI/'- INITIA'9'TES THE','REOUEST','FOR RETR'.'IEVAL CR'. 0000883 l'ITERIA T','O 8E ENT'y'ERED IN'y'THE FORM',': NAME R'9‘EL VALUE'I 0000884 DATA MSGZ/'- INITIA'y'TES THE'9'REQUEST','FOR A LO','GICAL EX', 0000885 l'PRESSION',‘ TO BE E'y'NTERED'g'USING LO'y'GICAL OP'.‘ERATORS.'/ 0000886 DATA MS63/'- INITIA','TES THE'y'SEARCH O','F A FILE'.’ BASED U'y 0000887 I'PON PREV‘y'IOUSLY E'y'NTERED','CONDITIO'.'NS AND L','OGIC. '/ 0000888 DATA MSG4/'- ALLOWS',‘ THE USE','R TO LIS'y'T SELECT'y'ED VALUE', 0000889 I‘S (VARIA'q'BLE NAME','S HILL"'BE ASKED',‘ FOR) IN',‘ A FILE.'/ 0000890 DATA MSGS/'- ALLOWS's‘ THE USE','R TO SEL','ECT OR C'.'HANGE TH'. 0000891 1 'E DATA B'y'ASE TO B','E USED.',3*' '/ 0000892 DATA MSGb/'- TERMIN','ATES THE',’ SYSTEM.‘,' ENTERIN'g'G IN R', 0000893 1'ESPDNSE','TO A PRO'y‘MPT WILL','ALSO STO'.'P THE SY'.'STEM. '/ 0000894 DATA MSG7/'- USED T‘y'O PRINT','ITEM NAM','ES, THEI'y'R TYPES', 0000895 I'AND DEFI','NITIONS’.'IN A '9'SELECTED'9' SET 0F','GROUPS.'/ 0000896 DATA MSGS/'- USED T','O OBTAIN',‘ THE ABO'p'VE COMMA','ND DEFIN', 0000897 I'ITIONS.’,5*' '/ 0000898 DATA MSG9/'- LISTS'u'THE FILE'y'S HHICH'1'HAVE BEE'o'N USED A'o 0000899 I'S HELL A"'S THE CO'.‘NDITIDNS','AND LOGI'.‘C ENTERE'.'D. '/ 0000900 DATA MSGIOI" PRINTS',‘ ALL ITE'y'MS PRESE','NT FOR E'9'ACH RECO',0000901 I'RD IN A','SELECTED',' FILE.','HA[TS AF','TER EACH',‘ N LINES'I -0000902 DATA MSGII/'- PROVID'.'ES FOR T','HE COMPU'9'TATI0N 0'9'F FUNCTI',0000903 I'ONS ON I"'TEMS IN','A DATA','SET (OR','FILE).'9' '/ 0000904 TYPE 10, (HUROS‘JIy‘TEXT‘IvJIvI=1v11I9J=1vllI 0000905 RETURN 0000906 10 FORMAT ('OTHE COMMANDS WHICH MAY BE ISSUED 'y'lAND THEIR MEANING) 0000907 lARE LISTED BELOW:'/(' '9A4,8A8/7X,3A8)) 0000908 END 0000909 GRASP SOFTWARE FUNCTION NAME: ICONV Purpose: ICONV is used to convert a number from unpacked- character form to numeric fixed-point form. Calling IVAL : ICONV (TANK,LNGTH,EXP, ERR) Arguments: TANK—Contains the number to be converted in unpacked- character form. LNGTH—The number of elements in TANK. EXP—The power of 10 to which the value returned must be raised to obtain the floating point value represented. ERR—An error flag which is turned on if an error is de- tected. Subroutines called: None sequence: G R A S P S O U R C E INTEGER FUNCTIONICONV(TANKyLNGTH.EXPyERRI 1I9DIGITSIIOIyBLANKyCOMMAyPOINTgEXP,VALUE INTEGER TANKI LOGICAL ERR DATA BLANK9COMMA9POINT/4H ,4Ho 14H2 .4H3 ,4H4 ,4H5 ,4H6 ERR=.FALSE. EXP=0 IF (LNGTH.GT.OI GO TO 10 ICONV=BLANK GO TO 90 VALUE=0 DO 20 K=1.LNGTH IF (TANKIK).NE.BLANK) GO TO 30 CONTINUE GO TO 80 00 70 J=K'LNGTH NEXT=TANKIJI IF INEXT.E0.COMMA) GO TO 70 IF (NEXT.EQ.8LANK) GO TO 80 IF (NEXT.NE.POINT) GO TO 40 EXP=1 GO TO 70 IF (EXP.GT.0) EXP=EXP+1 DO 50 L=1910 IF (NEXT.EQ.DIGITSIL)) GO TO 60 CONTINUE TYPE 1009 VALUEvEXPvITANKILI9L=IvL ERR=.TRUE. GO TO 90 VALUE=10*VALUE+L-1 CONTINUE EXP=MINO(0,1-EXPI ICONV=VALUE RETURN FORMAT END 10 20 30 40‘ 50 60 70 80 90 100 (15H BAD CHARACTER ,I11,15' SPECIFICATIONS 47 Common data referenced: None Cdeby:0PREP,PARSE,RELEXP Error checking and reporting: Each character is checked. If an invalid character is detected, an error message is typed, and the error flag is turned on. Program logic: If all characters are blank, blank is returned as a value. Leading blanks and imbedded commas are ignored, and a blank acts as a string deliminator. If a decimal point is sensed, the position counter EXP is initial- ized to 1 and incremented for each subsequent digit. As each digit is detected in a left-to-right fashion, the value is shifted left one digit, and the detected digit is added to the least significant part of the value. If all (LNGTH) char- acters have been scanned or if a trailing blank is detected, the scan is terminated and control is returned to the caller. PROGRAM 0000910 0000911 0000912 0000913 0000914 0000915 0000916 0000917 0000918 0000919 0000920 0000921 0000922 0000923 0000924 0000925 0000926 0000927 0000928 0000929 0000930 0000931 0000932 0000933 0000934 0000935 0000936 0000937 0000938 0000939 0000940 0000941 0000942 0000943 0000944 0000945 ,4H. ,4H7 lyDIGITS/4H0 ,4H8 '4H9 .4H1 / NGTH) 3X964A1. 48 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) SUBROUTIN E NAME: IF ILE Purpose: This subroutine is used to associate dynamically the name of an existing data set with a FORTRAN input unit number (that is, logical device number). When this subroutine is called, the file is opened and read; state- ments referencing the unit number given in the argument list are directed to the named file. The file may be closed by use of a rewind statement. Calling sequence: CALL IFILE (I,NAME) Arguments: I—An integer variable or constant specifying a logical de- vice number. NAME—Either a literal (hollerith) constant or a variable containing a file name consisting of five or fewer char- acters. Subroutines called: None Common data referenced: None Called by: BINTYP, DEFLST, FINDGP, OBEY, START, FILES Error checking and reporting: None Program logic: This is a DEC 1070, TOPS—10 system resi- dent routine. It provides the capabilities referred to in the Purpose section above. If the GRASP system is to be im- plemented on some other main frame, a comparable routine must be written or acquired. No listing is shown here. GRASP SOFTWARE SPECIFICATIONS SUBROUTINE NAME: INIT Purpose: INIT initializes a set of words and codes for future table lockup. The initialization assumes that the “linear quotient hash code” technique will be used for table lookup. See Bell and Kaman (1970) for a complete description of the technique. Calling sequence: CALL INIT(CHARS,CODES,M,SYMBOL, N) Arguments: CHARS—Table which is to contain the symbols in “hash- coded” order. CODES—Table giving the index of the symbol as it was stored in CHARS. M—Table size for CHARS and CODES. Note this must be a prime number. GRASP SOURCE SUBRDUTINE [NIT (CHARS,CODES, M,SYMBOL,N) INTEGER FlLLERyCHARSl l ) yCODESl ll 'SYMBOLl 1) DATA FILLER/'VUID'I DO 10 [=19M CHARS(I)=FILLER DO 40 ICODE=17N ICHAR=SYMBDL(ICODE) L=IABS(ICHAR) J=L/M l=L-M*J IF (MOD(J1M).EQ.0) J=1 IF (CHARS(I+1).EQ.FILLER) GO TO 30 I=MODlI+JyMl GO TO 20 CHARS(I+1)=ICHAR CODESlI+1l=ICODE RETURN END 10 20 30 40 49 SYMBOL—Table of words to “hash” into CHARS. N—Number of words in SYMBOL. Subroutines called: None Common data referenced: None Called by: LOGEXP, PARSE Error checking and reporting: None Program logic: 1. CHARS is filled with “VOID,” the flag for an empty loca- tion. 2. Each element of SYMBOL is then inserted into CHARS at the “hashed” address. “Collisions” are handled via the linear-quotient method. 3. As each element of SYMBOL is “bashed” into CHARS, its position is stored in CODES. 4. The initial probe address and collision displacement are de- termined from the contents of each element of SYMBOL. PROGRAM 0000946 0000947 0000948 0000949 0000950 0000951 0000952 0000953 0000954 0000955 0000956 0000957 0000958 0000959 0000960 0000961 0000962 0000963 50 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) SUBROUTIN E NAME: KEYBRD N——N umber of characters to be read. Subroutines called: None Purpose: KEYBRD accepts all user input in unpacked char- Common data referenced: None acter form and returns it to the caller. Calling sequence: CALL KEYBRD(&m,IMAGE,N) Called by: BDEF, COLPNT, CONDTN, DEFLST, FTNC, Arguments: LIST, LOGEXP, OPREP, QUIT, RETRVE, VLIST, Ina—Statement number (in caller) to which a branch is FILES: DRIVER, OBEY, PAUSE made if an EOF is found. Error checking and reporting: None IMAGE—Contains the user input in unpacked character Program logic: KEYBRD accepts input from the user (unit 5‘) form. and takes nonstandard return if EOF occurs. GRASP SOURCE PROGRAM SUBROUTINE KEYBRD(*,IMAGE1N) 0000964 DATA IX/' '/ 0000965 DIMENSION IMAGE(1) 0000966 READ(51209END=IC) (IMAGE(I)yI=1yN) 0000967 IF(IMAGE(1).EQ.IX) GO TO 10 0000968 RETURN 0000969 10 RETURN 1 0000970 20 FORMATl80Al) 0000971 END 0000972 GRASP SOFTWARE SPECIFICATIONS 51 SUBROUTINE NAME: LENGTH Purpose: LENGTH determines the number of leading non- blank words in a character string. Calling sequence: CALL LENGTH (VECT,N,L) Arguments: VECT—Array containing the character string to be ex- amined. GRASP SOURCE SUBROUTINE LENGTHiVECTvaL) INTEGER VECT(1) DATA IBLNKI' '/ L=0 DO 10 [=1 vN IF(VECT(I).EQ.IBLNK) GO TO 20 10 L=L+1 20 RETURN END N—The number of words to check. L—-The number of nonblank leading words. Subroutines called: None Common data referenced: None Called by: DEFLST, DUMPIT, NAME Error checking and reporting: None Program logic: The first full—word blank is searched for; its position is returned in L. PROGRAM 0000973 0000974 0000975 0000976 0000977 0000978 0000979 0000980 0000981 52 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) SUBROUTINE NAME: LIST Purpose: LIST is used as a driver for the listing of items from a selected file. LIST performs initialization common to both row and column forms of printout, and rewinds the input file after returning from the routine which created the printout. Calling sequence: CALL LIST(&n) Argument: n—Statement (in caller) that is branched to if an EOF is sensed in any of the called routines. Subroutines called: OPREP, KEYBRD, ROWPNT, COLPNT Common data referenced: None GRASP SOURCE SUBROUTINE LIST(*) COMMON /IOUNIT/ NFILE,IOFILE DATA IC.IR/'C','R'/ CALL OPREPI85018459NPAGE) 10 TYPE 60 CALL KEYBRD(850,IMy1) IF (IM.EQ.IC) GO TO 30 IF (IM.EQ.IRI GO TO 20 TYPE 70 GO TO 10 20 TYPE 80 CALL ROHPNT1850.NPAGEI GO TO 40 30 TYPE 80 CALL COLPNTKESOyNPAGE) 40 REWIND NFILE 45 RETURN 50 RENIND NFILE RETURN 1 60 FORMAT (' ENTER C FOR COLUMN OR R FOR RON PRINTING: 'v$l 70 FORMAT (' YOUR REPLY HAS NOT UNDERSTODDJ) Called by: DRIVER Error checking and reporting: The user response returned by KEYBRD is checked for validity. If in error, a message is typed, and the user is prompted again. Program logic: 1. The input file name and page size are set by a call to OPREP. 2. The user is prompted for “C” (column) or “R” (row) to establish the desired output form. 3. Either ROWPNT or COLPN T is called depending on the user’s response to step 2 above. 4. The input file is rewound prior to the return to DRIVER. PROGRAM 0000982 0000983 0000984 0000985 0000986 0000987 0000988 0000989 0000990 0000991 0000992 0000993 0000994 0000995 0000996 0000997 0000998 0000999 0001000 0001001 0001002 80 FORMAT (' AT 'EACH PAUSE PRESS CR KEY TO CONTINUE. 'y'TO ABORT ENTE0001003 1R Ao'l END 0001004 0001005 GRASP SOFTWARE SPECIFICATIONS 53 SUBROUTINE NAME: LOGEXP Purpose: This routine accepts a logical expression as user in- put via a call to KEYBRD and returns the encoded Re- verse-Polish form of the expressions. The logical expression may be composed of single—letter (A—Z) operands which refer to previously entered conditions, the logical operators “and,” “or,” “not,” and the grouping symbols (,). Each of the logical operators may be denoted in two ways, as fol- lows: .AND. or *, .OR. or +, .NOT. or —. Calling sequence: CALL LOGEXP(&n,POLISH,LPS,NCOND) Arguments: n—Statement number (in calling routine) to which a branch will be made if an EOF is sensed by KEYBRD. POLISH—Contains the encoded Reverse-Polish form of a logical expression. Let n denote the value of some ele- ment of POLISH. Then 16 n626 implies reference to the nth condition entered. If 29éné31, the logical op- perators OR, AND, NOT correspond to these three values. No other values will be assumed by elements of POLISH. LPS—Gives the number of elements in POLISH. NCOND—Gives the number of conditions which have been entered by a previous call to CONDTN. Subroutines called: INIT, KEYBRD, SCAN, FIND, PACK Common data referenced: LOGIC in /INPUT/ Called by: DRIVER Error checking and reporting: The logic expression entered is checked for syntactic correctness. Following are eight error messages which may be typed: 1. LOGICAL OPERATOR NOT PRECEDED BY A ) OR A LETTER (A—Z). 2. UNBALANCED PARENTHESIS. 3. LETTER (A—Z) NOT SEPARATED BY AN OPERA- TOR. 4. UNEXPECTED LEFT PARENTHESIS OR .NOT. OP- ERATOR (—). 5. INVALID CHARACTER IN EXPRESSION. 6. UNDETERMINED SYNTAX'ERROR. CONTACT PRO- GRAMMER. '7. LOGIC EXPRESSION REFERENCES A CONDITION (A—Z) WHICH WAS NOT ENTERED. GRASP SOURCE SUBRDUT INE LOGEXP( *1 POLISH9LPS.NCON0) COMMON IINPUT/ EXPRQLDGIC DOUBLE PRECISION ERRMSG(716),EXPR(49261.LOGIC(8) INTEGER POLISH( 1), ICOLS(33) ,CHARS(4119TDP'FC, STATE ,OR,AND,STACK(1 151'IMAT12 '6) qCODES (41) 'SYMBUL133).STRING(80 )yPERIODyBLANK LOGICAL CALLED 8. OPERATOR NOT ENCLOSED WITH PERIODS. RE- ENTER LOGIC. Program logic: 1. On the first call to LOGEXP, a call to INIT is made to “hash—code” the elements of SYMBOL into CHARS. CODES is used to save the original indices. 2. A prompt message is typed and a call to KEYBRD is made to get the input string which is then packed into LOGIC. 3. After initialization of pointers and counters, a call is made to SCAN to bracket the nonblank section of STRING. 4. At this point the actual algorithm begins. Transition mat- rix parsing is used with the following transition matrix (IMAT) : A—Z +' (——- ) blank . 1. f(1)/2 f(7)/1 f(4)/1 f(7)/2 f(2)/1 f(6)/1 2. f(7)/3 f(3)/1 f(7)/4 f(5)/2 f(2)/2 f(6)/1 where f (i)/ j means “do the i’th job and set the next row value to j.” The jobs are: f(1)——Insert character code into Reverse-Polish string and test to determine if there has been a condition entered for it. f (2) ——Go scan next character. f(3)—Pop stack into POLISH until value of topmost ele- ment is less than character code. Then do f(4). f (4)—Push down character code into stack. f (5)—Pop stack into POLISH until the value for ( is reached. Remove value for (. f (6)—-Period character sensed, find next matching period and determine logical operator. f(7)—Type the error message pointed to by the row value, then request reentry of logic. Each character of STRING is scanned using the subroutine FIND to obtain its code ICODE. ICODE is an index to ICOLS which then determines the proper column of IMAT. This element IFTN is then broken down into a function pointer JOB and a next row value STATE. Control is then passed to the function indicated by JOB. After the func- tion has been completed, the next character of STRING is scanned if an error was not detected. If an error was de— tected, the appropriate message is typed. PROGRAM 0001006 0001007 0001008 0001009 0001010 0001011 DATA SYMBOL/'A' 1'8.y'C"'D'y'E'Q'F'q'G'y'H".I'y.J."K'y'L'1'M'1'N0001012 1|,I0l,'pi,IQI,IRI’ISI"TI,IUO'IVO'INI,OXI'IYI’IZI'I(I,I’I'O§O'I*I'OOGIOI3 2|-n'u I'I.I/ 0001014 DATA ICOLS/26*193,412,2939516/y CALLED/.FALSE./ 0001015 DATA IMAT/12973171931’41p74;72152¢21,22v61,61/ 0001016 DATA ERRMSG/‘LOGICAL','OPERATOR'.’ NOT PRE','CEDED BY',' A 1 OR', 0001017 1'A LETTER',‘ (A‘Z1.','UNBALANC','ED PAREN',‘THESIS.',4*' '9 0001018 Z'LETTER ('y'A-Z) NOT'.‘ SEPARAT'p'ED BY AN',‘ OPERATO'.'R. 'y 0001019 3' 'y'UNEXPECT‘9'ED LEFT',‘PARENTHE','SIS OR .‘,‘NOT. OPE'y 0001020 4'RATOR (-'v'). 'y'INVALID','CHARACTE'.'R IN EXP'y'RESSION.'g 0001021 54 30 20 40 50 60 70 90 100 110 130 140 150 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) 53*' '.'UNDETERM'.'INED SYN'.'TAX ERRO'.'R. CONTA'.'CT PROGR'.0001022 6'AMMER.',' '/,0R,AND,NOT/'0R','AND'.'NOT'I 0001023 EQUIVALENCE (SYMBOL(32)yBLANK),(SYMBOL1331,PERIODI 0001024 IF (CALLED) GO TO 20 0001025 CALLED=.TRUE. 0001026 CALL IN1T(CHARS,CDDESy41,SYM80Ly33) 000102? TYPE 250 0001028 CALL KEYBRD18245,STRING,801 0001029 CALL PACK(STRING.LOGICy80.801 0001030 TOP=1 0001031 STACK(1)=0 0001032 LPS=0 0001033 STATE=1 0001034 CALL SCAN(830.STRING,1,IPT.LNGTH,2) 0001035 IPT=IPT~1 0001036 GO TO 40 0001037 RETURN 0001038 NSTATE=STATE 0001039 IPT=IPT+1 0001040 IF (IPT.GT.LNGTH) GO TO 230 0001041 CALL FIND‘C5O!STRING(IPT)vICODE1CHARSICUDE5941’ 0001042 GO TO 60 0001043 STATE=5 0001044 60 TO 220 0001045 IFTN=IMAT(NSTATEyICOLS(ICODEI) 0001046 J08=IFTNI10 0001047 STATE=IFTN-10*J08 0001048 00 T0 l70,40,909100.110.130.220),JDB 0001049 LPS=LPS+I 0001050 POLISH(LPSD=ICO0E 0001051 'IF IICUDE.LE.NCOND) GO TO 40 0001052 TYPE 270 0001053 GO TO 20 0001054 IF (STACK(TOP).LT.ICODE) GO TO 100 0001055 LPS=LPS+1 0001056 POLISHtLPSI=STACK(TOPI 0001057 TOP=TOP-I 0001058 IF (TOP.GT.0) GO TO 90 0001059 STATE=6 0001060 50 TO 220 0001061 Top=rgp+1 0001062 STACK(TUP)=ICODE 0001063 00 TO 40 0001064 IF (STACK1TOP).EQ.27) GO TO 120 0001065 LPS=LPS+I 0001066 POLISH(LPS)=STACK(TOP) 0001067 TOP=TOP-1 0001068 IF (TOP.GT.0) GO TO 110 0001069 STATE:2 0001070 00 T0 220 0001071 TOP=TOP-1 0001072 IF (TOP.GT.0) GO TO 40 0001073 STATE=2 0001074 GO TO 220 0001075 FC=IPT+1 0001076 D0 14c I=FCy80 0001077 IF (STRING(I).NE.BLANK) GO TO 160 0001078 CONTINUE 0001079 TYPE 280 0001080 GO TO 20 0001081 160 170 180 190 200 210 220 230 240 245 250 270 280 300 GRASP SOFTWARE SPECIFICATIONS FC=I DD 170 IPT=FCv80 IF (STRINGIIPT).EQ.PERIOD) GO TO 180 CONTINUE GO TO 150 NCHAR=3 ICODE=0 NCHAR=NCHAR—1 NCH=FC+NCHAR IOP=CHARS(21) CALL PACK(STRINGIFC),IOP,NCH-FC+1:4I IF (IOP.EQ.AND) ICODE=30 IF (IOP.E0.NOTI ICODE=31 IF (IOP.E0.0R) ICODE=29 IF (ICODE.NE.01 GO TO 200 IF (NCHAR.GT.1) GO T0 190 STATE=6 GO TO 220 IF (ICODE.LT.31) GO TO 210 IF (NSTATE.EQ.1) GO TO 100 STATE=4 GO TO 220 IF (NSTATE.EQ.21 GO TO 90 STATE=1 TYPE 3001 IERRMSGII,STATEI9I=1v7) GO TO 20 IF ITOP.EQ.1) GO TO 30 IF (STACKITOP).GT.28) GO TO 240 STATE=2 GO TO 220 LPS=LPS+1 POLISH(LPS)=STACK(TOPI TOP=TUP—1 GO TO 230 RETURN 1 FORMAT (' ENTER LOGIC: '1S) FORMAT (' LOGIC EXPRESSION REFERENCES A CONDITION IA-ZI'I' 1A5 NOT ENTERED-’1 55 0001082 0001083 0001084 0001085 0001086 0001087 0001088 0001089 0001090 0001091 0001092 0001093 0001094 0001095 0001096 0001097 0001098 0001099 0001100 0001101 0001102 0001103 0001104 0001105 0001106 0001107 0001108 0001109 0001110 0001111 0001112 0001113 0001114 0001115 0001116 0001117 WHICH H0001118 0001119 FORMAT (' OPERATOR NOT ENCLOSED WITH PERIODS. RE-ENTER LOGIC.‘) 0001120 FORMAT (' LOGICAL ERROR:'/1X,7A8) END 0001121 0001122 56 GEOLOGIC RETRIEVAL AND SUBROUTINE NAME: MEAN Purpose: MEAN provides for the computation of range, mean, sum, root mean square, and sum of squares for as many as five specified items in a specified file. Calling sequence: CALL MEAN (J,ISWTCH) Arguments: J—Pointer used to retrieve argument values from the com— mon area /FTNCOM/. ISWTCH—Switch indicating which of three parts (initiali- zation, body, postprocessing) of the code is to be executed. Subroutine called: UNCODE Common data referenced: ITYPE in blank common TAGS, IREC, ARGS, NARGS in /FTNCOM/ SYNPOSIS PROGRAM (GRASP) Called by: FDRIVE Error checking and reporting: None Program logic: The value of ISWTCH determines which of three sections of the code is executed If ISWTCHzl, sums and range values are initialized. If ISWTCH:2, the type for each argument value is deter- minted and its value is added to the appropriate sums. Range values are updated if required. If ISWTCH23, the final computations are performed and the results typed out to the user. The mean is determined using 3?:EX/N and the root mean square is determined using RMS:EX2/N, where N is the num- ber of nonblank values of X. G R A S P S O U R C E P R O G R A M SUBROUTINE MEAN(J.ISWTCH) 0001123 COMMON NAMES,ITYPE:IPTS:IDIM 0001124 COMMON /FTNCOM/ TAGS.IREC,APGS,NARGS,IFTN,NFTN 0001125 DIMENSION ARGS(6,5), NARGSlS), IFTN15), ITYPE(5001. IRECl5001y 0001126 IIPT51500)9NSUM(S,V SUMX(5)' SUMXS(5). VMAX(5), VMINIS) 0001127 DOUBLE PRECISION NAMESI50019TAGS(515) 0001128 INTEGER ARGS 0001129 EOUIVALENCE (IVALvVAL) 0001130 DATA IBLNKI' '/ 0001131 IFIISWTCH-Z) 5,15,25 0001132 5 K=NARGSIJD 0001133 00 10 I=1.K 0001134 SUMXII)=0. 0001135 VMAX(Il=-1.E30 0001136 NSUMIIl=O 0001137 VM[N(I)=1.E30 0001138 10 SUMXSlIl=O. 0001139 GO TO 55 0001140 15 K=NARGS(J) 0001141 00 20 [=19K 0001142 IVAL=IREC(ARGS(I+1:J1) 0001143 [F (IVAL.EQ.IBLNK) GO TO 20 0001144 NSUM(I)=NSUM(I)+1 0001145 VALUE=IVAL 0001146 IFIITYPE(ARGS(I+1le).EQ.5) VALUE=UNCOOE1VALgIQ) 0001147 IF (ITYPE(ARGS(1+1,J)).EQ.2) VALUE=VAL 0001148 [F (VALUE.LT.VMIN(I)’ VMIN(I)=VALUE 0001149 IF (VALUE.GT.VMAX(I)) VMAX(I)=VALUE 0001150 SUMX(11=SUMX(I)+VALUE 0001151 SUMXS(I)=SUMXS(Il+VALUE*VALUE 0001152 20 CONTINUE 0001153 GO TO 55 0001154 25 K=NARGS(J) 0001155 DO 50 I=loK 0001156 [F (NSUMII).EQ.O) GO TO 30 0001157 TYPE 60, TAGS‘IyJ),NSUM(I) 0001158 AMEAN=SUMXIIllNSUMIIJ 0001159 RMS=SUMXS(I)/NSUM(I) 0001160 [F (ITYPEIARGSlI+1yJ)).E0.21 GO TO 40 0001161 1F (ITYPE1ARGS‘I*19J)).EQ.51 GO TO 40 0001162 GRASP SOFTWARE SPECIFICATIONS 57 M1N=VMIN(I) 0001163 MAX=VMAX(I) 0001164 TYPE 80, MIN'MAXpAMEANvRM595UMXTIIoSUMXS(I) 0001165 GO TO 50 0001166 TYPE 70; TAGS(I,J1 0001167 GO TO 50 0001168 TYPE 80, VMIN(1),VMAX(I1,AMEAN,RMS,SUMX(ITySUMXSTI) 0001169 CONTINUE 0001170 RETURN 0001171 FORMAT (I' MEAN STATISTICS FOR 'yABy'WITH'yIéy' ITEM(ST.') 0001172 FORMAT (l' NO VALUES PRESENT FOR 'yA8) 0001173 FORMAT (' MIN=',1PG9.2.' MAX=',G9.2,' MEAN=',G9,2.' ROOT MEAN 0001174 ISQ.=',09.2/' SUM='.GIZ.5,’ SUM OF SQUARES=‘,GIZ.5) 0001175 END 0001176 58 SUBROUTINE NAME: NAME Purpose: NAME provides the user with the mechanism for examining the structure and content of the current data base. The user is permitted to select categories of interest. Item names, types, and descriptions for all entries in selected categories are printed. The user is also permitted to see the values which are assumed (in the current data base) by character- and multiple-choice-type items. Calling sequence: CALL NAME (&n) Argument: n~Statement number (in caller) to which a branch is made if the second nonstandard return is taken from DEFLST (EOF sensed by KEYBRD), or EOF is sensed in BDEF. Subroutines called: DEFLST, PAUSE, FINGDP, LENGTH, BDEF Common data referenced: None GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) Called by: DRIVER Error checking and reporting: None Program logic: 1. DEFLST is called to select those categories of interest (LIST). 2. For each category selected, a call is made to FINDGP to obtain the names, types, and descriptions for all items within that category. They are then printed under the cate- gory name. 3. Programmed pauses after each category or 30 lines of out- put are provided via calls to PAUSE. 4. A call to LENGTH is made to determine the number of nonblank words in the description. 5. Unit 20 (used by DEFLST and FINDGP) is rewound prior to returning to DRIVER. G R A S P S O U R C E P R O G R A M SUBROUTINE NAME(*) 0001177 INTEGER CAT(87171yDESC112'4511TYPEI45)'LIST(171 0001178 DOUBLE PRECISION NAMES(45) 0001179 CALL DEFLST(8509£601CATyNUMCoMCyLISTl 0001180 TYPE 70 0001181 CALL PAUSElESO) 0001182 DO 30 K=11NUMC 0001183 KNUM=LIST(K) 0001184 CALL FINDGP(850,KNUM,NUM,MAXL’NG.DESC) 0001185 READ (20) lNAMESlJl'TYPEIle(DESC(lleyI=1yMAXL19J=19NUMD 0001186 CALL LENGTHICAT‘lyKNUM)oMCyMCL’ 0001187 TYPE 90. (CATlI'KNUMl1I=1vMCL) 0001188 TYPE 91 0001189 LINE=0 0001190 10 LINE=LINE+1 0001191 [F (LINE.GT.NUM) GO TO 30 0001192 1F (MODILINE,30).NE.0) GO TO 20 0001193 CALL PAUSEISSO) 0001194 TYPE 90! (CATlviNUMlql=1.MCL) 0001195 TYPE 100 0001196 TYPE 91 0001197 20 CALL LENGTHIDESCI11LINE),MAXL.MXL) 0001198 TYPE 110' NAMESILINEinYPElLINE),(DESCII,LINE).I=1,MXL) 0001199 00 TO 10 0001200 30 CALL PAUSE1850) 0001201 50 RENIND 20 0001202 CALL BDEF(860) 0001203 RETURN 0001204 60 RETURN 1 0001205 70 FORMAT (' IN EACH CATEGORY, 1CRIPT IONS WILL BE'/' LISTED. THE ITEM NAMESv TYPE CODES:'/9X1'I = TYPE CODES,'y' AND 0E50001206 WHOLE NUMBERS'I90001207 2X,'R = NUMBERS WITH FRACTIONAL PARTS'l9X9'A = ALPHANUMERIC STRINGSOOOlZOB 3'I9Xv'B = MULTIPLE CHOICE TYPES'I9X9'Q = QUALIFIED NUMERIC VALUES'0001209 4 /‘ AT EACH PAUSE STRIKE CR KEY TO CONTINUE (STARTING NOH1.'1 0001210 9C FORMAT(' CATEGORY: ',9A5) 0001211 91 FORMAT(' NAME TYPE DESCRIPTION'/' “"'1 0001212 1 ' ---- ---------- '1 0001213 100 FORMAT('+'1T50,'(CON"T)'l 0001214 110 FORMAT (1X,A7.1XvA102Xy12A5) 0001215 END 0001216 GRASP SOFTWARE SUBROUTINE NAME: OBEY Purpose: OBEY associates input or output file names with FORTRAN unit numbers and provides a degree of file- name checking and protection. Calling sequence: CALL OBEY(&m,MSG,N) Arguments: m—Statement (in caller) to which a branch will be made if a protected (output) or unknown (input) file name is referenced. MSG—Contains (in packed—character form) either of the following: “EQUATE 11 name,” indicating input, “EQUATE 12 name,” indicating output. N—The number of words in MSG. Subroutines called: KEYBRD, IFILE, OFILE Common data referenced: IN, IOUT in /IOUNIT/ FNAMES, NUMI, WHICH in /FILNAM/ Called by: RETRVE, OPREP, FTNC Error checking and reporting: 1. Input file names are checked for recognition. 2. Output file names are checked to prevent writing on a “protected” file. GRASP SUBROUTINE OBEYl‘hMSGvN) COMMON lIOUNIT/ IN,IOUT COMMON IFILNAM/ FNAMES,NUMI'HHICH,PAD INTEGER MASTER,FNAME.FILES(20)9MSG(llvIMSGl‘t)oFNAMESlZOlyNHICH 1 ,PADl4l EQUIVALENCE (”450(4) ,FNAME) DATA 129NUMF,IEQ/'2'y0y 'EQUA'quYES/'Y'/ IF(N.NE.4) GO TO 100 IF(MSG(1).NE.IEQI GO TO 100 MASTER=FNAMESlWHICHl DO 2 I=1.4 2 IMSG(I)=MSGlll IFlIMSGl31.EQ.IZI GO TO 10 lN=23 IFlFNAME.EQ-MASTER1 GO TO 6 1FlNUMF.GT.0) GO TO 4 3 TYPE 500 CALL KEYBRDlSlOOnyll IF(I.EQ.lYESl GO TO 6 8 RETURN 1 4 DO 5 I=1oNUMF IFlFILESlI|.EQ.FNAME1 5 CONTINUE GO TO 3 6 CALL IFILElIN.FNAME1 60 T3 100 10 DO 11 J=1y5 K=4*(J-11 DO 11 [=19NUMI IF(FNAMES(K+I).EQ.FNAME) GO TO 14 11 CONTINUE GO TO 12 60 T06 SOURCE P SPECIFICATIONS 59 3. The total number of output files is checked against the maximum 20. Error messages for each of the above three checks are provided. Program logic: 1.. If the message length N is not 41 or if the first 4 char- acters in MSG are not EQUA, return is immediate. 2. MASTER is set to the current data-base name in /FILNAM/ and MSG is moved to IMSG. 3. If the third word of IMSG is not “2,” input is assumed, and the file name (last word of IMSG) is checked against the names of files created during this session. If no match is found, the user is informed and given the opportunity to exit and enter a new command. Finally, IFILE is called to associate unit 23 with the file name provided. 4. If the third word of IMSG is “2',” output is assumed, and the file name FNAME (last word in IMSG) is checked against the list of protected file names in /FILNAM/. If a match is found, the nonstandard return is taken. Otherwise, FNAME is added to the list FILES, and unit 24 is associ- ated with the file FNAME via a call to OFILE. ROGRAM 0001217 0001218 0001219 0001220 0001221 0001222 0001223 0001224 0001225 0001226 0001227 0001228 0001229 0001230 0001231 0001232 0001233 0001234 0001235 0001236 0001237 0001238 0001239 0001240 0001241 0001242 0001243 0001244 0001245 0001246 0001247 0001248 60 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) 14 TYPE 5011FNAME 0001249 GO TO 8 0001250 12 IF(NUMF.EQ.0) GO TO 20 0001251 00 13 I=1yNUMF 0001252 IF(FILES(I).EQ.FNAME) GO TO 21 0001253 13 CONTINUE 0001254 20 NUMF=NUMF+1 0001255 1F(NUMF.LE.20) GO TO 22 0001256 NUMF=20 0001257 TYPE 23 0001258 GO TO 8 0001259 22 FILES(NUMF)=FNAME 0001260 21 IDUT=24 0001261 CALL OFILE(IOUT,FNAME1 0001262 100 RETURN 0001263 23 FORMAT(' NO MORE THAN 20 FILES MAY BE CREATED IN ONE RUN.') 0001264 500 FORMATT' ATTEMPT TO REFERENCE A FILE NOT CREATED THIS RUN.'/ 0001265 1 ' DO YOU STILL WANT IT? (Y OR N): 'v$) 0001266 501 FORMAT‘IX'Aby'MAY NOT BE USED AS AN UUTPUT FILE NAME') 0001267 END 0001268 GRASP SOFTWARE SPECIFICATIONS 61 SUBROUTINE NAME: OFILE Purpose: This subroutine is used to associate dynamically the name of a new data set With a FORTRAN output unit num- ber (that is, logical device number). When this subroutine is called, the file is opened, and write statements referenc- ing the unit number given in the argument list are directed to the named file. The file may be closed by use of a rewind statement. Calling sequence: CALL OFILE (I,NAME) Arguments : I—An integer variable or constant specifying a logical de- vice number. NAME—Either a literal (hollerith) constant or variable containing a. file name consisting of five or fewer char- acters. Subroutines called: None Common data referenced: None Called by: COLPNT, OBEY Error checking and reporting: None Program logic: This routine is a DEC 1070, TOPS—10 sys- tem resident routine. It provides the capabilities referred to in the section “Purpose” above. If the GRASP system is to be implemented on some other main frame, a comparable routine must be written or acquired. Therefore, a listing has not been provided. 62 GEOLOGIC RETRIEVAL AND SYN’POSIS PROGRAM (GRASP) SUBROUTINE NAME: OPREP Purpose: This routine is used to prompt for and accept the name of a file and a page size. Calling sequence: CALL OPREP(&n,&m,NPAGE) Arguments: n—Statement number (in calling routine) to which a branch Will be made if an EOF is sensed by KEYBRD. m—Statement number (in calling routine) to which a branch will be made if the nonstandard return from OBEY is taken. NPAGE—Page size (in lines) as entered by user. Subroutines called: KEYBRD, OBEY, ICONV, PACK Common data referenced: NAMES, WHICH in /FILNAM/ G R A S P S O U R C E SUBROUTINE OPREP(*2*,NPAGEI COMMON /FILNAM/ FNAMES,WHICH,PAD INTEGER DBLNK,DFAULTyFNAMEyEQUATEl4)yTANKlSl,FNAMES(21IuWHICH 1 .PAD(4) LOGICAL BAD EQUIVALENCE (EOUATEl4lvFNAMEl DATA DBLNKyEQUATEl' TYPE 20 DFAULT=FNAMESlNHICHl CALL KEYBRD1815VTANK15) CALL PACK(TANK,FNAME,5,5) IFlFNAME.EQ.DBLNK) FNAME=DFAULT CALL OBEY(&18,EQUATE94) 10 TYPE 30 CALL KEYBRDlSlSyTANKyS) NPAGE=ICONVlTANK951lvBADl IF (BAD) GO TO 10 IF (NPAGE.EQ.0) NPAGE=10000000 RETURN 15 RETURN 1 18 RETURN 2 20 FORMAT 1' 3O FORMAT (' END ENTER NAME OF FILE: '.'EQUA','TE 1'7'1',' '/ ',$l ENTER NUMBER OF LINES/PAGE: Called by: DUMPIT, LIST Error checking and reporting: An error flag set by ICONV is tested. If set, the user is requested to reenter the value. Program logic: 1. A file name is prompted for and accepted (via KEYBRD). This name is then packed into FNAME and compared with blank. 2. If blank, FNAME is set to the default file name obtained in /FILNAM/. 3. FNAME is then passed to OBEY via EQUATE. 4. A page size is then prompted for and accepted in char- acter form. The numeric value is obtained by a reference to ICONV. If zero, then NPAGE is set to 10 million. PROGRAM 0001269 0001270 0001271 0001272 0001273 0001274 0001275 0001276 0001277 0001278 0001279 0001280 0001281 0001282 0001283 0001284 0001285 0001286 0001287 0001288 0001289 0001290 0001291 0001292 "$1 GRASP SOFTWARE SPECIFICATIONS 63 SUBROUTINE NAME: PACK Purpose: All user input to the GRASP system is in unpacked form (in other words, one single left-justified character per word). PACK is used to convert from this unpacked form to packed form. This is necessary because character data in files accessed by GRASP is in packed form to conserve space. Calling sequence: CALL PACK(SOURCE,DESTN,N,SIZE) Arguments: SOURCE—The array containing the unpacked character string. DESTN—The array which is to contain the packed char— acter string. GRASP SOURCE SUBROUTINE PACK(SOURCE,DESTN'N,SIZE) INTEGER SOURCE(1MDESTN(1)9$IZE ENCODE(SIZE9 11 DESTN) 1 FORMAT( 80A1) RETURN END (SOURCE(I)1I=19N) N—The number of characters to pack. SIZE—The size (in characters) of the area to receive the packed data. Subroutines called: None Common data referenced: None Called by: DRIVER, COLPNT, CONDTN,DECOMP, FILES, FTNC, LOGEXP, OPREP, PARSE, PNTER, RELEXP, RETRVE, VLIST Error checking and reporting: None Program logic: The ENCODE [statement is used to move the characters from the unpacked string (SOURCE) to the packed string (DESTN). PROGRAM 0001293 0001294 0001295 0001296 0001297 0001298 64 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) SUBROUTINE NAME: PARSE Purpose: PARSE converts arithmetic expressions to an en- coded Reverse-Polish form. Extensive syntax checking, con— version, and preliminary addressing are performed to facili- tate later evaluation by EVAL. The arithmetic expressions may contain the usual arithmetic operators (+,~,*,/), numeric constants, item names, and the following functions: ABS ( )——absolute Value; SQRT ( )—square root; LOG ( )—log base 10‘; SQR ( )—square; TEN ( )—power of 10. Parentheses may be used for grouping to control the order of evaluation. Calling sequence: CALL PARSE(EXPR,L,TYPE,POLISH, I,ERR) Arguments: EXPR—Arithmetic expression to be parsed in unpacked character form. L—The length of EXPR. TYPE, POLISH—Arrays which will contain the encoded Reverse-Polish form. See section on subroutine EVAL for additional encoding information. I—Length of TYPE and POLISH. ERR—Logical flag set if an error is detected. Subroutines called: INIT, FIND, BFIND, INCONV, PACK Common data referenced: NAMES, IPNTS, IDIM in blank common. Called by: PREVAL Error checking and reporting: The expression is checked for normal FORTRAN—like syntax (such as balanced paren— theses, binary operators bracketed by valid names or expres- sions, and correct spelling of function names). The message “ERROR IN EXPRESSION” is typed if an error is detected. If an operand or function name is not recognized, that message is typed. Program logic: The logical variable CALLED is tested. If it has not been set by a previous call, it is set to .TRUE. and INIT is called to “hash code” the elements of SYMBOL GRASP SOURCE SUBROUT INE PARSEIEXPRngTYPE, POL ISH,I,ERR) COMMON NAME S.ITYPE,IPNTS.IDIM into CHARS and CODES. Next, the variables ERR, ROW, TOP, I, and C are initialized. The remainder of the com- putation involves scanning EXPR, an element at a time. As in LOGEXP, a transition-matrix technique is used to parse the expression, converting it to reverse-Polish form. The transition matrix (TM) is given below: Ag; — + / * ( ) f(1)/2 f(8)/1 error error error f(10)/1 error . f(3)/2 f(4)/1 f(5)/I f(6)/1 f(7)/1 f(12)/1 f(9)/2 f(2)/2 f( 3)/2 error f(4)/1f(5)/I f(6)/1 f(7)/1 error f(9)/3 f(2)/3 f( 3)/3 where the f(i) are separate tasks as follows: f(1)—Start a name. f(2)—Go scan next character. f(3)—Append current character to name. f(4-7)—Binary arithmetic operator sensed: Set CODE to indicated operator; pop stack until CODE is less than the topmost stack element; push CODE down on stack. f (8)—Unary minus sensed: set CODE and push down on stack. f(9)—Right parenthesis sensed: pop stack until topmost element is code for left parenthesis (PAREN); decrease size of stack by one. f(10‘)—Left parenthesis sensed; push down parenthesis code PAREN. f (11)—Digit or period sensed: start a constant. NAME is used to contain the constant in character form. f(12)—Left parenthesis sensed in row 2: hence, the contents of NAME are assumed to be a function name. Check for validity and print an error message if invalid; otherwise, set CODE and push down on stack. The proper element of TM is selected by the variables ROW and COLUMN. The COLUMN value is determined by a lookup (via FIND) of the current character and the ROW value is set by the last element of TM referenced. Once the proper element of TM is selected, the next ROW value is set and a branch is made to the current task. This process is repeated until all elements of EXPR have been processed. See the section on subroutine EVAL for details of the encoding of TYPE and POLISH. blank 0—9 f(2)/1 f(11)/3 mug—I PROGRAM 0001299 0001300 DIMENSION NAMESISOO), PDLISH(1)9 STACKIQIIy EXPRII), NAME(81: TMI30001301 1’9), SYMBOLUtSIv COLSI45), TYPEII). ITYPE(500)7 IPNTS(50011 IFNCT50001302 2(5). CHARSI 471, CDDESI47) 0001303 DOUBLE PRECISION NAMES,VARBLE,DBLNK 0001304 INTEGER TMyTOPyROW yCOLUMN, ELEMNTySWITCHyTYPEo Cy COLSoSYMBOL. CHARS, C0001305 IODES,CHAR,EXPR LOGICAL POP,NUMy ERR. CALLED 0001306 0001307 DATA TM/12132y0981 ,2*4I,0,2*51.0.2*61,012*719101v121v2*0992y93v21c0001308 122v23v 113v 32v 33/ 0001309 DATA SYMBOL/IA.’OBC'ICI’IDO,IEI'IFI,CGI,'HI"II'IJI'IK"'LI,IMI'IN0001310 II'IOI'IPI'IQI'IRI,ISC,'TI'IUI’IV0,0HI’IXD’OYI'IZU’IDU’III’IZI'I30'0001311 2I4I’05"l69'07l,l8i"9"l.I,I I’l§I,I_I'O*I'l/l'l(l’0)I,I2I/ 0001312 DATA COLS/26*1,11*9,8,392759‘n6,7y 1/ 0001313 DATA IBLNKyCALLEDgDBLNKl' '9.FALSE..' '/ 0001314 DATA PARENQPLUS'DIFFQPRDDQDIV1UNARY/Oog‘loy‘2."309—4.1-5., 0001315 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 GRASP SOFTWARE SPECIFICATIONS DATA IFNCTS/‘ABS','SORT'y'LDG‘y'SQR‘,'TEN'/ IF (CALLED) GO TO 20 CALLED=.TRUE. CALL INITICHARS,CODES,47,SYMBOLy45) ERR=.FALSE. ROW=1 TOP=0 I=0 C=O C=C+1 IF (C.LE.LI GO TO 50 IF (ROW.EQ-13 GO TO 350 IF (POP) GO TO 250 IF (NUM) GO TO 40 SHITCH=1 GO TO 300 SHITCH=2 GO TO 320 CHAR=EXPR(CI CALL FINDI8350,CHAR.COLUMN,CHARSpCODESy47I COLUMN=COLSICOLUMN) ELEMNT=TMIRONyCOLUMNI JOB=ELEMNT/10 ROW=ELEMNT-10*JOB GO TO (60030970v130914011509160117001801230’2409270Ip JOB GO TO 350 NAMEI11=CHAR NCHAR=1 POP=.FALSE. NUM=.FALSE. GO TO 30 NCHAR=NCHAR+1 NAME(NCHAR)=CHAR GO T0 30 IF (POP) 60 TO 100 POP=.TRUE. IF (NUM) GO TO 90 SHITCH=3 GO TO 300 SHITCH=4 GO TO 320 IF (TOP-EQ.OI GO TO 120 IF (CODE.LT.STACK(TOPI) GO TO 120 SWlTCH=5 VALUE=STACKITOPI INDEX=VALUE GO TO 330 TOP=TDP~1 GO TO 100 TOP=TOP+1 STACKITOPI=CODE GO T0 30 CODE=DIFF 00 TD 80 CODE=PLUS GO TO 80 CODE=DIV GO TO 80 CODE=PROD 65 0001316 0001317 0001318 0001319 0001320 0001321 0001322 0001323 0001324 0001325 0001326 0001327 0001328 0001329 0001330 0001331 0001332 0001333 0001334 0001335 0001336 0001337 0001338 0001339 0001340 0001341 0001342 0001343 0001344 0001345 0001346 0001347 0001348 0001349 0001350 0001351 0001352 0001353 0001354 0001355 0001356 0001357 0001358 0001359 0001360 0001361 0001362 0001363 0001364 0001365 0001366 0001367 0001368 0001369 0001370 0001371 0001372 0001373 0001374 66 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) GO TO 80 CDDE=UNARY GO TO 120 IF (PUP) GO TO 200 POP=.TRUE. IF (NUM) GO TO 190 SHITCH=6 GO TO 300 SNITCH=7 GO TO 320 IF (TDP.EQ.0I GO TO 350 IF (PAREN.EQ.STACK(TOP)) GO TO 220 SWITCH=8 VALUE=STACKITOPI INDEX=VALUE GO TO 330 TOP=TOP-1 GO TO 200 TOP=TOP—1 GO TO 30 TOP=TOP+1 STACKITOPI=PAREN GO TO 30 NAMEI1I=CHAR NCHAR=1 NUM=.TRUE. POP=.FALSE. GO TO 30 IF ITOP.EQ.OI GO TO 370 SHITCH=9 VALUE=STACK(TOP) IF (VALUE.EQ.PAREN) GO TO 350 INDEX=VALUE GO TO 330 TDP=TUP-1 GO TO 250 IVAL=IBLNK IF (NCHAR.EQ.OI GO TO 350 CALL PACKINAME.IVAL,NCHAR,4I DO 280 J=1y5 IF (IVAL.EQ.IFNCTS(J)) GO TO 290 CONTINUE TYPE 410' IVAL GO TO 350 CODE=J~11 TOP=TOP+1 STACKITOPI=CODE GO TO 230 VARBLE=DBLNK CALL PACK(NAME.VARBLE,NCHAR18) CALL BFIND(5310.VARBLE.INDEX.NAMES.IPNTS.IDIM) GO TO 330 TYPE 390, VARBLE GO TO 360 VALUE=ICONVINAME,NCHAR,J,ERR) IF (ERR) GO TO 350 INDEX=O IF IJ.NE.OI VALUE=VALUF*10.**J 0001375 0001376 0001377 0001378 0001379 0001380 0001381 0001382 0001383 0001384 0001385 0001386 0001387 0001388 0001389 0001390 0001391 0001392 0001393 0001394 0001395 0001396 0001397 0001398 0001399 0001400 0001401 0001402 0001403 0001404 0001405 0001406 0001407 0001408 0001409 0301410 0001411 0001412 0001413 0001414 0001415 0001416 0001417 0001418 0001419 0001420 0001421 .0001422 0001423 0001424 0001425 0001426 0001427 0001428 0001429 0001430 0001431 0001432 330 340 350 360 370 380 39C 410 420 GRASP SOFTWARE SPECIFICATIONS I=I+1 IF (I.LE.15) GO TO 340 TYPE 420 GO TO 350 POLISHIII=VALUE TYPE(I)=INDEX GO TO (2509250110011001110,20012009210126019 SWITCH TYPE 380’ (EXPR(JDyJ=1yL) ERR=.TRUE. RETURN FORMAT (' ERROR IN EXPRESSION: '150A1/23X130Al) FORMAT (‘ UNDEFINED NAME '9A81 FORMAT (1X,A5,'IS NOT A PERMISSIBLE FUNCTION. HENCE') FORMAT (' MORE THAN 15 NAMES AND OPERATORS USED. HENCE') END 67 0001433 0001434 0001435 0001436 0001437 0001438 0001439 0001440 0001441 0001442 0001443 0001444 0001445 0001446 0001447 68 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) SUBROUTINE NAME: PAUSE the nonstandard return is taken from KEYBRD. S b t‘ ll d: KEYBRD Purpose: This routine is used to provide a system-generated u rou me ca e ' t t If b1 nk har etc i ent red b the Common data referenced: None e ll 11 . a non a C a 1' S e 32:: till; Zoitandard return is taken y Called by: BDEF, COLPNT, DUMPIT, NAME, ROWPNT Calling sequence: CALL PAUSE (&n) Error checking and reporting: None Argument: Program logic: PAUSE accepts a single (left-justified) char- n—Statement (in caller) to which a branch will be made acter from KEYBRD. If an EOF is sensed or the character if a nonblank character is returned by KEYBRD, or if is nonblank, take the nonstandard return. GRASP SOURCE PROGRAM SUBRJUTINE PAUSE(*) 0001448 DATA IBLNKl' '/9IBELL/"03400€000000/ 0001449 TYPE lyIBELL 0001450 CALL KEYBRD(810.I.1) 0001451 IF (I.EQ.IBLNK) RETURN 0001452 10 RETURN 1 0001453 1 FDRMATlleAll 0001454 END 0001455 GRASP SOFTWARE SPECIFICATIONS FUNCTION NAME: PNTER Purpose: PNTER is used to look up user-entered character- string or multiple-choice—type values in the value part of “conditions” statements. Lookup is performed in the appro- priate dictionary, and the value returned is a pointer to the particular dictionary item. If the value is not found, an error flag is set and zero is returned. Calling sequence: IPT = PNTER (VALUE, IDIM, NAME, ITYPE,ERR) Arguments: VALUE—Unpacked character-string value to be looked up. IDIM—Length of the string in VALUE. NAME—Item number for which the character string repre- sents a value. ITYPE—Item type of item pointed to by NAME. ERR—Error flag which is set if the value is not found in the dicitionary pointed to by NAME. Subroutines called: ACCESS, BINTYP, PACK Common data, referenced: None Called by: RELEXP GRASP SOURCE INTEGER FUNCT IONPNTER(VALUEy I DIMyNAME, ITYPEgERR) DOUBLE PRECISION LABEL(251 yBLABEL INTEGER VALUEI l l yTANKI 251 ,BITEMi 15,25) ,STRINGilZ) LOGICAL ERR EQUIVALENCE (STRINGillyflLABEL) DATA IBLNKI' '/ ERR=.FALSE. N=IBLNK [F (IDIM.EQ.0) GO TO 110 D0 10 1=1v12 STRINGIIl=IBLNK CALL PACK‘VALUEvSTRINGy101M960) DO 20 I=1y12 10 IF (STRINGI 13—1).NE.IBLNK1 GO TO 30 20 CONTINUE GO TO 110 LENGTH=l3-I IFIITYPE-31 35:35980 CALL ACCESS(NAME,K.TANK'NUM,31 N=0 IFiK.EQ.Ol GO TO 60 N=N+1 CALL ACCESS(NAME.K,TANK,NHORDS,4) IF (NNORDS.LT.LENGTH1 GO TO 40 DO 50 I=1oLENGTH 30 35 40 IF (TANK(11.NE.STRING(IH GO TO 40 50 CONTINUE GO TO 110 ERR=.TRUE. TYPE 130v (VALUEII).I=1:IDIM) GO TO 100 CALL BINTYP(NAME1LABEL1BITEM.L,Ml DO 90 N=1vM 60 80 IF (BLABEL.EQ.LABEL(NH GO TO 110 69 Error checking and reporting: If the character-string value is not found, a message is typed, the error flag is set, and zero is returned as the value of PNTER. Program logic: If the length of the string is given as zero, a value of blank is returned immediately. Otherwise, the string is packed into STRING. The value of ITYPE then determines whether the character-type dictionaries should be accessed (via ACCESS) or the multiple-choice-type dic- tionaries should be accessed (via BINTYP). If a character— type dictionary is indicated, a call to ACCESS is made, where the fifth parameter has a value of 31. This returns K as the pointer to the first dictionary item. ACCESS is then called, using the value 4 as the fifth parameter (which re- turns the K’th entry and updates K to point to the next entry), until all entries have been returned or until a match is found. If a match is found, the entry number is returned as a value. Otherwise, zero is returned as a value, and the nonstandard return is taken. If a multiple-choice—type dic- tionary is indicated, a call to BINTYP returns the possible values in LABEL. The string (equivalenced to BLABEL) is then compared with the items of LABEL. PROGRAM 0001456 0001457 0001458 0001459 0001460 0001461 0001462 0001463 0001464 0001465 0001466 0001467 0001468 0001469 0001470 0001471 0001472 0001473 0001474 0001475 0001476 0001477 0001478 0001479 0001480 0001481 0001482 0001483 0001484 0001485 0001486 0001487 0001488 0001489 70 90 100 110 130 140 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) CONTINUE 0001490 TYPE 140' BLABEL 0001491 ERR=.TRUE. 0001492 N=0 0001493 PNTER=N 0001494 RETURN 0001495 FORMAT (' CHARACTER TYPE VARIABLE DOES NOT ASSUME VALUE:'/1X160Al)0001496 FORMAT (' BINARY TYPE VARIABLE DOES NOT ASSUME VALUE '1A8, 0001497 END 0001498 GRASP SOFTWARE SPECIFICATIONS 71 SUBROUTINE NAME: PREVAL Purpose: PREVAL acts as an interface between the calling routine (VLIST) and the arithmetic-expression parsing routine PARSE. This interface allows a reduction in the number of dimensions for the variables in /EXPRNS/ which contain the Reverse-Polish form of the arithmetic expressions entered by the user. Calling sequence: CALL PREVAL(&n,IEXPR,L,KNT) Arguments: . . n—Statement (in calling routine) to which a branch is made if the routine PARSE sets an error flag. IEXPR—Contains (in unpacked character form) the ex- pression to be parsed. L—The length of IEXPR. GRASP SOURCE SUBROUTINE PREVAL‘*¢IEXPR,L,KNT) COMMON /EXPRNS/ PDLISHqITYPEyLPS DIMENSION POLI$H(15.8), LOGICAL ERR KNT=KNT+1 CALL PARSEl IEXPRyL yITYPEl 1vKNT). POLISHll. KNT! oLPS(KNT) 1ERR) 1F (ERR) RETURN 1 RETURN END ITYPE(15,8M KNT—Arithmetic expression counter. Subroutine culled: PARSE Common data referenced: /EXPRNS/ Called by: VLIST Error checking and reporting: Error flag returned from PARSE is testedf Program logic: 1. The expression counter KNT is incremented. 2.. Call PARSE, passing the input arguments IEXPR, L, and the KNT’th columns of ITYPE, POLISH along with the KNT’th element of LPS, and an error flag. 3. Take the nonstandard return if the error flag ERR has been set. POLISH, ITYPE, LPS in PROGRAM 0001499 0001500 0001501 0001502 0001503 0001504 0001505 0001506 0001507 LPS(81, IEXPR( 1) 72 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) SUBROUTINE NAME: QUIT Purpose: QUIT performs “wrap-up” processing prior to exiting from the GRASP system. This involves a typed statement regarding the disposition of files created during the current session. Calling sequence: CALL QUIT(OFILES,NFILES) Arguments: OFILES—List of output files created during this session. NFILES—The number of items in OFILES. Subroutines called: KEYBRD, RLIST Common data referenced: None Called by: DRIVER Error checking and reporting: The user’s response to prompts is checked for validity. Program logic: 1. The list of created file names is typed, and the user is asked if he would like to save any of them. 2. If so, he is asked to enter a list of numbers corresponding to those files he wishes to save. 3. The system then instructs him how to delete the files he does not wish to save. This routine is provided primarily for bookkeeping. The file-maintenance functions can be per- formed at the program level on those systems having this capability. G R A S P S O U R C E P R O G R A M SUBROUTINE QUITIOFILE59NFILES) 0001508 DIMENSION IMAGEi30leIST(201v KLISTIZO) 0001509 INTEGER OFILES(20),FILEyYE59REPLY 0001510 DATA YESyNOI'Y'y'N'/ 0001511 IF (NFILES.EQ.0) GO TO 100 0001512 TYPE 110. (I.OFILES(II.I=19NFILES) 0001513 TYPE 120 0001514 10 TYPE 130 0001515 CALL KEYBRDIEIOOoREPLYyll 0001516 IF (REPLY.EQ.YES) GO TO 40 0001517 IF (REPLY.EQ.NDI GO TO 20 0001518 TYPE 150 0001519 GO TO 10 0001520 20 NKILL=NFILES 0001521 DO 30 I=1,NFILES 0001522 30 KLISTII1=I 0001523 GO TD 80 0001524 40 IF (NFILES.EQ.1) GO TO 100 0001525 TYPE 160 0001526 50 CALL KEYBRDIEIOOglMAGEy30) 0001527 CALL RLISTI850,1MAGE9LIST9NSAVE920) 0001528 IF (NSAVE.E0.0) GO TO 20 0001529 NKILL=0 0001530 00 70 I=19NF1LES 0001531 DD 60 J=lyNSAVE 0001532 IF (LIST(J).EQ.I) GO TO 70 0001533 60 CONTINUE 0001534 NKILL=NKILL+1 0001535 KLISTINKILLI=I 0001536 70 CONTINUE 0001537 IF (NKILL) 100,100.80 0001538 8C TYPE 9O,(DFILES(KLIST(II)pI=l9NKILLI 0001539 100 RETURN 0001540 9C FORMATI' ISSUE .DEL COMMANDS FOR THE FOLLOWING FILES:'/(1X.10A61) 0001541 110 FORMAT (/l' IESSIDN:'/(1595X,A611 120 FORMAT (l/' 130 FORMAT (' 150 FORMAT (' 160 FORMAT (' ENTER A LIST OF 1 YOU WISH TO SAVE (IE. END (ENTER YES OR NO): 1-315).'i THE FOLLOWING FILES HAVE BEEN CREATED DO YOU WISH TO SAVE ANY OF THEM?'I ',$l YOUR REPLY HAS NOT UNDERSTOOD.') NUMBERS CORRESPONDING TO 'I' ','DURING THIS 50001542 0001543 0001544 0001545 0001546 THOSE FILE50001547 0001548 0001549 GRASP SOFTWARE SPECIFICATIONS 73 SUBROUTINE NAME: RELEXP Purpose: This subroutine is used to decode the “condition” appearing in IMAGE into the components NAMEPT, RCODE, and IVAL. If it is unsuccessful, an error message is typed and an error flag is set. The “condition” is in unpacked character form and is assumed to be a name followed by a relation followed by a value. Name must be an item name in the current data base (as established by the file command). Relation must be one of the following: EQ, equal; LT, less than; GT, greater than; LE, less than or equal; GE, greater than or equal; NE, not equal; BE, be- tween. Value must be a number, number pair, character string, set of qualifiers, permissible multiple-choice acronym, or blank. The following table gives valid construCtions for “conditions” : Item type Relation Value Integer or real--- EQ, LT, GT, LE, GE, NE__ Numeric. ____________ Numeric pair. Character _______ EQ, LT, GT, LE, GE, NE__ Any printable string. BE ____________ Printable string con- . taining comma. Multiple choice __ EQ, NE ____________________ Multiple-choice acronym. Qualified real _-__ EQ, LT, GT, LE, GE, NE__ Numeric. BE ____________ Numeric pair. EQ, NE ____________ Qualifier set1 in parentheses. 1Qualifier set is one or more of the following characters, each of which occur, at most, once; G, H, N, T, or blank. Calling sequence: CALL RELEXP(&n,IMAGE,NAMEPT, RCODE,IVAL,ERR) Arguments: n—Statement number (in calling routine) to which a branch will be made if an all-blank condition is detected. IMAGE—Contains “condition” in unpacked-character form. NAMEPT—Returned pointer to item name. RCODE—Returned encoding of relation having the follow- ing possible values: 1—7 corresponding to the relations EQ, LT, GT, LE, GE, NE, BE. 11 or 16 corresponding to the relations EQ or NE, applied to a set of qualifiers. IVAL—Returned as one of the following: 1. Integer or real value. 2. Pointer to a particular entry in the character dictionary associated with the item pointed to by NAMEPT. 3. Bit encoding, giving the position of a particular multiple- choice acronym in the file containing possible acronym values for the item pointed to by NAMEPT. 4. Pointer to the number pair in the common block BTWN which will be used by this instance of the BE relation. 5. Bit encoding of a qualifier set. ERR—Returned error flag that is set if an error is detected. Subroutines called: SCAN, BFIND, ICONV, PNTER, PACK Common data referenced: GRASP SOURCE SUBROUT [NE RELEXPI*11MAGEv NAMEPTyRCODE . IVAL qERR) COMMON NAMEsleYPEypNTERSqlDIM COMMON [BTWN] IVALSvNBE DIMENSION RELSlle IVALSlZlely IMAGEl 80h NAMES, ITYPE, PNTERS, IDIM in blank common IVALS, NBE in /BTWN/ Called by: CONDTN Error checking and reporting: 1. A11 testing is performed to insure conformity to the table of valid constructs appearing in the preceding “purpose” section. 2. An error flag that may be set by the routines ICONV or PNTER is tested. 3. A nonstandard return from BFIND indicates an invalid name. An error message is printed reporting any of the following errors: a. Unable to find relation (that is, EQ, LT, GT, LE, GE, NE, BE). b. Incorrect qualifier set. c. Qualifier codes are referenced in forms other than EQ or NE. d. Invalid name as first syntactic unit of condition. e. No comma separating a value pair used with the BE relation. Program logic: 1. A call to SCAN is made to bracket the name as the first syntactic element. If the image is all blank, the nonstandard return is taken. 2. The name is packed into NAME via ENCODE, and BFIND is used to do the lookup. If the name is not found, a message is typed, and the error flag is set. 3. The next call to SCAN brackets the relation. It is packed into REL and tested against the list of valid relations. Note that RCODE is used as the index. If invalid, a message is typed, and the error flag is set. 4. The value part of the condition is then bracketed via the next call to SCAN. If the value field is blank, IVAL is set to blank. 5. Otherwise, the type of name is determined using ITYPE in blank common. 6. The logical variable BE (indicating the “between” rela- tion) is determined. If set, the second value is determined and stored in the BTWN common area, and IVAL is set to point to the BTWN location. The second value deter- mination is logically similar to the first which is described in step 8. 7. If BE was not set, the value element is tested as a qualifier set. If it is one, the appropriate tests are made, and IVAL is bit encoded to show which codes are present. RCODE is, also, incremented by 10 as a flag indicating comparison of qualifier cores. 8. If the value element was not a qualifier set, and the rela- tion was not BE, IVAL is set via a call to ICONV, if type was numeric. Note that for real values, VAL (equivalenced to IVALL) is set. IVAL is then set by IVALL, which shares storage with VAL. For character and multiple-choice types, IVAL is set using the external function PNTER. PROGRAM 0001550 0001551 0001552 0001553 ITYPE( 500i v NAME Si 500) 74 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) DOUBLE PRECISION NAMEoNAMESyDBLNK 0001554 INTEGER FCNAME.FCOP.FCCON.RCODE.PNTER,BLANK,PNTERS(500),E.REL'REL50001555 lyCOMMAyRPARENleUAL16) 0001556 LOGICAL ERRyBE 0001557 EQUIVALENCE (VAL,1VALL).(IQUAL,BLANK1 0001558 DATA COMMA,RPAREN,LPAREN.DBLNK,IQUALI','p')’.'(‘p' 'p 0001559 1 ' '1'G'v'H'v'L'1'N'v'T'/p 0001560 1 RELS/‘EQ','LT','GT',‘LE','GE'.‘NE','BE'/ 0001561 ERR=.FALSE. 0001562 I=1 0001563 CALL SCAN18295.IMAGE,I,FCNAME,LCNAME,11 0001564 NAME=DBLNK 0001565 CALL pACK11MAGE1FCNAME1vNAMEvLCNAME‘FCNAME+1181 0001566 CALL BFIND!8110yNAMEyNAMEPTyNAME55PNTERS,IDIM) 0001567 I=LCNAME+1 0001568 CALL SCAN(820.1MAGE,IoFCOPydyl) 0001569 IF(J.NE.FCOP+1) GO TO 20 0001570 REL=BLANK 0001571 CALL PACK(IMAGE(FCOP),REL,2,4) 0001572 00 10 RCODE=1’7 0001573 IF(RELS(RCODE).EQ.REL1 GO TO 30 0001574 10 CONTINUE 0001575 20 TYPE 320 0001576 00 TO 144 0001577 30 I=J+1 0001578 CALL SCAN(84091MAGEpIqFCCONyLCCON’Z) 0001579 00 TO 140 0001580 40 IVAL=8LANK 0001581 GO TO 210 0001582 295 RETURN 1 0001583 140 J=ITYPE(NAMEPT) 0001584 8E=RCODE.EQ.7 0001585 IF(BE) GO TO 220 0001586 1F([MAGE(FCCON).NE.LPAREN) GO TO 150 0001587 IF(J.NE.5) GO TO 150 0001588 1F(IMAGE(LCCON).EQ.RPAREN) GO TO 142 0001589 1405 TYPE 141 0001590 00 TO 144 0001591 142 IFIRCODE.EQ.1) GO TO 145 0001592 IF(RCODE.EQ.6) GO TO 145 0001593 TYPE 143 0001594 144 ERR=.TRUE. 0001595 GO TO 210 0001596 145 RCODE=RCODE+10 0001597 IVAL=0 0001598 FCCON=FCCON+1 0001599 LCCON=LCCON-1 0001600 00 147 K=FCCONpLCCON 0001601 J=IMAGE(K) 0001602 00 146 1:116 0001603 IF(IQUAL(I1.EQ.J1 GO TO 147 0001604 146 CONTINUE 0001605 00 TO 1405 0001606 147 1VAL=IVAL+2**(1-1) 0001607 GO TO 210 0001608 110 TYPE 3109 NAME 0001609 00 TO 144 0001610 150 GO TO (160,170,190y19C91701’J 0001611 160 1VAL=IC0NV(1MAGE‘FCCON’1LCCON‘FCC0N+lvaERR1 0001612 00 TO 200 0001613 GRASP SOFTWARE SPECIFICATIONS 170 K=ICDNVIIMAGEIFCCON1,LCCON—FCCON+11E0ERR) IF (K.NE.BLANK) GO TO 180 [VAL=K GO TO 200 180 VAL=K*10.**E IVAL=1VALL GO TO 200 190 IVAL=PNTER(IMAGE(FCCDN)yLCCDN-FCCON+1yNAMEPT,J,ERR) 200 IF (ERR) GO TO 210 IF (.NDT.BEI GO TO 210 NBE=NBE+1 IVALS(1yNBE)=IVAL IVAL=NBE 210 RETURN 220 00 230 I=FCCDN9LCCON IF (IMAGE(I).EQ.COMMA) GO TO 240 230 CONTINUE TYPE 330 GO TO 144 240 GO TO (250,260.28092801260),J 250 IVAL=ICONV(IMAGE(I+1).LCCON—I,E,ERR) GO TO 290 260 K=ICONVIIMAGEII+11cLCCDN—I:E,ERR) IF (K.NE.BLANK) GO TO 270 IVAL=K GO TO 290 270 VAL=K*10.**E IVAL=IVALL GO TO 290 280 IVAL=PNTER(IMAGEII+11qLCCDN-I,NAMEPT,J,ERR) 290 IF (ERR) GO TO 210 IVALSI2yNBE+11=IVAL LCCON=I-1 GO TO 150 141 FDRMAT(' QUALIFIER CODES MUST BE ONE OR MORE OF "L. vNyT,G,H"', 1 ' AND ENCLOSED IN PARENTHESIS.') 143 FDRMAT(' ONLY EQINE CAN BE USED WITH QUALIFIER CODES.') 310 FORMAT (' INVALID NAME = '9A8) 320 FORMATI' UNABLE TO FIND RELATION (LTchyLEvGEvEQvNEIBEI'1 330 FORMAT (' NO COMMA SEPARATING CONSTANTS FDLLOHING BE DPERATDR') END 75 0001614 0001615 0001616 0001617 0001618 0001619 0001620 0001621 0001622 0001623 0001624 0001625 0001626 0001627 0001628 0001629 0001630 0001631 0001632 0001633 0001634 0001635 0001636 0001637 0001638 0001639 0001640 0001641 0001642 0001643 0001644 0001645 0001646 0001647 0001648 0001649 0001650 0001651 0001652 0001653 0001654 76 GEOLOGIC RETRIEVAL AND SUBROUTINE NAME: RETRVE Purpose: RETRVE is used to retrieve records from a selected file and write them on some other selected file. User-specified encoded retrieval criteria are passed to RETRVE via its argument list. Calling sequence: CALL RETRVE(&n,&m,IFILES,OFILES, NFILES,POLISH,LPS,VARS,CODES,VALS,NCOND) Arguments: n—Statement number (in calling routine) to which a branch is made if previously undetected (by LOGEXP) errors are encountered. m—Statement number (in calling routine) to which a branch is made if KEYBRD senses an EOF. IFILES, OFILES—Arrays of input and output file names, respectively. NFILES—Number of elements in either IFILES or OFILES (IFILES and OFILES are of equal size). POLISH—Array containing the Reverse-Polish form of the logic expression to be used for data retrieval. LPS—The number of elements in POLISH. VARS, CODES, VALS—Arrays that give an encoding of the conditional expressions entered by the user. NCOND—Number of elements in VARS, CODES, and VALS. Subroutines called: KEYBRD, OBEY, GETPUT, COMP, UNCODE, PACK Common data referenced: FNAMES, SELECT in /FILNAM/ ITYPE in blank common INPUT, OUTPUT in /IOUNIT/ Called by: DRIVER Error checking and reporting: 1. Check to assure entry of retrieval criteria (that is, LPS>0) 2. Check to assure absence of undetected errors in the re- trieval criteria. GRASP SOURCE SYNPOSIS PROGRAM (GRASP) Messages are typed corresponding to the two error situa- tions above. Program logic: 1. LPS is checked to insure that retrieval criteria have been entered. 2. The elements of VALS are moved to IVAL to allow the equivalencing necessary for mixed modes (in particular, integer and real) possible in retrieval criteria. . Input- and output-file names are prompted and accepted, then associated with FORTRAN unit numbers via calls to KEYBRD and OBEY. The new file names are added to IFILES and OFILES. . The input file is then read, one record at a time (via GETPUT), until the nonstandard (EOF) return is taken. After each call to GETPUT, the record (IREC) is tested against the retrieval criteria indicated in POLISH, VARS, CODES, and VALS via the logical valued push-down stack technique described as follows: Any element of POLISH less than 27 points to one of the conditional expressions en- coded in VARS, CODES, VALS. That expression is evalu- ated via a reference to the logical function COMP, and the result is placed in the push-down stack. If POLISH (I)>26, it points to one of the logical operators “and,” “or,” “not” (denoted by *, +, —). If the operator is —, the “not” op- eration is performed on the topmost stack element. If the operator is * or +, the operation is performed on the top- most two stack elements, the size of the stack is decremented, and the resultant (:logical) value replaces the new topmost stack element. After the last element of POLISH has been processed, the size of the stack should be 1, and the value of this element indicates whether or not the record meets the retrieval criteria. If so, it is added to the output file by a call to GETPUT. Counts of records read and records retrieved are kept and typed at the end of retrieval process- mg. PROGRAM SUBROUTINE RETRVE(*.*.IFILESyOFILE59NFILESyPOLISHyLPSyVARSpCDDESv 0001655 IVALSoNCOND) 0001656 COMMON NAMES,ITYPE91PTS,IDIM 0001657 COMMON IIOUNIT/ INPUTyOUTPUT 0001658 COMMON /FILNAM/ FNAMESySELECTyPAD 0001659 DIMENSION VALI26) 0001660 DOUBLE PRECISION NAMESiSOG) 0001661 INTEGER IVAL(26)gCODESI1),VARS(1D,HHICHgDRECISOO)vOUTPUTyTOPyPOL 0001662 IISH(11yITYPElSOOlvONEyTNOyVALSi1"EQUATEl4iylPTSISOOTpSELECT 0001663 INTEGER FILEIDyDBLANKyIFILES(11,0FILE5(1)oDFAULT'FNAMES(211 0001664 LOGICAL COMPvVALUEosrACKIZO)1EVAL’PADI4) 0001665 EQUIVALENCE (IRECyRECIylIVAL,VAL).(STACKgEVAle(EQUATEi4ivFILEID) 0001666 DATA ONE,THO,DBLANKI'1'.'2',' 'l, EQUATE/‘EQUA'1'TE 1"2*' ‘/ 0001667 IF (LPS.GT.0) GO TO 10 0001668 TYPE 180 0001669 GO TO 170 0001670 10 DO 20 I‘lvNCUND 0001671 20 IVALII)=VALS(I) 0001672 DFAULT=FNAMES(SELECT) 0001673 NRECS=0 0001674 NFOUND=0 0001675 30 40 50 60 62 63 64 66 7o 80 90 100 110 120 130 GRASP SOFTWARE SPECIFICATIONS NFILES=NFILES+1 EQUATE131=DNE TYPE 190 CALL KEYBRD(£17510REC.S) CALL PACK(DREC,FILEID,5:51 IF (FILE10.EQ.0BLANK) FILEID=DFAULT 1FILES(NFTLES)=FILEID CALL OBEY(81701EQUATE94T EQUATE(3)=THO TYPE 210 CALL KEYBR018175908ECv5) CALL PACKiDREC,FILEIDv5v5) 0FILES(NFILES)=F1LEID CALL DBEY(81709E0UATE¢4) CALL GETPUT18140.DREC.1) NRECS=NRECS+1 TOP=0 DO 110 J=1.LPS INDEX=POLISHTJ| 1F (INDEX.GT.261 GO TO 70 TOP=TDP+1 HHICH=VARS1INDEX1 1=ITYPE(WHICH) IREC=DREC1HHICH1 GO TO (40.50940160y62)’ I STACK(TOP)= COMP1IREC gIVALTINDEX),01902yCDDESTINDEX1o1) GO TO 110 STACK1TOP1= COMPIIDlg1029REC ,VAL(INDEX).CODES(INDEXTvZ) GO TO 110 STACK1TOPI= COMPIIREC yIVAL(1NDEX7!DI!DZ,CODES(INDEX793’ 60 T0 110 IF(IREC.NE.DBLANK) GO TO 63 IF(CODES(INDEX).LT.11) GO T0 50 STACK(TOP)=.FALSE. GO TO 110 REC=UNCODE(REC.IQ) IF(CODES(INDEX)-11) 50.64.66 STACK(TOP)=MOD(IVAL(lNDEX)/2**(IQ-1)y2).EQ.1 GO TO 110 STACKTTUP)=MDD(IVAL(INDEX1/2**(10—1112).EQ.0 GO TO 110 VALUE=STACK1TOP1 IF (INDEX-30) 809901100 TOP=TOP~1 STACK1TOP)=STACK(TOP).DR.VALUE GO TO 110 TDP=TOP-1 STACK(TOP)=STACK(TOP).AND.VALUE GO TO 110 STACKTTOP)=.NOT.STACK(TOP) CONTINUE IF (TOP.EQ.1) GO TO 120 TYPE 220 RENIND INPUT RETURN 1 IF (EVAL) GO TO 130 GO TO 30 CALL GETPUT18140yDRECv2) NFOUND=NFOUND+1 GO TO 30 77 0001676 0001677 0001678 0001679 0001680 0001681 0001682 0001683 0001684 0001685 0001686 0001687 0001688 0001689 0001690 0001691 0001692 0001693 0001694 0001695 0001696 0001697 0001698 0001699 0001700 0001701 0001702 0001703 0001704 0001705 0001706 0001707 0001708 0001709 0001710 0001711 0001712 0001713 0001714 0001715 0001716 0001717 0001718 0001719 0001720 0001721 0001722 0001723 0001724 0001725 0001726 0001727 0001728 0001729 0001730 0001731 0001732 0001733 0001734 0001735 78 140 150 160 170 175 180 190 210 220 230 240 250 GEOLOGIC RETRIEVAL AND SYNPOSIS PROGRAM (GRASP) TYPE 230v NRECSyIFILESINFILES) [F (NFOUND.GT.O) GO TO 150 TYPE 240 NFILES=NFILES-1 GO TO 160 TYPE 250v NFOUNDoFILEID REHIND INPUT REHIND OUTPUT RETURN RETURN 2 FORMAT (' LOGIC MUST BE SUPPLIED BEFORE A RETRIEVAL CAN BE MADE') FORMAT (' ENTER INPUT FILE NAME: ',$) FORMAT (' ENTER OUTPUT FILE NAME: 'y$) FORMAT (' ERROR IN LOGIC EXPRESSION') FORMAT (' ALL'.16,' RECORDS OF'.A6,' SEARCHEDo') FORMAT (' THERE ARE NO RECORDS HHICH SATISFY THE REQUEST') 0001736 0001737 0001738 0001739 0001740 0001741 0001742 0001743 0001744 3001745 0001746 0001747 0001748 0001749 0001750 0001751 FORMAT (1109' RECORDS FOUND WHICH SATISFY THE REQUEST.‘/' THEY HAV0001752 1E BEEN STORED IN '9A6) END 0001753 0001754 GRASP SOFTWARE SPECIFICATIONS SUBROUTINE NAME: RLIST Purpose: RLIST is used to convert an unpacked character string representing a list of user-entered numbers into a corresponding numeric list. Calling sequence: CALL RLIST(&n,IMAGE,LIST,NUMC, MOST) Arguments: n—Statement number (in caller) to which a branch will be made if an uncorrectable error in the entered character string is detected. IMAGE—The unpacked character-string form of the list of numbers. LIST—The list of numbers which are returned in numeric form. NUMC—The number of items in LIST. MOST—The maximum number of items that LIST may contain. Subroutines called: None Common data referenced: None Called by: DEFLST, QUIT Error checking and reporting: If an illegal character (not 0~9, dash, or comma) is detected, an error message is typed and the nonstandard return is taken. If the list isn’t GRASP SOURCE SUBRDUT INE RLIST(*.IMAGEpLISTvNUMCyMOST) INTEGER LISTIl),IMAGE(301¢CHAR(12)9C0L(121,IMATI‘MBI DATA 1MAT/12,22'14y24yoc31g0,4190,33.0yO/yIBLNKl' '1 DATA CHAR/'0'.‘1'9'2'9'3'9'4'0'5'9 DATA COL/10*112v3/ NUMC=C LA5T=1 IP=0 IRUN=1 IP=IP+1 IF (IP.GT.301 GO TO 120 ICHAR=IMAGE11P1 IF (ICHAR.EQ.IBLNK) GO TO 10 DO 20 I=1v12 IF IICHAR.EQ.CHAR(I11 GO TO 40 CONTINUE TYPE 1509 RETURN 1 TYPE 14591VAL GO TD 30 IFS=IMATIIROH9CULIIll IF IIF$.NE.0) GO TO 60 TYPE 160 GO TO 30 IROH=MODIIFS’101 IFS=IFS/10 GO TO (70,8019001001’ IVAL=I—1 GO TO 10 IVAL=10*IVAL+I—1 GO TO 10 10 20 ICHAR 30 35 4O 50 60 IFS 7O 80 79 composed of numbers or number ranges separated by com- mas, an error message is typed, and the nonstandard re- turn is taken. Program logic: Each character of the unpacked string is proc- essed via the following transition matrix: 0—9 _ 1 f (1) / 2 error error 2 f(2)/2 f(3)/1 f(3)/3 3 f (1 ) /4 error error 4 f(2)/4 f(4)/1 error f(i) are defined as: f (1)—Start a number value. f (2)——Build number value by adding digit on right. f(3) —Number built, add it to list. f(4)—-Fill LIST with values up to and including current value. Rows 1 and 2 are used to process a single list element or the first of a number-range pair. Rows 3 and 4 are used to process the second of a num- ber-range pair. Blanks are completely ignored. The columns of the transition matrix are associated with the indicated characters, and the rows correspond to indi- vidual states. PROGRAM 0001755 0001756 0001757 0001758 0001759 0001760 0001761 0001762 0001763 0001764 0001765 0001766 0001767 0001768 0001769 0001770 0001771 0001772 0001773 0001774 0001775 0001776 0001777 0001778 0001779 0001780 0001781 0001782 0001783 0001784 0001785 I60,'7I'I8I'09I,0,',l_l/ 80 GEOLOGIC RETRIEVAL AND SNYPOSIS PROGRAM (GRASP) 90 NUMC=NUMC+1 0001786 IF (NUMC.GT.MOSTI GO TO 110 0001787 1F(IVAL.GT.MOST) GO TO 35 0001788 LISTlNUMC1=IVAL 0001789 GO TO (10,1401: LAST 0001790 100 LIST(NUMC+1)=LIST(NUMC)+1 0001791 NUMC=NUMC+1 0001792 1F (NUMC.GT.MO$T) GO TO 110 0001793 1F(IVAL.GT.MOST) GO TO 35 0001794 [F (LIST(NUMC).LT.IVAL1 GO TO 100 0001795 GO TO (10.14019 LAST 0001796 110 TYPE 170' MOST 0001797 NUMC=MOST 0001798 GO TO 140 0001799 120 IF (NUMC.NE.O) GO TO 130 0001800 [F (IRON.EQ.1) GO TO 140 0001801 NUMC=1 0001802 LIST(1)=IVAL 0001803 GO TO 140 0001804 130 IF (MOD([RON12).NE.0) GO TO 50 0001805 LAST=2 0001806 IROW=IROWI2 0001807 60 TO (90.100), IRON 0001808 140 RETURN 0001809 145 FORMAT(15o' DOES NOT CORRESPOND TO A CATEGORY. RE-ENTER NUMBERS') 0001810 150 FORMAT (1X,A1y' IS AN ILLEGAL CHARACTER. RE-ENTER NUMBERS.'1 0001811 160 FORMAT (' EACH NUMBER OR NUMBER RANGE (1E. 4-7) EXCEPT '1'THE LAST0001812 I MUST BE FOLLOWED BY A COMMA.'/' RE-ENTER NUMBERS.') 0001813 170 FORMAT (' TOO MANY NUMBERS GIVEN. ONLY THE FIRST'pIBy' WILL BE USE0001814 10.') 0001815 END 0001816 GRASP SOFTWARE SPECIFICATIONS 81 SUBROUTINE NAME: ROWPNT Purpose: ROWPNT prints selected items or expressions, one to a line, from a selected file. Calling sequence: CALL ROWPNT(&n,NPAGE) Arguments: n—Statement (in caller) to which VLIST will branch when an EOF is encountered by KEYBRD. NPAGE—Number of lines per page of printed output. Subroutines called: VLIST, ACCESS, GETPUT, PAUSE, EVAL, UNCODE, BINTYP, BLIST Common data referenced: POLISH, ICODE, LPS in /EXPRNS/ ITYPE in blank common Called by: LIST Error checking and reporting: None Program logic: 1. The list of items to be printed is obtained via a call to VLIST. 2. A call to ACCESS is made to initialize the lookup of dic- tionary-type items. 3. Each record of the ,selected file is obtained via GETPUT and a line counter is incremented and tested. If it exceeds the page size, a pause is generated via PAUSE. 4. After each record is obtained, the selected items are evalu- ated (if necessary) and printed. G R A S P S O U R C E P R 0 G R A M SUBROUTINE ROWPNT(*,NPAGE) 0001817 COMMON NAMESyITYPEyIPTSpIPAD 0001818 COMMON IEXPRNS/ PDLISH:ICODE.LPS 0001819 DOUBLE PRECISION NAMEScLABELqNAMESI.VNAMES(20) 0001820 INTEGER BLANKgTANKolPTSISOO)'IQUAL(61,POLISH(151819 0001821 1 ICODE(15v81yLPS(8) 0001822 DIMENSION ITYPE150017 BITEM(15v25), ITEMSIZO), IREC(5001, 0001823 1 REC(5001. NAMES(500). TANK1251, LABEL1251' LISTIZS) 0001824 LOGICAL ERR 0001825 EQUIVALENCE (REC,IREC).IIVAL9VAL1,(LISToTANKIyIBLANK,IQUAL) 0001826 DATA IQUALI' '9'G'Q'H'v'L'y'N'1'T'l 0001827 KOUNT=0 0001828 CALL VLIST(£270,VNAMES'ITEMS,NUM1 0001829 IF (NUM.E0.01 GO TO 260 0001830 TYPE 280 0001831 CALL ACCESS(II,IVAL,TANK,J,1) 0001832 120 CALL GETPUTIfiZéOyIREqu) 0001833 KOUNT=KOUNT+NUM+1 0001834 IF (KOUNT.LE.NPAGEI GO TO 130 0001835 KOUNT=0 0001836 CALL PAUSElfiZbO) 0001837 130 D0 240 JJ=1yNUM 0001838 II=ITEMS(JJ) 0001839 IFlII.GT.01 GO TO 135 0001840 II=-II 0001841 NAMESI=VNAMESIJJ1 0001842 VAL=EVALIIPEC11CODE1121111P0LISH1111119LP$111)vERR) 0001843 IFiERR’ GO TO 240 0001844 GO TO 150 0001845 135 IVAL=IRECI111 0001846 IF (IVAL.E0.8LANK) GO TO 240 0001847 KIND=ITYPElIII 0001848 NAMESI=VNAMESlJJ1 0001849 GO TO (1401150v16002209170’9 KIND 0001850 140 TYPE 300v NAMESI,IVAL 0001851 GO TO 240 0001852 150 TYPE 310. NAMESIyVAL 0001853 GO TO 240 0001854 160 CALL ACCESSiIIgIVALyTANKcJ921 0001855 TYPE 320, NAMESIoITANK(I)pI=1,J1 0001856 GO TO 240 0001857 82 170 220 230 240 260 270 280 290 300 310 320 330 340 GEOLOGIC RETRIEVAL AND SNYPOSIS PROGRAM (GRASP) VAL=UNCODETVAL9101 TYPE 3109NAMESIVVA1910UALTIQ) GO TO 240 CALL BINTYPTIIvLABELyBITEMvKIM) KUUNT=KOUNT+1 TYPE 3401 NAMESI CALL BLIST(LIST.NUM511VAL1 DO 230 I=lyNUMS KOUNT=KUUNT+1 J=L1$T111 TYPE 330g LABEL1J1oTBITEM(L9J)7L=vi) CONTINUE TYPE 290 GO TO 120 RETURN RETURN 1 FORMAT 1///’ FORMAT (1X,3(8H********11 FORMAT 11X1A811H=9191 FORMAT (1X9A891H=11PG12.5’A1) FORMAT (IX’A891H=v12A5/15Xv12A5) FORMAT (5X9A8,15A41 FORMAT (IXyA891H=) END 0001858 0001859 0001860 0001861 0001862 0001863 0001864 0001865 0001866 0001867 0001868 0001869 0001870 0001871 0001872 0001873 0001874 0001375 0001876 0001877 0001878 0001879 0001880 0001881 GRASP SOFTWARE SPECIFICATIONS SUBROUTINE NAME: SCAN Purpose: This subroutine is used to set character-position pointers for the syntactic elements of a condition (name, relation, value(s) ) or a logical expression. Calling sequence: CALL SCAN(&n,IMAGE,IS,I1,12,IT) Arguments: n—Statement number (in calling routine) to which a branch will be made if IMAGE is all blanks. IMAGE—String of unpacked left-justified characters. IS—Starting position of the scan. Il—Pointer to first character of syntactic element. IZ—Pointer to last character of syntactic element. IT—Embedded blank switch. GRASP SOURCE SUBROUT INE SCANH‘. IMAGEvISv 11,12911') INTEGER IMAGE(1) DATA IBLNKI' '/ DO 1 I=I5980 IFIIMAGE(11.NE.IBLNK) GO TO 2 1 CONTINUE RETURN 1 11=I J=11+1 GO TO (31101'IT DO 4 I2=JoSO IFIIMAGE(IZ).E0.IBLNK) CONTINUE IZ=81 5 [2:12‘1 RETURN DO 11 I=Jy80 IZ=80-I+J IF(IMAGE(IZI.NE.IBLNK) GO TO 6 CONTINUE GO TO 6 END GO T05 83 Subroutines called: None Common data referenced: None Called by: RELEXP, LOGEXP Error checking and reporting: None Program logic: 1. The position of the first nonblank character is determined. 2. If all characters after the IS’th are blank, the nonstandard return is taken. 3. If no embedded blanks are permitted (IT=1), the position of the last nonblank character to the right of the position found in step 1 is determined, and control passes to the caller. 4. Otherwise (if, IT=2), the position of the first nonblank character to the left of position 80 is determined and control returns to the caller. PROGRAM 0001882 0001883 0001884 0001885 0001886 0001887 0001888 0001889 0001890 0001891 0001892 0001893 0001894 0001895 0001896 0001897 0001898 0001899 0001900 0001901 0001902 0001903 84 GEOLOGIC RETRIEVAL AND SNYPOSIS PROGRAM (GRASP) SUBROUTINE NAME: START Called by: DRIVER E h k' d ‘ : Purpose: START determines availability of data bases and P3321218; 917:9 an reportmg None the“ associated files. 1. The name GFILE is associated with FORTRAN input Calling sequence: CALL START unit 20_ Arguments: None 2. A welcoming massage is typed and records of GFILE are Subroutine called: IFILE read to fill the /FILNAM/ common area. Common data referenced: 3. As each record is read, parts of it are output to the termi- All variables in /FILNAM/ except NUMF nal. GRASP SOURCE PROGRAM SUBROUTINE START 0001904 COMMON IFILNAM/ MASTER,MASK,DEFTN.DFILE.BFILE,NUMF,NUM1,IDIMS 0001905 DOUBLE PRECISION CONTNTIé) 0001906 INTEGER MASTERl4IoMASK(4)oDEFTN(4).DFILE(4I.BFILE(41oIDIMS(4) 0001907 CALL IFILEIZOy'GFILE') 0001908 NUMF=1 0001909 TYPE 1 0001910 10 READIZOyIIyEND=ZOI MASTERINUMF),CONTNT,MASK(NUMFI, 0001911 1 DEFTN(NUMF)10FILE(NUMF)oBFILETNUMF)yIDIMSINUMFI 3001912 TYPE 129MASTER(NUMF17CONTNT 0001913 NUMF=NUMF+1 0001914 GO TO 10 0001915 20 NUMF=NUMF-1 0001916 RENIND 20 0001917 TYPE 2 0001918 RETURN 0001919 1 FORMATll' WELCOME TO THE USGS GRASP RETRIEVAL SYSTEM.'/ 0001920 1 ' AT THE CURRENT TIME THE FOLLOWING DATA BASES ARE AVAILABLE='1 0001921 2 FORMATll' BEFORE ANY OF THESE DATA BASES MAY BE ACCESSED,'/ 0001922 1 ' THE "FILE" COMMAND SHOULD BE USED TO IDENTIFY THE DATA', 0001923 2 ' BASE OF INTEREST.') 0001924 11 FORMAT(A5,1X,4A1094(lvaSI.141 0001925 12 FORMAT(/1XoA69'- '.4A101 0001926 END 0001927 GRASP SOFTWARE SPECIFICATIONS 85 FUNCTION NAME: UNCODE Common data referenced: None _ , Called by: COLPNT, DUMPIT, FIT, MEAN, RETRVE, Purpose: UNCODE breaks down each qualified real value into ROWPNT, EV AL a. real value and a qualifier code. Error checking and reporting: None Calling sequence: VALUEZUNCODE (VAL’ID) Program logic: The type REAL argument VAL may be visual- Arguments: ized as composed of both whole and fractional parts. ID is VAL—Packed value and qualifier. set to the 3 low-order bits of the whole part. The whole part ID—Encoding of qualifier value. is then shifted right 3 bits, and the result is added to the Subroutines called: None fractional part to form the value returned by the function. GRASP SOURCE PROGRAM FUNCTION UNCODElVAL,ID) 0001928 RSIGN=SIGN11.0.VAL) 0001929 VAL=ABSlVAL) 0001930 lPART=VAL 0001931 REST=VAL-IPART 0001932 lD=MODlIPART.8) 0001933 UNCODE=RSIGN*((lPART/8)+REST) 0001934 RETURN 0001935 END 0001936 86 SUBROUTIN E NAME: VLIST Purpose: VLIST prompts for and accepts (via KEYBRD) a set of item names or arithmetic expressions that will be printed by the caller. The user is provided the ability to reuse. the list that was last entered. Calling sequence: CALL VLIST(&m,VNAMES,LIST,N) Arguments: m—Statement number (in caller) to which a branch will be made if an EOF' is sensed by KEYBRD. VNAMES—List of item names to be printed. LIST-Item numbers corresponding to VNAMES. N—Number of elements in VNAMES and LIST. Subroutines called: KEYBRD, SCAN, BFIND, PREVAL, PACK Common data referenced: NAMES, PNTS, IDIM in blank common Called by: BDEF, COLPNT, ROWPNT Error checking and reporting: None Program logic: 1. The user is asked if he wishes to enter a new list of item GRASP SOURCE SUBROUTINE VLIST(*9VNAMES.LIST,N1 COMMON NAMES:TYPE,PNTS,IDIM DOUBLE PRECISION NAMES(500),NAME'B INTEGER PNTS1500’vTYPE15001yL15T11 ,LSAVE(201 EQUIVALENCE (HALVES(11yNAME1 DATA lEQUAL,BLANK/'='y' 'lyNS DATA EXPHDGI'1.'.‘2.','3.','4.'.'5 lo'v'12.'y'l3.'y'14.','15.','16.',' IF(NSAVE.EQ.0) GO TD 5 TYPE 4 CALL KEYBRDTEIIO’Iyll IF(I.NE.N0) GO T0 5 N=NSAVE DO 6 l=lyNSAVE LIST111=LSAVE111 VNAMES(11=VSAVE(12 GO TO 130 N=0 KNT=O TYPE 120 N=N+1 TYPE 140, EXPHDGTN) CALL KEYBR018110.EXPR,801 CALL SCAN15115,EXPRo1,L19L921 IF(L-L1.GT.61 GO TO 60 L2=L1+6 NAME=BLANK CALL PACK(EXPR(L11.NAMEyLZ-L1+1'81 CALL BFINDTEbOyNAMEvIvNAMEsaPNTSyI VNAMESTNiéNAME LIST(N1=I IF (N.EQ.201 RETURN GO TO 10 1 10 30 GEOLOGIC RETRIEVAL AND SNYPOSIS PROGRAM (GRASP) names. If not, the values of the arguments as set by a pre- vious call are returned. Otherwise, a new list is processed as below. . The routine requests the user to enter a set of item names or expressions. Each entry is processed as specified below until a blank entry is detected. Control is then returned to the caller. . An entry is accepted via KEYBRD, and leading and/or trailing blanks are eliminated via SCAN. If the entry length is greater than seven characters, an expression is assumed. If the entry is not found to be an item name (via BFIND), an expression is assumed. If the entry is determined to be an item name, that name is stored in VNAMES, and the corresponding item number is stored in LIST. . If the entry is an. expression, a call to PREVAL is made to parse it into Reverse-Polish form for later evaluation. KNT points to the ReversePolish form (stored in /EXPNS/ by PREVAL) , and the negative of KN T is returned in LIST. The values of the arguments are saved for future calls if required. PROGRAM 0001937 0001938 0001939 0001940 0001941 0001942 AVE9ND/09'N'l 0001943 .‘y'6.'.'7.','8.'.'9.‘,'10-'y'110001944 17.','18.','19.'y'20.'/ 0001945 0001946 0001947 0001948 0001949 0001950 0001951 0001952 0001953 0001954 0001955 0001956 0001957 0001958 0001959 0001960 0001961 0001962 0001963 0001964 0001965 0001966 0001967 0001968 0001969 0001970 LANKyVNAMEST1hVSAVE1201 l.EXPR(80hHALVES(21uEXPHDG‘ZO) DIM) 60 70 80 90 100 110 115 GRASP SOFTWARE SPECIFICATIONS IF (KNT.EQ.8) RETURN NAME=BLANK DO 70 K=1210 IF (EXPR(K+L1-1).EQ.IEQUAL) GO TO 80 CONTINUE K=1 HALVES(2)=EXPHDG(N) GO TO 90 J=MINO(K-l,7)*L1-1 CALL PACK(EXPR(L1)9NAME,J *L1+1,8) K=K+1 CALL PREVAL(8100,EXPR(K+L1-11vL'KNTD =~KNT GO TO 30 N=N-1 GO TO 10 RETURN 1 N=N'1 NSAVE=N DO 116 I=1yNSAVE LSAVE(1)=LIST(I) 116 VSAVE(I)=VNAMES(I) 130 RETURN 4 FORMAT(' DO YOU WISH TO ENTER A NEW LIST OF NAMES OR'v 1 ' EXPRESSIONS? (YES OR NO): 'yS) 120 FORMAT (' ENTER THE NAMES OF ITEMS OR THE EXPRESSIONS '9 1'HHICH YOU WANT PRINTED.') 140 FORMAT (1X.A4v$) END 87 0001971 0001972 0001973 0001974 0001975 0001976 0001917 0001978 0001979 0001980 0001981 0001982 0001983 0001984 0001985 0001986 0001987 0001988 0001989 0001990 0001991 0001992 0001993 0001994 0001995 0001996 0001997 0001998 0001999 75- 7 34541555 Trioctahedral Smectite in the Green River Formation, Duchesne County, Utah EARTH CIENCES .IBRARY GEOLOGICAL SURVEY PROFESSIONAL PAPER 967 MENTS DEPARTMENT DOCU JUL 9 1976 UBRARY UNNERSH‘! OF CALIFORNIA m 141976 '4} 5’ S494 Trioctahedral Smectite in the Green River Formation, Duchesne County, Utah By JOHN R. DYNI GEOLOGICAL SURVEY PROFESSIONAL PAPER 967 An X—ray-dszraction study of clay minerals in some lacustm'ne rocks of Eocene age exposed in the southwestern part of the Uinta Basin UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON : 1976 UNITED STATES DEPARTMENT OF THE INTERIOR THOMAS S. KLEPPE, Secretary GEOLOGICAL SURVEY V. E. McKelvey, Director Library of Congress Cataloging in Publication Data Dyni, John R. Trioctahedral smectite in the Green River Formation, Duchesne County, Utah. (Geological Survey Professional Paper 967) Includes bibliographical references. Supt. of Docs. No.: 119.16:967 1. Smectite—Utah—Duchesne Co. 2. Rocks, Sedimentary—Analysis. 3. Geology, Stratigraphic—Eocene. 4. X—rays—Diffraction. I. Title. II. Series. United States Geological Survey Professional Paper 967. QE391.SGD95 552’.s 76—43 For sale by the Superintendent of Documents, US. Government Printing Office Washington, DC. 20402 Stock Number 024—001—02821—2 CONTENTS Page Page Abstract .............................................. 1 Mineralogy—Continued Introduction ........................................... 1 Other minerals .................................... 10 Geologic setting ........................................ 1 Origin of smectite ..................................... 10 Method of investigation ................................. 4 Stratigraphic mineralogy ............................... 11 Mineralogy ............................................ 5 Basin lithofacies ....................................... 13 Clay minerals ...................................... 5 References cited ...................................... 13 FIGURE TABLE 1. 559°!“ ILLUSTRATIONS Page Map showing locations of composite measured section ....................................................... 2 Composite measured section of part of the Green River Formation ............................................ 3 Photograph showing units 3 — 18 exposed in roadcut along Utah Highway 33 ................................... 4 Photograph showing exposures of units 24 — 71 near South Fork Avintaquin Creek ............................. 4 Photograph showing exposures of units 54 —— 118 near South Fork Avintaquin Creek ............................ 4 X-ray diffraction patterns of oil shale sample 49 ............................................................. 6 X-ray diffraction patterns of claystone sample 64 ........................................................... 7 X-ray diffraction patterns of purified clay fractions of several oil-shale samples ................................. 8 Distribution of structural charge for some trioctahedral smectites ............................................. 9 Diagram showing the eastward increase in resistivity of some Green River rocks ................................ 12 Schematic cross section of rock types and clay minerals ...................................................... 13 TABLE S Page Qualitative mineralogy of 17 samples of oil shale and claystone .................................................. 5 Chemical analysis of oil-shale sample 17 ...................................................................... 9 Semiquantitative spectrographic analysis of oil-shale sample 17 ................................................. 9 Semiquantitative determination of zinc in 17 whole-rock samples of oil shale and claystone ......................... 12 III TRIOCTAHEDRAL SMECTITE IN THE GREEN RIVER FORMATION, DUCHESNE COUNTY, UTAH By JOHN R. DYNI ABSTRACT Abundant trioctahedral smectite of probable authigenic origin and smaller amounts of discrete illite comprise a simple assemblage of clay minerals in a basin-edge sequence of lacustrine claystones and oil shales in the Eocene Green River Formation, southwest Uinta Basin, Duchesne County, Utah. Small amounts of chlorite and mixed-layer clay are present in a few samples; kaolinite is absent. The smectite is well crystallized and has a high expandable-layer content. Chemical analysis of partly purified material suggests an iron-bearing hectorite which contains notable quantities of zinc (740 ppm (parts per million)), lithium (670 ppm), copper (175 ppm). and fluorine (0.85 percent). No significant differences were noted in the clay—mineral content of the claystones and oil shales. However, of the associated carbonate minerals. calcite predominates in the oil shales, whereas dolomite strongly predominates in the claystones. This relationship suggests that the interstitial waters of the claystones during deposition and (or) lithification were more saline than those of the oil shales. The trioctahedral smectite evidently formed in a magnesium-rich lacustrine environment and appears to be part of a basin edge—to-centér series of clay mineral facies—mixed detrital clay minerals to smectite to illite. INTRODUCTION This report presents the results of an X-ray-diffrac— tion study of 17 samples of oil shale and claystone collected from a sequence of lacustrine rocks in the Eocene Green River Formation, which is exposed on the southwest side of the Uinta Basin, Duchesne County, Utah. The objective of this study was to identify the clay and associated minerals in some claystones and oil shales. and to determine the differences, if any, in the mineralogy of the two rock types. Earlier studies (Hite and Dyni, 1967; Hosterman and Dyni, 1972) suggest that several major lithofacies of the Green River Formation and associated Tertiary rocks in Colorado and Utah have characteristic assemblages of clay minerals. For example, in the deeper part of the Piceance Creek basin, east of the Uinta Basin, in western Colorado, the oil-shale facies of the Green River Formation contains moderate to sparse amounts of illite, abundant dolomite, and sodium carbonate minerals, whereas an underlying lithofacies composed of dark kerogenaceous shale contains abundant illite and lesser amounts of carbonate minerals. Older fluvial mudstones and associated rocks of the Wasatch Formation on the margins of the basin contain a mixed assemblage of kaolinite, illite, and mixed—layer clay. Overlying the oil-shale facies, a thick sequence of chiefly fine-grained elastic rocks contains smectite and illite. These data suggest the possibility that clay mineral assemblages in a time-stratigraphic sequence of rocks of the Green River Formation change progressively from a shoreward fluvial environment to a saline lacustrine environment at basin center. With this idea in mind, a sequence of lacustrine rocks was picked for study that was thought to represent shallow-water basin-margin sediments (claystones) that alternate with deeper water anoxic lacustrine sediments (oil shales). Acknowledgment—The capable assistance of Robert W. Brown and Jerry D. Tucker in the sample preparation and X-ray work is gratefully acknow- ledged. GEOLOGIC SETTING The sequence of rocks studied includes most of the Parachute Creek Member of the Eocene Green River Formation as used by Dane (1955) in the southwestern part of the Uinta Basin, Utah. These rocks were described and sampled by the author and W. B. Cashion in 1969 (unpub. data) at two localities (fig. 1) near Utah Highway 33 in the vicinity of the divide between the Price River drainage to the southeast and the Strawberry River drainage to the north, about 25 miles (40 km) southwest of the village of Duchesne. In this area, the Parachute Creek Member forms a belt of outcropping rocks several miles wide that dip 2° -4° NE. beneath younger Eocene rocks into the Uinta Basin. The member forms a moderately rugged topography of steep shaly slopes and thin laterally persistent ledges of resistant oil shale. The composite section of rocks exposed at the two localities is 416 feet (126.8 m) thick (fig. 2). At the base is an asphaltic sandstone which forms the uppermost part of the delta facies of Bradley (1931). The top of 1 TRIOCTAHEDRAL SMECTITE IN THE GREEN RIVER FORMATION, UTAH 110°45’ R.8 Wt. Ouray Duchesnea \o I ‘ T. 6 S. UTAH u MAP LOCATION 39°55' 1 L ' T.11S. 39°52’30” FIGURE 1.—Localities of composite measured section in southwest Duchesne County, Utah. Localities photographed in figures 3, 4, and 5 are shown by < - Base from U.S. Geological Survey Gray Head Park and Jones Hollow topo- graphic quadrangles, scale 1:24.000, 1969 and 1968. GEOLOGIC SETTING _ Cumulative thickness, Thickness rSiZS Fm. Member . ' Unit Lithologic description In feet (and metres) Feet Metres ’13 118 1.7 0.52 Tuff e x L0 115—117 6.6 2.01 Oil shale E 8 93 40° _ —~~ 114 2.0 0.61 Limestone 3 5 a; (121-92) .._ 113 4.3 1.31 Claystone u, a, g _"’ 112 11.5 3.51 Limestone and claystone, interbedded g ‘5 Q \111 . 6.0 1.83 Claystona N f, ‘: 110 1.6 0.49 Limestone E 3 m ‘13 3'3? 8175212" '— D' OE.) 12.0 3.66 Claystone E 3.4 1.04 Oil shale and claystone 350 0.8 0.24 Oil shale (106.68) 8.4 2.56 Dolomite, siltstone, and claystone 5.4 1.65 Mostly claystone 8.4 2.56 Siltstone and claystone 9.0 2.74 Claystone 1.0 0.30 Dolomite 17.0 5.18 Claystone 5.2 1.59 Mostly oil shale; several thin tuffs 18.6 5.67 Siltstone and claystone, interbedded 300 12.8 3.90 Claystone and marlstone 4.8 1.46 Oil shale (91-44) — .. 82 4.9 1.49 Claystone and marlstone ' 80 81 4.8 1.46 Oil shale __' 79 0.2 0.06 Tuff 76—78 5.3 1.62 Marlstone and claystone Tar 75 0.5 0.15 on shale 7‘ 72—74 10.0 3.05 Marlstone 66—71 4.9 1.49 on shale. f5 J-TJ— 65 2.4 0.73 Marlstone g 250 64 4.0 1.22 Claystone .— (76.20) J— J— 63 0.7 0.21 Oil shale m- 7‘ ’13:!— 62 8.4 2.56 Claystone g g ‘1— —r 0.02 2.006 Tuff ': 12.0 .66 Claystone “J E g ‘L T 'L 58—59 3.8 1.16 Claystone, marlstone, and oil shale ._ a, 55—57 6.6 2.01 Siltstone; 0.3 ft (9.1 cm) tuff in middle 2 g E 54 4.3 1.31 Claystone : 5': a, /51—53 5.9 1.80 Siltstone, claystone and oil shale E 50 8.6 2.62 Claystone o "E: x (620036) ‘57.. _/ 02) 0.8 0.24 Oil shale uJ : 3’, ' _.,._ 48 2.3 0.70 Marlstone 3 5 47 2.0 0.61 Claystone \- Q .—.'—':—. 46 0.8 0.24 Tuff 0 g 45 10.7 3.26 Claystone .c ‘ 1- 1. TA 44 2.5 0.76 Limestone g 43 2.5 0.76 Concealed 5 42 10.9 3.32 Siltstone '3- 40—41 5.1 1.55 Siltstone, dolomite, and claystone 150 _ 39 7.0 2.13 Claystone (45.72) -- 38 0.2 0.061 Tuff - 7.0 2.13 Claystone — 0.1 0.031 Tuff 3.3 1.01 Claystone 4.8 1.46 Dolomite and limestone 9.4 2.87 Claystone 0. 0.24 Oil shale . 0.031 Tuff 6.10 Claystone 10° — 0.18 Tuff (30-48) 0.091 011 shale 1.52 Dolomite, crayey 2.62 Claystone 0.091 Tuff 1.89 Claystone, marlstone, and oil shale 0.40 Claystone and oil shale 1.68 Siltstone 4.88 Dolomite, clayey 50 _ 0.21 Oil shale (15.24) 0.67 Dolomite and marlstone ‘ 1.40 Clavstone 0.30 Oil shale 1.25 Dolomite; 0.1 ft (3.0 cm) tuff in middle 0.61 Claystone 1.74 Limestone, clayey 0.031 Oil shale 5.12 Dolomite 0.61 Siltstone, asphaltic 0 Delta facies 2.04+ Sandstone, asphaltic of Bradley (1931) FIGURE 2.—Composite measured section of part of the Green River Formation. Units 1— 28 were measured in a roadcut along Utah Highway 33 near the top of the divide between the drainages of Willow Creek and Left Fork Indian Canyon in the NE1/4SW‘I4 sec. 12, T. 11 S., R. 10 E., Duchesne County, Utah. Units 29— 118 were measured near South Fork Avintaquin Creek in sec. 27, T. 6 S., R. 8 W., Duchesne County. The section was measured and described by J. R. Dyni and W. B. Cashion in 1969 (unpub. data). The circled numbers indicate the units that were studied during the present investigation. 4 TRIOCTAHEDRAL SMECTITE IN THE GREEN RIVER FORMATION, UTAH the section is a bed of volcanic tuff, 1.7 feet (0.52 m) thick, which is the basal unit of the tuff zone of the Parachute Creek Member (Dane, 1955). The composite section of rocks consists of claystone and marlstone in beds commonly 2 —20 feet (0.6—6.1 m) thick; scattered siltstone beds of similar thickness; limestone and dolomite, especially in the lower 55 feet (16.8 m) of the section; and scattered thin ledge-form- ing beds of dark oil shale commonly a few tenths of a foot to a few feet thick. Throughout the section are many thin beds of orange-weathering tuff. The rocks exposed at the two localities are shown in figures 3-5. METHOD OF INVESTIGATION Seventeen fist-size samples of selected beds of oil shale and claystone collected at the two localities shown in figure 1 were prepared for X-ray-diffraction analysis as follows: One-half of the sample was ground to minus-60 mesh in a hammermill. A 5-g (gram) portion of this material was pelletized at 20,000 lb/in2 (pounds per square inch) in a hydraulic press and analyzed by conventional X-ray-diffraction methods. Another 15-g portion was prepared for clay-mineral analysis of the <2-1um (micrometre) fraction by the methods of Jackson (1973), which were modified as necessary to shorten sample preparation time. The <2-um fraction of each sample was treated in several ways to obtain specimens that were (1) magnesium saturated. (2) magnesium saturated and glycolated, (3) potassium saturated, and (4) potassium saturated and heated at 525°C. X-ray-diffraction traces were made of these specimens oriented on glass slides and of randomly oriented magnesium-saturated speci- FIGURE 3.—Units 3— 18 exposed in roadcut along the south side of Utah Highway 33 in sec. 12, T. 11 S., R. 10 E., Duchesne County, Utah. Units measured here but not shown in the photograph are equally well exposed in the same roadcut. FIGURE 4.—Exposures of units 24-71 on the north side of an unnamed valley tributary to South Fork Avintaquin Creek in sec. 27, T. 6 S., R. 8 W., Duchesne County, Utah. Most of the thin laterally persistent dark ledges are oil shale. FIGURE 5.—Exposures of units 54-118 on the north side of an unnamed valley tributary to South Fork Avintaquin Creek in sec. 27, T. 6 S., R. 8 W., Duchesne County, Utah. Most of the thin laterally persistent dark ledges are oil shale. mens. The <2-um fraction of sample 17 was additionally tested for trioctahedral smectite by the method of Greene-Kelly (described by MacEwan, in Brown, 1961, p. 190) and was semiquantitatively analyzed for major elements by X-ray spectrometer. The instrumental conditions for X-ray diffraction were nickel-filtered CuKa radiation generated at 34 kilovolts and 18 milliamperes with 1° scatter and divergence slits, 0.010-inch receiving slit, a 20 scan speed of 2° per minute, scintillation detector, and a counting rate of 3,000 counts per second. MINERALOGY 5 TABLE 1.—Qu.alitative mineralogy of oil-shale and claystone samples from Duchesne County, Utah [A, abundant; X. moderate to small amount; Tr., trace; 7. doubtfully present; . . .. absent] Clay minerals (determined in 2pm fraction) Other minerals (determined in whole-rock fraction) Smectite-to-illite Sam 1e . . Mixed Potassium Sodium , Dolomite-to-calcite No. p __—_—rat_lo___—__ Chlorite layer Quartz feldspar feldspar Analcime ratio Method A Method B Oilshale 7 ..... 85 15 86 14 A X ‘7 80 20 13 ..... 89 11 92 8 A X x 50 50 17 ..... 80 20 96 4 X1 A x 55 45 27 ..... 84 16 87 13 A X 21 79 49 ..... 91 9 91 9 A X 28 72 63 ..... 79 21 79 21 A X 14 86 68 ..... 80 20 74 26 A X X 21 79 80 ..... 82 18 80 20 A X X 22 78 104 ..... 86 14 76 24 A X X 62 38 Claystone 9 ..... 77 23 92 8 x2 . . A x X x 27 73 15 ..... 96 4 96 4 X i A Tr 100 0 25 ..... 79 21 90 10 X A X 80 20 37 ..... 87 13 88 12 A X 89 11 60 ..... 89 11 90 10 A T Tr. 73 27 64 ..... 79 21 92 8 A Tr Tr. 72 28 95A.... 95 5 92 8 A Tr. 100 0 109 ..... 81 19 90 10 . . . ? A X X 72 28 1 Mixed-layer clay computed with smectite and illite. 2 Chlorite computed with smectite. MINERALOGY The X—ray mineralogy of the 17 samples of oil shale and claystone is summarized in table 1. The clay minerals were determined in the <2-pm fraction, whereas the remaining minerals were determined from X-ray—diffraction analyses of the whole-rock sample. CLAY MINERALS Trioctahedral smectite and illite are the principal clay minerals found in both rock types. Figures 6 and 7 show X-ray-diffraction patterns of the < 2-pm fraction of typical samples of oil shale and claystone which were magnesium saturated, magnesium saturated and glycolated, potassium saturated, and potassium saturated and heated to 525°C. The smectites in both rock types behave similarly with various treatments; the basal 001 reflection of magnesium—saturated specimens expands from 14 A to 17 A on glycolation, and the potassium-saturated specimens show a 001 reflection at 12.1 — 12.6 A, which collapses to about 10 A on heating to 525°C. Applying the method of Reynolds and Hower (1970), the smectite was found to contain expandable 17 A layers ranging from 70 percent to usually in excess of 85 percent.1 The smectite in most samples shows good crystallinity, judging from the sharpness of the 001 reflection. The smectite was determined to be trioctahedral by its 060 reflection, which is close to 1.53 A in most samples (figs. 6, 7, and lower pattern of fig. 8), and by reexpansion of the basal 001 spacing to 17 A after heating and glycolation of a lithium—saturated specimen (fig. 8, upper pattern). This technique (Greene-Kelly, 1953) shows that lithium-saturated trioctahedral smectites and beidellite will reexpand to 17 A on glycolation after they have been heated between 200° and 300°C; lithium—saturated diocta- hedral montmorillonites collapse irreversibly on heating and will not reexpand on glycolation. The trioctahedral smectites of this study are distinguished from beidellite by the 060 reflection; for beidellite, this reflection has a spacing of about 1.49—1.50 A. The species of trioctahedral smectite in the samples studied could not be determined from the X-ray data alone. The diffraction pattern of an unoriented composite specimen of the <2—pm fraction free of nonclay impurities, which was prepared from oil—shale 1In table 3 of Reynolds and Hower (1970, p. 84), the d spacing for the 003 reflection for loo-percent expandable 17 A layers in the randomly stratified in the IM—ordered columns should read 5.67 A instead of 5.61 A, according to J. B. Hayes. Marathon Oil Co. (oral commun., 1973). TRIOCTAHEDRAL SMECTITE IN THE GREEN RIVER FORMATION, UTAH Q S 001/1001 Fand? Potassium saturated and Q S 002/1 002 heated at 525° C; oriented EXPLANATION S, smectite I, illite Q, quartz F, feldspar Fand? Potassium saturated; oriented Magnesium saturated, glycolated;oriented S 001 14.5 A Intensity (counts per second) 600 Magnesium saturated; oriented S 001 300 Magnesium saturated; random orientation S 13, 20 2.56-2.60 A Q Degrees 20 FIGURE 6.—X-ray diffraction patterns of 2- [.4 m fraction of oil-shale sample 49 treated in several ways. MINERALOGY S 001/1001 9.8 A Q 8003/1003 ~3.29 A\ S 002/1002 4.98 A Potassium saturated and heated at 525°C; oriented S 005/I 005 1.99 A EXPLANATION S, smectite I, illite Q, quartz D, dolomite F, feldspar . . I 002 Potassrum saturated; oriented 4 98 A Q 02- s 002 Q Q 5198 A S 001 Q ~ 17.0 A Magnesium saturated, glycolated; oriented S 006 ? 2.79 A D? s 003 Q 1002 5.57 A Intensity (counts per second) 600 I 005 300 0 Magnesium saturated; S 13 20 random orientation 2.58 A S 005 Degrees 20 FIGURE 7.— X—ray diffraction patterns of <2-um fraction of claystone sample 64 treated in several ways. 8 TRIOCTAHEDRAL SMECTITE IN THE GREEN RIVER FORMATION, UTAH S 001 17.0 A Sam le 17 EXPL N I P S A A.T ON Lithium saturated, glycolated, and , smectlte o . . . heated at 250 C; oriented I, 1111te S 002 8005/1003 8.8 A [001 Intensity measured on 3-35 A \ log scale (not shown) 8006 Intensity (counts per second) 600 300 27.6 A\ Composite of samples 13, 17, 27, and 49 001 23.2 A Sodium saturated; random orientation 12.6 A \20-5 A\ 0 02, 11 13, 20 06, 33 2.58 A ~ 003 1.53A 15, 24, 31 3-15-3-33A 1.50 A\ 1.7 A 1 L l | | | I | I I | | | | | | i | I l I l | | | I I | | I I I | | 70 60 40 20 2 Degrees 20 FIGURE 8.—X-ray diffraction patterns of purified <2- um clay fractions of several oil-shale samples. samples 13, 17, 27, and 49, is shown in figure 8 (lower pattern). Several low-angle reflections at 3°—5° 20 suggest some mixed-layer clay. The X-ray pattern resembles in a general way some patterns for high- magnesium— and low-aluminum-bearing trioctahedral smectites (hectorite and stevensite) reported by Faust, Hathaway, and Millot (1959). The low-angle reflections just mentioned suggest the possible presence of stevensite; however, the d spacings for the 060 reflections published by the aforementioned authors are consistently lower (1.52 A) than for the smectites reported here (1.53 A). According to X-ray data given by MacEwan (in Brown, 1961, p. 192) for some trioctahedral smectites, other possibilities include aluminous and iron-riCh saponite. The possibility of more than one species of trioctahedral smectite cannot be ruled out as suggested by the broadness (multiple peaks?) of the 02 reflection in many X-ray patterns. (See the patterns for the unoriented magnesium-saturated specimens in figs. 6 and 7.) A chemical analysis of a partly purified sodium-saturated specimen of the <2-ym fraction of oil-shale sample 17 is given in table 2, and an emission spectrographic analysis of a portion of the same specimen is given in table 3. Assuming that all of the K20 (potassium oxide) in the chemical analysis of sample 17 is in mixed-layer and discrete illite, and assuming an average composition of illite as determined by Weaver and Pollard (1973, table 3, p. 9), about 20 percent of the chemically analyzed specimen is illite and the remainder is trioctahedral smectite and probably some silicate impurities. (See ratios of smectite to illite for sample 17 in table 1.) The amounts of silica, lithium oxide, and fluoride in the calculated chemical composition of the illite-free smectite (table 2, rightmost col.) are close to the values reported for these constituents in stevensite by Bradley and Fahey (1962, table 1). However, the amounts of iron oxide and alumina reported here are considerably higher than the amounts found by Bradley and Fahey. A structural formula for the smectite calculated from the data in table 2 (rightmost col.) gave -0.04 —0.47 +0.47 41.43 [(A1 0.21F8327 Mgzna Zn 0.01 Lio.05)(Si 3.96Alo.o4 )010 (0H,F)z] (Na 0.45Cao.01) 2.57 MINERALOGY 9 TABLE 2.—Chemical analysis of a partly purified sodium- saturated specimen of the < 2- pm fraction of oil-shale sample 17 [Chemical analysis by Wayne Mountjoy, Violet Merritt, Johnnie Gardner. and G.T. Burrow] Oxides from fir“ Calculated composition column ass1gnable - of smectite free of 0x1de Percent to illitel illite (percent) (percent) 810 2 ............. 53.4 9.79 54.4 A1203 ............ 7.5 5.18 2.9 Total Fe as Fe203 . . 5.00 .98 4.99 MgO .............. 15.5 .54 18.7 030 .............. .17 .06 .14 N320 ............. 2.55 ... 3.18 K20 .............. 1.38 1.38 . . . F ................. .85 . . . 1.06 ZnO .............. .09 . . . .11 Li20 .............. .14 17 CuO .............. .02 . . . .02 Loss on ignition. . . . 13.2 1.69 14.4 Total ........ 99.8 . . . 100.1 1Composition of illite is based on the average of chemical analyses of 24 illites calculated by Weaver and Pollard (1973, table 3, p. 9). T ABLE 3.—Emission spectrographic analysis of a portion of the < 2-pm fraction of oil-shale sample 17 [Semiquantitative six»step spectrographic analysis by Nancy M. Conklin] Parts per Parts per Element Percent Element million Element million S1 .......... 10 Zn ........... 700 N1 ........... 20 Al .......... 5 L1 .......... 1 .500 Pb ........... 70 Fe ......... 3 Cu ........... 150 Sc ........... 7 Mg ......... 10 Mn .......... 70 Sr ........... 70 Ca ......... 15 Ag ........... 1 V ............ 150 Na ......... 3 B ............ 150 Y ............ 10 K .......... 3 Ba ........... 70 Zr ........... 30 T1 .......... 1 Be ........... 1 Ga ........... 20 Co ........... 10 Yb ........... 1 Cr ........... 70 NOTE—Looked for, but not found: As. Au. Bi Cd, La. Mo, Nb, Pd, Pt. Sb, Sn, Te, U, W, P, Ce, Ge, Hi, In, Re, Ta, Th. Tl, Eu. Because of the likelihood of impurities in the smectite and because of the lack of data for ferric and ferrous iron, the preceding formula should be treated circumspectly. Despite these limitations, the formula shows that the amount of aluminum substitution in the tetrahedral sheet is small and that occupancy of cation sites in the octahedral sheet is low. If these relationships are true, they indicate a stevensite- hectorite smectite rather than a saponite, which has considerably more aluminum substituting for silicon in the tetrahedral sheet and, consequently, also has a higher negative tetrahedral charge of about 0.25 to 0.85 (fig. 9). Saponite also differs from stevensite and hectorite by generally having a positive octahedral charge and a cation occupancy of the octahedral sheet close to 3.00. The occupancy of the octahedral sites as shown by the formula seems unusually low, although a value as low as 2.30 for a zincian trioctahedral smectite (sauconite) was reported by Weaver and Pollard (1973, table 40, analysis 3). Q EXPLANATION o Stevensite O Hectorite I Saponite V Sauconite TETRAHEDRAL CHARGE (-l A_i_\_.A-.7 0 O O O O O O . . o o 0 ‘d> '—> 0 19. 'v. v ‘9 '2 1.0 FIGURE 9.—Distribution of structural charge for some triocta- hedral smectites. The data were compiled from published analyses. 1, Stevensite of Bradley and Fahey (1962); 2, smectite of this report. Faust, Hathaway, and Millot (1959) have shown that stevensite and hectorite are closely related species of trioctahedral smectite but that they differ in chemical composition, in structural charge, and in other properties. Stevensite is essentially a magnesium silicate and has a low negative layer charge owing to a small deficiency of divalent octahedral cations. Hectorite is similar in composition to stevensite, but it also contains essential amounts of lithium, aluminum, and fluorine. Hectorite has a larger negative layer charge than stevensite because of substitution of lithium for divalent octahedral cations and because of a deficiency of octahedral cations as in stevensite (fig. 9). The distinction between the two minerals becomes less clear when the occurrence of stevensite in the 10 TRIOCTAHEDRAL SMECTITE IN THE GREEN RIVER FORMATION, UTAH Green River Formation reported by Bradley and Fahey (1962) is considered. This smectite contains appreciable aluminum, lithium, and fluorine, like hectorite, but it has a structural layer charge of only —0.06, which is typical of stevensite. The relatively high layer charge (-0.47) for the smectite of this study, as well as the presence of lithium, aluminum, and fluorine, suggest closer affinities to hectorite than to stevensite. It is concluded, therefore, that the smectite of this study is probably a hectorite. The nonexpandable layer clay mineral having a basal 001 spacing of 10 A is identified here as illite. The few broad reflections suggest that most of this material is illite in the sense of Hower and Mowatt (1966), but some of it may be a true mica, possibly biotite, which is commonly found in many of the thin beds of volcanic tuff scattered through the composite measured section ( J. R. Dyni, unpub. data, 1972). The relative abundance of illite to smectite was determined in two ways. In the first method (table 1, method A) the ratio of the relative intensity of the 001 reflection of loo-percent illite to 100-percent smectite in the glycolated magnesium—saturated specimens was assumed to be 1:45. A similar ratio was used indirectly by Schultz (1964, p. OH) to calculate the proportions of aluminous illite and aluminous mont- morillonite in the Pierre Shale. A similar 001 peak-height ratio was found in X-ray-diffraction analysis of some synthetic mixtures of illite, mont- morillonite, and kaolinite by Jarvis, Dragsdorf, and Ellis (1957, fig. 1). In the second method (table 1, method B), adopted from Droste (1961, p. 13, 15), the intensity of the 001 reflection of illite at 10 A is assumed to remain unchanged after heating to 525°C. (Droste heated his samples to 400°C.) The intensity of the reflection at 10 A is assumed to increase in proportion to the amount of smectite present when its crystal structure collapses to a basal spacing of about 10 A on heating. The percentage of illite is calculated from the ratio of the area of the 10 A reflection of the glycolated specimen (which provides maximum separation of the 001 reflections of illite and smectite) to the area of the same reflection after heating to 525°C multiplied by 100. In both methods smectite was calculated as the difference between 100 percent and the calculated percentage of illite. The results (table 1) show the preponderance of smectite, commonly 80 percent or more, in the clay mineral fraction in most samples. Mixed-layer clay was detected in a few samples. Because the mixed-layer clay collapsed to 10 A on heating to 525°C, all of it can be assumed to be mixed-layer illite-smectite. Chlorite was noted in one sample. Kaolinite was not detected in any of the samples. OTHER MINERALS Quartz is relatively abundant in all of the samples. Potassium feldspar and, in lesser amounts, sodium feldspar are present in small to trace amounts in many of the samples. Analcime was detected in two samples. Dolomite and calcite are major mineral components of the oil shales and claystones. Their relative abundance to each other was calculated by the method of Royse, Wadell, and Petersen (1971) by using peak heights as measured on the diffraction trace rather than by fixed time. Calcite is relatively more abundant in the oil shales than in the claystones; in five of nine samples of oil shale, calcite exceeds dolomite in abundance. Conversely, dolomite is more abundant in the claystones than in the oil shales; in all but one sample of claystone, dolomite greatly exceeds calcite in abundance. ORIGIN OF SMECTITE Trioctahedral smectite, which is considered to be sparse in sedimentary rocks (Weaver, 1967, p. 43), may be more abundant in sediments of continental closed basins than has been previously suspected. Papke (1970, 1972) described playa deposits of saponite at Ash Meadows, Nev., and sepiolite and associated saponite in Amargosa Desert, southern Nevada. Droste (1961) found smectite and illite to be the principal clay minerals in sediments from many playa lakes in the Mojave Desert, southern California. Although Droste (1961, p. 15) noted the presence of both dioctahedral and trioctahedral smectite, he did not distinguish between them but inferred that the latter type comprised a substantial part of the clay fraction at least in a few playas. Saponite in dry-lake sediments in western Australia was reported by Graham (1953), and hectorite from some Tertiary lake beds in eastern Morocco was described by Millot (1949) and Jeanette (1952). (Also see Faust, and others, 1959, p. 343.) ' In the Green River Formation, stevensite. first reported by Bradley and Fahey (1962), occurs in widely spaced thin layers through the saline facies of the Wilkins Peak Member in southwest Wyoming (Bradley and Eugster, 1969, p. B16). Tank (1969) reported abundant smectite and illite in oil shales and shales of the underlying Tipton Shale Member of the Green River Formation in the same area, but he did not distinguish between types of smectite. Considering these occurrences and the occurrences described in this report, trioctahedral smectite may be much more abundant in the formation than has been recognized. The trioctahedral smectite in the rocks studied may STRATIGRAPHIC MINERALOGY . 11 be authigenic. Magnesium-rich interstitial water, as evidenced by the presence of abundant (and probably authigenic) dolomite, may be an important factor in the origin of the smectite, as it seems to be for sepiolite and probably for associated saponites in a playa deposit described by Papke (1972, p. 213 — 214). The simple clay-mineral assemblage of smectite and illite, with the virtual exclusion of other clay minerals, might suggest an authigenic origin, although lack of clay minerals can be explained several ways. Therefore, it would be of value to know whether the discrete illite is also trioctahedral, which, if it is, would suggest a common origin or precursor. If the smectite is allogenic, its origin is problematic inasmuch as large quantities of trioctahedral smectite in probable source rocks are not known. Cretaceous marine shales and younger rocks on the San Rafael Swell are likely source rocks, but these probably contained dioctahedral smectite and other clay minerals, as they do elsewhere in the Rocky Mountain region (Schultz, 1964, p. C6; Weaver, 1961). Volcanic ash is a probable source for some of the smectite; however, the ash beds present in the composite section amount to only 1 or 2 percent of the total sequence. Even if the entire hydrographic basin of the Tertiary lake occupying the Uintah Basin is considered, the total amount of ash that could have been contributed to the lake by streams, plus that deposited directly in the lake, probably would not amount to more than two or three times the amount of ash now present in the composite section. Unless a source area which contained unusually large amounts of trioctahedral smectite is found, it becomes necessary to consider the formation of the smectite by reactions with other minerals or possibly by transformation of dioctahedral to trioctahedral smectite. Levinson and Vian (1966) synthesized probable trioctahedral smectite by reacting dolomite, quartz, kaolinite, and water at moderate temperatures, and Graham (1967) subsequently proposed that saponite in a surficial playa deposit near Marchagee, Western Australia, was formed by this method at ambient temperature and pressure. Transformation of dioctahedral to trioctahedral smectite has not, to the author’s knowledge, been demonstrated experiment- ally, but it would be worth trying. STRATIGRAPHIC MINERALOGY In Indian Canyon in the Uinta Basin, the sequence of rocks overlying unit 118 of the measured section in figure 2 to the base of the Horse Bench Sandstone Bed in the upper part of the Green River Formation is 324 feet (98.8 m) thick and consists predominantly of lacustrine claystones, thin beds of oil shale, and associated rock types similar to those of the measured section. The X-ray mineralogy of the rocks in this 324-foot-thick (98.8-m-thick) sequence, similar to that of the measured section, consists of dolomite, calcite, feldspars, quartz, illite, and abundant smectite. The smectite is probably trioctahedral, but this was not confirmed by X-ray analysis. Minor amounts of other clay minerals are also present. The smectite in the entire 740-foot-thick (225.6-m-thick) sequence of rocks from the base of the measured section to the base of the Horse Bench Sandstone Bed can be divided, with few exceptions, into five distinct units, on based on the spacing of the 001 reflection of the smectite, as follows: Feet (metres in parentheses) above Auemge spacing (A) base of total sequence of001 reflection of smecttte Base of Horse Bench Sandstone Bed. 721—740 (2198-2256) ................... 14.7 540—721 (164.6—219.8) ................... 12.6 299—540 (91.1—164.6) ................... 14.7 173—299 (52.7— 91.1) ................... 12.6 0—173 (0 — 52.7) ................... 14.7 The change in spacing in the basal reflection of smectite, if real, may reflect several episodic changes in the major interlayer cation vertically through the above sequence of rocks—that is, from calcium (?) smectite with a basal spacing of 14.7 A to sodium smectite with a spacing of 12.6 A. Possibly, this alternation in spacing may have been in response to changes in the salinity of the lake waters. As salinity increased, the dominant exchangeable cation became sodium, and as salinity decreased, calcium became the dominant cation. The sequence of rocks between the probable Mahogany oil-shale bed (units 80 —81 of the measured section in fig. 2, according to W. B. Cashion, oral commun., 1971) and the base of the Horse Bench Sandstone Bed is 484 feet (147.5 m) thick in Indian Canyon and can be traced in the subsurface in oil and gas test wells eastward to the vicinity of Ouray, Utah. This sequence of rocks, which contains abundant smectite and illite in Indian Canyon, thins and grades laterally into a lithofacies composed entirely of oil shale that contains no smectite, little or no illite, and, locally, abundant nahcolite and shortite. The Horse Bench Sandstone Bed thins eastward and is absent in the Ouray area; however, its approximate position in wells can be determined from geophysical well logs. The same sequence of rocks in the Western Oil Shale Corp. core hole EX—l, in the SW1/4SE1/4 sec. 36, T. 9 S., R. 20 E., Uintah County, about 57 miles (92 km) east of Indian Canyon, has thinned to 448 feet 12 TRIOCTAHEDRAL SMECTITE IN THE GREEN RIVER FORMATION, UTAH (136.6 m) and lies between the depths of about 1,854 and 2,302 feet (565.1 and 701.6 m). Smith, Trudell, and Robb (1972), who studied in detail the lithology and mineralogy of the core from this well, found that the sequence consists entirely of oil shale. They reported that throughout the sequence, as at Indian Canyon, dolomite, calcite, potassium feldspar, sodium feldspar, and quartz are abundant but, significantly, that smectite is absent and illite evidently is present only in the lower 250 feet (76.2 m) of the sequence. The last reported occurrence of illite upward in the sequence coincides roughly with the first appearance of nahcolite as in the Piceance Creek basin in Colorado (Hite and Dyni, 1967, fig. 4). N ahcolite becomes locally abundant in the nodules, masses, and thin beds in the upper 120 feet (36.6 m) of the sequence. Some shortite (Nag 003-20aCO3) and sparse neighborite NaMgF3) and searlesite (NaBSi206-H20) are also found in the nahcolite-bearing part of the sequence. The top of the nahcolite- and shortite-bearing oil shale extends several tens of feet higher than the sequence of rocks considered here. The eastward change from claystones and associated rocks to oil shale is reflected by increasing resistivities on electric logs of wells that penetrate these rocks (fig. 10). The predominantly claystone sequence in the vicinity of Indian Canyon shows a shale-like response on electric logs (Gulf 1 Indian Canyon well, in fig. 10), averaging about 20 ohms or less. The resistivity increases gradually eastward toward the Western Oil Shale Corp. core hole EX—l, and in the Sun 1 South Ouray well, a few miles to the northwest, the average apparent resistivity of these rocks has increased to about 156 ohms. This increase in resistivity is attributed to decreasing clay mineral content with a corresponding increase in kerogen and carbonate minerals. The area of the oil shale and associated sodium minerals marks a chemical depocenter of Lake Uinta during Green River time, which is younger than nahcolite-bearing oil-shale deposits in the Piceance Creek basin in Colorado and older than another saline sequence of lacustrine rocks at the base of the Uinta Formation near Duchesne, Utah. Cook (1973) and Vine and Tourtelot (1969) have noted that the amounts of many of the trace elements found in oil shale from the Green River Formation are less than would be expected for an organic-rich sedimentary rock. On the other hand, a few elements, including lithium, beryllium, and fluorine, show a twofold to tenfold increase over crustal abundance. The distribution and abundance of some trace elements found in oil shale may be controlled in part by basin lithofacies and the geochemical conditions of a)". .ELU Em U". ”TEE 8W3: Umwm gl—Dm E W 1"" r: ,_ 7%.),3. ’7 160p LUV-(50m w 00& ”F >0 - wamEd ‘5‘“ - , 'mw . > x; wEIS’Tée 5“ t g. «avg: 5.. > 120 A m_ Iggggs go — A . 9' l\. m A . ’— :§ § Sm Eiwl‘m‘” u} 5V)" U) 0 J‘s: m? CWL~ V a: qu—. —(/) >5 1' 0(1) 20):“ «a: OEN U) . o H a_ HP) N Lug gm LL03 00°) «1,; . 2.5015). [II 80 u. .4: 31,; E. 3 EU: =::' g 0 [US 5,“ 3. 2L0 :0 IL:O. cn in ._ O ‘— V . V F2 so I“? gm 2. 035m 2— 51—“ ..l- mg; (g 0.35“ w W- :5 m 4 95m [I w—F :1“ w EwN~ 40 - 0' m 0 <1 35 g .::‘_ gm 9L- 03 3’2 ”’ng v m < v e 0 l I l l O 10 20 30 40 SOMI O 20 40 60 80KM FIGURE 10.—Diag'ram showing the eastward increase of the averaged resistivity determined from the 16-inch normal curve for the sequence of Green River rocks between the Mahogany oil- shale bed and the Horse Bench Sandstone Bed in 10 wells between Indian Canyon and Ouray, Utah. their depositional environments. As noted earlier, trioctahedral smectite contains notably large amounts of lithium, zinc, and copper. These elements, and others, may have been trapped during the formation and deposition of the smectite before they could migrate basinward into the oil-shale facies. This could explain why copper and zinc are found in quantities less than anticipated in the oil shale. For example, zinc in the whole-rock fraction of the 17 samples listed in table 1 ranges from 32 to 121 ppm (table 4), a threefold to ninefold increase in abundance over that of pyrolyzed Green River oil shale analyzed by Cook (1973). If smectite served as a chemical trap for selected metals, perhaps such elements could be T ABLE 4.—Semiquantitative X—ray fluorescence analysis of zinc in whole-rock samples of oil shale and claystone [Analyses by J. R. Dyni} Oil Shale Claystone Zinc Zinc Sample No. (ppm) Sample No. (ppm) 7 ................... 102 9 ................... 108 13 ................... 38 15 ................... 32 17 ................... 88 25 ................... 36 27 ................... 75 37 ................... 55 49 ................... 43 60 ................... 42 63 ................... 71 64 ................... 54 68 ................... 121 95A ................. 41 80 ................... 59 109 ................... 76 104 ................... 74 REFERENCES CITED 13 Environment Fluvial Near-shore lacustrine Offshore lacustrine Chemical depocenter Salinity 0f Fresh Fresh to moderately saline Saline Hypersaline sediment waters . . Moderately to . Eh Oxidizing Mildly reducmg Strongly reducmg strongly reducing R k t Variegated mudstone, oc ype siltsto ne, and drab sandstone Cl minerals Various smectite, illite, mixed- ay layer chlorite and kaolinite Principal authigenic . . Calcite(?) matrix minerals Green claystone, some marlstone, and thin beds of oil shale Smectite (trioctahedral) and illite Mixed dolomite and calcite Oil shale and evaporites Kerogenaceous marlstone and oil shale Little illite or lllite . no clay minerals Dolomite, dawsonite and quartz Dolomite FIGURE 11.-—Schematic cross section of rock types and associated clay and authigenic matrix minerals found in the Green River Formation and related rocks, Uinta and Piceance Creek basins, Utah and Colorado. concentrated in economic amounts. In this regard, there are a number of units of high electrical conductivity in rocks above and below the Mahogany bed in several wells between Indian Canyon and Ouray. For example, in the Shamrock 2 Walton well, in the NW1/4NE1/4 sec. 14, T. 9 S., R. 16 E., Duchesne County, several such conductive zones are present between the depths of 2,420 and 3,130 feet (737.6 and 954.0 m). The top of the Mahogany bed in this well is at a depth of 2,769 feet (844.0 m). Perhaps these high conductivities are attributable to metal-bearing smectites or possibly to sulfide minerals, although sulfide minerals did not seem especially abundant in a few samples of rotary cuttings of one conductive unit that were examined. BASIN LITHOFACIES Regardless of the process responsible for the large quantities of trioctahedral smectite in the Green River Formation, the fact remains that this study, together with other published data, suggests a zonation of clay minerals from basin edge to center (fig. 11). The clay facies shown in figure 11 are not necessarily synchronous; the cross section simply shows their relative positions in the basin as they are now known. REFERENCES CITED Bradley, W. H., 1931, Origin and microfossils of the oil shale of the Green River formation of Colorado and Utah: U.S. Geol. Survey Prof. Paper 168, 58 p. Bradley, W. H., and Eugster, H. P., 1969, Geochemistry and paleo- limnology of the trona deposits and associated authigenic minerals of the Green River Formation of Wyoming: U.S. Geol. Survey Prof. Paper 496—B, 71 p. Bradley, W. H., and Fahey, J. J ., 1962, Occurrence of stevensite in the Green River Formation of Wyoming: Am. Mineralogist, v. 47, nos . 7—8, p. 996—998. Brown, G., 1961, The X-ray identification and crystal structures of clay minerals [2d ed.]: London, Mineralog. Soc. 544 p. Cook, E. W., 1973, Elemental abundances in Green River oil shale: Chemical Geology, v. 11, p. 321—324. Dane, C. H., 1955, Stratigraphic and facies relationships of the upper part of the Green River Formation and the lower part of the Uinta Formation in Duchesne, Uintah, and Wasatch Counties, Utah: U.S. Geol. Survey Oil and Gas Inv. Chart OC—52. Droste, J. B., 1961, Clay minerals in the playa sediments of the Mojave Desert, California: California Div. Mines Spec. Rept. 69, 19 p. Faust, G. T., Hathaway, J. C., and Millot, Georges, 1959, A restudy of stevensite and allied minerals: Am. Mineralogist, v. 44, nos. 3-4, p. 342-370. Graham, J ., 1953, An examination of clays from Marchagee and Cardabia, Western Australia: Royal Soc. Western Australia Joun., v. 37, p 91—95. Graham, J ., 1967, Formation of saponite from kaolinite, quartz and dolomite: Am. Mineralogist, v. 52, p. 1560—1561. Greene-Kelly, R., 1953, The identification of montmorillonoids in clays: Soil Sci. Jour. (London), p. 233-237. Hite, R. J ., and Dyni, J. R., 1967, Potential resources of dawsonite and nahcolite in the Piceance Creek basin, northwest Colorado: Colorado School Mines Quart, v. 62, no. 3, p. 25-38. Hosterman, J. W., and Dyni, J. R., 1972, Clay mineralogy of the Green River Formation, Piceance Creek basin—A prelim- inary study, in Geological Survey research 1972: U.S. Geol. Survey Prof. Paper 800-D, p. D159-D163. Hower, John, and Mowatt, T. C., 1966, The mineralogy of illites and mixed-layer illite/montmorillonites: Am. Mineralogist, v. 51, nos. 5-6, p. 825-854. Jackson, M. L., 1973, Soil chemical analysis—Advanced course [2d ed.]: pub. by the author, Madison, Wisconsin Univ., Dept. Soil Sci., 894 p. Jarvis, N. L., Dragsdorf, R. D., and Ellis, Roscoe, Jr., 1957, Quantitative determination of clay mineral mixtures by X-ray diffraction: Soil Sci. Soc. America Proc., v. 21, p. 257—260. 14 TRIOCTAHEDRAL SMECTITE IN THE GREEN RIVER FORMATION, UTAH Jeanette, Andrew-Pierre, 1952, Argiles smectiques et Rhassoul, in Geologies des Gites Minéraux Monocains: Internat. Geol. Cong., 19th, Algiers, 1952, Regional Mon. Ser. 3, Monoc no. 1, p. 371-382. Levinson, A. A., and Vian, R. W., 1966, The hydrothermal synthesis of montmorillonite group minerals from kaolinite, quartz, and various carbonates: Am. Mineralogist, v. 51, no. 3— 4, p. 495 — 498. Millot, Georges, 1949, Relations entre la constitution et la genese des roches sédimentaires argileuse: Geol. app. et prospection miniére, Nancy, v. 11, p. 199— 201. Papke, K. G., 1970, Montmorillonite, bentonite, and fuller’s earth deposits in Nevada: Nevada Bur. Mines Bull. 76, 47 p. 1972, A sepiolite-rich playa deposit in southern Nevada: Clays and Clay Minerals, v. 20, no. 4, p. 211—215. Reynolds, R. 0., Jr., and Hower, John, 1970, The nature of inter- layering in mixed-layer illite-montmorillonites: Clays and Clay Minerals, v. 18, no. 1, p. 25—36. Royse, C. F., Jr., Wadell, J. S., and Petersen, L. E., 1971, X-ray determination of calcite-dolomite—An evaluation: Jour. Sed. Petrology, v. 41, no. 2, p. 483—488. Schultz, L. G., 1964, Quantitative interpretation of mineralogical composition from X-ray and chemical data for the Pierre Shale: U.S. Geol. Survey Prof. Paper 391—0, 31 p. Smith, J. W., Trudell, L. G., and Robb, W. A., 1972, Oil yields and characteristics of Green River Formation oil shales at WOSCO EX — 1, Uintah County, Utah: U.S. Bur. Mines Rept. Inv. 7693, 145 p. Tank, Ronald, 1969, Clay mineral composition of the Tipton Shale Member of the Green River Formation (Eocene) of Wyoming: Jour. Sed. Petrology, v. 39, no. 4, p. 1593— 1595. Vine, J. D., and Tourtelot, E. B., 1969, Geochemical investigations of some black shales and associated rocks: U.S. Geol. Survey Bull. 1314 —A, 34 p. Weaver, C. E., 1961, Clay mineralogy of the Late Cretaceous rocks of the Washakie Basin, in Wyoming Geol. Assoc. 16th Ann. Field Conf. Guidebook, Symposium on Late Cretaceous rocks of Wyoming: p. 148 — 154. 1967, The significance of clay minerals in sediments, Chap. 2, in Nagy, B., and Columbo, U., eds., Fundamental aspects of petroleum geochemistry: Amsterdam and New York, Elsevier Publishing Co., p. 37 — 75. Weaver, C. E., and Pollard, L. D., 1973, The chemistry of clay minerals—Developments in sedimentology 15: New York, Elsevier Sci. Publishing Co., 213 p. trus, GOVERNMENT PRINTING OFFICE: 1976—677-340/77 3123/ 0% EARTH SCIENCES mm w 7 DAYS The Paleontology of Rostroconch Mollusks and the Early History of the Phylum Mollusca GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 ‘ ENT LleARY umvrmlrv or nAHrnnNM WM The Paleontology of Rostroeonch Mollusks and the Early History of the Phylum Mollusca By JOHN POJETA, JR, and BRUCE RUNNEGAR GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 An analysis of the functional morphology, taphonomy, stratigraphic ranges, phylogenetic relationships, and taxonomic variability of rostroconch mollusks, and the evolutionary history of other early mollusks UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON 2 1976 UNITED STATES DEPARTMENT OF THE INTERIOR THOMAS S. KLEPPE, Secretary GEOLOGICAL SURVEY V. E. McKelvey, Director Library of Congress Cataloging in Publication Data Pojeta, John The paleontology of rostroconch mollusks and the early history of the phylum Mollusca. (Geological Survey professional paper ; 968) Bibliography: p. Includes index. 1. Rostroconchia. 2. Mollusks—Evolution. 3. Paleontology—Paloezoic. I. Runnegar, Bruce, joint author. II. Title. III. Series: United States. Geological Survey. Professional paper ; 968. QE814.P64 564 76—18911 For sale by the Superintendent of Documents, U.S. Government Printing Office Washington, D.C. 20402 Stock Number 024-001-02822-1 CONTENTS Abstract . Introduction Acknowledgments _________________________________ Functional morphology ____________________________ Orientation ___________________________________ Larval shell __________________________________ Metamorphosis Subsequent shell growth _______________________ Summary of shell growth ______________________ Opening of the valves ________________________ Musculature __________________________________ Pedal musculature ________________________ Pallial musculature _______________________ Alimentary canal _____________________________ Feeding structures ____________________________ Cleaning the mantle cavity ____________________ Water currents and gills ______________________ Function of the hood __________________________ Taphonomy Phylogeny Origin of the Mollusca ________________________ The ancestral mollusk _________________________ The oldest known fossil mollusks ________________ Early Cambrian univalves _____________________ Radiation of the Monoplacophora _______________ Phylogeny—Continued Origin of the Rostroconchia—Continued Origin of the Pelecypoda ______________________ Bivalved condition in the Mollusca __________ Limits of convergence in shell form ________ Rostroconchs as ancestors of the. Pelecypoda“ Accompanying modification of body form _____ Radiation of the Pelecypoda ________________ Origin of the Scaphopoda ______________________ Mattheva and Stenothecoida (Probivalvia) ______ Molluscan subphyla ___________________________ Glossary of morphological terms ____________________ Systematic paleontology ____________________________ Synoptic classification of known rostroconch mollusks Ribeirioida ___________________________________ Ribeiriidae ________________________________ Technophoridae ___________________________ Ischyrinioida _________________________________ Ischyriniidae Conocardioida Eopteriidae ______________________________ Conocardiidae _____________________________ Bransoniidae Hippocardiidae Incertae sedis References cited __________________________________ Index ____________________________________________ ILLUSTRATIONS [Plates follow intdex J Origin of the Gastropoda ______________________ Origin of the Cephalopoda _____________________ Origin of the Rostroconchia ____________________ Radiation of the Rostroconchia ____________ PLATE 1. Yochelcionella. 2. Heraultipegma. 3. Watsonella, Wanwam'a, and Ribeirina. 4—8. Ribeim'a. 9. Ribeiria and Pinnocaris. 10. Pinnocaris, “Technopho'rus,” Myocam's, Oepikila, and Technophorus. 11—13. Technophoms. 14‘. Technophorus and Tolmacho'via. 15. ?Ma,c'roscenella. 16. Tolmachovia. 17. Tolmachom'a and Anabarella. 18. Anisotechnophorus and Ischyrinia. 19. Ischym'm'a. 20. Pseudotechnophorus. 21. Pseudoeuchasma, Wanwanoidea, and Wanwanella. 22-26. 27—29. Eopteria. Euchasma. III Page 40 40 40 41 42 43 43 44 44 45 47 47 48 48 56 62 62 65 65 69 71 74 77 78 85 IV PLATES 30, 31. 32. 33. 34. FIGURE TABLE 35'. 36. 37. 38, 39. 40. 41. 42!. 43. 44. 45—49. 50. 51. 52~54. PWflQFPWP?‘ CONTENTS Ribe’im. Pseudoconocurdium, Hippoca’rdz'a, and Bransonia. Hippocardia. Mulceodens. Mulceodens and Bigalea. Bigalea. Bigalea and Conoca'rdium. Conocardiu'm. Conocardium and Pseudoconocardium. Pseudoconocaroh'um. Pseudoconocardium and Arceodomus. Arceodomus. Arceodomus and Hippocwrdia. Hippocardia. Hippoca/rdv'a and Bransom'a. Bransom'a and Conoca'rdium. Brunsom'a. Page Paper cutout models to illustrate the growth of the ribeirioid shell ______________________________ 6 Schematic diagrams illustrating the posterior growth of conocardiid and pelecypod shells ____________ 8 Musculature of various genera of Ischyrinioida and Ribeirioida __________________________________ 16 Composite diagram of musculature of Eopterm ventricosa. _______________________________________ 17 Musculature of various genera of the Conocardiacea _____________________________________________ 18 Origin of the ribeiriid rostroconchs ___________________________________________________________ 25 Variation in shell form of Cambrian univalves _________________________________________________ 26 Shell muscle insertions of cyclomyan monoplacophorans ___________________________________________ 2.7 Reconstructions of Helcionella and Yochelcionella _______________________________________________ 28 Reconstruction of Scenella ___________________________________________________________________ 29 Diversity of form in the Rostroconchia ________________________________________________________ 36 Reconstructions of Fordilla ___________________________________________________________________ 42 Schematic view of the origin of molluscan classes _______________________________________________ 45 Historical record of the initial radiation of the Mollusca ________________________________________ 46 TABLES Page Range chart showing the known stratigraphic distribution of all species of Ribeiriidae recognized herein _________________________________________________________________________________ 49 Range chart showing the known stratigraphic distribution of all species of Technophoridae recog- nized herein ____________________________________________________________________________ 57 Range chart showing the stratigraphic distribution of all species of Ischyriniidae and Eopteriidae‘ rec- ognized herein .......................................................................... 63 THE PALEONTOLOGY OF ROSTROCONCH MOLLUSKS AND THE EARLY HISTORY OF THE PHYLUM MOLLUSCA By JOHN POJETA, JR., and BRUCE RUNNEGAR 1 ABSTRACT The class Rostroconchia is known in the fossil record from the Early Cambrian to the Late Permian; the taxa herein included within it have not previously been recognized as a biological entity and are grouped together for the first time. We functionally analyze the morphology of rostroconchs as to orientation, modes of growth, method of opening the valves, musculature, feeding structures, and so forth, and conclude: that the group has a common biological pattern which indi- cates a commonality of descent. Thus, the Rostroconchia are treated here as a separate and extinct class of the phylum Mollusca. Phylogenetically, it is possible to show that mollusks began to diversify and radiate in the Early Cambrian and that at this time the Monoplacophora gave rise to the Gastropoda and the Rostroconchia. We present evidence that the helcionella— cians, formerly regarded as gastropods, are monoplacophorans and that they gave rise to the Rostroconchia. The Rostrocon- chia in turn gave rise to the Pelecypoda in the late Early Cambrian and possibly to the Scaphopoda in the Ordovician. Tyhe rostroconchs underwent a major radiation in the Early Ordovician, at which time they were as diverse as the Pele- cypoda. Only one order of rostroconchs survived the Ordovi- cian, a fact that we attribute to the competition between ros- troconchs and pelecypods for living space and food. All the known Cambrian and Ordovician species of rostro— conchs are described, discussed, and documented stratigraphi- cally. We did not have adequate collections of post-Ordovician material to analyze all species, and we limit our systematic and stratigraphic considerations to those forms that show the generic diversity of the Conocardioida, the single known post- Ordovician order. For the practicing stratigrapher, we present a new tool, a group of organisms heretofore neglected stra- tigraphically because they were not recognized as a biological entity. The following new taxa are proposed: families—Bran- soniidae and Hippocardiidae; genera—Anisotechnophoms, Arceodomus, Bigalea, Bransom'a, Heraultipegma, and Mulceo- dens; species—Ribeim'a austmliensis, R. taylo'ri, Pinnocaris americana, Technophorus marija, T. millem', Tolmachovia? jelli, Eopteria conocardiformis, Eucha‘sma jonesi, E. mytili— forme, Conooardium pseudobellum, Bransom'a wilsom', B. ala- bamensis, B. cress’mani, Mulceodens jaanussom', Hippocardia coope'ri, Bigalea yangi, B. ohioensis, and B. m'sbyens'is. 1 University of New England, Armidale, New South Wales, Australia. INTRODUCTION Rostroconchs are a small but widespread and per- sistent Paleozoic faunal element. To date, they have been little studied because of the lack of adequate material and because they were not recognized as a separate molluscan lineage. It has been general prac- tice to treat the older members of the class as arthro- pods and the younger members as unusual pelecypods. The oldest known rostroconchs are assigned to the genus H emultipegma. ( =Heraultia Cobbold) and are from Lower Cambrian rocks in France (Cobbold, 1935). According to Waterhouse (1967), the rostro- conch "Conocardium" occurs in the Makarewan Stage of New Zealand, which is placed at the top of the Permian System. Thus, rostroconchs range throughout the Paleozoic. We agree with Morris (1967) that the Triassic species placed in Conocm'di- um by Healy (1908) probably belong to the pelecy- pod groups Poromyacea or Burmesiidae. Herein, we review the paleontology of the rostro- conchs throughout their stratigraphic range and in- dicate the importance of early rostroconchs in the phylogeny of the Mollusca. In order to study any group throughout its stratigraphic range, it is neces- sary to have sizable collections. It was obviously not possible to collect the entire Paleozoic throughout the world, so we turned to the museums of four con- tinents and borrowed specimens from more than 30 institutions. We examined more than 3,600 speci- mens ranging in age from Early Cambrian to Late Permian. The study of what are now called rostroconchs be- gan when Martin (1809) described one species and Sowerby (1815) described two species. Martin’s work was subsequently declared invalid for nomen- clatural purposes (Hemming, 1954, ICZN Opinion 231). Bronn (1835) named the genus Conocardium, citing a single species name, Cardium elongatum 2 PALEONTOLOGY 0F ROSTROCONCH MOLLUSKS Sowerby, which is the type species of the genus by monotypy. The name Conocardium was subsequently used for almost all Silurian-Permian species and some Ordovician species. At least 275 species were assigned to the genus. Herein we subdivide the genus Conocardium into seven generic-level taxa. To date, the study of rostroconchs has largely con- sisted of the description of species, little attempt having been made at interpretation above this level. Major monographs, summaries of species, or bibli- ographies of rostroconchs include: Babin (1966), Barrande (1881), Beushausen (1895), Branson (1942a, b; 1948; 1966), Fletcher (1943), Hall (1885), Hind (1900), Kobayashi (1933), LaRocque (1950), Paul (1941), Pohl (1929), Schubert and Waagen (1904), Ulrich (1894), Weller (1898), and Whidborne (1892). It was not recognized until recently that conocar- diaceans are allied to eopteriids (Pojeta, 1971) and ribeirioids (Morris, 1967; Pojeta and, others, 1972) and that all three groups are neither pelecypods nor arthropods. Previously, the conocardiaceans had consistently been treated as pelecypods and usually allied to the cardiids. In the Treatise on Invertebrate Paleontology, Branson, LaRocque, and Newell (1969) regarded them as pelecypods, but placed them in their own order and regarded the subclass assignment as uncertain. Poj eta (1971) placed Cona- cardium, Eopteria, and Euchasma in a separate sub— class of pelecypods and noted that the rostroconchs formed an enigmatic group whose pelecypod affini- ties were not well established. In 1972, Poj eta, Run- negar, Morris, and Newell made the rostroconchs a separate class of mollusks, which then included four genera; they also noted that rostroconchs were allied to the ribeirioids. Although most ribeirioids and their allies have at one time or another been regarded as mollusks, usually pelecypods, in the 20th century they have been consistently placed with the arthropods. Schu- bert and Waagen (1904) argued against a pelecy- pod placement of Ribeiria and came to the conclusion that it was an apodid arthropod that has been much compressed laterally. Kobayashi (1933) accepted the arthropod placement of Ribeirr'a. and its allies and treated them as notostracans. The molluscan nature of ribeirioids is indicated by the presence of a protoconch (pl. 41; pl. 47, figs. 13—15), by the cal— careous shells which have growth lines (pl. 5, figs. 13, 14), and by the prominent muscle scars which also show the growth increments (pl. 6, figs. 1, 4, 14) . Because of the need to accumulate large numbers of specimens from many museums, the names of the museums are subsequently abbreviated as follows: AMS, Australian Museum, Sydney; AM, American Museum of Natural History, New York; BM, British Museum (Natural History), London; BMR, Aus- tralian Bureau of Mineral Resources, Canberra; FM, Field Museum of Natural History, Chicago, 111.; GB, Institute of Geological Sciences, London, England; GSC, Geological Survey of Canada, Ottawa, Ont.; MCZ, Museum of Comparative Zoology, Harvard University, Cambridge, Mass; MU, Miami Univer- sity, Oxford, Ohio; NYSM, New York State Muse- um, Albany; PRI, Paleontological Research Institu- tion, Ithaca, N.Y.; ROM, Royal Ontario Museum, Toronto; SM, Sedgwick Museum, University of Cambridge, England; SMNH, Swedish Museum of Natural History, Stockholm; UCB, University of California, Berkeley; UCM, University of Cincin- nati Museum, Cincinnati, Ohio; UI, University of Illinois, Urbana; UM, University of Michigan, Ann Arbor; UMN, University of Minnesota, Minneapolis, Minn.; UNE, University of New England, Armidale, New South Wales, Australia; UO, University of Oslo, Norway; UOK, University of Oklahoma, Nor- man; UQ, University of Queensland, Brisbane, Aus- tralia; USNM, United States National Museum, Washington, D.C.; YU, Yale University, New Haven, Conn. ACKNOWLEDGMENTS We are deeply indebted to museum curators and collection managers on four continents. Without their active help and cooperation we would not have been able to pursue this study. Large numbers of specimens were lent to us from many institutions, each of which we gratefully acknowledge. We would like to thank the following persons for placing specimens and type material at our dis- posal: A. Ritchie, Australian Museum, Sydney; N. D. Newell, American Museum of Natural History, New York; A. A. Opik and J. H. Shergold, Australi- an Bureau of Mineral Resources, Canberra; L. F. Hintze, Brigham Young University, Provo, Utah; R. J. Cleevely and N. J. Morris, British Museum (Natural History), London; J. Wyatt Durham and J. H. Peck, University of California, Berkeley; C. L. Forbes, C. P. Hughes, R. B. Rickards, and H. B. Whittington, University of Cambridge, England; T. E. Bolton, M. J. Copeland, and W. T. Dean, Geologi- cal Survey of Canada, Ottawa; K. E. Caster and R. A. Davis, University of Cincinnati, Cincinnati, Ohio; ACKNOWLEDGMENTS 3 S. J. Gould, Bernhard Kummel, and R. D. K. Thomas, Museum of Comparative Zoology, Harvard Univer- sity, Cambridge, Mass.; J iri Kriz, Central Geological Survey, Prague, Czechoslovakia; K. Krueger and M. H. Nitecki, Field Museum of Natural History, Chicago, 111.; Jochen Helms, Humboldt University, Berlin, Deutsche Demokratische Republik; A. W. A. Rushton, Institute of Geological Sciences, London, England; D. B. Blake, University of Illinois, Ur- bana; A. S. Horowitz, Indiana University, Blooming- ton; E. C. Wilson, Los Angeles County Museum, Los Angeles, Calif.; T. E. Yancey, University of Malaya, Kuala Lumpur; J. H. Marak and J. K. Pope, Miami University, Oxford, Ohio; D. B. Macurda, J r., University of Michigan, Ann Arbor, Mich.; P. K. Sims, Minnesota Geological Survey, St. Paul; R. E. Sloan, F. M. Swain, and D. Wallace, University of Minnesota, Minneapolis; R. H. Flower, New Mexico Institute of Mining and Technology, Socorro; B. M. Bell, Robert Conrad, and D. W. Fisher, New York State Geological Survey, Albany; S. M. Bergstrom and W. C. Sweet, Ohio State University, Columbus; T. W. Amsden, Oklahoma Geological Survey, Nor- man; R. M. Bonem and C. C. Branson, University of Oklahoma, Norman; D. H. Collins, M. A. Fritz, and John Monteith, Royal Ontario Museum, Toronto; D. L. Bruton and G. Henningsmoen, University of Oslo, Norway; K. V. W. Palmer, Paleontological Research Institution, Ithaca, N.Y.; Robert Hansman, Prince- ton University, Princeton, N.J.; D. Hill and J. S. J ell, University of Queensland, Brisbane, Australia; J. F. Bokkelie and Valdar Jaanusson, Swedish Mu- seum of Natural History, Stockholm; T. Kobayashi and T. Hanai, University of Tokyo, Japan; R. J. Ross, US. Geological Survey, Denver, 0010.; Leon- ard Alberstadt, Vanderbilt University, Nashville, Tenn.; and A. L. McAlester, Yale University, New Haven, Conn. For helpful conversations, correspondence, or for reading the manuscript we would like to thank; J. H. Shergold and J. Gilbert-Tomlinson, Australian Bu- reau of Mineral Resources, Canberra; John Taylor, British Museum (Natural History), London; J. W. Wells, Cornell University, Ithaca, N.Y.; David Nicol, University of Florida, Gainesville; C. M. Yonge, University of Glasgow, Scotland; L. S. Kent, Illinois Geological Survey, Urbana; E. R. Trueman, University of Manchester, England; A. Graham and V. Fretter, University of Reading, England; A. R. Palmer, State University of New York, Stony Brook; A. J. Boucot, Oregon State University, Corvallis; J. C. Brower, Syracuse University, Syracuse, N.Y.; D. Herm, University of Tiibingen, Federal Republic of Germany; M. E. Taylor and E. L. Yochelson, US. Geological Survey, Washington, D.C.; E. G. Kauff— man, US. National Museum, Washington, D.C.; Anders Martinsson, University of Uppsala, Sweden; D. L. Clark and Klaus Westphal, University of Wis- consin, Madison. For the photograph of Anabarella plana we are indebted to Robin Godwin and S. C. Matthews, Uni- versity of Bristol, England. S. C. Mathews provided specimens of H emultipegma for this study and Douglas Lorenz, University of California, Los An- geles, supplied two specimens of Technophoms from Edenian rocks of Kentucky. Scanning electron microscope photographs of H eraultz'pegma were sup- plied by the University of Cambridge, England. FUNCTIONAL MORPHOLOGY ORIENTATION In any group of extinct metazoans, the problem of morphological orientation exists. Because rostro- conchs are extinct mollusks, the correct orientation of their skeleton is not immediately obvious. Their orientation can be established by comparing rostro- conchs with other groups of mollusks and by consid- ering their probable phylogenetic relationships. Pre- vious discussions of this topic have assumed that rostroconchs were either pelecypod mollusks or arthropods, whereas we treat rostroconchs as a sep- arate class of mollusks. Various recent workers (Branson, 1965, 1966; LaRocque, 1950; Nicol, 1970; Wilson, 1970) have oriented the Conocardiacea as we do. Rostroconchs are bilaterally symmetrical about the commissural plane which separates the valves into mirror images (pl. 40, figs. 4—7). By analogy with other bivalved mollusks, we regard the hinged part of the shell as dorsal (pl. 40, fig. 5). Our phylo- genetic conclusions, which indicate an evolutionary relationship of the rostroconchs to the Monoplaco- phora on the one hand and the Pelecypoda on the other, reinforce this interpretation. Many older workers (for example, Ball, 1913) regarded the rostrate end of the conocardiacean shell as anterior. This was by analogy with pelecypods, at least in part. In most pelecypods, the direction of coiling of the umbos is toward the anterior, and in conocardiaceans, the umbos coil toward the rostrate end. However, most recent workers regard the ros- trate end of rostroconchs as posterior and the gaping end as anterior (LaRocque, 1950; Wilson, 197 0). Conocardiaceans have a relatively large gape at one end of the shell (pl. 40, figs. 6, 11), and a narrow 4 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS tubular rostrum at the other (pl. 43, figs. 5, 6). The most logical explanation of the large gape is that it allowed for the protrusion of the foot from the shell. The large gape is therefore anatomically anteroven- tral and functionally anterior. This interpretation is supported by the existence of analogs among the pelecypods, many of which have large anterior pedal gapes. In addition, most primitive rostroconchs pos- sess a large anterior plate or pegma that connects the valves dorsally (pl. 4, figs. 20—22) . By analogy with scaphopods, pelecypods, and some gastropods, the rostrum of rostroconchs allowed wa- ter and excretory products to enter and leave the mantle cavity; it was most likely posterior. The morphology of advanced conocardiaceans, when traced back through morphologically gradational rostroconchs to the most primitive members of the class, reinforces the interpretation of the orientation of the anterior—posterior axis on the basis of the postulated phylogenetic connection of the Ros- troconchia with the Monoplacophora and the Pelecypoda. As in pelecypods (Cox and others, 1969), ventral is more difficult to define. If it be taken to coincide with the sole of the protracted foot, then the large (anterior) shell or pedal gape of many rostroconchs should be anatomically ventral. We adopt a more geometric view and term the margin of the shell op- posite the hinge as ventral. We thus treat the four coordinates, anterior, posterior, dorsal, and ventral, as mutually orthogonal directions in the commissural plane. All are related to the hinge, which is fixed as dorsall Like pelecypods, rostroconchs may have umbos that coil anteriorly (prosogyral, pl. 28, fig. 13) or posteriorly (opisthogyral, pl. 40, fig. 5). The direc- tion of coiling is related to the geometry of shell growth (as in the Pelecypoda) and has nothing to do with the orientation of the shell. As a general rule, conocardiaceans are anteriorly expanded, have an anterior gape, and a posterior rostrum (pl. 40, figs. 5—14) ; most other rostroconchs are posteriorly expanded and have an anterior peg- ma (pl. 11, fig. 22). LARVAL SHELL The valves of rostroconchs have only a single beak because growth originates from a single cap-shaped larval shell situated between the umbos of the juve- nile shell (pl. 41; pl. 47, figs. 13—15). This structure, termed the protoconch (Pojeta and others, 1972), is normally destroyed in conocardiaceans by crushing caused by inrolling of the umbos. In ribeirioids, it is usually visible at the apex of well-preserved speci- mens (pl. 10, fig. 16; pl. 12, fig. 17). The boundaries of the protoconch are not easily identified, but in most forms the protoconch appears to have been about 300—600 microns in diameter. Its size com- pares favorably with the larval shell of other mol- lusks; the prodissoconch II (veliconch) of pelecy- pods is 200—600 microns in diameter (Cox and oth— ers, 1969, p. N95), and the protoconch of bivalved gastropods is about 250 microns in diameter (Ka- waguti, 1959). In the conocardiaceans, the protoconch is a limpet- shaped structure, which is separated from the ad- jacent valves by shallow concave areas formed as a necessary consequence of the change from a uni- valved to a bivalved shell (pl. 41, figs. 1—5). Radial ribs first appear after this change has taken place (pl. 41, fig. 5), and the protoconch seems to have been relatively smooth. In the Ribeiriidae and Tech- nophoridae, the valves are less inflated than in cono- cardiaceans, so that the distinction between the pro- toconch and the juvenile shell is less clear (pl. 4, figs. 20, 21, 23). Well-preserved internal molds of all genera often have a small cone-shaped elevation in the center of the hinge area; we interpret this to be a natural cast of the interior of the protoconch and early juvenile shell (pl. 22, fig. 11; pl. 23, fig. 1). The protoconch or its natural mold has been ob- served in the following genera: Anisotechnophorus (pl. 18, figs. 7—9); B‘ransom'a (pl. 52, figs. 3—5); Eoptem‘a (pl. 22, fig. 11); Pseudoconocardium (pl. 41, figs. 1—5) ; Pseudotechnophorus (pl. 20, figs. 13— 15) ; Ribeim'a (pl. 4, figs. 20, 21, 23) ; Technophorus (pl. 12, fig. 17) ; and Hippocardia? (pl. 47, figs. 13— 15). METAMORPHOSIS We suggest that the bivalved postlarval shell be known as the dissoconch to conform with the termi- nology applied to pelecypods. There is no clearly de- fined junction between the protoconch and the disso- conch except posteriorly in some conocardiaceans, where an obvious transverse cleft separates the pro- toconch from the rostral area of the shell (pl. 40, fig. 5). Metamorphosis apparently occurs by accelerated growth of the left and right flanks of the protoconch, producing the two valves of the dissoconch. Because rostroconchs characteristically have no adductor muscles, the two newly formed valves must grow rapidly to encompass the body. One specimen of the conocardiacean Bromsoma wilsom’ shows steplike growth increments on the posterior and lateral flanks FUNCTIONAL MORPHOLOGY 5 of the immediately postlarval shell (pl. 52, fig. 4), indicating that the juvenile shell remained pyra- midal in shape to a size of at least 1.5 mm. Other specimens show that the anterior edge of the post- larval shell developed a shallow sulcus generated by a sinus in the anterior commissure (pl. 47, figs. 14, 15) . As growth continued, the lobes on either side of the sulcus enlarged to form the snout region of the valves. Lateral growth occurred at the same time to pro- duce the flanks of the valves, but because the juve- nile flanks are rotated dorsally by subsequent growth at the ventral commissure, even small shells (pl. 47, fig. 15) are deformed Where the larval shell becomes obviously bivalved. In the Conocardiacea, this de- formation is extreme, and it frequently obliterates the protoconch. The protoconch survives in forms like Bransom'a wilsom' because the outermost longitudinal clefts separate the protoconch from the carinal areas of the shell (pl. 51, fig. 17). In contrast to conocardiaceans, the valves of technophorids, ribeiriids, and Eoptem‘a are less inflated, so the protoconch is more easily preserved (pl. 22, fig. 11). The larval and juvenile growth of advanced rostro- conchs like Brimsonia and Pseudoconocardium re- flects the broad outlines of the phylogeny of the class. The protoconch resembles the monoplacophoran shell; it is succeeded by a simple bilobed shell like that of some ribeirioids, which in turn grows into the inflated radially ribbed shell characteristic of conocardiaceans. SUBSEQUENT SHELL GROWTH Adult rostroconchs range in size from 2 mm to 150 mm in length. The postlarval growth of many rostroconch shells produces complex skeletal struc- tures which have no counterparts in other previous— ly described mollusks. Because some of these struc- tures are restricted to one or two genera and because homologies between structures are uncertain, it is convenient to describe the growth of several differ- ent forms separately. In this section, we describe the growth of the skeletal elements of Ribeim'a, Euchas- ma, Pseudoconocardium, Hippocardia, and Arceo- domus. 1. Growth of the ribeiriid shell (fig. 1). Our knowl- edge of the growth of the ribeiriid shell is based mainly on thin-section studies of the shell of Ribeim’a apusoides Schubert and Waa— gen from the Ordovician of Bohemia and on silicified exteriors of Ribeim‘a calcifera, Billings from the Ordovician of Ontario. The ribeiriid shell grew from a protoconch situated nearer the anterior end of the dorsal margin (pl. 6, figs. 7, 11, 13). Growth lines on the shell of Ribeiria apusoides show that the valves separated a little during growth and that the longer posterior dorsal margin func— tioned as a hinge. A strong plate, termed the pegma, extends posteroventrally from the apex of the shell (pl. 5, fig. 4). The pegma is at- tached to both valves (pl. 4, figs. 22, 23; pl. 5, figs. 2—4); it divides the apical area of the shell into anterior and posterior cavities. In front of the beak, the dorsal margin of the shell drops abruptly, and a second short hinge region occurs between the beak and the dorsal edge of the anterior gape (pl. 5, figs. 9, 12) . This is best seen in the silicified replicas of Ribeim’a calcifem. Because the two hinge axes of Ribeiria are approximately parallel, but not colinear, a curved tensional fracture developed below the beak (pl. 4, fig. 9). The ventral edges of each fracture enlarged as growth continued so that the two hinge axes remained the same proportional distance apart, irrespective of the size of the shell. We use the term “anterior clefts” for the right and left ends of this ten- sional fracture. In some individuals of Ribeiria calcifem, the anterior gape extends posteriorly along the anterior dorsal margin for some distance (pl. 4, fig. 8). If this embayment continued as far as the beak, the need for the anterior clefts would disappear as it does in pelecypods and the bivalved gastropods. In these shells, the valves rotate about the ligament and separate at other parts of the dorsal margin. Serial thin sections cut perpendicular to the anterior-posterior axis of Ribeim’a apusoides show that the shell consists of three main lay- ers—a thin outer layer of relatively uniform thickness (pl. 30, fig. 5) formed at the com- missure by the outer surface of the outer mantle fold, and thicker middle and inner shell layers separated by a discontinuity formed by the myostracum of the linear lateral muscle bands. The middle and inner shell layers were secreted by the outer surface of the mantle, and they lapped against the inner surface of the outer shell layer (pl. 30, fig. 5). All layers are continuous across the hinge of the shell. When first formed, the outer shell layer and the immediately underlying middle shell layer were bent in acute angle at the anterior and PALEONTOLOGY OF ROSTROCONCH MOLLUSKS — ~“-—_-_-—-—" ~ ‘~ Anterior cleft Protoconch Growth line insertions FIGURE 1.—Paper cutout models to illustrate the growth of the ribeirioid shell. Drawings are based on Ribeiria lucan (Walcott). Black areas represent muscle insertions; stippled areas show thickness of shell along dorsal margin of completed models. Compare models to see how anterior clefts enlarge mechanically during growth because the shell at the ventral edges of these clefts cannot be resorbed. Growth line on lower model is same size as entire upper model. Instructions: 1, photocopy page; 2, cut reach model from photocopy, cut along all dashed lines; 3, use transparent adhesive tape to fix edges marked “B“ together; 4, staple points marked “A” to- gether; flatten area below protoconch by gently pushing taped area inwards. posterior ends of the dorsal margin (pl. 31, figs. 2, 5). Subsequent deposition of thick inner and middle shell layers cemented the hinge into an inverted U-shape (pl. 30, fig. 4), and the early formed layers were forced apart. This allowed for growth of the shell. The growing edge of the pegma is concave posteriorly so it appears as two lobes project- FUNCTIONAL MORPHOLOGY 7 ing from the inner surfaces of the valves in more posterior sections (pl. 31, fig. 3). These lobes coalesce in more anterior sections (pl. 31, fig. 1). Because the posterior face of the pegma is the site of a large muscle insertion (pl. 5, fig. 4; pl. 6, fig. 8), it is marked with growth lines showing successive positions of the ventral edge of the muscle. These growth lines mark the boundaries between successive increments of the middle shell layers forming the pegma. Consequently, sections through the pegma show that the growth increments of the middle shell layers intersect its upper face (pl. 31, fig. 3). The inner shell layer was deposited behind the linear lateral muscle bands as they mi- grated ventrally during growth. This layer buried the old muscle insertion areas above the dorsal edges of the muscles. It appears in transverse section as a series of overlapping layers which extend laterally from the dorsal margin (pl. 31, fig. 3). Near the apex of the posterior cavity, the whole posterior face of the pegma is covered with inner shell layer (pl. 30, fig. 2). Obviously, if the early formed parts of the valves were separated during growth, the peg- ma would be subjected to tensional stress. Ten- sional fractures parallel to the commissural (symmetry) plane are visible in thin sections of the early formed parts of the pegma in the two specimens of Ribez’m’a apusoides that were fully examined (pl. 30, figs. 2, 3; pl. 31, fig. 1) . Because these fractures do not penetrate. sub— sequently formed shell layers, it is clear that they were not produced after the death of the organism. 2. Growth of Euchasma. This interpretation of the growth of Euchasma is based mainly on infor- mation obtained from silicified replicas of the shells of E. jonesi n. sp. and E. mytiliforme n. sp. from the Ordovician of Malaysia. Unlike Ribez’ria, Euchasma has a strongly inflated shell, which is more or less flattened anteroventrally (pl. 29, figs. 6-13). Because of the inflation, the valves must separate more during growth, and prominent umbos appear on either side of the protoconch region. As in Ribeiria, the valves are connected by an an- terior plate (pl. 29, figs. 3, 11, 14, 15), which also effectively blocks a reduced, almost circu- lar anterodorsal aperture (pl. 28, figs. 15, 16). Also as in Ribeim'a, there are two hinge axes, so anterior clefts are well developed (pl. 27, fig. 13; pl. 28, fig. 17). The posterior dorsal margin is the main hinge axis. Posterior clefts may also form (pl. 29, fig. 10) ; see discussion under 3 for an explanation of posterior clefts. Because both small and large specimens of Euchasma are similar in form and because all have apertural plates (modified pegmas) , it is difficult to understand how they grew when the valves were held rigidly together by the aper- tural plate. Unlike the pegma of Ribeim'a, the apertural plate of Euchasma is attached to the valves at only four points, two dorsal and two lateral (pl. 29, figs. 3, 15). The dorsal attach— ments could remain more or less static during life without affecting shell growth, but the right and left lateral attachments appear to have moved ventrally as the shell grew. Ap- parently shell was added to the ventral edge of each lateral attachment and at the same time resorbed from its dorsal edge. In this way, the valves could separate relatively widely during growth, while remaining rigidly joined by the modified pegma or apertural plate. 3. Growth of Pseudoconocardium (fig. 2). Pseudo- conocardium lanterna (Branson) has an in- flated shell with a huge anterior gape (pl. 40, figs. 6, 11) but no complex internal skeletal elements. It is known from more than 100 un- distorted specimens from the Pennsylvanian of north-central Texas. Growth started from a well-defined proto- conch (pl. 41, figs. 1—5) , which merged into the juvenile shell. Growth lines on the shell show that subsequent growth moved the juvenile shells farther and farther apart, as in other invertebrates having paired calcareous valves. By the time the shell was sufficiently large, umbos formed on either side of the protoconch (pl. 41, fig. 2). The shape of the growth lines on Pseudo- conocardium lanterna (pl. 40, fig. 3) shows that the valves rotated about the anterior dor- sal margin as the shell became larger. We term this the hinge axis. Because the valves are joined along this axis, the hinge must either bend or break as the valves grow. An anterior view of the shell (pl. 40, fig. 6) shows that the dorsal margin forms a smooth U-shaped curve at the commissure. Equivalent earlier formed parts of the shell were progressively deformed as growth proceded, and were cemented into their deformed shape by subsequently depos- Rostral orifice Ventral orifice Ventral orifice FIGURE 2.—Schematic diagrams illustrating the posterior growth of conocardiid rostroconch and pelecypod shells. A. Hippoca’rdia hibemica (Sowerby). Rostral orifice is almost colinear with the hinge axis of the shell, and growth lines (w—f) rotate about rostral orifice as shell grows. Only minor deformation of the upper surface of the rostrum occurs dur— ing growth. Note how the dorsal edges of the hood overlap to hide the hinge axis and protoconch. B. Hippocardia 62mm (Conrad). Rostral orifice is below hinge axis, and orifice generates hoodlike rostral structure through growth. Growth lines (ct—d) continue radially to protoconch. C, Pseudocono- cardium lanterna. (Branson) or Bransonia wilsom' n. sp., rostral orifice is below hinge axis, but no rostral structure PALEONTOLOGY OF ROSTROCONCH MOLLUSKS Ventral orifice is produced. Instead, tension fractures (rostral clefts, ex- aggerated in this diagram for comparison with B) form be- tween the loci of the ventral edge of the rostral orifice and the sides of the tubular rostrum. These clefts do not pene— trate through the shell because new shell layers are con- tinually added internally. Growth lines on such shells are interrupted where they cross the rostral clefts. D, The pele- cypod Hecuba sco'rtum (Linnaeus) (modified from Carter, 1967, pl. 7, fig. 17), which has spinose carinae. The spines are formed at the posteroventral part of the: commissure and are moved outwards during growth in the same way that the rostroconch hood is formed. FUNCTIONAL MORPHOLOGY ited prograding layers of inner shell material. ‘ These layers in turn were deformed by addi- tional growth at the anterior end of the shell, and this growth also redeformed the previously formed layers. Each inner shell layer was thus slightly less strained than the next older one. The effects of this deformation can be seen in thin sections of the hinge of Pseudocono- cardium lanterna (pl. 32, fig. 2). In the speci- men figured, tensional stresses caused by the bending of the shell have fractured the earlier formed shell layers, and the damage has been repaired by the youngest layers. Occasionally the whole anterior hinge may rupture during life (Pojeta and others, 1972, fig. 1). In Pseudoconocardmm lanterna and other conocardiaceans that we have sectioned, the outer shell layer is not continuous across the hinge (pl. 32, figs. 1, 2), and thus was not normally deformed during growth. However, because the outer shell layer is relatively thin, the hinge of P. lanterna is almost as thick as the flanks of the valves (pl. 32, fig. 2). Because the anterior dorsal margin of cono- cardiaceans is the hinge axis, the posterior dor- sal margin must have compensated for valve rotation in some other way. All well-preserved specimens of P. lanterna have a series of sym- metrical fissures, termed clefts, behind the um- bos (pl. 40, figs. 5, 7), which result from the periodic failure of the shell across and on each side of the tubular rostrum. In the past, these clefts were thought to be ligament grooves (Hind, 1900), but they do not penetrate the shell, were not connected to the secretory man- tle, and thus could not have contained liga- mental material. In transverse section, they appear as V-shaped fissures which penetrate at right angles to the growth lamellae. On some well-preserved exteriors of P. lanterna, the outer clefts cut previously formed growth lines. Because the clefts do not extend as far on small shells that are similar in size to the growth lines cut by the clefts on larger shells, we conclude that the clefts are fractures which enlarged as the shell grew. They are compara- ble to the anterior clefts of Ribeim’a and Euchasma, and formed for much the same reason. The clefts are best developed in inflated shells such as P. lanterna where the rostrum occurs below and at an angle to the hinge axis (pl. 40, fig. 8). In these shells, the posterior 9 commissure is again defined by the growing edge of the inner shell layers. It extends dor- sally from the ventral orifice and continues around the tip of the rostrum. It does not ex- tend along the dorsal surface of the rostrum (pl. 40, fig. 5) as it does in Hippocardia czmea (Conrad) (Case '4). This difference between these shells accounts for the presence of the rostral clefts in Pseudoconocardium lantev'na but not in Hippocardia cunea. The need for the clefts is easier to understand if Bransom'a wil- som' n. sp. (pl. 52, figs. 1—5) is considered as an intermediate form. In lateral View, the ventral surfaces of the rostra of Hippocardia cunea (pl. 48, fig. 8) and Bromsom'a wilsoml (pl. 51, fig. 1) are similar in general form. However, their upper surfaces are quite different because B. wilsom’ has a pair of deep clefts on either side of the rostrum (pl. 51, fig. 17) and in addition has smaller clefts near the protoconch. Between the large clefts, the rostrum is relatively uniform in width when viewed dorsally (pl. 51, fig. 17 ). The outer edges of the large rostral clefts of Bramsonia wilsoml are topologically equivalent to the angular edges of the rostral structure of Hippocardia cunea (pl. 48, figs. 10, 11), and growth lines on each type of shell intersect these edges in an acute angle which opens an- teriorly. In Hippocardia cunea, the growth lines are immediately reflected toward the pro- toconch, and the upper surface of the rostral structure is covered with growth lines that radiate from the beak. Thus, the dorsal rostral surface of Hippocardz’a. cunea. grows like the lunule of a venerid pelecypod. In contrast, the dorsal surface of the rostrum of shells with a rostrum like Bransonia wilsom’ and Pseudoconocardium lanterna is marked with growth lines that cross the dorsal margin at right angles to the commissural plane. In these shells, the rostrum can only grow distal- ly, ventrally, and internally. In shells where the upper surface of the rostrum is in line with the hinge axis, there is no problem, and the rostrum grows like the posterior wing of various pteriacean pelecypods (pl. 43, figs. 1, 5). But in those forms where the rostrum is below and at an angle to the hinge axis, the rostrum Would split dorsally as the valves grew. This is avoided in forms like Pseudoconocardium lanterna and Bran- som’a wilsom’ by the formation of tensional 10 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS fractures, clefts, on either side of the rostrum. These occupy the space filled by additional shell secreted along the dorsal margin of the ros- trum in Hippocardia cunea. There are two kinds of rostral clefts in Pseudoconocardium lantema and most other conocardiids: symmetrical longitudinal clefts resulting from tensional stresses more or less parallel to the rostrum, and one or more trans- verse clefts which apparently compensate for the required rotation of the rostrum within the commissural plane (pl. 40, figs. 5, 7). If al- lometric changes in the angle the rostrum makes with the shell are to be avoided, the rostrum must be gradually raised during growth. Apparently the transverse clefts closed slowly during growth and allowed for this rotation. The first-formed transverse cleft separates the protoconch from the rostral area of the dissoconch in this and many other cono- cardiids (pl. 40, fig. 5). In thin section, the shells of Pseudoconoca/r- dimn lanterna and Bromsonia wilsom' are formed of two obviously different layers. The outer shell layer is relatively transparent and coarsely prismatic (pl. 32, fig. 4). The inner shell layers are darker and well laminated, though they show no obvious microstructure (pl. 32, fig. 1, 2, 4). They may have been nacreous, cross-lamellar, or homogeneous. Because the exterior of the shell is covered with narrow closely spaced radial and comar- ginal walls (pl. 40, fig. 12), thin sections sometimes give the impression that the outer shell layer contains a series of roofed pits or vacuoles. Similar structures have been re— ported from other conocardiids (Panella and MacClintock, 1968, pl. 8, fig. 5). These views can result from oblique sections through inter- secting radial and comarginal walls, and there is no evidence that the intervening pits were roofed over in P. lantema. In at least some specimens of P. lanterna, small secondary ribs run between the vertical lamellae of the co- marginal markings in the interspaces between primary ribs (pl. 42, fig. 6). The intersection of the vertical lamellae and the secondary ribs produces a grid pattern that significantly re- duces the space open to the exterior between adjacent lamellae. The inside of the anterior aperture of Pseu- doconocardz‘um lanterna. and many other ros- troconchs is lined with a series of blunt pro- jections (pl. 40, figs. 6, 11) that we have termed commissural (or marginal) denticles. These are formed of the prismatic outer shell layer secreted at the edge of the mantle. Simi- lar smaller denticles line the inside of the rest of the commissure as far as the base of the rostrum. As the shell grew, these commissural denticles were buried by prograding inner shell layers secreted by the external surface of the mantle. As growth continued, the commissural denticles generated internal ribs which were totally submerged by the inner shell layers. Sometimes solution of the inner shell layers before lithification may expose the ribbed in- ternal surface of the outer shell layer, thus giving the impression that the interior of the shell was ornamented with radial ribs (pl. 45, fig. 4) . The inner shell layers that bury the sub- merged ribs are relatively uniform in thick- ness, and so they form concentric folds over the ribs of the outer shell layer. When the inner shell layers became thick enough, they became relatively flat (pl. 42, figs. 13, 14). The growth of the anterior dorsal margin of P. lanterna also illustrates features found in many conocardiids. Well-preserved specimens normally have two subcircular depressions, one on either side of the middorsal line, just inside the anterior commissure (pl. 40, fig. 11; pl. 42, fig. 5). These depressions are separated by a wall of shell, best shown by views of the growth layers in transverse section (pl. 32, fig. 1). The function of the depressions is uncer- tain, but they may have been muscle insertions because they are filled by subsequent deposits of inner shell material. The abrupt contact be— tween these latter deposits and the base of the depression may represent the myostracal layer of the muscle insertion. 4. Growth of Hippocardia (fig. 2). Hippocardia czmea (Conrad), the most common rostroconch in the northeastern United States Devonian, has a spectacular hood attached to the carinal area of the shell (pl. 48, fig. 2). The species is well known from external molds in New York State, silicified replicas of the outer shell layer from the Falls of the Ohio River along the Kentucky-Indiana border, and original shells from Ohio (pls. 48, 49). The hood of Hippocardia. was secreted by a tubular extension of the mantle at the ventral orifice, which is located midventrally on the posterior face of the shell (pl. 49, figs. 5, 6). FUNCTIONAL MORPHOLOGY 11 The ventral surface of the hood forms a con- tinuous curved surface with the posterior flanks of the valves, but it lacks radial ribs (pl. 49, figs. 4, 10). Both upper and lower surfaces of the hood are ornamented with fine closely spaced growth lines that are continuous with growth lines on the rest of the shell. The growth lines indicate that new shell material was periodically added to the hood along the: entire length of the tubular extension of the ventral orifice as the valves gradually sepa- rated. Thus, the hood represents a surface generated by a tube at the posterior ventral commissure and was inevitably produced if the tube was maintained throughout the life of the animal. It follows that the hood may have had no function; it may only represent the loci of the edges of the tube. Alternatively, the only function of the tubular extension of the third aperture may have been to generate the hood, and the hood may be the functionally impor- tant structure. There is no evidence for allometric changes in the size of the hood during growth. Conse- quently, in shells that have inflated valves, the edges of the hood on left and right valves may have interfered with one another as the umbos enrolled. This problem appears to have been solved, as it is in living heart cockles, by hav- ing the hood of one valve slightly ahead of that of the other. Serial transverse sections of the hood of Htppocardia cunea show that it is formed of concave lamellae which are separated by open spaces in the distal part of the body (pl. 32, figs. 5, 6). Proximally, these spaces are filled with prismatic outer shell material (pl. 32, fig. 5). Thus, the whole of the hood is constructed of the outer shell layer. Because the outer shell layer of mollusks is normally only secreted at the mantle edge by the outer surface of the outer fold of the mantle, we believe that a hy- pertrophied part of the outer mantle fold formed the hood of H ippocardia. The rostral structure of Hippocardia cunea simulates a second hood (pl. 48, figs. 10, 11). Like the hood, it is a curved surface generated through growth by a tube at the commissure. There are, however, two important differences. First, it is clear by comparison with other conocardiaceans that it is the tube, not the structure generated by the tube, that is func— tionally important. And second, again by com~ parison with other conocardiaceans, it is cer- tain that the tube (rostrum) is formed of all shell layers (see 3). 5. Growth of Arceodomus. Arceodomus is best known from recrystallized original shells of A. glabrata (Easton) from the Mississippian and Pennsylvanian of Montana, Nevada, and Texas (pl. 42, figs. 8-10; pl. 43, figs. 1—4; 7—12) and from silicified replicas of A. langenheimi (Wil- son) from the Permian of California. (pl. 43, figs. 13—15). Arceodomus resembles Conocardi- um (pl. 38) but lacks radial ornament on the body of the shell. Externally, the growth of Arceodomus is similar to that of Pseudocono- cm-dium, and small rostral clefts are visible in A. glabmta (pl. 43, fig. 10). We use these two species to illustrate the formation and growth of the anterior longitudinal shelves (pl. 43, figs. 12, 1.3). Anterior longitudinal shelves are curved plates composed of the outer (prismatic) shell layer (pl. 43, fig. 13) that project more or less horizontally across the snout region of Arceo— domus and Conocardium. As growth continued, the older parts of the shelves were buried in thick deposits of inner shell layer (Wilson, 1970, figs. 10—18). Wilson (1970) recognized the microstructural difference between the shelves and the inner shell layers, but tenta- tively interpreted the shelves as myostracal layers. The shelves are unusually enlarged commis- sural denticles, and clear transitions can be seen in several morphological series, for ex- ample, Pseudoconocardium lanterna (pl. 40, figs. 6, 11), Mulceodens jaomussoml n. sp. (pl. 34, figs. 3—5), Hippocardia zeilem‘ (Beushaus- en) (pl. 47, figs. 8, 9), and Arceodomus gla- bmta. (pl. 43, fig. 12). Significance of the shelves lies in the fact that they were formed of the outer shell layer, implying that they were formed by the mantle edge. We conclude that the anterior part of the mantle of A7060- domus (and Conocardium) was enlarged, that it was complexly folded when withdrawn into the shell, and that the edges of the mantle must have been located at the growing edges of the anterior shelves when the mantle was at rest in the shell. SUMMARY OF SHELL GROWTH All rostroconchs grew a bilobed shell (dissoconch) from a univalved protoconch or juvenile shell. Primi- tive rostroconchs (Ribeirioida) have all shell layers 12 PALEONTOLOGY 0F ROSTROCONCH MOLLUSKS continuous across the dorsal margin, but in advanced rostroconchs (Conocardiacea), the outer shell layer is dorsally discontinuous, except in the region of the protoconch. Primitive rostroconchs have the valves connected by a stout anterior pegma (Ribeirioida, some Eopteriidae) or anterior and posterior pegmas (Ischyrinioida) which is deformed (Ribeim'a) or partially resorbed (Euchasma) to allow for valve separation during growth. Parts of the dorsal shell margin of all rostroconchs function as a poorly elastic hinge during growth, but during day-to-day living, the valves were held rigid- ly together. If the shells are anteriorly elongated (Pseudotechnophorus, Conocardiacea), the anterior dorsal margin functions as the hinge. In posteriorly elongated shells (Ribeiriidae, Technophoridae), the reverse is true. Because the dorsal shell margin is rarely straight, transverse and longitudinal tension- al fractures called clefts form between the main hinge and the less elevated parts of the dorsal mar- gin. In subequidimensional shells (Eoptem'a, Euchas- ma), clefts are present on both sides of the proto- conch; anteriorly or posteriorly elongated shells nor- mally have clefts only at the shorter end of the shell. The distribution of the clefts can be explained by the geometry of shell growth. All rostroconchs that have visible shell structure have an outer (often prismatic) shell layer formed by the outer edge of the mantle. One or more inner shell layers are lapped against the internal surface of the outer shell layer by the outer surface of the mantle. The hood of Hippocardia, is formed of outer shell material only; the commissural denticles and anterior shelves of conocardiaceans are also part of the outer shell layer, implying that the edge of the mantle could be withdrawn into the anterior part of the shell in some forms. OPENING OF THE VALVES The valves of pelecypods and bivalved gastropods are joined dorsally by an elastic structure called a ligament. Owen, Trueman, and Yonge (1953) showed that the simplest pelecypod ligament con- sists of three layers (periostracum, lamellar, and fibrous layers), which are continuous with compara— ble shell layers in the right and left valves. The probable structure of the ligament of the Early Cam— brian pelecypod Fordilla (Pojeta and others, 1973) and the nature of the ligament of the bivalved gas— tropod Berthelinia (Kawaguti and Yamasu, 1961) support this observation. Thus, the valves and liga- ment are part of a single structure (the shell) and differ only in the degree of calcification of the pro- tein matrix. This explanation is supported by the ontogeny of living pelecypods, whereby a single lar- val shell gland secretes “a saddle-shaped cuticular pellicle, which becomes calcified at two symmetrical points, right and left of the middle line” (Pelseneer, 1906, p. 245). In engineering terms, the ligament can be de- scribed as a spring, because it stores energy supplied by contractions of the adductor muscles attached to each valve. This energy is released when the ad- ductors relax, and experimental studies show that the elastic efficiency of the ligament can be estimated from the size of the hysteresis .loop obtained by load- ing and unloading a freshly killed individual (True- man, 1953; Hunter and Grant, 1962). Pelecypods having a mechanically inefficient ligament use mus- cular energy transmitted hydrostatically through the foot or mantle cavity to open the valves (Hunter and Grant, 1962). The pelecypod ligament is strained when the valves are closed. Above the hinge axis, the strain is tensile; below it, the strain is compressional. Nor- mally the ligament is constructed so that the junc- tion between the functional parts of the lamellar and fibrous layers more or less coincides with the hinge axis, because each layer is resistant to only one kind of stress. Only some of the energy stored in the pelecypod ligament comes from the adductor muscles. Galtsoff (1964) noted that if the adductor muscle of an oyster is cut, the valves open farther than they do when‘the oyster is narcotized and the adductor fully relaxed. The origin of this extra energy remains unclear. Trueman (1949) found that the functional part of the ligament of Tellina tennis is not, as would be expected, the most recently formed part; he sug- gested that additional secretion of fibrous layer in the middle part of the ligament stretches the lamellar layer so that early formed parts of the ligament remain functional for long periods of time. If this did not happen, only the last formed part of the liga- ment could function, because the strain on the older parts would be gradually released as the valves opened slowly during growth. In fact, in many pelecypods, the anterior part of the ligament is torn apart as each valve grows in a separate helical spiral (Perkins, 1969, p. N756). Thus, Trueman (1950) found that in Mytilus edulz's, the whole of the original ligament of a shell 16 mm long is destroyed when the mussel has grown to a length of 70 mm. Thus, only the posterior part of the ligament of M. edulis func— tions at any one time. Trueman concluded that the FUNCTIONAL MORPHOLOGY 13 ligament of M. edulz's becomes functional when the lamellar layer is subjected to tensile strain by the growth of the fibrous layer beneath it. This strain is increased when the valves are closed, but a signifi- cant part of the opening force is generated bio- chemically by the mantle during the formation of the ligament. This extra energy may assist in. open- ing the valves for locomotion, burrowing, and feed— ing, but equally importantly, it enables the valves to open slowly during growth. Bevelander and Nakahara (1969) reported that the fibrous layer of the ligament of M. edulis is formed of long euhedral pseudohexagonal needles of aragonite dispersed in a homogeneous organic ma- trix. The diameter of each needle increases away from the calcification front, and it may be this ex— pansion in volume that stretches the outer layer and activates the ligament. Alternatively, quinone tan- ning of the protein forming the lamellar layer may cause it to shrink and thus compress the underlying fibrous layer. A closer analog to the rostroconchs is found in the extraordinary living pelecypod Pinna. In Pinna, the valves are joined rigidly by a long simple ligament, which has both lamellar and fibrous layers impreg- nated with calcium carbonate. Yonge (1953b) re- ported that the ligament of Pimw is not elastic and has no opening thrust on the valves. When the ad- ductor muscles contract, the valves are flexible enough to be pulled together, but the ligament does not bend appreciably (Yonge, 1953b, p. 338). Growth lines on the shell of Pimw. show that earli- er growth increments gape more widely than later formed ones, so that the valves must have opened slowly during the growth of the shell. The force that causes the valves to gape during growth has not been documented, but the ligament may generate this opening moment. If Pinna has a self-opening shell, rostroconchs may have functioned in a similar manner. In contrast to most pelecypods and the bivalved gastropods, all rostroconchs had valves that were rigidly joined dorsally. In some genera like Cano- camdium (pl. 38, figs. 1, 3, 6, 11, 14) and Arceo- domus (pl. 43, figs. 9, 11, 14), the ventral and pos- terior shell margins are tightly apposed, but in others like Ribez'm'a (pl. 6, figs. 3, 5, 6) Eopteria (pl. 24, figs. 14, 15, 20) and Pseudoconocamdium (pl. 40, figs. 4, 6, 11), there are prominent shell gapes. Growth lines on both kinds of shells show that the valves opened slowly during life, so that the ventral edges of early increments may eventually gape at angles of 180° or more. Because the valves of rostroconchs are joined dor- sally, energy was needed to separate the valves so that new increments could be added at the commis- sure. This energy could have been supplied in sever- al, not necessarily mutually exclusive, ways. Al- though applied in small amounts for long periods of time, the energy was sufficient to rupture parts of the dorsal shell margin in almost all rostroconchs. The most obvious primary source of mechanical energy in any animal is its musculature. Energy generated by the contraction of muscles could be transmitted hydrostatically to the shell either through fluids in the body cavity, particularly blood in the pedal haemocoele, or by the fluid (sea water) in the mantle cavity. If the volume of blood in the foot of a rostroconch could be: kept constant by means of a Keber’s valve or some comparable struc- ture, the foot could be protracted between the ven- tral valve edges by contraction of its intrinsic trans— verse muscles and then inflated by means of the pedal retractors inserted on the shell. This would force the valves apart. Alternatively, the foot and (or) hypertrophied mantle tissue could be with- drawn into the shell by appropriate muscles, and if no sea water were allowed to escape, hydrostatic pressure would tend to open the valves. We prefer the former explanation for forms like Ribeim‘a, Eopteria, and possibly Bremsom'a. and Pseudoconocardium, as it is difficult to see how they could have effectively sealed all shell gapes to con- serve water in the mantle cavity. However, it is equally difficult to envisage that Arceodomus and Conocardium had a foot large enough to open the valves, as the anterior gape is almost completely blocked by the anterior longitudinal shelves (pl. 43, fig. 13). In these latter forms, it seems more likely that the withdrawal of mantle tissue increased the hydrostatic pressure in the mantle cavity if muscu- lar energy were used to open the valves. Muscular energy may not have been the prime or only force that opened the valves of rostroconchs. Other possibilities include growth pressure resulting from the addition of new cells to the body mass, osmotic pressure in the body mass, or the unex- plained opening moment of the pelecypod ligament which may be generated by crystallization pressure in the fibrous layer of the ligament. MUSCULATURE The principal muscles of shelled mollusks serve the foot and the mantle edges; other smaller muscles may be used to move the head, jaw apparatus, gills, and visceral mass. Most of the muscles are attached 14 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS to the shell at surfaces known as insertion areas (muscle scars) where minute microvilli reinforced by cytoplasmic fibrils fill tiny pits in the surface of the shell (Hubendick, 1957). Each muscle insertion moves toward the commissure and increases in size during growth; it also generates a thin shell layer with characteristic ultrastructure and mineralogy known as the myostracum (Oberling, 1964; Taylor and others, 1969; Batten, 1972). If the muscle inser- tion is linear, the corresponding myostracal layer will be planar, but if the insertion area is circular or oval, it generates a linear piece of myostracum (Waller, 1972). Growth lines reflecting the shape of the commis— sural side of the muscle are normally produced on the surface of the insertion area by fluctuations in the rate of deposition of the underlying shell layers or the myostracum. As the trailing edge of the muscle moves toward the commissure during growth, new shell layers wholly or partly obliterate the areas where the muscle was previously inserted. If these shell layers are thinner across the old muscle-inser- tion areas than they are in other parts of the shell, a smooth concave muscle track is visible on the in- side of the shell. Sometimes a muscle inserted on the shell may run parallel to it for some distance, for example, the siphonal retractor muscles of siphonate pelecypods. In such a case, some individuals in a population may mold shell around the muscle bundles so that a series of linear concave depressions is formed on the inside of the shell (Runnegar, 1972). We refer to these structures as muscle impressions; they give evidence for the direction of action of the muscle. Muscle insertions in fossil mollusks are best ob- served on natural or artificial internal molds of the shell. Growth lines on the insertion areas or well- defined muscle tracks unequivocally identify muscle- insertion areas; the shape of the leading and trailing edge of the insertion is also important. Character- istically, the leading edge will be a smooth, poorly defined curve; the trailing edge is usually well de- fined and is often scalloped. If the insertion is too poorly preserved to show any of these features, a thin section of a shelled specimen may reveal a myostracal layer or a discontinuity representing a thin myostracum. It must be pointed out that only some of the original muscle insertions may be preserved. Many specimens of Paleozoic pelecypods have well-pre- served adductor and pedal muscle-insertion areas but show no trace of a pallial line. Only one specimen in many may show this structure, and one should be extremely cautious in using such negative evidence. Thus, although many ribeirioids have well-preserved pedal muscle insertions, only a few specimens show traces of the pallial line. We feel that these differ— ences are due to preservation and that a pallial line was probably present in many rostroconchs. Because there are no living rostroconchs, no mod- ern analogs can be used to predict what rostroconch musculature looked like. Muscle-insertion areas must be recognized in one or more of the ways indicated above, and it is necessary to be most rigorous in assessing the significance of depressions on the in- teriors of shells. Yonge (1953a) observed that pelecypods probably differ from primitive ancestral mollusks in having the peripheral part of the mantle attached to the shell by a series of radial muscles, the insertions of which coalesce to form the pallial line. He suggested that pallial attachment probably became necessary as lateral compression enlarged the mantle and shell relative to the foot and visceral mass. When the bivalved condition finally developed, the adductor muscles were formed by cross-fusion of the distal ends of the anterior and posterior radial muscles of the mantle. Yonge therefore made an important dis- tinction between an inner series of shell or pedal muscles that control the foot and support the Vis- ceral mass, and an outer set of pallial muscles that control the edges of the mantle and serve to close the valves. It is the enlarged mantle cavity of rostroconchs and pelecypods that creates the need for strong pal— lial retractor muscles and hence a well-defined pallial line. Small scattered pallial retractor muscles are present in N eopilz'na, and it also has a broad roughly circular zone where cells in the mantle epithelium are attached to the shell (Lemche and Wingstrand, 1959). Batten (1972) described myostracal shell layers in several Pennsylvanian gastropods which he attributes to muscles attaching the mantle to the shell. Thus, although N eopz'lina and the fossil gastro- pods lack what is conventionally called a pallial line, they have small muscles or appropriate mantle epithelial cells in comparable positions. Presumably these could be hypertrophied to form a well-defined pallial line if the proper conditions arose. Yonge’s twofold classification of the shell-attached muscles of the pelecypods can be conveniently ap- plied to the Rostroconchia. We distinguish pedal and pallial muscles in many forms. Normally, the pedal muscles are more deeply inserted into the shell and are therefore more commonly preserved than the pallial muscles. FUNCTIONAL MORPHOLOGY 15 PEDAL MUSCULATURE The oldest known rostroconchs that have well- preserved muscle-insertion areas are Early Ordo- vician ribeirioids (fig. 3) and eopteriids (fig. 4). A simple muscle array is shown by Tolmachovia? jelli n. sp. (fig. 30, D; pl. 14, figs. 9—19). This species has subequal oval insertion areas on the anterior and posterior sides of the umbonal cavity (pl. 14, figs. 9—16, 18, 19) which are connected ventrally by linear insertion areas on the left and right umbonal flanks (pl. 14, figs. 11, 12). Because this ribeirioid is ap- proximately equilateral, we feel that the foot pro- jected ventrally and was formed mainly of subequal anterior and posterior retractor muscles inserted on the shell at the two oval depressions. The foot proba- bly also contained circular and transverse intrinsic muscle fibers that were not attached to the shell. These muscles could oppose the longitudinal re- tractor muscles through the hydrostatic skeleton of the pedal haemocoele to lengthen or broaden the foot for probing and pedal anchorage. The linear inser- tion areas connecting the two pedal retractor inser- tions may have contained the ends of longitudinal muscles forming the sides of the foot, or they may have been the areas of attachment of muscles sup- porting the gills. The anterior and posterior pedal retractor inser— tions lie across the midline of the shell. We term these the anterior and posterior median insertions of the anterior and posterior median pedal retractor muscles, and distinguish them from right and left pedal retractor muscle insertions arranged sym- metrically in both valves in advanced rostroconchs (pl. 22, figs. 5, 6) and most pelecypods. As we are uncertain of the function of the linear muscles con- necting the median muscles, we name them descrip- tively as the right and left linear insertion areas of the corresponding side muscles. Anterior and posterior median and left and right linear muscle insertions are present in several ribeirioid genera (fig. 3). In those forms having a well-developed pegma, the anterior median retractor is inserted on the posterior face of the pegma (pl. 5, fig. 4), and all muscle insertions are confined to the posterior umbonal cavity. In Ribeim’a lucom (Wal- cott) , a series of discrete circular— to kidney—shaped insertions replace the linear insertions of the left and right side muscles (fig. 3E; pl. 8, fig. 14). Complex right and left linear muscle insertions also are present in the ischyrinioid Ischym'nia nor— vegica Soot-Ryen (fig. 3A; pl. 19, figs. 10—14). It is not known whether I schyrim'a. has large anterior and posterior median insertions, but because of the pres- ence of side muscles, we assume that the anterior and posterior retractors of the foot originated on the inner sides of the two pegmas found in Ischym‘nia (pl. 18, figs. 22, 25). The protoconch of another ischyrinioid, Pseudo- technophorus typicalis Kobayashi, has a tiny anteri- or median muscle insertion and an equally small linear insertion area preserved only on the left side of the specimen (fig. 3F; pl. 20, figs. 13—15). We View these structures as the insertions of muscles that were larger in the ancestors of Pseudotechno- phoms but that became limited to the shell apex when new right and left lateral pedal muscles evolved; the relict structures also indicate a phylo- genetic relationship of Pseudotechnophoms to the ribeirioids. The lateral pedal muscles of Pseudoteah— nophoms were attached at one large insertion and several smaller insertions on the right and left flanks of the valves (fig. 3G; pl. 20, fig. 8). We term the large lateral pedal muscle, the primary pedal re- tractor, and the small lateral pedal muscles the “secondary retractors,” to diflerentiate them from the median retractors of ribeirioids and the anterior, umbonal, and posterior pedal retractors of pelecypods. Eopteria, like Pseudotechnophorus, has a large lateral pedal muscle insertion (pl. 22, figs. 5, 6) and three smaller secondary insertions on each valve (pl. 22, figs. 5, 6). A reconstruction of the musculature of Eopteria is shown in figure 4. The entire muscula- ture of this genus is not known from any one speci- men, and the reconstruction is based on several speci- mens from the same locality, each of which shows some of the muscle insertions. The interpretation of the musculature of Eoptem’a given here differs from that of Pojeta, Runnegar, Morris, and Newell (1972), in that three small dorsally located inser- tions shown on the earlier diagram are now thought to be discontinuous muscle insertions forming the dorsal part of the anterior end of the pallial sinus (pl. 22, figs. 3, 4, 7, 8) . Figures 3 and 4 on plate 22, show that in this specimen the pallial sinus is dis- continuous dorsally but not ventrally, and figures 7 and 8, on plate 22, show three discontinuous small dorsal muscles in the same approximate position as the dorsal part of the pallial line of the other speci- men. The similarity of the adult musculature of Eoptem'a and Pseudotechnophorus suggests a phylo- genetic link between the ischyrinioids and the conocardidids. Some of the few known specimens of conocardia- ceans that have muscle-insertion areas preserved (fig. 5) have a large primary pedal-retractor inser- 16 amm PALEONTOLOGY OF ROSTROCONCH MOLLUSKS pmm pmm E FIGURE 3.—(See explanatinn on facing page.) pmm FUNCTIONAL MORPHOLOGY am? FIGURE 4.-—-Composite diagram of the muscle in- sertions of the left valve of Eopteria ventricosa. (Whitfield); axm, adductor muscle insertion(?); mi, muscle impression; mt, muscle track; ppr, primary pedal-retractor muscle insertion; ps, pallial sinus; pl, pallial line; spr, secondary pedal-retractor muscle insertions. tion in the umbonal cavity of each valve (pl. 38, figs. 21, 22, 24). Smaller insertions in this area are re- garded as secondary pedal retractors (pl. 53, figs. 21—23). Some conocardiaceans show no obvious pedal-retractor-insertion areas (pl. 53, figs. 1—4). Some pelecypods have muscles inserted on the shell that serve to protract the foot. In some forms, the distal ends of these muscles run transversely around the foot, forming a sphincter which is used to confine blood to the pedal haemocoele. Contraction of the intrinsic transverse muscles of the foot then protracts the foot if the shell-inserted pedal retrac- tors are simultaneously relaxed. The living solenid Ensis operates its foot in this way (Trueman, 1967). Other pelecypods (unionids, arcaceans, trigoniids) have a pedal muscle inserted on the shell below the insertion of the anterior adductor. In these animals, the muscles act in directions that enable them to move the foot anteriorly and ventrally. Such muscles are termed “pedal-protractor muscles.” FIGURE 3.—Musculature of various genera of the Ischyrinioida and Ribeirioida; amm, anterior median muscle insertion; mt, muscle track; 17mm, posterior median muscle insertion; ppr, primary pedal-retractor muscle insertion; mm, pallial re- tractor muscle insertion; sm, side muscle insertion; spr, sec- ondary pedal-retractor muscle insertions. A, Composite diagram of left valve of Ischym'm'a winchelli Billings and I . norvegica Soot-Ryen. B, Left valve of Tolmachovia con- centrica Howell and Kobayashi. C, D, Dorsal and left-valve views of Tolmachom'a? jelli n. sp. E, Left valve of Ribeiria lucan (Walcott). F, G, Enlargement of larval musculature and a diagram of adult musculature of Pseudotechnophorus typicalis Kobayashi. H, Composite diagram of left valve of Technopho'ms sp. 1, J, dorsal- and left-valve views of Ribeiria pholadiformis Sharpe. 17 None of the primary and secondary pedal-muscle l % 1nsertlons of rostroconchs are low enough on the shell to have functioned as direct pedal protractors, and we cannot determine whether any of them served as sphincters. Thus, in rostroconchs, pedal protraction seems to have been accomplished only by hydrostatic means. PALLIAL MUSCULATURE The Ordovician ribeirioid Wcmwam'a shows traces of a pallial line (pl. 3, fig. 7). We believe that most ribeirioids, with the possible exceptions of H erault’i- pegma and Watsonella, had pallial lines where the peripheral parts of the mantle were attached to the shell by radial mantle muscles. As noted, many speci- mens of many species of Paleozoic pelecypods pre- serve impressions of the adductor and pedal muscles, but only a very few specimens preserve the pallial line. The internal molds of ribeirioids that we have studied are generally not well enough preserved to show the shallow insertions of the small radial mus- cles of the mantle. Several specimens of Eoptem‘a preserve parts of the pallial line (fig. 4; pl. 22, figs. 1, 2, 3, 4; pl. 23, figs. 2, 3). There is an obvious anterior pallial sinus, and one specimen shows shallow impressions of the radial muscles of the mantle within the pallial sinus (pl. 22, figs. 5, 6). The absence of visible impressions on other parts of the shell indicates that the radial mantle muscles were both larger and longer in the area of the sinus. If so, the anterior part of the mantle was also enlarged in this area, and, by analo- gy with pelecypods, could be extended beyond the limits of the shell. We term these enlarged radial muscles of the mantle pallial retractor muscles. The pallial line and associated radial muscle tracks are best preserved in one specimen of Euchasma from the Lower Ordovician of Newfoundland (pl. 27, fig. 9). The bending of the pallial line laterally in the anterodorsal region of the shell shows that an anteri- or pallial sinus was also present in Euchasma, but this part of the specimen is not well preserved, and no radial mantle muscle impressions are visible. Pseudotechnophorus has a large muscle insertion in the isolated cavity of the shell above the anterior pegma (fig. 3G; pl. 20, figs. 10, 11). A muscle originating in this position is unlikely to have been connected to the foot, because the pegma effectively separates this cavity from the main mantle cavity. We assume that Pseudotechnophorus had radial mantle muscles, even though a pallial line is not pre- served on any of the specimens seen by us; and we interpret the muscle in the anterior umbonal cavity 18 PALEONTOLOGY 0F ROSTROCONCH MOLLUSKS FIGURE 5.—(See explanation on facing page.) FUNCTIONAL MORPHOLOGY as an enlarged radial mantle muscle which func- tioned as a pallial retractor. We do not have many conocardiaceans that have preserved pallial muscle insertions. In the few speci- mens we do have, the pallial line is smooth (fig. 5; pl. 53, figs. 1—4, 21-23). In Bransom'a? sp. (fig. 5E, F; pl. 53, figs. 1—4), and B. robustum (Fletcher) (fig. 5D, G; pl. 53, figs. 21—23) , the anterodorsal part of the pallial line is Y-shaped and has anterior, pos- terior, and ventral branches; in these two species, the junction of these three branches (the pallial junction) is the site of a larger muscle insertion (pl. 53, figs. 1, 21, 23). In B. cressmani n. sp. (pl. 54) and Conocardium elongatum (Sowerby) (pl. 38, fig. 22), the anterior part of the pallial line is not Y— shaped. In Conocardz'um alifo'rme? (Sowerby) the posterior branch is not connected with the anterior and ventral branches (fig. 5H; pl. 51, fig. 11). This species has well-defined radial mantle impressions emanating from the anterior and possibly the pos- terior branch of the pallial line (fig. 5H; pl. 51, fig. 11), indicating that as in Eopteria, these parts of the pallial line had pallial retractor muscles which could withdraw mantle tissue extended from the aperture of the shell. The larger muscle insertion at the junction of the branches of the pallial line in Bromsonia? sp. and B. robustum probably housed a muscle that performed the same function. Several conocardiacean species have circular to elongate depressions on both sides of the hinge just inside the anterior aperture of the shell (fig. 5B, O ; pl. 40, fig. 11; pl. 47, fig. 12). These depressions were filled by subsequently deposited inner shell layers as the shell grew. We interpret them to be muscle in- sertions and suggest that they housed muscles used to protract mantle tissue from the anterior aperture of the shell (pallial protractor muscles). This inter- pretation helps to explain their unusual position on the shell and suggests the method used by cono- cardiaceans to protract the mantle. No similar in- sertions are known in either Eoptem'a or Pseudo- technophorus, and we assume that mantle protrac- FIGURE 5.—Musculature of various genera of the Conocar— diacea; ab, anterior branch of the pallial line; mi, muscle impression; pb, posterior branch of the pallial line; 127', pallial junction; pl, pallial line; ppm, pallial-protractor muscle insertion (?); pp’r, primary pedal-retractor muscle insertion; pr'm, pallial retractor muscle insertion; 810?, sec- ondary pedal retractor muscle insertion. A, Reconstruction of Conoca/rdium elongatum (Sowerby). B, C, Dorsal and left-lateral views of Hippocardia zeilem' (Beuhausen). D, G, Left and dorsal views of Bransom'a robustum (Fletcher). E, F, Dorsal and left—lateral views of Bransom’a? sp. H, Left-lateral view of Conocardium aliforme? (Sowerby). 19 tion in these forms was accomplished entirely by hydrostatic means. Pojeta, Runnegar, Morris, and Newell (1972) argued that no rostroconchs had cross-fused radial mantle muscles and therefore no adductor muscles. This statement was based on the premise that cross- fusion could not have occurred because the mantle lobes were not embayed dorsally. We now feel that the large insertion at the posterior terminus of the pallial line in Eoptem’a (fig. 4; pl. 22, figs. 1, 2) may have housed an adductor muscle that served to flex the shell and to create water currents in the mantle cavity. No other rostroconch is known to have had a. comparable structure. ALIMENTARY CANAL Rostroconchs are regarded as having an anterior mouth and a posterior anus. This conclusion is sup- ported by the lack of any evidence of torsion in ros- troconchs and by the likelihood that they are de- scended from monoplacophorans and were ancestral to pelecypods. In primitive rostroconchs like Her- aultipegma and Ribeim‘a, the mouth was probably close to the anterior gape, but in such highly special- ized genera as Conocardium and Arceodomus, the mouth was probably situated at the anterior end of the mantle cavity, near the junction of the body and snout. We assume that the mouth moved posteriorly (in a relative sense) as the feeding structures of the snout became increasingly more complex. Before the development of a prominent posterior rostrum, the anus probably was near the dorsal side of the posterior shell gape. Because of the small di- ameter of the rostrum in many forms, the anus was probably not at its distal end. We conclude that in rostrate forms, the anus was near the proximal end of the rostrum and that water currents generated by cilia removed feces from the mantle cavity. FEEDING STRUCTURES Various rostroconchs have features indicating that structures could be protruded from the anterior shell gape. These features are: (1) the anterior pallial sinuses of Eopteria (pl. 22, figs. 1—4; pl. 23, figs. 2— 3), Euchasma (pl. 27, fig. 9), and perhaps Wam- wam’a (pl. 3, fig. 7); (2) the impressions of en- larged radial muscles of the mantle at the anterior end of the pallial line of Eoptem’a (pl. 22, figs. 5, 6) and Conocardium (pl. 51, fig. 11) ; (3) the apertural denticles of all conocardioids, probably formed by folds in the enlarged mantle as it was withdrawn into the shell (pl. 34; figs. 9—10) ; and (4) the muscle I insertions interpreted as pallial protractor muscles 20 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS that occur inside the edge of the anterior aperture in conocardiaceans (pl. 47, fig. 12). We believe that rostroconchs used this enlarged and extendible man- tle tissue for deposit feeding. Scaphopods and palaeotaxodont pelecypods use cephalic tentacles (captaculae and palp proboscides) in deposit feeding to collect food particles; similar structures are‘used for the same purpose by various prosobranch gastropods. It seems likely that many primitive rostroconchs used cephalic outgrowths to gather food. However, in the Conocardiidae, at least, these structures seem to have been superseded by outgrowths of pallial (mantle) tissue. In Conocardi- um and Arceodomus, the snout was effectively blocked by several pairs of enlarged marginal denti- cles, called longitudinal shelves (pl. 43, figs. 12, 13; pl. 44, figs. 2, 4). These shelves contained complex folds of the mantle, because the growing edges of the shelves, which oppose one another at the midline, are formed of outer shell layer. This composition im- plies that the mantle edge rested along the edges of the longitudinal shelves when the mantle was with- drawn. Because the mantle epithelium of mollusks is characteristically ciliated, we feel that the mantle resting on these anterior longitudinal shelves formed a complex sorting structure which collected and sorted food for eventual transmission to the mouth. The mouth was at the end of an elongate passage that penetrated between the various pairs of shelves (pl. 43, fig. 13). Because another characteristic of the molluscan mantle is the widespread occurrence of tentacles fringing the mantle edge, it is reason- able to believe that similar pallial tentacles may have assisted in the food-gathering process of rostro- conchs. We thus View the feeding structure of cono- cardiids as a set of forward-opening ciliated cones that acquired food by means of an inhalent water current and the manipulative abilities of fringing pallial tentacles. The food particles were moved pos- teriorly by cilia lining the cones, sorted, and trans- ferred to the mouth through a small aperture on the posteroventral side of the cones. Inhaled sediment was transferred anteriorly and exited along a re— jection tract at the ventral edge of the feeding aperture. The shape of conocardiids suggests that they both lived and fed infaunally. The scaphopods are good functional analogs for such genera as Conocardz’um (pl. 38, figs. 2, 8) and Arceodomus (pl. 43, figs. 5, 15) , except that they use cephalic captaculae rather than mantle tissue to collect food. Conocardium and Arceodomus are highly special- ized rostroconchs. More primitive rostroconchs have less complex anterior skeletal structures, and we speculate that they had a primitive version of the conocardiid feeding apparatus. Probably they used enlarged flaps of the mantle to collect and funnel food to the mouth. Some forms may have had a structure analogous to a pelecypod siphon which projected from the anterior gape. We envisage such a structure as being present in the ischyriniid Pseudotechnophorus because the anterodorsal gape is oval or kidney shaped, and because an insertion of a large muscle, which probably retracted the mantle (fig. 3G), is dorsal to the anterior pegma. Living tellinacean pelecypods use their posterior siphons for deposit feeding in this way (Pohlo, 1969). Most technophorid rostroconchs as well as the genus Ischym'nia have no anterior gape. These forms are laterally compressed, have a posterior rostrum, and probably lived infaunally. It seems likely that they were filter feeders which used cilia on the gills or mantle to pump water and suspended food in the posterior shell apertures. The mode of life of the eopteriid Euchasma is more difficult to interpret. All species of this genus have the anterior end of the shell reduced. The shells of E. blumenbachii (pl. 27, figs. 1—16) and E. mytili— forme (pl. 29, figs. 6—8) resemble those of living epifaunal mytilid and dreissenid pelecypods. E. jone- set has a small anterior lobe, and the shell is more or less similar to that of living modioliform pelecy- pods (pl. 28, figs. 12—15). By analogy with living pelecypods (Stanley, 1972) , the mytiliform Euchasmas probably lived epi- faunally, and the modioliform Euchasmas may have lived semi-infaunally. The narrowness of the antero- ventral shell gape (pl. 29, figs. 8, 9, 12) of Euchasma and the shell shape suggest that this genus was sessile. This conclusion is supported by the narrow- ness of the ventral shell gape, which would make it difficult for a foot to project ventrally. If, however, Euchasma is compared with the epifaunal cowrie gastropods, which it also approximates in form, a different interpretation results. This comparison shows that a large and effective foot can project through a narrow shell aperture, so Euchasma may have been a motile epifaunal or semi-infaunal ani- mal. The presence of marginal denticles lining the ventral gape indicates that mantle tissue at least was probably extended through this aperture. Euchasma has a sizable circular shell aperture above the anteroventral gape. This circular hole is formed by the edges of both valves (pl. 28, figs. 15, 16). It is effectively blocked by the pegma (pl. 29, figs. 3, 11, 14, 15), although there are small holes FUNCTIONAL MORPHOLOGY on either side between the pegma and the valves (pl. 29, fig. 15). Euchasma has an anterior pallial sinus (pl. 27, fig. 9) , so mantle tissue could probably have been extended from and withdrawn into this area of the shell. This mantle tissue may have been used for deposit feeding, as it was in other rostroconchs, and the pegma may have blocked sediment from entering the shell. The circular aperture is anatomically an- terior but functionally ventral, as during life, it would have been apposed to the substrate. Thus, Euchasma may have been an epifaunal to semi-in- faunal deposit feeder which “vacuumed” organic matter from the sediment-water interface. This un- usual mode of life may explain why it could not com- pete successfully with epifaunal suspension-feeding pelecypods, which first became abundant in the Mid- dle Ordovician. An alternative explanation of the mode of life of Euchasma is that the circular anterior aperture con- tained a structure for attaching the animal to the substrate. The attachment structures may have been similar to the byssus of pelecypods or the pedicle of brachiopods, and the animal may have suspension- fed from the posterior shell gape. Because rostro- conchs probably had hypertrophied anterior pallial structures, Euchasma may have been attached by one or more hypertrophied pallial tentacles. These tentacles could have been manipulated by contained fluid and pallial retractor muscles attached at the anterior pallial sinus. On the whole, we prefer the explanation that Euchasma was a mobile epifaunal or near epifaunal deposit feeder. CLEANING THE MANTLE CAVITY Suspension-feeding organisms that have an en- closed mantle cavity have the problem of eliminat- ing unwanted particulate matter (pseudofeces) swept into the mantle cavity along with the food. Pseudofeces continually accumulate in enclosed shells and must be continually removed. In most pelecypods, pseudofeces fall from the gills and man- tle to the floor of the mantle cavity and are then ejected by sudden contractions of the adductor mus- cles (Cox and others, 1969, p. N19). In brachiopods, reversal of the frontal cilia of the lophophore trans- ports the pseudofeces to the mantle, and then mantle cilia move them to the mantle edge. They are ex- pelled when the valves are adducted (Rudwick, 1970, p. 121). Some pelecypods have the ventral edges of the mantle extensively fused together, leaving only the two posterior siphonal orifices and a relatively small aperture for the foot. Many such animals also have ‘21 a small fourth aperture between the pedal and si- phonal orifices (Yonge, 1948; Runnegar, 1972). In active burrowers like Ens'is, this aperture acts as a safety valve to lower the fluid pressure in the mantle cavity which peaks with adduction in the digging cycle (Trueman, 1968). In more passive burrowers, the fourth aperture functions as an outlet for pseudofeces carried to it by ciliated tracts on the mantle and visceral mass (Yonge, 1948). Nearly all rostroconchs that have reduced pos- terior and ventral shell apertures retain a small cir- cular ventral orifice between the posterior end of the anterior gape and the rostrum (pl. 40, fig. 7). This orifice probably functioned in the same way as the fourth aperture of less actively burrowing pele- cypods—as an outlet for pseudofeces. Such a ventral orifice does not occur in the Conocardiidae, but, as mentioned previously, these rostroconchs may have had effective sediment screens in their anterior gapes and thus may not have been troubled by the accumulation of pseudofeces in the mantle cavity. WATER CURRENTS AND GILLS The helcionellacean univalves, like N eopilina (Lemche and Wingstrand, 1959) , probably drew wa- ter in under the anterolateral eaves of the shell and passed it out posterolaterally (figs. 6, 9) . We assume that the gills were laterally disposed in these ani— mals and that cilia on the gills and epithelium of the mantle cavity generated the water currents. In the univalve genera Yochelcionella (pl. 1, figs. 1—7) and Anabarella (pl. 17, fig. 8), the water current proba- bly entered anteriorly (figs. 6, 9). When the shell became modified into the ribeiriid shape, water was drawn in through the anterior gape and left the shell posteriorly (fig. 6). This water current was used for feeding as well as respiration. Ribeiria may have had a single pair of gills, the blood vessel being connected with the heart through the discontinuity in the side muscles seen in some species (fig. 3J). A similar water flow may have taken place in the eopteriids and conocardiceans. However, it is equally likely that the Conocardiacea drew water for respir- ation in through the posterior rostrum. Scaphopods use the posterior shell aperture to obtain water for respiration, although they deposit feed anteriorly. In the conocardiaceans, the exhalent current may also have flowed out the rostral orifice, or it may have left through the ventral orifice. Perhaps both aper- tures were used for this purpose. We suggest that all rostroconchs had gills because they all have an ex- panded mantle cavity. The technophorids and Ischyrinia have no anteri- 22 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS or shell apertures. We conclude that water entered and left the mantle cavity via the two posterior shell apertures. FUNCTION OF THE HOOD Some Ordovician through Mississippian conocardi- aceans have a hood attached to the carinal areas of the valves (pl. 45, figs. 10—13; pl. 48, fig. 2). There are two obvious possible explanations for the exist- ence of the hood, and either or both may have func- tional significance. First, the significant structure may be the hood, which could have been used to sup- port the shell in a soft substrate, or it could have de- flected water currents to or away from the rostral orifice. Some hooded rostroconchs resemble some liv- ing tropical cardiid pelecypods that are flattened in an anterior-posterior direction because the body tis- sue contains symbiotic algae which receive sunlight through the thin shell (Kawaguti, 1950). These clams lie exposed and are metabolically connected with the algae in their tissue. However, the resem- blance of t ese cardiids to hooded rostroconchs is only superficial. The hood is a totally external struc- ture, composed of the outer shell layer, which con- tained living tissue only along its central axis, the elongated ventral orifice. Thus, the hood probably had some other function. The second possibility is that the final structure, the hood, is not of primary functional significance; rather the structure that forms it, the elongated ven- tral orifice, is the functionally significant structure. If rostroconchs evolved a long thin tube at the pos- teroventral commissure, they could maintain such a structure during growth only by generating a planar structure on each valve. We speculate that the pro- longation of the ventral orifice, for whatever func— tional reason, may have had more functional sig- nificance than the finished hood. Once the hood formed, it may have provided support, enabling the animal to live in soft substrates, but the hood may have just been the necessary consequence of the elon- gation of the ventral orifice. TAPHONOMY Post-Ordovician rostroconchs (Conocardioida) are most common in marine shales and silts and reef limestones. Older rostroconchs are presently known most commonly from the carbonate sequences of epicontinental seas. J ameison (1971, p. 1334) noted that “Conocardium” has only been found in marginal reef deposits in the Devonian of western Canada, “and is therefore considered indicative of shallow, turbulent, open marine conditions.” Similarly, cono- cardiids are common in Viséan (Mississippian) shoreline cliff talus of the Wagon Creek Breccia (Veevers and Roberts, 1966) in northwestern Aus- tralia (John Roberts, oral commun., May 1972) . In contrast, Pseudoconocardium and Arceodomus are most common in low-energy marine shales of the Pennsylvanian of north-central Texas. Bransom'a occurs in a similar environment in coastal outcrops of the middle Permian Wandrawandian Siltstone. in New South Wales, Australia. At one important 10- cality, a recent shore platform at the town of Ulla— dulla, many specimens of Bransonia robustum (Fletcher) occur in silty beds. They are associated with many other fossils, most specimens of which have been preserved in situ. These include life-ori- ented productoid and spiriferoid brachiopods; shal- low-burrowing, free-swimming, and endobyssate pelecypods; collapsed but articulated crinoids; large unbroken colonies of lacy fenestrate bryozoans; and discoidal poriferans. The section of the Wandrawandian Siltstone ex- posed at Ulladulla contains several thin sands that vary in thickness from a few centimetres to tens of centimetres. These sands are also fossiliferous, but most of the fossils they contain are transported. The sands appear to have formed during rare high- energy events and therefore contain disoriented skeletons of organisms that (1) inhabited the sur- face of the silt and were light enough or were suffi- ciently loosely attached to be transported with the sand; and (2) that were unable to disinter them- selves after burial in the sand. The sand beds also contain life-oriented burrowing pelecypods that rapidly recolonized the substrate after each high- energy event. Extensive collecting has shown that there are few if any specimens of Bra/mania in the sand beds. Nor are there any byssate pelecypods or attached echino- derms. At least three alternative explanations are possible: (1) Bransonia burrowed so deeply that it was never disinterred by the high—energy currents that deposited the sands; this is unlikely, as there is no evidence of an elongate rostrum or large posteri- or siphons in Bransom’a. (2) Bransom'a was attached to the substrate by some structure comparable with the pelecypod byssus; this too is unlikely, as Brcm- som’a shows no anterior reduction, a feature seen in the epifaunal species of Euchasma. (3) Bramsonia. was sufficiently mobile to tunnel out of the sand after transportation and burial; this seems to be the most reasonable alternative and is the one we prefer. We have searched for specimens of rostroconchs encrusted with other organisms that might provide some clue to the life habits of the rostroconchs. Such PHYLOGENY 23 specimens are difficult to find in museum collections, and it is always difficult to prove conclusively that the encrusting organisms lived during the life of the rostroconch rather than encrusting it after death. The most useful specimen we have found is an individual of Hippocardia encrusted by an aulo- porid tabulate coral (pl. 33, figs. 1—2). The coral colony on this specimen seems to have been broken at least three times by the growth of the hood of the Hippocardia, suggesting that both organisms were growing simultaneously. The coral growth in this specimen also suggests that in life this species of Hippocardz’a had the whole of the dorsal surface of the hood exposed and was at most semi-infa‘unal. A more equivocal example is a bryozoan holdfast attached to the rostral area of the right valve of a specimen of Bransonia wilsoni (pl. 52, fig. 9). On this specimen, a matching mark on the correspond— ing part of the left valve suggests, but does not prove, that the holdfast was attached to the rostro- conch While the rostroconch was alive. Similar hold- fasts on the interiors of productid brachiopod valves at the same locality show that bryozoans were grow- ing on some of the dead organisms. In summary, we have very limited paleoecological evidence. Most of the specimens for this study were gathered from museums, and paleoecological infor- mation must usually be gathered in the field. What information we do have tends to confirm the con- clusions reached on comparative and functional mor- phology, that is, that most rostroconchs were mobile members of the shelf benthos and lived wholly or partially buried in the sediment. We have been un- able to devote much time for research in this area and suggest it as a profitable and challenging direc- tion for further enquiry. PHYLOGENY Living and fossil mollusks constitute the second largest and most variable invertebrate phylum (Bar- rington, 1967; Stasek, 1972). Most mollusks can be described as free-living metazoans that utilize a dorsal calcareous exoskeleton to provide structural support for a muscular foot (or its specialized de- rivative) and to provide an enclosed space outside the body (mantle cavity) that is used for feeding, respiration, and sometimes, locomotion. Because mollusks are so variable, no single unique character is present in all members of the phylum; they are unified by morphological gradations between differ- ent forms, by embryonic similarities, and by fossil evidence of their evolutionary history. We recognize eight classes of mollusks and refer these to four subphyla: Phylum MOLLUSCA Cuvier, 1797 Subphylum ACULIFERA Hatscheck, 1891 Class APLACOPHORA von Ihering, 1876 Subphylum PLACOPHORA von Ihering, 1876 Class POLYPLACOPHORA de Blain— Ville, 1816 Subphylum CYRTOSOMA Runnegar and Pojeta, 1974 Class MONOPLACOPHORA Wenz, 15940 Class GASTROPODA Cuvier, 17 97 Class CEPHALOPODA Cuvier, 1797 Subphylum DIASOMA Runnegar and Pojeta, 1974 Class ROSTROCONCHIA Pojeta, Runnegar, Morris, and Newell, 1972 Class PELECYPODA Goldfuss, 1820 Class SCAPHOPODA Bronn, 1862 We do not doubt that the forms Yochelson (1966, 1969) placed in the classes Mattheva and Stenothe- coida (=Probivalvia Aksarina, 1968) are mollusks, but we prefer to assign them to other molluscan classes. They are discussed in subsequent parts of this section. Tentaculites, lapworthellids, cornulitids, hyoliths, and hyolithellids probably belong to other phyla (Fisher, 1962; Matthews, 1973; Runnegar and others, 1975). ORIGIN OF THE MOLLUSCA Stasek (1972) produced a thoughtful review of the data pertinent to this problem. We agree with his conclusion that the mollusks evolved from a pre- annelid stock of small ciliated acoelomate, vermiform organisms that had a diverticulated gut, longitudi- nal nerve cords, and a series of dorsoventral body muscles. None of the known Ediacaran fossils of soft-bodied organisms of late Precambrian age (Glaessner, 1971) resemble this hypothetical an- cestor, but it is obviously similar to known living turbellarian flatworms. THE ANCESTRAL MOLLUSK Nineteenth century biologists visualized the com- mon ancestor of mollusks as a bilaterally symmetri- cal untorted snail-like animal that had a limpet— shaped shell and a posterior anus opening into a small mantle cavity containing a pair of simple ctenidia (Pelseneer, 1906) . This concept of an arche— typical mollusk was derived mainly from studies of 24 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS the comparative anatomy of living forms, but it was also widely accepted by paleontologists. When the living monoplacophoran N eom'lina Lem- che was discovered (Lemche, 1957), it was hailed as a living archetype. The only significant difference be- tween Neopilina and the theoretical ancestral mol- lusk lies in the structure of the gills, which are ar- ranged in a series lateral to the foot in N eopilma. Because of the close similarity of Neopilma and early Paleozoic Monoplacophora, N eopilina is often considered to resemble the ancestor of all other molluscan classes. This archetypical concept has been criticized by Yochelson (1963) , Horny (1965), Harry (1969), and Stasek (1972). An alternative View advocated by some authors is to derive the Monoplacophora and the other molluscan classes from nonshelled organisms that may have existed before and with early monoplacophorans. In this scheme, the differences between the various classes are produced in the nonshelled organisms, and calci- fication occurs after the characters of each class have been attained. Stasek (1972) argued that the Aplacophora, Poly- placophora, and Monoplacophora were derived se- quentially from an evolving ancestral stock. At present little evidence from fossils supports, or con- tradicts this suggestion. However, the Polyplaco- phora may have been derived from monoplacophor- ans that evolved multiple centers of calcification (Runnegar and Pojeta, 1974). Our study is largely concerned with the Monoplacophora and the five molluscan classes that we suggest were derived from it. This radiation began in the earliest Cambrian and is adequately shown in the fossil record (Runnegar and Pojeta, 1974). THE OLDEST KNOWN FOSSIL MOLLUSKS Russian stratigraphers divide the Early Cambrian of the Siberian Platform into four stages: from oldest to youngest Tommotian, Atdabanian, Botomi- an, and Lenian (Zhuravleva, 1970). The Tommotian deposits predate the first trilobites in the Siberian succession, and they contain a characteristic biota of archaeocyaths, mollusks, hyoliths, algae, and Problematica (Rozanov and others, 1969). In the fossil record, the base of the Tommotian appears to reflect the first appearance, in abundance, of animals that had calcareous skeletons, which is one definition of the beginning of the Cambrian (Zhuravleva, 1970; Webby, 1973). Tommotian mollusks are small or minute limpet- shaped planispiral or helically coiled univalves (Rozanov and others, 1969). They include forms re- sembling the widely known Cambrian genera Scen- ella Billings, Helcionella Grabau and Shimer, and Pelagiella Matthews, which are variously regarded as monoplacophorans, gastropods, or representatives of other primitive classes of mollusks (Knight, 1952; Rasetti, 1957; Horny, 1965; Yochelson, 1963, 1967). Rozanov and others (1969) referred all these uni- valves to the superfamilies Helcionellacea Wenz, 1938, and Pelagiellacea Knight, 1956, and considered them to be gastropods. One of the Tommotian mollusks, the genus Ana- barella Vostokova, is a laterally compressed plani- spiral univalve having a ventral margin that is obvi— ously curved when the shell is Viewed laterally (figs. 6, 7; pl. 17, fig. 8). Anabarella is intermediate in shell form between more typical Cambrian univalves like Helcionella, Latouchella Cobbold, and Igorella Missarzhevsky, and the first ribeiriid rostroconch, H eraultipegma n. gen. (=Hemultia Cobbold), from the Lower Cambrian of France (fig. 6; pl. 2). If H emultipegma is derived from Anabarella, it is un- likely that Anabarella was a gastropod, as the Late Cambrian and Ordovician descendents of H eraulti- pegma show no evidence of torsion. It is therefore pertinent to examine the biological placement of Anabarella and other Early Cambrian univalves. EARLY CAMBRIAN UNIVALVES The class Gastropoda comprises animals that have a distinct head, a solelike foot adapted for creeping, a radula, and a visceral mass that is apparently ro- tated 180° about a vertical axis so that the anus and organs of the mantle cavity are above the head. This twisting of the visceral mass is known as torsion; in living gastropods, it occurs in early ontogeny by rapid contraction of the asymmetrical right larval retractor muscle and by differential growth (Fretter, 1969). The torsion seen in all primitive and most advanced gastropods isolates them from their pre- sumed ancestors, the Monoplacophora (Knight and others, 1960). Most gastropods have the body contained in a calcareous univalved shell which coils posteriorly away from the head and is therefore termed “endo- gastric.” With the exception of the specialized lim- pets and a few aberrant forms, living gastropods have the «shell coiled in a helical spiral. Normally this coiling is orthostrophic and dextral, but rare individuals or species have hyperstrophic (ultradex- tral) or sinistral shells. Planispiral (isostrophic) shells resembling gas- tropods in external ornament and other features oc- cur as fossils from the earliest Cambrian to the PHYLOGENY 25 Early Triassic (Knight and others, 1960). These are now normally referred to the gastropod suborder Bellerophontida, though there has been a long de- 1 bate as to whether they were torted (gastropods) or untorted (monoplacophorans) (Yochelson, 1967). Recent studies have suggested that externally simi- lar planispiral shells housed both monoplacophorans and gastropods (Rollins and Batten, 1968). If the shells have several bilaterally symmetrical muscle insertions, they are believed to be untorted and hence monoplacophorans; all others have been considered to be gastropods until proved otherwise. This argu- ment is supported by the obvious asymmetry of the shell musculature in the otherwise symmetrical bi- valved snails (Kawaguti and Yamasu, 1960), and by the presence of only one pair of pedal muscles in some bellerophontids (Knight, 1947 ; Peel, 1972). If no muscle insertions are preserved, other cri- teria have been used. Knight (1952) and others sug- gested that the presence of (1) an anal slit or sinus, (2) secondary inner shell layers (parietal deposits) covering the exterior of earlier formed parts of the coil, and (3) an elongate trail, could be used to dis— tinguish torted bellerophontids from untorted mono- placophorans (Rollins and Batten, 1968). The most compelling argument related to the anal slit or sinus, as it was believed that these structures only became §\\\\\\ necessary when torsion juxtaposed the anus and ‘\ 'll . S b t1 , R ‘11' d B tte (1968) Ellioived utlfg’duirhey Devbnligji a14)Illanisp2ilraln univalve \\\\\\\\\\\\\\\\\\\\\\\ . . . . . Smmtopszs Perner has a series of bllaterally sym- metrical muscle insertions (fig. 8) as well as a deep sinus, and they concluded that it was a sinus-bearing monoplacophoran. They speculated that a posterior anal sinus was probably advantageous in achieving maximum separation of respiratory currents and excretory products. They discouraged the use of a sinus as a tool for recognizing gastropods and em- phasized the criteria of parietal deposits and pos— terior trails. No Early Cambrian univalves have parietal de— posits, but some have the concave side of the shell ....... \\\\\§‘ \\\\\\\ FIGURE 6.——Speculative View of the origin of the ribeiriid mstroconchs (A—B) from Early Cambrian helcionellacean monoplacophorans. Arrows indicate probable path of water currents through. the mantle cavity. The extent of the shell aperture is shown by the thick black line. Dotted shading in A represents the radial mantle muscles attached to the shell at the pallial line. Pedal muscle insertions in D are modeled from the Devonian cyclomyan monoplacophoran shown in E. A, Ribeiria, Late Cambrian-Ordovician; B, E H emultipegma, Early Cambrian; C, Anabarella, Early Cam- brian; D, Latouchellu, Early-Late Cambrian; E, Cyrtonella, Middle Devonian;— PALEONTOLOGY OF ROSTROCONCH MOLLUSKS 26 . é. \ i \. £sz , 5 ,~ WW .UWWW . Xummmwmvfim... awn.“ Wm” 5» , ,3.» z. zfimzzua «kg } FIGURE 7.—(See explanation on facing page.) PHYLOGENY 27 A v. B C FIGURE 8.—She11 muscle insertions of cyclomyan monopla- cophorans. A, Sinuitopsis, data from Rollins and Batten (1968); B and C, Cyrtolites, modified from Horn}; (1965). expanded to form a trail (fig. 9). Knight (1952) argued that a trail would impede the maneuverabili- ty of the protracted head of the animal and con- cluded that such shells must have been coiled endo- gastrically away from the head (fig. 9A, modified from Knight (1952)). For this reason, Knight re- ferred Helcionella, and related Early Cambrian uni- valves to the Gastropoda. No muscle insertions have been seen in these forms. In 1954, Rasetti illustrated internal molds of a small limpet-shaped shell from the Middle Cambrian Mt. Whyte Formation of British Columbia. These specimens have a number of small muscle insertions that are effectively bilaterally symmetrical. Rasetti referred the specimens to Scenella and concluded that they were monoplacophorans. A reconstruction of the body, modeled from Neopilz‘na (fig. 10), vin- dicates this decision. Knight and others (1960) referred Scenella and the enigmatic genus Palaeacmaea, Hall and Whitfield to a separate family of the Monoplacophora. Roza— nov and others (1969), however, referred the ex- ternally similar genus Tannuella Missarzhevsky to the Helcionellacea, and there is a gradation in exter- nal shell form from Helcionella through Bemella Missarzhevsky, Ginella Missarzhevsky, and Tamm- ella to Scenella (Rozanov and others, 1969). We therefore believe that the Helcionellacea are mono- placophorans, not gastropods. Additional support for the monoplacophoran placement of the Helcionellacea comes from the Aus- tralian Cambrian helcionellid, Yochelcionella Run- negar and Pojeta, 1974 (pl. 1), which differs from other helcionellids by having a tube attached to the concave side of the shell. Figure 9 shows normal and tube-bearing heliconellids reconstructed as gastro- pods and monoplacophorans. The first reconstruc- tion (fig. 9A) is modified from Knight (1952), who described it as a “harmonious and plausible picture.” We agree. However, if the tube-bearing helcionellid is reconstructed in the same way (fig. 9C) , the tube has no apparent function. By analogy with other mollusks, the tube probably carried water in or out of the mantle cavity. It could do this if the animal were an exogastric monoplacophoran (fig. 9D), an endogastric monoplacophoran (fig. 9F), or an exo- gastric gastropod (fig. 9E). Because of torsion, and hence by definition, all gastropods are endogastric (Knight and others, 1960). The gradations in shell form between H elci- onella and Scenella, and Helcionella and Hemulti- pegma indicate that the helcionellids Were exogas- tric (shell coiled forward over the head). We con- clude that the Helcionellacea were exogastric mono- placophorans, not endogastric gastropods. RADIATION OF THE MONOPLACOPHORA Horny (1965) divided the Monoplacophora into two groups designated by the terms Tergomya and Cyclomya. As the names imply, tergomyan monopla- cophorans normally have a series of discrete muscle insertions on each side of the shell (as in Scenella and Neopilina), whereas the cyclomyans have the muscle insertions more or less fused into a ring. The FIGURE 7.—Variation in shell form of Cambrian univalves. The shells are shown in left-lateral profile. Heavy lines show approximate shape of generating curves (apertural shape). A, Anabarella plana Vostokova, from pl. 17, fig. 8. B, Igorella ungulata Missarzhevsky, modified from Rozanov and others, 1969, pl. 4, fig. 21. C, Latouchella insulcata (Rasetti), modified from Rasetti, 1957, pl. 122, fig. 11. D, Helcionella carinata Rasetti, modified from Rasetti, 1957, pl. 122, fig. 5. E, Hypseloconus besseme'rense (Ulrich, Foerste, and Miller), modified from Stinchombe and Echols, 1966, pl. 79, fig. 13. F, Helcionella? rugosa var. comleyensis Cobbold modified from Cobbold, 1921. G, Scenelm sp., show- ing muscle-insertion areas, modified from Rasetti, 1954, pl. 12, figs. 5—8. 28 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS E FIGURE 9.—(See explanation on facing page). PHYLOGENY Radula __ ,' “N71 |!||| l,||||||| ll \|1I_ “Shell apex Anus FIGURE 10,—Rasetti’s Middle Cambrian specimens of Scenella reconstructed as a monoplacophoran, using Neopilina for a model. Stippled ring represents incipient pallial line indi- cated by change in slope of shell. Radial fluting of shell outside pallial line probably reflects weak radial mantle muscles shown here on left side only by short stippled lines. Other features shown are radial and circular muscles of the foot and muscles controlling head (anterior pair hy- pothetical). Arrows indicate probable direction of water flow through the right mantle cavity. Data from Rasetti (1954); from Runnegar and Pojeta (1974, fig. 2). Copy- right 1974 by the American Association for the Advance— ment of Science, published with permission. tergomyans are normally limpet shaped, and the cy- clomyans are taller coiled shells. Little distinguishes Ordovician, Silurian, and Devonian cyclomyan monoplacophorans like Cyrto- lites Conrad (Horny, 1965), Yochelsonellz's Horny (Horny, 1965), and Cyrtonella Hall (Rollins, 1969) from helcionellids like Latouchella (Cobbold, 1921) and I gorella. (Rozanov and others, 1969) , except for the anterior trail in some forms. The external orna- FIGURE 9,—Helcionella and related tube-bearing helcionellid Yochelcionella Runnegar and Pojeta reconstructed as an endogastric gastropod (A, C), an exogastric monoplaco- phoran (B, D), an exogastric gastropod (E), and an en- dogastric monoplacophoran (F). B and D are considered correct. See text for further explanation. A, Modified from Knight (1952). 29 ment of Scenella and of some species of H elcz’onella, Anabarella, and Cyrtonella is quite similar (Knight, 1941; Rozanov and others, 1969), consisting of fine radial threads between comarginal ribs, rugae, or other elements. We suggest that the post-Cambrian cyclomyan monoplacophorans are derived directly from the Helcionellacea. Most students of molluscan phylogeny derive the cyclomyan monoplacophorans from apparently more primitive limpet-shaped shells. The Tommotian fos- sil record suggests that the reverse may have been true. The oldest zone of the Tommotian Stage yields the relatively tall shells Bemella, I gorella, Anabel/rel- la, and Latouchella (Rozanov and others, 1969, table 9). The intermediate form Gimylla appears in the two succeeding zones, and the limpet-shaped shells Tannuella and H elcionella [sic] are absent until the base of the overlying Atdabanian stage. As the tall- er, coiled helcionellaceans are closer to the earliest gastropods, rostroconchs, and cephalopods than the limpet—shaped shells are, we suggest that the ter- gomyan monoplacophorans are not the ancestral stock but were secondarily adapted for benthic grazing. ORIGIN OF THE GASTROPODA Most malacologists consider the Bellerophontacea to be intermediate between the planispiral monopla- cophorans and helically coiled primitive gastropods (Knight, 1952; Knight and others, 1960; Morton and Yonge, 1964; Batten and others, 1967 ; Stasek, 1972). In this scheme, planispiral coiling precedes torsion and helical coiling follows it. The model implies that the shell and visceral mass rotate 180° with respect to the head and foot so that the gut becomes twisted, the left and right gills come to lie on right and left sides of the body, and the nervous system forms a figure of eight. It has always been difficult to imagine how this process could occur phylogenetically and to explain its adaptive significance (Knight and others, 1960; Ghiselin, 1966; Stasek, 1972). Stasek (1972) summarized two long-held theories: (1) that torsion had adaptive significance for the swimming veliger larva by bringing the mantle cavity into a position Where the tender vellum could be more easily re- tracted into the shell; and (2) that torsion would be beneficial to the adult, as it would move the sensory osphradia and gills away from water dirtied by the locomotion of the animal. Stasek rejected both ex— planations and suggested that torsion resulted from the temporary need of juvenile and adult monopla- cophorans to be able to twist the body to provide space for a protractible head. This, he suggested, 30 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS resulted in muscular asymmetry that was transfer- red to the larval stage in gastropods. Ghiselin (1966) had a different explanation of the adaptive significance of torsion, which is more con- sistent with the early Tommotian fossil record. He argued that planktonic larvae with a helically coiled exogastric shell would have problems settling be— cause the spire would interfere with their locomo- tion. Unfortunately, Ghiselin relied mainly on deduc- tions from the morphology of the protoconch and shell of N eom‘lina galatheae for his functional inter- pretation. This species was reported to have a helically coiled protoconch by Lemche (1957) and Lemche and Wingstrand (1959). Other species of N eom'lma taken subsequently have a bilaterally sym— metrical bulbous protoconch (N. W. Riser, written commun., 1974) , similar to that found on compara- bly shaped early Paleozoic forms (pl. 15). Conse- quently, we believe that all known living and extinct monoplacophorans are bilaterally symmetrical at all stages of growth. In a rejoinder to Ghiselin’s paper, Batten, Rollins, and Gould (1967) attempted to re- late Ghiselin’s model to the fossil record. They sug- gested that Ordovician-Devonian planispiral cyclo- myans were the ancestors of the gastropods. Pelagiella Matthew is a small helically coiled shell that is widely distributed in Lower Cambrian rocks (Knight and others, 1960). Although it resembles younger gastropods in shape, most authors have pre- ferred to regard it as an end-product of early mol- luscan experimentation rather than a primitive member of the class Gastropoda (Wenz, 1938; Knight, 1952; Knight and others, 1960; Yochelson, 1963). The oldest zone of the Tommotian Stage yields ribbed and smooth helically coiled pelagiellids refer- red to species of Aldomella Vostokova (Runnegar and Pojeta, 1974; Rozanov and others, 1969). Coarsely plicated helcionellids called Latouchella memorabilis and Latouchella korobkovi occur in the same beds. Both genera are preserved as minute phosphatic internal molds; we conclude that they were closely related. Specimens of Pelagiella from the English Lower Cambrian (Runnegar and Poj eta, 1974) are intermediate in form between Aldomella and Latouchella. If Latouchella were a monoplacophoran, it had an anterior mouth, posterior anus, and probably one or more pairs of gills attached to the lateral or postero- lateral flanks of the body (fig. 6D) . All body struc- tures were bilaterally symmetrical; Latouchella was untorted. Latouchella has a bilaterally symmetrical shell; in life, its plane of symmetry was probably vertical (fig. 6D). Aldomella has an asymmetric shell coiled in a low dextral spiral (Runnegar and Pojeta, 1974; fig. 13). Although the shells of Aldomella are very small (maximum diameter about 2 mm), they are significantly larger than the larval shells of living prosobranch gastropods that remain for an unusual- ly long time in the plankton (Fretter and Graham, 1962, p. 462—463) . Unless Aldomella was adapted for postlarval planktonic life, therefore, it would probably have settled out of the plankton when the shell reached a size of 300—400 microns (J. Taylor, oral commun., 1973) and had one or two whorls of coiling. Some species of Pelagiella are much larger (Knight, 1941), suggesting that the whole group lived benthonically as adults. During its planktonic larval life, the shell of Alda- nella could have been coiled exogastrically over the head. During settling, this orientation would be awk- ward, because in living prosobranchs the foot is poorly formed at this stage (Fretter and Graham, 1962; Ghiselin, 1966), and the larva would have difficulty balancing the shell vertically. Furthermore, to creep along the substrate, it would have to carry the spire and visceral mass instead of dragging them behind (Ghiselin, 1966). Between periods of activi- ty, the shell probably rested on one side; the newly settled larva of the living archaeogastropod H aliotis rests on its posttorsional left side during this period of its development (Fretter and Graham, 1962, p. 435). The dextrally coiled shells of Aldcmella would probably have fallen onto their umbilical side. Other dextrally coiled species referred to Pelagiella are significantly flattened on the side away from the um- bilicus (Robison, 1964, pl. 92, figs. 7—10; Hill and others, 1971, pl. 2, figs. 25—28), suggesting that the opposite side of the coil may have been the resting surface. Knight and others (1960, p. 323) reported that some species of Pelagiella contain both dextral and sinistral individuals, indicating that the side on which the larval shell rested may not have been rigidly fixed. Aldamella is consistently dextrally coiled and prob- ably rested with its umbilicus downward during and after settlement. If the body of Aldomella was or- ganized in the same way as the body of Latouchella, the head and foot of Aldwnella would need to rotate about 90° in the shell aperture to compensate for the change in orientation with respect to the substrate. This would allow the animal to protract its foot over the functionally ventral edge of the shell aperture and to move the coil from an anterior to a posterior PHYLOGENY 31 position (fig. 13). The unlikely alternative is that the animal protracted its foot by contracting the intrinsic circular muscles, anchored the end of the foot in the substrate, and then contracted the shell- inserted pedal retractor muscles to lift the shell into a vertical position above the head. We prefer the former explanation because: (1) it would have adap- tive significance for the settling larva and juvenile animal; (2) it shows how torsion could have origi- nated as a result of a small change in life orienta- tion; (3) it implies that pelagiellids are primitive gastropods, as their shape suggests; and (4) it al- lows us to derive the Gastropoda directly from the Helcionellacea in the Early Cambrian. Because of the asymmetry of the spire of Aldomel- la, the side of the aperture that is homologous with the left side of the shell of Latouchella is relatively enlarged, and the aperture is now asymmetric. This differential growth may have shifted the anus in the opposite direction to the mouth in the shell aperture. Aldomella crassa (Rozanov and others, 1969, pl. 3, fig. 16) has a small apertural sinus above the periph- eral part of the whorl, suggesting that the anus had moved in this way. We conclude that various parts of the body of Aldomella were probably rotated between 30° and 90° in two directions with respect to their positions in Latouchella. We note that the peripheral part of the shell of Aldanella is homologous with the convex edge of the shell of Latouchella; in a geo- metric sense, both shells are exogastric.1 If Aldanella is oriented like a primitive gastropod (fig. 13), however, the shell would be described as endogastric (coiling away from the head), and the body is partly or completely torted. The anus lies above and slightly left of the head, the left gill on its right side. The nervous system forms a figure of eight, and the gut is bent into a simple U. We con- clude that Aldanella and Pelagiella were primitive gastropods. Only a small readjustment is needed to produce the organization found in living pleuroto- mariid gastropods (Knight, 1952; Knight and oth- ers, 1960) . We conclude that earliest Cambrian planispiral exogastric monoplacophorans evolved directly into helically coiled, torted, primitive gastropods (Pela- giellids) when the orientation of the shell with re- spect to the substrate changed. During the initial period of experimentation, both sinistral and dextral forms developed, depending on whether the left or 1The distinction between endogastric and exogastric gastropods is not the same as the distinction between endogastric and exogastric cephalo- pods. In gastropods, both shells coil the same way, but the orientation on the head-foot differs; in cephalopods, the two types of shells are believed to have coiled in opposite directions. right side of the ancestral monoplacophoran came to lie on the substrate. When this occurred, torsion be- came necessary because the head-foot had to operate in a direction away from the coil instead of beneath it. This relatively small change converted the ani- mals from untorted exogastric monoplacophorans to torted endogastric gastropods. In living primitive gastropods, torsion is caused by a separation in the time of development of left and right larval retractor muscles (Fretter and Graham, 1962; Fretter, 1969). The pretorsional veli- ger develops only the right retractor muscle; when torsion begins, this muscle contracts rapidly (within 3—6 hours) and rotates the dorsal side of the velum to the pretorsional right side of the shell (Fretter and Graham, 1962, fig. 227). This muscle becomes the posttorsional left pedal retractor when the velum is lost. The right pedal retractor muscle develops later, during the period when differential growth completes the torsional process. It is not only the delayed development of the post- torsional right pedal retractor muscle which allows the initial 90° rotation of the velum to occur; most of the torque exerted by the pretorsional right larval retractor results from the way the distal ends of the fibers of this muscle run around the velum and are inserted on its pretorsional left side (Fretter and Graham, 1962, fig. 227 ; Morton and Yonge, 1964, fig. 3). If this did not happen, it would be difficult or impossible for a shell-inserted retractor muscle to rotate the velum in the plane of the shell aperture. Ordovician, Silurian, and Devonian planispiral monoplacophorans have simpler shell musculature than Neopilina, Scenella, and most other limpet— shaped tergomyans (Horny, 1965; Rollins and Eat- ten, 1968; Lemche and Wingstrand, 1959). Presum- ably this difference is related to shell form; as the shells became taller, the insertions of the longitudi- nal (retractor) muscles of the foot would coalesce or be reduced in number, and the circular muscles of the foot that are inserted on the shell in N copilma, Tryblidium Lindstrom, and Scenella (Lemche and Wingstrand, 1959) would no longer be attached to the shell. It would be difficult for any of these mus- cles to rotate the headnfoot in the shell aperture. Stasek (1972) has suggested that the delayed de- velopment of the left muscles of one or more pairs of pedal retractors would have allowed the cyclomyan monoplacophorans to twist the shell and visceral mass on the head-foot and so become preadapted for torsion. We disagree; even if these muscles were asymmetrically developed (spatially or temporally), they could only retract the head-foot into the shell. 32 PALEONTOLOGY 0F ROSTROCONCH MOLLUSKS In the tall, helically coiled shells, where the muscle insertions are about 90° of coiling from the shell aperture, it would be impossible for differential con- traction of the longitudinal muscles of the foot to rotate the head-foot in the shell aperture. An alternative explanation is that the pelagiellids were virtually untorted when the foot was with- drawn. It was only when the foot was protracted by contraction of its circular muscles that the shell assumed a posterior position. Individuals that could twist the head-foot efficiently would be selected for, particularly if the torsion was visible before the larva settled. Thus, torsion may have had adaptive significance for-both the larva (as suggested by Garstang, 1928) and the adult. The limited infor- mation does not allow us to suggest whether the pelagiellids were functionally (temporarily) or mor- phogenetically (permanently) torted; they may have been both. A corollary of this explanation for the origin of torsion and the Gastropoda is that the planispiral Bellerophontacea are no longer required as inter- mediates and may not have been torted (Runnegar and Pojeta, 1974). Our explanation only allows them to be torted if they are secondarily symmetrical, having descended from primitive helically coiled forms. As mentioned previously, Rollins and Batten (1968) used three criteria to identify bellerophonta- ceans as gastropods: (1) a long trail on the concave side of the shell, said to impede the maneuverability of a protracted head (Knight, 1952) ; (2) secondary inner shell layers (parietal deposits) covering the exterior of earlier formed parts of the coil; and (3) paired muscle insertions limited to the left and right sides of the columella (Knight, 1947) . By treating the Helcionellacea as monoplacophor- ans, we dispute criterion 1; possibly the trail pro- vided structural support for a sessile head and radu- lar apparatus as in Neopilina (Lemche and Wing- strand, 1959). Cowries and other gastropods secrete secondary shell layers on all parts of the shell; the argument that the parietal deposits of bellerophon- taceans could not be secreted by epithelium near the head is questionable (N. J. Morris, oral commun., May 1973). The difference in shell musculature in externally similar planispiral univalves is more problematical. Starobogatov (1970) suggested that the main muscles of a planispiral gastropod would be inserted on the columella of the spire to counter- balance its weight over the posterior part of the foot. Planispiral monoplacophorans would need their main muscles on the opposite side of the shell be- cause the spire was suspended over the head. We agree that it is unlikely that planispiral gastropods could have muscles on the outer side of the shell, but there is no reason why similarly shaped monoplaco- phorans could not have had their main pedal muscles attached to the columella. Morris (oral commun. May, 1973) has suggested that planispiral monopla- cophorans that had lateral gills could have had their main pedal muscles attached posteriorly; those that had more posterior gills may have emphasized the anterior musculature. We conclude that the small dextrally coiled Early Cambrian shells Aldomella and Pelagiella are the first gastropods; they gave rise to the sinuopeids, raphistomenids, and eotomariids of the Late Cam- brian (Knight and others, 1960). ORIGIN OF THE CEPHALOPODA Cephalopods have the mouth and anus juxtaposed, but the body is not torted; it is still bilaterally sym- metrical. Most living cephalopods lack a calcareous exoskeleton and are thought to be derived from more primitive shelled forms (Teichert, 1967) . Apart from the enigmatic fossil Vologdinella Balashov (Ruzhentsev and others, 1962) , no septate shells that could be cephalopods have been found in rocks older than the Late Cambrian (Teichert, 1967; Yochelson and others, 197 3). These primitive cepha- lopods, referred to the family Plectronoceratidae Kobayashi, have elongate, straight, or endogastrical- 1y curved shells, many closely spaced septa, and large ventral siphuncles (Flower, 1964; Teichert and oth- ers, 1964). The apparently oldest and most primitive genus is Plectronocems Ulrich and Foerste, in which the shell expands rapidly towards the aperture. As noted previously, most monoplacophorans have exogastrically coiled shells. However the Late Cam- brian and Early Ordovician genera Hypseloconus Berkey and Yochelsom'ella Flower, are tall, laterally compressed shells, which appear to have been endo- gastrically coiled (Knight and others, 1960; Stinchcombe and Echols, 1966; Yochelson and oth- ers, 197 3) . Such shells first appear in the early Late Cambrian (Lochman and Duncan, 1944, pl. 12, figs 37—38). Some forms have apical septa, and Yochel- son, Flower, and Webers (1973), suggested that they became primitive cephalopods when they de- veloped a siphuncle. We believe that the hypseloco- nids were derived from Early Cambrian orthocones like Tannuella (Rozanov and others, 1969) ; as Yoch- elson, Flower, and Webers (1973) suggested, the class Cephalopoda probably did not appear before the Late Cambrian. PHYLOGENY 33 Flower (1964, 1968), Teichert (1967), and Teich- ert and others (1964) have discussed the subsequent radiation of the Cephalopoda. ’ ORIGIN OF THE ROSTROCONCHIA The Early Cambrian genus H eraultipegma n. gen. (=Hemultia Cobbold) is a simple laterally com- pressed shell that has gaping anterior, ventral, and posterior margins (pl. 2, figs. 1—13). It is the oldest known rostroconch. We envisage a laterally com- pressed monoplacophoran such as Anabarella (fig. 6; pl. 17, fig. 8) giving rise to H emultipegma by the middle Early Cambrian. Internally, Hemultipegma is poorly known, but it clearly has a small pegma (pl. 2, figs. 7, 8) produced by a fold in the shell between the dorsal edge of the anterior gape and the anterior slope. We assume that H eraultipegma gave rise to a Ribeirz’a—like animal when the fold of shell beneath the beak thickened internally to form a transverse plate or pegma. Heraultipegma is known only from ferruginous internal molds which abound at the type locality (Thoral, 1935, pl. 1, fig. 3). It had only one center of calcification because the growth rugae cross the dorsal margin at right angles to the midsagittal plane (pl. 2, fig. 12). Watson‘ella Grabau is another laterally compressed shell from the Lower Cambrian of Massachusetts, which is probably allied to H eraultipegma; however, it is known only from the type specimens which are not well preserved (pl. 3, figs. 1—4) and yield little additional information about this type of animal. The change in shell form from a univalved mono- placophoran to a pseudobivalved ribeiriid like He- multipegma was probably accompanied by a change in life habits. Both Harry (1969) and Stasek (1972) described hypothetical animals which they felt must have existed as intermediates between monoplaco- phorans and pelecypods; both authors accurately de- scribe the morphology of Anabarella, Heraultipeg- ma, and other ribeiriid rostroconchs. Stasek’s fuller explanation (1972, p. 31—32 is par- ticularly pertinent: By and large, the monoplacophorans and primitive gastropods (Helcionellacea) of the early Cambrian seem to have been sluggish grazers of surface films or larger benthic algae. It was earlier inferred that within the herbivorous adaptive» zone, and while the phylogenetically fertile Monoplacophora were still less than 1 cm long, some side groups were experiencing anatomical trends toward increased eflic‘iency of individual pairs of their pseudometamerous organ systems. In [relation to the heightened form of the body and mantle cavity, some of these monoplacophorans had already successfully reduced of the gill filaments continued to function in creating a res— piratory water current and in removing particulate matter from it and the gills. This unwanted material undoubtedly in— cluded detritus and living plankton; that is, it comprised a quantity of material drawn from the same bank of organic energy that, from their earliest history, entire other phyla, especially the sponges and brachiopods, had tapped as a source of food. It is not surprising that one or more of the archaic monoplacophoran populations should have gradually come to exploit the same bank for its food supply, since a collecting me- chanism already existed in the ciliary cleaning device of its gills. The source, but not the kind of food would have changed, for the original filter-feeding types probably retained the essentially herbivorous habits of their ancestors. Some gastropods, such as Crepddula, utilize similar mech- anisms for collecting food, but having arisen late, found their potential for radiation somewhat stifled by preexisting and highly diversified filter-feeding mollusks. The Cambrian filter- feeding types, on the other hand, had entered an adaptive zone that had been untried by previous members of the phylum. The filter-feeding Monoplacophora [?Anabarella, Heraulti— pegma, Ribe'i'ria] are envisaged to have undergone trends toward increasing the length of the gill axes and of the fila- ments upon them in correlation with. ventral expansion of the eaves of the mantle and shell, which housed the body cavity. Passage of mucus-bound material anteriorly may originally have been by way of ciliary tracts on the surface of the body, but channeling devices and flaps of the body wall soon fun- neled potential food into the mouth. These flaps, the labial palps, later expanded and acquired a sorting mechanism based on the relative sizes of the particles gathered by the ctenidia. Retrospectively, the Bivalvia (Pel‘ecypoda) are descendents of these hypothetical monoplacophorans. Eventually the ribeiriids became adapted for in- faunal life, becoming deposit feeders or filter feeders rather than browsers or grazers. This allowed them to diversify and to exploit the soft-sediment environ- ment. We speculate that decephalization may have accompanied this change. RADIATION OF THE ROSTROCONCHIA In so far as possible, we rely on the stratigraphic succession in placing the gradations in morphology between various rostroconch taxa in their proper phylogenetic sequence. Thus, primitive characters are those known to have arisen first and are found in the oldest members of the class; advanced charac- ters occur in younger forms thought to have evolved from primitive members. Admittedly, this approach presents some difficulties. New discoveries may change present thoughts on correlations, they may extend the ranges of critical taxa, or they may pro- vide specimens that show morphological features not previously known. Sometimes a late-surviving primi- tive form may provide more insight into phylo- genetic relationships than stratigraphically older forms. Nevertheless, the stratigraphic succession of the number of ctenidia to one pair. Ciliation upon the surfaces 1 organisms iS basic to our notions of primitive and 34 advanced features and ancestors and descendents; each instance where it is not used is individually justified. Rostroconchs are not common fossils. Yet, in spite of this, we can demonstrate close morphological gradations between all major taxa, and in most cases, we can relate these changes to the stratigraphic suc- cession. It is the kind of paleontological situation which is theoretically called for, but which all too often cannot be observed in the fossil record. This situation makes systematic subdivision of the class difficult, because the taxa (both major and minor) that we recognize are parts of a continuum and are therefore difficult to define in the Linnean hierarchi— cal system. The only Early Cambrian rostroconchs known are H e'rcmltipegma vamensalense (Cobbold) from south- ern France (pl. 2, figs. 1—13), and Watsonella cros- byi Grabau from Massachusetts (pl. 3, figs. 1—4). Both occur with trilobites and are therefore Atda— banian or younger in age. They are small laterally compressed pseudobivalves, which gape anteriorly, ventrally, and posteriorly. In the Late Cambrian, Ribeiria taylori n. sp. is known from Trempealeauan rocks of New York State (pl. 8, figs 12, 13) ; R. austmliensis (pl. 4, figs 26—29) is present in the Mindyallan rocks of Queens- land, Australia; Womwamz'a cambm'ca Kobayashi (pl. 3, figs 5, 11—14) occurs in the Cambrian Tsinam'a Zone of Manchuria; and Oepikila cambrica (pl. 10, figs 14, 15) is found in the Idamean rocks of Queens- land. These Late Cambrian ribeiriids retain the prominent comarginal ornament of Hemultipegma. and Watsonella, the gaping margins, and the domi- nant posterior growth component. They are larger than Hemultipegma and Watsonella. and have a larger and more prominent pegma. What we know of Cambrian rostroconchs shows that a minimum radiation of the group took place at that time: two families, five genera, and six species. In the Early Ordovician, rostroconchs underwent their greatest radiation, diversifying into four known families, 14 genera, and about 43 species. In comparison, only one family, genus, and species of Cambrian pelecypods are known. By the Early Ordo- vician there are six families, 16 genera, and about 45 species of pelecypods. All mollusks, and indeed many invertebrates, radiated rapidly in the Early Ordovician (Tremadocian-Arenigian; Canadian), although they are known from few forms in the Cambrian. In the Middle and Late Ordovician, rostroconchs remained at about the same level of diversity as in PALEONTOLOGY 0F ROSTROCONCH MOLLUSKS the Early Ordovician: five families, 10 genera, and about 40 species. In contrast, pelecypods continued to radiate explosively and are represented by about 16 families, 140 genera, and 1,400 species in the same period of time. In the remainder of the Paleozoic, rostroconchs are represented by two families, seven genera, and about 275 known species. Pelecypods of the same age are referred to approximately 75 families, many hundreds of genera, and thousands of species. Thus rostroconchs form one of the smaller classes of mol- lusks, comparable in the number of named taxa to the Aplacophora, Monoplacophora, and Scaphopoda. The dominant rostroconchs of the Ordovician were ribeiriids, technophorids, and eopteriids; ischy- riniids and bransoniids were present in smaller num- bers. Technophorids, eopteriids, and ischyriniids show a melange of primitive and advanced features, indicating that the Ordovician was a time of adap— tive radiation for the class; various marine habitats were invaded, and several modes of life were evolved. Of the various combinations of morphology known in Ordovician rostroconchs, only the combination seen in the bransoniids (Conocardiacea) survives the end of the period. Perhaps only the conocardiaceans evolved a morphology that enabled them to exist with the far more efficient pelecypods during the latter part of the Paleozoic. The last rostroconchs occur in some of the youngest Permian deposits known (Newell, 1940; Waterhouse, 1967, p. 178) ; their fossil record terminates at the close of the Paleozoic. We now review the history of the class in greater detail. The most primitive family is the Ribeiriidae; it is the first to appear, its oldest members grade morphologically into primitive monoplacophorans, and all species referred to it have a simply con- structed shell, little different from the earliest mem- ber of the family. Most ribeiriids are posteriorly elongated (fig. 11), have a well-developed pegma, and have a shell that gapes anteriorly, ventrally, and posteriorly (primitive forms: H emultipegma, Ri- beim'a, Ribeirma) , or only anteriorly and posteriorly (advanced forms: Ribeiria, Wanwam'a, Pinna- com's). Assuming that all ribeiriids had similar shell musculature, the foot was attached to the shell by anterior and posterior median pedal retractors and possibly by left and right side muscles; the mantle was attached to the shell along a pallial line that had a shallow sinus near its anterior end. Ribeiriids probably had one pair of laterally disposed gills which created anterior to posterior water currents in the mantle cavity. The animals were motile, lived infaunally, and obtained food by a combination of PHYLOGENY 35 deposit and filter-feeding methods. Although the peg- ma may have provided needed structural support for the anterior retractor muscle of the foot, its pres— ence created problems during growth, and as a result the shell and mantle cavity remained narrow. Primi- tive ribeiriids are therefore laterally compressed; only in the advanced species Wanwcmia cambm‘ca Kobayashi (fig. 113' ) does the shell become inflated, thus showing a trend toward the eopteriid Euchasma. Like ribeiriids, technophorids also have a large and prominent anterior pegma (pl. 11, figs 21, 22), anterior and posterior median muscle insertions, and left and right linear muscle insertions (fig. 3H, 11%). The occurrence of these primitive features and the general similarity in shell form indicates a close phylogenetic relationship between techno- phorids and ribeiriids; this relationship is reflected in the systematics of the class by placing both fami- lies in the same order (Ribeirioida) . Technophorids have advanced beyond the ribei- riid stage in that, except for Myocaris Salter, they no longer have an anterior or ventral shell gape (pl. 11, figs 10, 11). In most forms, the posterior gape of the ribeiriids has been reduced to two small orifices formed by opposing folds (plicae) of the posterior part of the shell. Some technophorids (Technophor- us) developed a primitive rostrum, formed by an extension of the posterior dorsal margin of the shell (pl. 14, figs 6, 7) ; this rostrum terminated in a rela- tively larger dorsal posterior orifice (pl. 11, fig. 9). We homologize the smaller lower orifice, with the more distinct ventral orifice of younger conocardia- ceans; this structure appears for the first time in the Technophoridae. The organization of the body of ribeiriids and technophorids appears to have been similar, but the shell of the technophorids is effectively closed an- teriorly and ventrally, except perhaps when the foot forced the valves apart. We conclude that techno- phorids could not have moved around easily and that water entered and left the mantle cavity through the two posterior orifices. By analogy with younger conocardiaceans, the dorsal rostral orifice was the inhalant (and possibly exhalant aperture) ; the ven- tral orifice may have served as an exit for pseudo- feces. We regard technophorids as infaunal suspen- sion feeders, functionally analogous to coeval and younger pelecypods. As in the Ribeiriidae, the peg- ma inhibited the inflation of the mantle cavity. This and the other restrictions of an inflexible hinge reduced their chances of competing successfully with the Pelecypoda. Two technophorids, Tolmachom’a? jelli n. sp. and Tolmachovia concentrica Howell and Kobayashi, are almost equally expanded anteriorly and posteriorly (figs. 3B—D, 11k) . In Tolmachom’a, the cavity of the shell that contained the visceral mass is bounded anteriorly and posteriorly by transverse Shelly par- titions—anterior and posterior pegmas. These sub- equilateral species connect the technophorids to the genus Ischym’m’a Billings of the late Middle and Late Ordovician (fig. 3A). Significantly, there is a close resemblance between the Middle Ordovician Talmu- chom‘a concentrica and the oldest species of Ischy- rim‘a, I . norvegica Soot-Ryen, from the upper Middle Ordovician of Norway. The ischyriniids are the first rostroconchs to have a dominant anterior growth component, so that the umbos came to lie at the center of the hinge, or pos- terior to the center (fig. llo-q). They retained the anterior pegma, and in Pseudotechnophoms Koba- yashi this pegma has evolved into a greatly elongated structure (pl. 20, figs. 10, 11). In addition to the anterior pegma, all ischyriniids have a posterior peg- ma. Like technophorids, Ischyrinia lacks anterior and ventral shell gapes, and the posterior gape is restricted to two discrete orifices. Pseudotechno- phorus has an oval or kidney-shaped anterior dorsal aperture and a small posterior rostrum. In Ischym‘nia, the more dorsal of the two posterior orifices forms the aperture of the projecting rostrum, well differentiated in Ischyrmia winchelh’ Billings (pl. 18, figs. 22—25). In two species of Ischyrim'a, the ventral orifice coincides with a posterior carina and is a widely flaring transverse aperture that ap- pears to form late in ontogeny (fig. 3A; pl. 18, fig. 23) . The musculature of Ischyrima is apparently ribeir- iidlike. It consists of a linear muscle insertion con- necting the ends of the two pegmas (fig. 3A), and presumably of anterior and posterior median pedal retractor insertions. Pseudotechnophorus has more advanced lateral pedal insertions, like those of the conocardiacean Eoptem'a, and another large muscle insertion which forms a ring or a horseshoe in the cavity in front of the anterior pegma. For reasons explained elsewhere, we interpret this latter inser- tion as the attachment point of hypertrophied radial muscles of the mantle, termed pallial retractor mus- cles. Significantly, the protoconch of Pseudotechno- phoms has a tiny anterior median muscle insertion and an equally small linear insertion that passes from the anterior median muscle posteriorly behind the beak. We view these structures as the insertions ; of atrophied ribeiriid pedal muscles and conclude 36 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS ISCHYRINIIDAE 6’ RIBEIRIIDAE V3- 9 §fi§ fig , b M ON OPLA COPH ORA , a FIGURE 11,—(See explanation on facing page.) PHYLOGENY 37 that the functional pedal muscles of Pseudotechno- phoms were a new development. Ischyrinia and Pseudotechnophorus are referred to the same family (Ischyriniidae) and order (Is- chyrinioida) because of their similarities in shell form. However, Pseudotechnophorus is more ad- vanced than Ischyrinia. in its shell musculature. It cannot have been derived from Ischyrinia. if Ischy- rinia is descended from Tolmachovia, because Pseu- dotechnophorus is Early Ordovician in age. Pseudo- technophorus may have evolved from Eoptem'a, which is also found in the Early Ordovician and which resembles Pseudotechnophorus in shell mus- culature and to some extent in shell form (fig. 11). We realize the Ischyriniidae may be a nonphylo- genetic grouping, but find it convenient at this stage of knowledge. It is difficult to place too much em- phasis on stratigraphic occurrences at this time be- cause the ischyriniids are known from only a hand- ful of specimens from a few localities. Both Ischym’nia and Pseudotechnophorus have relatively few primitive characteristics and have departed farther from the ribeiriid stem stock than have the technophorids. Primitive features are the presence of an anterior pegma and the ribeiriidlike musculature of Ischym‘nw and the protoconch of Pseudotechnophoms; advanced features include the closing of the anterior and ventral parts of the shell, the development of a rostrum and specialized ventral orifice, the addition of a second pegma, and the domi- nant anterior growth component. Despite these rela- tively complex features, which to some extent mimic features found in younger conocardiaceans, we be- lieve that the ischyrinioids, like the technophorids, are a side branch of the rostroconch evolutionary tree. Throughout their history, conocardiaceans re— tained an anterior gape and lacked a pegma. We regard the less elaborate Eopteriidae as the an- cestors of the Conocardiacea. In the Early Ordo- vician radiation of rostroconchs, several lineages developed comparable structures by parallel evolu- tion; such parallel features include the closing of the ventral shell gape, the development of a rostrum, FIGURE 11.—Diversity of form in the Class Rostroconchia; Arrows indicate probable paths of evolution. Genera belong- ing to the same family are shaded in the same way. a, Latouchella; b, Anabarella; c, Heraultipegma; d—f, Ribeir- ia; g—h, Pinnocam's; i—j, Wanwania; k—l, Tolmachovia; m, Myocom's; n, Technophorus; o—p, Ischyrinia; q, Pseudo- technophorus; r—s, Eopteria; t, Wanwomella; u—w, Euch- asma; x—z, Bransom'a; (w, Hippocardia; bb, Conocardium. and restriction of the posterior shell gape to one or two apertures. The inhomogeneity of the Ischyrinioida is also re- flected in our interpretation of the life habits and soft-part morphology of Ischyrinia and Technophor— us. Ischyrinia probably developed the suspension feeding habit of its technophorid ancestors. It was probably buried to the depth of the carina, remained stationary, and fed and obtained oxygen through its posterior orifices. In contrast, we believe that Pseudotechnophorus was motile, had an anterior to posterior water cur- rent, and used protractible mantle tissue to deposit feed from the sediment-water interface. This pre- sumed life habit also places Pseudoteclmophoms closer to Eopteria' than to Ischyrinia, but the simi— larities may be due to convergence. All other rostroconchs are referred to the order Conocardioida. The Early and Middle Ordovician Eopteriidae are the oldest members of this order, and they are connected through Wanwomella Koba- yashi (Eopteriidae) and Wanwam'a (Ribeiriidae) to the Ribeiriidae (fig. 11). The Eopteriidae includes the genera Eopteria Billings (pls. 22—26) and Euchasma Billings (pls.27— 29) , which are the first known rostroconchs to have marginal denticles (pl. 23, figs. 4, 5; pl. 24, fig. 14; pl. 29, fig. 4) and external radial ribs (pl. 24, figs. 12, 18; pl. 28, figs. 12—15), features seen on all subse- quent conocardiaceans. The ribeiriid affinities of the Eopteriidae are shown by the presence of an anteri- or pegma in Euchasma (pl. 29, figs. 3, 11, 15) and Wanwanella (pl. 21, figs. 18—20), and by the fact that the eopteriids still have gaping anterior, ven- tral, and posterior valve margins (pl. 24, figs. 14, 15, 20). The shell musculature of Eopteria is well known from E. ventricosa (Whitfield) (fig. 4; pl. 22, figs. 1— 6; pl. 23, figs. 2, 3). Like Pseudotechnophorus, Eop- terz'a has the pedal muscles inserted laterally on each valve, but there is no trace of the relict ribeiriid mus- culature seen in Pseudotechnophorus. The pallial line is well preserved; there is an anterior sinus which housed pallial retractor muscles, and there is possi- bly a posterior adductor muscle. This latter muscle may have been used to flex the valves and so create water currents in and out of the mantle cavity. Some Middle Ordovician species of Eopteria show a dominant anterior growth component (pl. 26, figs. 12—18) at about the time that conocardiaceans first appear in the fossil record. Both Euchasma, and Eop- teria developed an incipient rostrum at the posteri- 38 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS or end of the shell (pl. 24, figs. 11, 12; pl. 28, figs. 12, 13), and all eopteriids had a more inflated shell and mantle cavity than the ribeiriids and technophorids. Thus, the presence in the eopteriids of a combination of ribeiriid and conocardiacean features suggests that they are descended from the former and are ancestral to the latter. The loss of the pegma in Eoptem'a and the modifi- cation of its edges in Euchasma allowed the valves to grow in tighter spiral, thus enlarging the mantle cavity. This lateral expansion of the shell produced umbones on either side of the beak of Euchasma, as it did in all members of the Conocardiacea. At this evolutionary level, the shells of rostroconchs, al- though still starting growth from a univalved proto- conch, became clearly bivalved and thus superficial- ly resemble the shells of pelecypods. The inflation of the mantle cavity also affected the topography of the hinge. It became increasingly diffi- cult for rostroconchs to maintain an approximately rectilinear hinge; tensional fractures called clefts developed between topographically high and low parts of the hinge. These are visible in front of the beak of some ribeiriids (pl. 4, fig. 9), but they are much more obviously developed in the eopteriids (pl. 27, fig. 13) and conocardiaceans (pl. 40, figs. 5, 7). In general, those forms having a dominant posterior growth component like Euchasma have well-devel- oped anterior clefts (pl. 27 , fig. 13) ; those having a dominant anterior growth component (Conocardia- cea) have posterior clefts (pl. 34, figs. 6—8); and those having subequilateral shells (Eoptem'a) may have clefts on both sides of the beak. The need for the clefts is explained by figure 1. As well as being strongly inflated, most species of Euchasma are flattened anteroventrally, and one species, Euchasma mytiliforme n. sp. has the exter— nal shell forms of epibyssate mytilid and dreissenid pelecypods (pl. 29, figs. 6-15). This suggests that Euchasma lived epifaunally. Although the shell of Euchasma has a narrow ventral to posterior gape (pl. 29, figs. 8, 9), the only sizable shell aperture is a circular hole formed by the edges of both valves at the dorsal anterior margin, just in front of the peg- ma (pl. 28, figs. 15, 16). If Euchasma were orien- tated as an epifaunal animal, this hole would face the substrate. We speculate that it contained a struc- ture formed by hypertrophy of one or more pallial tentacles and that was manipulated by pallial re- tractor muscles attached in an anterior sinus in the pallial line (pl. 27 , fig. 9). Euchasma may have lived attached to the substrate by this structure and may have suspension fed from water entering and leaving the mantle cavity through the posterior gape. Euchasma may have been a sessile epifaunal suspen- sion feeder. By contrast, Eopteria seems to have been a motile semi—infaunal deposit feeder which collected food from the sediment-water interface, using hyper- trophied mantle tissue and possibly pallial tentacles for the collecting structure. This tissue was extruded from the anterior gape and withdrawn by pallial retractor muscles located in the anterior pallial sinus. A narrow pelecypodlike foot was used for locomo- tion, and a posterior adductor muscle may have been used to clean the mantle cavity. Eoptem‘a is a suitable ancestor for the Conocardia- cea, and in fact, only the presence of a well—developed posterior rostrum distinguishes the Ordovician cono- cardiacean Bromsonia. cressmami n. sp. (pl. 52, figs. 10—14; pl. 53, figs. 6—21) from Eoptem'a (pl. 26, figs. 12—18). Eoptem’a is similar to the conocardiaceans in retaining an anterior gape, in lacking a pegma, in its musculature, and in having well-developed ex— ternal ribs and commissural denticles. It differs from conocardiaceans in having a continuous posteroven- tral gape instead of a rostrum and a discrete ven- tral orifice. Eoptem'a is also more variable in shape. Only some species are anteriorly expanded (pl. 26, figs. 12—18) like the Conocardiacea; others are sub- equilateral (pl. 26, figs. 1, 2) or posteriorly expanded (pl. 25, fig. 15). The technophorids and Ischym'mia. difier from eopteriids, ribeiriids, and conocardiaceans in that they have a closed anterior end; they disappear by the end of the Ordovician. Only the anteriorly gap— ing Conocardiacea are found in the post-Ordovician Paleozoic. There is evidence that most of the advanced ros- troconch genera having anteriorly gaping shells had hypertrophied mantle tissue which could be pro- tracted through the anterior gape. The three im- portant indications of the existence of this tissue are: (1) commissural denticles, formed by folds in the enlarged mantle as it was withdrawn into the shell (functionally analogous to the denticles lining the apertures of living cowrie shells) ; (2) the an- terior pallial sinuses of Wamvama, Eoptem’a, and Euchasma; and (3) the impressions of the radial muscles of the mantle seen in species of E'optem'a and Bransonia n. gen. In more primitive rostroconchs like Eoptem’a and Pseudotechnophorus, the mantle tissue seems to have been protracted hydrostatically. In the Conocardiacea, pallial protractor muscles in- sert-ed just inside the anterior end of the hinge probably pulled the mantle tissue out of the shell. PHYLOGENY 39 In the highly complex rostroconchs Conocardium Bronn and Arceodomus n. gen., the anterior gape is largely obstructed by internal calcareous shelves (pl. 43, fig. 13). These are formed of outer shell layer, implying that the outer edge of the mantle formed them. We conclude that the mantle was com- plexly folded in these areas in these forms. We homologize the shelves with the marginal denticles found in all other conocardiaceans, because these are also formed initially of the outer shell layer. In all conocardiaceans, the posterior shell gape is reduced to the small aperture at the end of the ros- trum, and in most an even smaller ventral orifice. The Conocardiidae have only the rostral orifice. We therefore doubt that any of these animals could have been posterior suspension feeders and conclude that they were all anterior deposit feeders using hyper- trophied mantle tissue and perhaps pallial tentacles to accumulate food. The more primitive forms prob- ably operated like Eopteria; the most advanced forms (Conocardiidae) may have had complex sort- ing structures formed by ciliated mantle surfaces resting on the anterior shelves. The inflated mantle cavity of all conocardiacean rostroconchs suggests that all genera had gills. This may account for the most striking difference in form between species like Conocardium elongatum and its functional analog, the tusk-shaped scaphopods. We assume that these gills pumped water in and out of the rostrum to supply oxygen to the organism and to remove body wastes. The ventral orifice may have been used to remove pseudofeces from the mantle cavity; its absence in the Conocardiidae may reflect the sophisticated sorting devices at the anterior end of the shell which prevented anything but food from entering the mantle cavity. We conclude that all conocardiaceans were deposit feeders. The filter- feeding rostroconchs (technophorids, Ischym’nia, and Euchasma) became extinct by the end of the Ordo- vician. The Ordovician was a time of major expan- sions in the suspension-feeding pelecypods and brachiopods, and competition for this mode of life may have led to the extinction of suspension-feeding rostroconchs. As mentioned above, the unspecialized eopteriid Eoptem'a gave rise to the most primitive conocardia- ceans, the Bransoniidae, in the Middle Ordovician. The bransoniids rapidly diverged into two long- ranging types, the hooded and nonhooded forms. Bransonia n. gen. (pl. 51, figs. 1-10, 12—16) is a simple conocardiacean having a large anterior gape, small commissural denticles, a small rostrum sharp- ly delimited by posterior clefts, coarse full body rib- bing, and a well-defined ventral orifice. Hippocardia Brown has all of the conocardiacean features of Bransonia, but has in addition a hood (pl. 48, fig. 2) attached to the umbonal areas of the valves, an ex- tended ventral orifice where the left and right sides of the hood meet, and enlarged anterior marginal denticles or small anterior shelves. The hooded lineage begins with Hippocm'dz'a, which gives rise to the Silurian-Devonian genus Bigalea n. gen. This form possesses two small hoods, one anterior to the other (pl. 37, fig. 4). Each hood has an aperture along the ventral margin, so that Bigalea has four commissural orifices, rather than the three usually found in conocardiaceans. In Biga- lea, the hoods are always small and never reach the enormous size of some species of Hippocardia, (pl. 48, fig. 2). The most complex rostroconchs belong to the fami- ly Conocardiidae. These have elaborate longitudinal shelves in the anterior aperture and an elongate shell clearly separable into three regions—rostrum, body, and snout (pl. 43, figs. 7, 5, 13, 15). Two gen- era are placed in this family, Conocardium Bronn (pl. 38) and Arceodomus n. gen. (pl. 43). They differ principally in the ornament on the body of the shell and are clearly closely related; Conocardium is the older of the two and presumably ancestral to Arceodomus. As noted above, the longitudinal shelves are prob- ably enlarged commissural denticles. In the bran- soniid genus Mulceodens n. gen., the denticles in the ventral part of the aperture are enlarged so that those from opposite sides are in contact (pl. 34, figs. 9—14). The denticles project into the aperture and are elongated anteroposteriorly (pl. 34, figs. 3—5). Further enlargement of such denticles could easily lead to longitudinal shelves, and the bransoniids probably gave rise to the conocardiids. Pseudoconocardium Zawodovsky is a bransoniid having an anterior gape that occupies almost the entire anterior face. The gape is not restricted to an anterodorsal position as it is in other conocardiace- ans. It seems likely that Pseudoconocardium was de- rived from Bransonia by an enlargement of the an- terior aperture in a ventral direction. In summary, ribeiriids are looked upon as the ros- troconch stem .stock, which in the Ordovician gave rise to the technophorids, ischyriniids, and eop- teriids. The ribeiriids continued until the end of the Ordovician. In the Middle Ordovician, Eopteria produced the first conocardiacean, which rapidly di- versified into the hooded and nonhooded lineages of the Hippocardiidae and Bransoniidae. The most spe- 40 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS cialized rostroconchs, the Conocardiidae, were de- rived from the bransoniids in the middle Paleozoic. The class became extinct at the end of the Permian. The Ordovician was the period of greatest radiation and diversification of the Rostroconchia; during this time, rostroconchs became adapted for infaunal and epifaunal suspension feeding as well as infaunal de- posit feeding. Only the deposit feeders survived the end of the Ordovician. ORIGIN OF THE PELECYPODA BIVALVED CONDITION IN THE MOLLUSCA Several kinds of mollusks have a bivalved shell, and the bivalved condition is the distinctive feature of the classes Rostroconchia, Pelecypoda, and the enigmatic group Stenothecoida. A few gastropods have a bivalved shell, and at least one species of octopus habitually inhabits discarded pelecypod shells. Other mollusks have shells composed of two or more parts (chitons, gastropods having opercula, pholad pelecypods having accessory plates, cephalo- pods having aptychi) , which are normally separate, but which in rare ”cases may be joined by a flexible structure resembling the pelecypod ligament. Such structures occur in some of the plates of the living chiton Schizoplax Dall (Dall, 1878; Knight and oth- ers, 1960; Harry, 1969) and in the junction between the operculum and shell of the Virgin Islands land snail Thyrophorella (Girard, 1895; Boettger, 1962; Harry, 1969). None of these shells are bivalved, but they demonstrate that two or more centers of calci- fication have arisen independently in many different kinds of mollusks. Some mollusks may be secondarily univalved, hav- ing descended from bivalved ancestors. Some clava- gellid pelecypods clearly fit this category, and there is some embryological evidence that the ancestors of the Scaphopoda had a bivalved shell (Lacaze- Duthiers, 1856—57 ; Yonge, 1957). By contrast, the oldest rostroconchs are morphologically intermediate between univalved monoplacophorans and younger bivalved forms. In this case, there is no clear distinc- tion between bivalved and univalved shells. The phylogenetic and behavioral changes accom- panying the attainment of a bivalved shell may or may not be reflected by major changes in the organi- zation of the body. Thus, the body and habits of the living bivalved opisthobranch snails (J uliidae) differ very little from those of related univalved opisthobranchs (Kay, 1968) ; the oldest pelecypod, Fordilla Barrande (Pojeta and others, 1973), how- ever, had its body organized quite differently from its univalved monoplacophoran ancestor. Despite these differences, the shells of the simplest juliid Berthelz’nia Beets and Fordilla are remarkably simi- lar in external form. The orientation of the valves on the body also varies. Rostroconchs, pelecypods, and bivalved opis- thobranch gastropods have the valves disposed on right and left sides of the body, the junction between the valves being in the anatomically dorsal position. The gastropods and rostroconchs retain this primi- tive orientation of the organism with respect to the substrate, but many pelecypods lie on left or right valves or even on the hinge (Cox and others, 1969) . Oysters, for example, developed valves that are func- tionally dorsal and ventral as in some productid brachiopods (Grant, 1966). Tridacnid clams have the plane of symmetry vertical but lie on the hinge (Yonge, 1953a). The stenothecoids (Yochelson, 1969) and the bi- valved limpets Hippom‘x antiquatus (Linnaeus) (Yonge, 1953b) and Cheilea, equestris (Linnaeus) have two subequal valves, but these are anatomically and functionally dorsal and ventral. The soft-part morphology of Stenothecoides Resser and related genera is not easily reconstructed, but the ventral valve of Hippom'x is formed by the sole of the foot, and thus is in no way analogous to the right and left valves of other mollusks. LIMITS OF CONVERGENCE IN SHELL FORM All pelecypods have a shell that is bivalved from the time that calcification begins (Raven, 1958). Before this stage, there is a single dorsal uncalcified cuticle (shell gland) on the larva. When calcification begins, it starts at two points on left and right sides of the saddle-shaped cuticle; these points eventually become the beaks of the valves. The intervening un- calified zone becomes the ligament, or the ligament may suddenly appear in this region in later ontogeny (Chanley and Chanley, 1970, pl. 3). Initially the hinge is long and straight so that the early larval shells of most pelecypods are character- istically D-shaped. As inflation of the valves pro- ceeds, umbos form on either side of the hinge, and the ligament becomes proportionally shorter. In most adult pelecypods, the ligament is relatively short, so that the valves are circular or oval. The growth lines reflecting the edge of the mantle curve towards the beaks and cross the hinge at the pos— terior and functionally anterior ends of the ligament. In the Rostroconchia, the larval shell is univalved, there is only one center of calcification, and the bi- valved condition arises through postlarval accentu- PHYLOGENY 41 ated growth of the lateral lobes of the shell. Because the valves are always effectively closed, there is no need for deep anterior and posterior embayments, so the growth lines are not recurved towards the beak. In the bivalved opisthobranch gastropods, the larva has a helically coiled shell and an operculum that is eventually shed. When the bivalved shell be- gins to form, it is not symmetrical because the liga- ment forms on the right side of the helical proto- conch (Kawaguti, 1959). This is a third situation in which there is originally a single center of calcifica- tion (the protoconch) and a second center of calci- fication (the right valve), develops subsequently. The adult shell is superficially bilaterally symmetri- cal but has the asymmetrically placed protoconch and a different arrangement of muscles attached to each valve (Kawaguti and Yamasu, 1960). For the pelecypod or gastropod ligament to func- tion efficiently, it must be relatively straight. Be- cause the generating curve of primitive pelecypod and bivalved snail shells is approximately circular or elliptical, it would be mechanically inefi‘icient to have the hinge on both sides of the beaks. In primi- tive pelecypods, the ligament is always behind the beaks (opisthodetic) and grows from its posterior end; in the bivalved opisthobranch snails, the re- verse is true, and the ligament is entirely prosodetic. The resulting shells are quite similar in external form except that the left valve of a primitive pelecy- pod like Fordilla. resembles the right valve of the bivalved gastropod Berthelz‘nia. The origin of the bivalved condition in the Steno- thecoida is not well understood, but the growth lines on each valve run completely round the beak, produc- ing a good deal of interumbonal growth (Yochelson, 1969, fig. 3). Analogous growth increments are found in many brachiopods, where the valves are totally separate structures growing at all edges; growth lines also occur in the shells of the bivalved limpets. This growth pattern reinforces the View that stenothecoids are only remotely related to ros- troconchs and pelecypods. ROSTROCONCHS AS ANCESTORS OF THE PELECYPODA Fordilla troyensis Barrande from the Early Cam— brian is the oldest known pelecypod (Pojeta and others, 1973; Pojeta and Runnegar, 1974; Pojeta, 1975). It has a laterally compressed shell, and promi- nent, but not rugose, comarginal ornament. Fordilla is about the same size and age as the ribeiriid H er— autipegma (pl. 2, figs. 1—13), and the two genera have similar lateral profiles. Fordz'lla. has a bivalved larval shell, a simple ligament-insertion area, ad- ductor muscles, pelecypodlike pedal muscles, and a well-developed pallial musculature. There are no shell gapes; when the adductors contracted, the valve margins were tightly closed. In contrast, Heraultipegma has a univalved larval shell, a pseu- dobivalved adult shell, and an anterior through pos- terior shell gape; by analogy with younger ribeiriids, H eraultipegma probably lacked adductor muscles. Several Cambrian-Ordovician ribeiriids resemble coeval pelecypods in external form, and decalcifica- tion of the posterior dorsal margin in successive generations could have produced the primitive opis- thodetic parivincular ligament found in early pele- cypods. Such a ligament could only function if the anterior embayment of the shell extended as far as the beak, and both evolutionary changes may have progressed simultaneously. Increasing flexibility of the posterior hinge would have enabled the valves to open wider anteriorly, and less shell may have been secreted along the anterior dorsal margin. Obviously the pegma would have inhibited valve movement, so forms like H eraultipegma that have a small pegma are more likely to have developed a flexible hinge. The main difference, however, between pelecypods and rostroconchs is that in pelecypods, the shell is bivalved from the very beginning of its growth. Ros— troconchs have a univalved larval shell and only one primary center of calcification. This is a funda- mental difference; there can be no intermediates. Once a flexible ligament was established in the larval shell, the adult would inevitably resemble a pelecy- pod. The hinge would remain relatively short to re- tain flexibility, and each center of calcification would have a generating curve that terminated at the ends of the hinge. By contrast, the ribeiriids could never produce a pelecypodlike shell because the univalved protoconch remains attached to both valves. The conchological differences between rostro- conchs and early pelecypods like Fordz'lla, Redom’a Rouault, Babinka Barrande, and Lyrodesma Con- rad are considerable (Pojeta, 1971) , but most of the differences relate to the geometric effects of growth from two centers of calcification. The shape of the valves and the presence of the ligament on the longer side of the shell are a direct consequence of two growth centers, as precisely the same structures are found in primitive pelecypods and bivalved opistho- branch gastropods. We believe that a single muta- tion producing a flexible hinge in the larval shell would be sufficient to convert a ribeiriid into a pelec- ypod, if the difference were judged on shell form 42 4: D FIGURE 12..—Four possible explanations of the shell muscle insertions of the Early Cambrian pelecypod Fordilla. Ad- ductor muscles are cross hatched; radial pallial muscles are stippled; muscles extending from the shell to the inner surface of the mantle are diagonally shaded; pedal muscle PALEONTOLOGY OF ROSTROCONCH MOLLUSKS alone. This change probably occurred in the Atda— banian Stage of the Early Cambrian, when Fordilla evolved from H emultipegma or some closely related form. ACCOMPANYING MODIFICATION OF BODY FORM Judging from N eom'lina and the muscle insertions visible on fossil monoplacophoran shells, Cambrian- Ordovician monoplacophorans were untorted snail- like animals with a head, a visceral mass, laterally disposed gills, and a ventrally flattened creeping foot attached to the shell by muscles inserted in a continu— ous or discontinuous ring. Tiny muscles controlling the edges of the mantle occur in N eopilina, but none of the ancient or modern shells shows a well-differ- entiated‘ pallial line. In marked contrast, ancient pelecypods resemble their modern counterparts in having a reduced head and a laterally flattened probing foot attached to the shell mainly above the anterior and posterior ad- ductor muscles. The radial muscles in the mantle are greatly enlarged to form a continuous pallial line, and the adductor muscles which close the valves are believed to be hypertrophied radial muscles which have been cross-fused in the anterior and posterior embayments (Yonge, 1953a) . The pedal musculature of Fordilla is particularly significant, as the main pedal retractors are inserted anteriorly and posteri- orly, implying that the extrinsic muscle fibers of the foot were arranged in a geodetic net as in modern pelecypods (Trueman, 1967). Thus, the foot of Fordilla must have been used for burrowing rather than creeping. Fordz’lla also has an unusually large set of muscle insertions forming the posterior part of the pallial line (fig. 12). None of the explanations of the func— tion of these muscles offered by Pojeta, Runnegar, and Kriz (1973) (siphonal retractors, accessory ad- ductors, muscles retracting the inner surface of the mantle) would be logical if Fordilla were a mono- placophoran having a bivalved shell. We conclude that Fordilla and similar Ordovician genera were al- ready well adapted for life as bivalved organisms. They were pelecypods and not bivalved monoplaco- phorans, \ We have already noted that the ribeiriids are tran- sitional between the Monoplacophora and Pelecy— poda in shell form. The ribeiriid shape would allow the animal to become adapted for life in a bivalved insertions are black. Arrows indicate possible water flow in and out of the mantle cavity. Note that if B were correct, the whole of the posterior end of the shell would be effec- tively sealed. A and B are less likely, C and D more likely. PHYLOGENY 43 shell before the truly bivalved condition developed. Presumably, Heraultipegma and Watsonella were infaunal animals capable of deposit or filter feed- ing. They may have lacked a well—formed head, and their foot was probably already adapted for probing rather than creeping. Thus, they may have had the soft-part morphology of early pelecypods in an ef- fectively univalved shell. A single mutation produc- ing two centers of calcification in the larval shell would then produce a pelecypod, preadapted for ex- ploiting the new shell form. We have no information on the musculature of H emultipegma or Watsonella, and ribeiriids in Which the muscle insertions are known postdate Fordilla by some 50—70 million years. Ribeiriids are almost as rare as pelecypods throughout the Cam- brian, so our interpretation of the evolution of the pelecypods from the monoplacophorans via the ribeiriids is based as much on comparative morpholo- gy as on stratigraphic sequence. However, we be- lieve that the Ordovician representatives of all three groups retain the fundamental features of their Cambrian ancestors, just as living fish, reptiles, and mammals reflect their Mesozoic and early Cenozoic counterparts. If this be so, some Cambrian ribeiriids, like their Ordovician descendents, had probably developed a pallial line. The pedal musculature still formed a ring on the shell, as in the Cambrian-Ordovician mono- placophorans. Eventually the muscles on the midline were enlarged to operate a pelecypodlike foot. When the truly bivalved condition was attained, the an- terior and posterior radial muscles of the mantle cross-fused to form adductors, and the anterior and posterior median muscles split to form the paired pedal retractors attached above the adductor inser- tions on each valve. The remaining parts of the pedal musculature, already fragmented in some ribeiriids, formed the small visceral/pedal muscles found in Fordilla and many Ordovician pelecypods. It is the adductor muscles and well-developed hinge teeth of Cambrian-Ordovician pelecypods that so clearly separate them from coeval rostroconchs, but both of these structures would normally be unneces- sary until a flexible ligament evolved. Both are pres- ent in bivalved snails, suggesting they can form rapidly when the need arises. As some univalved snails have an adductor muscle that is used to pump water in and out of the mantle cavity (Marcus and Marcus, 1956), some rostroconchs may have devel- oped adductors for the same purpose. Eoptem‘a, for example, seems to have a posterior adductor muscle. If some primitive ribeiriids had adductor muscles, they would obviously have been more successful when the ligament evolved. RADIATION OF THE PELECYPODA Apart from the curiously enlarged pallial muscles, Fordilla troyensis is a suitable ancestor for all the known subclasses of the Pelecypoda (Pojeta and others, 1973; Pojeta and Runnegar, 1974). Recent discoveries summarized by Pojeta (1975) show that the pelecypod subclasses visible in the Ordovician (Pojeta, 1971) could easily have stemmed from a single Cambrian stock. We conclude that the class appeared from the ribeiriid rostroconchs in the At- dabanian Stage of the Early Cambrian, remained an almost insignificant component of the biosphere un- til the Tremadocian, and then radiated rapidly into all the existing subclasses by the late Middle Ordo- vician (Pojeta, 1971). ORIGIN OF THE SCAPHOPODA Scaphopods have a tubular shell which is often gently curved and is invariably open at both ends. Growth proceeds by the deposition of new shell at the larger end of the tube and simultaneous resorb- tion at the opposite end. Coarsely silicified replicas of tusk-shaped shells resembling later undoubted scaphopods are known from the Ordovician of the United States. We be- lieve the class was probably well differentiated by this time. However, Yochelson (oral commun., 1973) maintains a more conservative View of the range of the class, preferring to accept only Devonian and younger tusk-shaped shells as scaphopod mollusks. The presence of a slit in the smaller shell aperture is probably the best evidence for distinguishing scapho- pods from similarly shaped worm tubes or other or— ganisms. Unfortunately, the Ordovician specimens are too poorly preserved to show this feature. The ontogeny of the living scaphopod Dentalium shows that the larval mantle and shell first appear dorsally and then grow left and right lobes which eventually coalesce ventrally to produce the tubular juvenile and adult shell (Lacaze-Duthiers, 1856—57 ) . This embryological observation has led to the belief that scaphopods are more closely related to the Pe- lecypoda than they are to any other group of extant mollusks. We also rely on the embryological evidence to pos- tulate that the Scaphopoda may have evolved from the ribeiriid rostroconchs. If the inner edges of the mantle lobes of a ribeiriid fused ventrally, the shell could still grow normally, as it does in pelecypods that have ventrally fused mantle margins. Subse- 44 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS quent fusion of the outer edges of the mantle could have produced a ventrally fused shell. As soon as this happened, the postlarval shell would become tubular, and all subsequent growth would proceed as in living scaphopods. The result would be the im- mediate production of a scaphopodlike shell; there could be no morphological intermediates between the two growth forms. If our reasoning be correct, we are unlikely to discover fossils that prove the phylo- genetic connection between the two classes. Intermediate forms could occur if ventral fusion of the shell first occurred in late ontogeny and was subsequently transferred to the larval shell. For ex— ample, the juvenile bivalved shell is preserved on the dorsal side of the tubular “adventitious” shell of the clavagellid pelecypod Brechites (Purchon, 1960) . So far, no fossil ribeiriids showing similar features have been discovered. If we can demonstrate a connection between the monoplacophorans and ribeiriids, it would allow us to argue that the ancestral group would have shared primitive anatomical features found in living scapho- pods. In particular, living scaphopods have a radula. If the scaphopods developed from the ribeiriids, we can conclude that some or all ribeiriids also had a radula. The ribeiriid Pinnocaris (pl. 9, figs. 11—24) has a shell form approaching that of scaphopods but still has a prominent pegma. MATTHEVA AND STENOTHECOIDA (PROBIVALVIA) Yochelson (1966, 1968, 1969) and Aksarina (1968) placed two small groups of enigmatic Early Cambrian fossils in separate molluscan classes called Mattheva Yochelson 1966 and Stenothecoida Yochelson 1968 or Probivalvia Aksarina 1968. The names Stenothecoida and Probivalvia Aksarina are objective synonyms; Yochelson’s name was published in an abstract in August 1968, and Aksarina’s at an unspecified time in 1968. As Harry (1969) also used Probivalvia in a different sense, and the name has some phylogenetic connotations, we suggest that Yochelson’s name be used for this group of organisms. The class Mattheva is based on a single genus, Matthevz’a. It is known from two, co—occurring, sub- equal, massive conical plates that are flattened on one side and that have two tapering cavities on the side that was attached to the animal. The plates show growth lines and probably formed part or all of the exoskeleton of a primitive mollusk (Yochelson, 1966). Runnegar and Pojeta (1974) suggested that M atthevia is a primitive chiton. The class Stenothecoida is more diverse; it proba— bly includes the following genera: Stenothecoides Resser, Bagenovia Radugin, Cambridium Horny, Bagenoviella Aksarina, Sulcocam'na Aksarina, Kaschkadakia Aksarina, and Makamkia Aksarina (Aksarina, 1968; Yochelson, 1969). Stenothecoids are demonstrably or inferentially bivalved shells that are normally found disarticulated. 'The valves resemble coeval limpet—shaped tergomyan monopla- cophorans in shape, except that they are slightly asymmetrical, and valves that look like the right and left valves of some pelecypods can usually be dis- tinguished (Poulsen, 1932; Yochelson, 1969). The few articulated specimens known (Aksarina, 1968; Yochelson, 1969) are slightly inequivalv‘ed. The shell morphology is best known from Steno- thecoides (Rasetti, 1954; Horny, 1957; Robison, 1964; Yochelson, 1969). The valves are unorna- mented except for obvious comarginal growth lines and a subangular carina that runs from the beak to near the midpoint of the opposite part of the com— missure. Apparently well preserved internal molds have a relatively smooth median zone that appears to coincide with this external carina, and a series of branching elevations that reflect grooves in the shell that run away from the central zone on both sides of the valve. These grooves may reflect bilaterally ar- ranged canals or tubes in the mantle; if so, these grooves branch towards the margin of the shell, and they all appear to join the central zone. Externally, stenothecoids vary from relatively smooth shells to elongate oysterlike forms orna- mented by divergent angular folds in the shell (Ak- sarina, 1968). These folds interlock at the valve margins and may be homologous with radial mark- ings on the interiors of the smoother shells. Yochel- son interpreted Stenothecoides as a brachiopodlike mollusk. Runnegar and Pojeta (1974) offered the alternative suggestion that it may have been a bi- valved monoplacophoran, the lower (smaller?) valve being formed by the sole of the foot. A few living limpets form a second valve in this way, al- though in the limpets, the lower valve is cemented to rocks. MOLLUSCAN SUBPHYLA Stasek (1972) theorized that the extant mollusks are the progeny of three lineages that separated be- fore the phylum was well established. He noted that no known intermediate forms, fossil or living, bridge the “enormous gaps between any two of the three lineages.” He therefore treated each as a separate subphylum. They are: (1) the subphylum Aculifera, PHYLOGE NY 45 containing only the class Aplacophora, derived from the most primitive of the ancestors of the Mollusca; (2) the subphylum Placophora, containing only the class Polyplacophora and emphasizing the pseudo- metamerism of its more advanced premollusk an- cestor; and (3) the subphylum Conchifera, contain- ing the class Monoplacophora and the other classes derived from it. We have no expert knowledge of the aplacophoran and polyplacophoran mollusks, but we agree with Stasek that major differences exist between these organisms and other mollusks. We adopt his basic subdivisions of the phylum but suggest that the Conchifera can itself be separated into two major lineages worthy of the rank of subphylum (fig. 13). The fossil record indicates that the Monoplacophora gave rise to the Gastropoda, Cephalopoda, and Ros- troconchia, and that the Pelecypoda and Scaphopoda are derived from the Rostroconchia. These last three classes thus form a lineage that diverged from the Monoplacophora in the Early Cambrian. They em- phasized a shell that in all groups is primitively open at both ends, allowing the gut to remain rela- tively straight, and having an anterior mouth and posterior anus. We coined the term Diasoma (through-body) for the subphylum containing the three classes Rostroconchia, Pelecypoda, and Scapho- poda (Runnegar and Pojeta, 1974). The remaining three classes (Monoplacophora, Gastropoda, and Cephalopoda) emphasize a conical univalved shell, usually twisted into a spiral. The relatively small single shell aperture forces the anus to lie close to the mouth, and the gut is bent into a U. We use the name Cyrto‘soma (hunchback-body) for the subphy— lum containing these three classes (Runnegar and Pojeta, 1974). Strictly speaking, the cyrtosomes are the ancestors of the diasomes, but in fact both sub- phyla appeared and began to diversify within a few millions of years in the Early Cambrian. The Cam- brian-Ordovician record of molluscan higher taxa is shown in figure 14. GLOSSARY OF MORPHOLOGICAL TERMS Perpendicular to hinge or almost so. adductor muscle. Muscle used to draw the tw0 halves of a bivalved shell together; believed to have developed by crossefusion of distal ends of opposing pallial retractor muscles (pl. 22, figs. 1, 2). anterior. Edge of shell having gape and commissural den- ticles (Eopteriidae, Conocardiidae) (pl. 24, fig. 14; pl. 40, fig. 6); or end of shell having pegma (Ribeiriid‘ae, Tech— novphoridae) (pl. 6, fig. 7; pl. 11, fig. 22). anterior branch. Part of pallial line extending dorsally from the pallial junction toward the midline (pl. 53, fig. 1). anterior clefts. Tension fractures formed during growth on either side of shell in front of beak (pl. 4. fig. 9; pl. 28, fig. 17). anterior gape. Opening at anterior end of shell (pl. 2, fig. 12; pl. 6, fig. 5; pl. 24, fig. 14; pl. 28, fig. 15; pl. 40, fig. 6; pl. 43, fig. 13). anterior umbonal cavity. Part of um‘bonal cavity of ribeiri- oids in front of pegma (pl. 5, fig. 9). acline. Ald SUBPHYLUM DIASOMA V mea . ' L to ch ll innocam's g..fi m m w x .c or: - / g E m 3e ‘J K —> g {g . I PRIMITIVE m E MONOPLA COPHORA MOLL USK / {‘3 a ' Ribeirw I Plagioglypta S ”3 ~. m S h M ”WWW”, g s: ’ , , / ” o E a” - — fl 8 :5) Tannuella \ E - Fordilla E01” term U3 Knightoconus Anabarella HeTaultzpegma I Cycloczmfiha a K Plectronoceras a /’Wm/’” ‘ %* S UBPHYL UM C YR TOSOMA 5": PELEC YPODA FIGURE 13.——-Schematic View of the origin of the univalved and bivalved molluscan classes. Most drawings are based on internal molds of the shells. Thick lines show extent of shell apertures; stippled areas represent muscle insertions; shaded areas show probable position of gut, and mouth is indicated by asterisk. From Runnegar and Pojeta, 1974, fig- ure 4; Copyright 1974 by the American Association for the Advancement of Science, published with permission. 46 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS apertural plate. Internal subcircular disk attached to both body of shell. Inflated part of shell between snout and valves behind the anterior gape of some eopteriids; a modi- rostrum (pl. 43, fig. 5). fied pegma (Pl- 29, figs. 14, 15 )- carina. Angulation of umbo (pl. 29, figs. 6, 7; pl. 50, fig. 38). apertural (longitudinal) shelves. Curved plates of outer shell layer projecting horizontally across the snout region of some conocardioids (pl. 43, figs. 12, 13). beak. Projecting juvenile part of shell (pl. 6, fig. 7). bivalved shell. Shell having two obviously expanded lateral commissural (marginal) denticle. Visible part of submerged comarginal. Feature on exterior surface of shell parallel to growing margin. commissure. Growing edge of shell. lobes (valves), not necessarily distinguished in early on- rib immediately inside commissure (pl. 34, figs. 9, 13). togeny. I dissoconch. Postlarval shell. Bug gig ‘1 g , m N m w l ‘ ‘ s \x 4% A \\ .\ _. LATE ORDOVICIAN is § § § § E \ « 3 MIDDLE ORDOVICIAN 8 § § E §\ E: x M s . \ >-'z ARENIGIAN §\:§ § §§\§\/ s§§ ééz—i . m\:§ .\ es E\ RE 50; E \ N E \ “G 2\ "i \ :5 § 9 TREMADOCIAN § § § § g\ E: 8\ E \ m § "‘ § § § 3 \ w ‘ \ . TREMPEALEAUAN 5 E 8 \ \ E \\ § 2 ~ 2 s \ \ 8 s s E: 3 \ \ s :3. 5 IDAMEAN N § § 1% g fig MINDYALLAN : 3% a g S MIDDLE CAMBRIAN \ §\ § E i .3 § § . z LENIAN Ed \ \ 8’ S S \ \ § 8 g BOTOMIAN E ‘1: § § E E E 2 D In \ \ s a 5 E] g 6 § ;\\ Q YES 5 ATDABANIAN 5 _ E E a \/ E ’1 s a a g é “- TOMMOTIAN g 3 \ §\ §-/ VENPSIEKEJ’t-xEMDBl/llglAhlliAN E V f/ .9 Bunyem'chnus 7 FIGURE 14.—Historica1 record of the initial radiation of the Mollusca, scaled against time divisions based primarily on the succession of fossil archaeocyaths and trilobites. The two largest molluscan subphyla (Cyrtosoma, fine-shaded columns, and Diasoma, coarse-shaded columns) separated in the Early Cambrian. Modified from Runnegar and Pojeta, 1974, fig- ure 1. GLOSSARY OF MORPHOLOGICAL TERMS 47 divaricate. Exterior ornament which is neither simply radial nor simply comarginal (pl. 11, fig. 16; pl. 14, fig. 2). dorsal. Fused junction of midsagittal plane passing between valves. height. Distance between two planes parallel to hinge axis and perpendicular to plane of symmetry, which just touch most dorsal and ventral parts of shell. hinge. Dorsal margin of she-ll which rotates during growth. hinge axis. Imaginary line about which the valves rotate during growth. hood. Curved lamellose plates. connected to carinae in Cono- cardiidae; growing edges form tubular extension of ven- tral orifice (:collar, schleppe, eventail, fringe, Kragen) (pl. 45, fig. 14; pl. 47, fig. 1; pl. 48, fig. 2). inflation. Distance between two planes parallel to mid- sagittal plane, which just touch the lateral edges of the shell. insertion, insertion area. Place where a muscle is attached to the shell (:muscle scar). length. Distance between two planes perpendicular to hinge axis and just touching anterior and posterior extremities of shell. longitudinal clefts. Rostral clefts that are subparallel to the rostrum (pl. 40, fig. 5; pl. 43, fig. 10). median muscles. Single anterior and posterior pedal re- tractor muscles inserted across the dorsal midline of the shells of ribeirioids (pl. 6, figs. 4, 14). muscle impression. Mold of muscle bundle on interior of shell. muscle track. Depression of inner surface of shell caused by thinning of shell layers over underlying myostracum; shows direction of movement of muscle insertion during growth. myostracum. Shell layer formed at muscle-insertion area. opisthodetic. Wholly behind the protoconch. pallial junction. Junction of anterior and posterior branches of pallial line (pl. 53, fig. 1). pallial line. Linear, continuous or discontinuous insertion area of radial muscles of mantle (pl. 22, figs. 1, 2, 3, 4, 13). pallial muscles. Radial muscles of the mantle attached to the shell. pallial protractor muscles. Radial muscles of the mantle that serve to protract the mantle edge (pl. 47, fig. 12). pallial retractor muscles. Radial muscles of the mantle that serve to retract the mantle edge (pl. 20, figs. 10, 11). pallial sinus. Embayment of pallial line due to retreat of pallial muscle insertions away from commissure (pl. 22, figs. 1, 2, 3, 4, 13). pedal muscles. Muscles of the foot (pl. 5, fig. 4; pl. 6, fig. 14; pl. 22, figs. 5, 6). pegma. Plate connecting right and left valves in umbonal part of shell (pl. 4, figs. 21, 22; pl. 5, figs. 2, 4); supports large muscle in ribeirioids. posterior. End of shell opposite that having anterior gape and (or) pegma. posterior branch. Posterior part of pallial line extending dorsally from the pallial junction along the anterior slope (pl. 53, fig. 1). posterior clefts. Tension fractures formed during growth on either side of the shell behind the beak (pl. 40, fig. 5). posterior gape. Relatively large opening at posterior end of shell (pl. 6, fig. 6). posterior umbonal cavity. Part of umbonal cavity of ribeiri- oids behind pegma (pl. 6, fig. 15). primary pedal retractor muscles. Relatively large bilateral- ly paired pedal muscles inserted on the body of the shell of advanced rostroconchs (pl. 22, figs. 1, 2, 5, 6). prosocline. Shells having demarcation line inclined pos- teriorly. prosodetic. Anterior to protoconch. protoconch. Larval shell (pl. 41; pl. 47, figs. 13—15). rostral area. Area surrounding rostrum, bordered by hood, carina, or prominent rib. rostral clefts. Elongate tension factures bordering or cross— ing the dorsal part of the rostrum (pl. 40, fig. 5). rostral orifice. Hole in commissure at end of rostrum (pl. 43, fig. 11). rostral structure. Curved, hoodlike structure generated at rostral orifice of some conocardioids (see hood). rostrum. Tubular extension of posterodorsal part of shell (pl. 39, fig. 3; pl. 43, fig. 5). secondary pedal retractor muscles. Relatively small bilat- erally paired pedal muscles inserted on the body of the shell of advanced rostroconchs (pl. 22, figs. 5, 6). shell muscles. Muscles inserted on the shell that are used to control the foot and support the visceral mass. side muscles. Lateral pedal and (or) visceral muscles of ribeirioids; insertions form left and right linear connec- tions between anterior and posterior median muscle in- sertions (pl. 6, fig. 8; pl. 7, fig. 1; pl. 8, fig. 14; pl. 12, figs. 13, 17). snout. Enlarged anterior part of shell, separated from body by sulcus and differences in sculpture (pl. 38, fig. 2; pl. 43, fig. 7). - submerged ribs. Ribs generated by commissural denticles and cove-red by growth of inner shell layers (pl. 43, fig. 1; pl. 45, fig. 4; pl. 50, fig. 2). transverse clefts. Clefts that cross the rostrum (pl. 40, fig. 5). umbo. Dorsal projection of valve above protoconch (pl. 42, fig. 1). ventral. Part of shell opposite fused dorsal margin. ventral orifice(s). Small aperture(s) in commissure between rostrum and anterior gape (pl. 24, fig. 15). SYSTEMATIC PALEONTOLOGY SYNOPTIC CLASSIFICATION OF KNOWN ROSTROCONCH MOLLUSKS Phylum MOLLUSCA Subphylum DIASOMA Class ROSTROCONCHIA Order RIBEIRIOIDA Family RIBEIRIIDAE Ribeim‘a H emultipe gma Pinnocaris Ribeirina Wanwomia Watsonella Family TECHNOPHORIDAE Technophorus Anisotechnophorus Myocm‘is Oem'kila Tolmachovia 48 Order ISCHYRINIOIDA Family ISCHYRINIIDAE Ischgrim'a Eoischyriria Pseudotechnophorus Order CON OCARDIOIDA Superfamily EOPTERIACEA Family EOPTERIIDAE Eopteria Euchasma Wanwariella Wanwanoidea Superfamily CONOCARDIACEA Family CONOCARDIIDAE Conocardium Arceodomus Family BRANSONIIDAE Bransonia Malzceodems Pseudoconocardium Family HIPPOCARDIIDAE Hippocardia Bigalea Rostroconchia incertae sedis Euchasmella Myona Pseudoeuchasma Phylum MOLLUSCA Cuvier, 1797 Subphylum DlASOMA Runnegar and Pojeta, 1974 Class ROSTROCONCHIA Poiela, Runnegar, Morris, and Newell, 197Z Diagnosis.———Mollusks with an uncoiled and un- torted univalved larval shell which straddles the dorsal midline, and a bivalved adult shell with one or more shell layers continuous across the dorsal margin so that a dorsal commissure is lacking. Stratigraphic distribution.—Lower Cambrian (Georgien)—Upper Permian (Makarewan). We agree with Morris (1967) that the Triassic species placed in Conocardium by Healy (1908) are pelecy- pods probably belonging to the Poromyacea or the Burmesiidae. Order RlBElRlOlDA Kobayashi, 1933 Diagnosis.—Rostroconchs with all shell layers continuous across the dorsal margin, an anterior pegma, and a dominant posterior growth component; musculature consists of anterior and posterior me- dian pedal retractor muscles connected by right and left side muscles. Stratigraphic distribution—Lower Cambrian (Georgien)—Upper Ordovician (Ashgillian). Discussion—This order contains two families and 11 genera and includes the stratigraphically oldest, l PALEONTOLOGY OF ROSTROCONCH MOLLUSKS morphologically simplest, and phylogenetically most primitive rostroconchs. All the forms included here have previously been classified as bivalved (concho- stracan) arthropods (Ulrich and Bassler, 1931; K0- bayashi, 1933; Salter, 1864; Etheridge, 1878). Mol- luscan nature of these forms is indicated by the presence of a protoconch, comarginal growth incre- ments growth increments on the muscle scars, and a pallial line in some forms. Family RlBElRllDAE Kobayashi, 1933 Diagnosis.—Ribeirioids with anterior and posteri- or shell gapes and lacking radial ornament. Stratigraphic distribution—Lower Cambrian (Georgien)—Upper Ordovician (Ashgillian) . The stratigraphic range of each species is shown in table 1. Discussion—Our concept of this family differs significantly from that of Kobayashi (1933) in that he included all nonconocardiacean rostroconchs in the Ribeiriidae. In 1936, he maintained the same con- cept of the family but used the name Eopteriidae Miller (1889). As used herein, this family contains six known genera and approximately 23 known spe- cies. It is presently known from all continents except Antarctica, and is known from a greater number of geographic localities in North America than on the other continents. Genus RIBEIRIA Sharpe, 1853 Plates 4—9, 30, 31 Ribeiria Sharpe, in Ribeiro, Geol. Soc. London Quart. Jour., v. 9, p. 157. Ribeiria Sharpe [partim], Billings, Palaeozoic Fossils, v. 1, p. 339. Ribeiria Sharpe, Tromelin, Soc. Linnean Normandie Bull, ser. 3, v. 1, p. 35. Ribei'rm Sharpe [partim], Whitfield, Am. Mus. Nat. History Bull., v. 1, p. 343. Ribeiria Sharpe [partim], Miller, Geology and Paleontology, p. 566. Ribeiria Sharpe [partim], Etheridge, Woodward, and Jones. British Assoc. Adv. Sci., Rept. 59th Mtg. 1889, p. 66. Ribeiria Sharpe [partim], Cleland, Bulls. Am. Pale- ontology, v. 3, no. 13, p. 20. Ribeiria Sharpe [partim], Schubert and Waagen, K. K. Geol. Reichsanstalt Jahrb., v. 53, p. 41. [Non] Ribeiria Clarke, New York State Mus. Mem. 6, pt. 2, p. 406. Ozomia Walcott, Smithsonian Misc. Colln., v. 67, no. 9, p. 531. Ribeiria Sharpe [partim], Kobayashi, Tokyo Imp. Univ., Fac. Sci. Jour., sec. 2, v. 3, pt. 7 p. 289. Ribeiria Sharpe, Thoral, Contr. étude paléont. Ordovi- cian inferieur "‘** Montagne Noire, p. 200. Ribeiria Sharpe, Yang, Acad. Sinica, Inst. Paleontology, p. 320. 1853. 1865. 1877. 1886. 1889. North American 1890. 1900. 1904. 1904. 1924. 1933. 1935. 1957. SYSTEMATIC PALEONTOLOGY 49 fl: 3 we 3 .2 g 3 s S '§ S .3 w 33 g SPECIES as)? E «E E 6 'g s E: g“ TIMESCALE 3::sggguN:§:N§u§g§§: §§~§§§§§s§33§ssais§sgafi§§s3 mpkkkké‘mfidfifiimflifififiEnifidmodnzkmm' 7 7 2 Z / < .7 § 7 > / 8 Cincinnatian / u: (Caradocian-Ashgillian) / O *— < / ? ? 5 Wildernessian—Shermanian ? 7 2 (Caradocian) / / > o / n: O / / 5 A Z O Marmorian-Porterfieldian 7 Q (Llanvirnian—Caradocian) ? I E // 7 z I / ZZZ S / ? 9 I 7 / 8 (TSanadian-Whiterockian /é / a: (remadocian-Arenigian) o I 1/? z l I .v / / ’ / 5 /% %7%%77? /,// /,// // A” A h ’% z Z A E n: g 5 Mindyallan-Trempealeauan Lu‘ ’3: .1 7 A MIDDLE CAMBRIAN EARLY CAMBRIAN /A 7 7 TABLE 1.—Range chart showing the known stratigraphic distribution of all species of Ribeiriidae recognized herein [Species roughly arranged according to stratigraphic order, from oldest to youngest. for each genus] 50 1965. ['2] Technophorus Branisa, Bolivia Serv. Geol. B01. 6, p. 76. Type species.—Ribeiria pholadiformis Sharpe, 1853 (p. 158) by monotypy. Diagnosis.—Posteriorly elongated ribeiriids in which the dorsal and ventral margins are not sub- parallel and which lack rugose comarginal ornament. Stratigraphic distribution.—Upper Cambrian (Mindyallan) —Upper Ordovician (Ashgillian) . Discussion.——It is not our intention to describe each of the species that we regard as belonging to Ribeiria. We have not been able to obtain specimens of certain species, and for knowledge of these spe cies we are totally dependent upon an aging litera- ture. The known material of some species is not well preserved, and detailed description would provide little more information than the figures provided on our plates. In the following section, we list all the species known to us, figure and diagnose those for which we have been able to obtain adequate material, make various comments, and provide descriptions only for those species that are newly named herein. Ribeiria pholadifonnis Sharpe, 1853 Plate 7, figures 3—16 Diagnosis.—-Ribeiria with shell thickening above posterior median pedal retractor muscle, producing a prominent notch in the posterior dorsal margin of internal molds. Also in internal molds, apical part of shell usually projecting well above rest of dorsum. Types and materials—This is the type species of the genus. It is based upon a syntypic suite of which we choose as the lectotype the specimen figured by Sharpe, 1853, on his plate 9, figures 17 b-c (GB 7798) , which is figured herein on plate 7, figures 3—7. The paralectotypes are figured herein on plate 7, figures 8—10 (BM PL 4176a, b; BM PL 4177). We had three other specimens of this species (pl. 7, figs. 11—16) in addition to the type material. Stratigraphic distribution.—On the basis of the geographic information given by Sharpe (1853, p. 158) and the associated mollusks, many of which were redescribed by Babin (1966) , the type material of R. pholadiformis is probably Llandeilian (Middle Ordovician) in age from Portugal. Two specimens of the species from Normandy, France (pl. 7, figs. 11, 12) , are at the Sedgwick Museum, University of Cambridge, England, and are listed as middle Are- nigian (Early Ordovician) in age. PALEONTOLOGY OF ROSTROCONCH MOLLUSKS Ribeiria apusoides Schubert and Waagen in Perner, 1903 Plate 5, figures 1—14; plate 6, figures 1—12, 14, 15; plate 7, figures 1, 2; plate 30, figures 1—5; plate 31, figures 1—5 Diagnosis—Large Ribeiria with gently concave dorsal margin lacking a prominent posterior dorsal notch. Types and materials.—R. apusoides is based upon a syntypic suite, of which we choose as the lectotype the specimen figured by Schubert and Waagen, in Perner, 1903, plate 49, figures 18—20 (Schubert and Waagen, 1904, pl. 1, fig. 9). A plastotype (USNM 209402) of this specimen is shown herein on plate 6, figures 1—4. Plastotypes of paralectotypes of the spe- cies are shown herein on plate 5, figures 9—11, 13, 14. In addition to the plastotypes, we had 16 other speci- mens of R. apusoides to examine. Stratigraphic distribution—According to Ki‘iz (oral commun., July 1973), the type material of this species is from the Caradocian (Middle-Late Ordo- vician) of Bohemia, Czechoslovakia. Also according to Kriz, the specimen shown herein on plate 6, figure 5, may be Llanvirnian (Middle Ordovician) in age; however, this is uncertain. So far as known, the species is limited to the Middle and Upper Ordo- vician rocks of Bohemia. The specimen assigned to R. apusoides by Termier and Termier (1950, pl. 184, fig. 45) from the Llanvirnian of Morocco does not belong to this species as it has a gently convex dor- sal margin. Ribeiria australiensia n. 1]). Plate 4, figures 26—29 Description—Small Ribeiria with straight to gently concave dorsal margin, posteriorly attenu- ated; shell gaping posteriorly, ventrally, and an- teriorly, with anterior gape extending dorsally to protoconch; protoconch terminal. The only internal feature presently known is a small pegma. Types.—The holotype (BMR CPC 14670) is shown on plate 4, figures 27, 29; it is 9.4 mm long and 5.9 mm high. One paratype (BMR CPC 14671) is shown on plate 4, figures 26, 28; it is 9.6 mm long and 5.7 mm high. Type locality—All specimens of this species are presently known only from one locality, Australian Bureau of Mineral Resources locality G 128 (Opik, 1967) ; western Queensland, Australia, lat. 22°, 17 ’ 8., long 139°, 01’ E., Glenormiston 1:250,000 Geo- logical Series Sheet SF 54—9 (Casey and others, 1965). The locality is indicated on the Glenormiston sheet. The specimens are from the Mungerebar Limestone. SYSTEMATIC PALEONTOLOGY 51 Stratigraphic distribution.—-According to Opik (1967) the age of G—128 is Mindyallan (early Late Cambrian). R. australiensis n. sp. is only one of six known species of Cambrian rostroconchs and is the oldest known species of the genus Ribeiria. Etymology.——The species name is derived from Australia. Ribeiria bussleri Kolnyaslli, 1933 Discussion—This species is known from only one specimen which was figured by Kobayashi (1933, pl. 4, figs. 4a, b). Although we have seen a replica of this specimen, it is not well executed, and we can do little more than to list the species here. Stratigraphic distribution.—-Kobayashi (1933, p. 292) listed the species as coming from the “Wan- wankou dolomite; Wan-wan-kou in the Niuhsintai Basin, South Manchuria.” On p. 259, he indicated that the Wanwankou Dolomite is assigned to the Wanwanian Stage (Early Ordovician). Ribeiria bussacensis Tromelin, 1877 Discussion—This species name is listed by Tro- melin (1877, p. 35) and Etheridge, Woodward, and Jones (1890, p. 67) as a synonym of R. pholadifor- mis Sharpe and is credited to Sharpe. The Sharpe reference given by both papers that cite R. bassa- cehsis is exactly the same as that for R. pholadifor- mis down to plate and figure number. Apparently R. bussacensis was intended as a substitute name for R. pholadiformis. To our knowledge, Sharpe never used R. bussacensis in print, and the reason for the possible substitute name is not given by either Tro- melin or Etheridge, Woodward, and Jones. Ribeiria calcifera Billings, 1865 Plate 4, figures 1—24 Diagnosis.——Ribeiria with convex dorsal margin and anterior clefts. Types and materials.—This species is also based on a syntypic series, of which we choose the speci- men shown here on plate 4, figures 4—6 as the lecto- type (GSC 469). Paralectotypes are shown on plate 4, figures 1—3, 7-12 (GSC 469a, b, d). In addition to the type suite, we figure four other specimens of the species (pl. 4, figs. 13-24) . Stratigraphic distribution—R. calcifera is known from several localities in the Beekmantown Group (Lower Ordovician) of Ontario, Canada. Recent work by Yochelson and Copeland (1974) indicates that the beds containing this species are latest Ca- nadian (late Early Ordovician) in age in Ontario. Two specimens from the Tanyard Formation of Texas (lower Lower Ordovician) are herein also assigned to R. calcifera (USNM 127908, 127909); these specimens were previously figured by Cloud and Barnes (1948). The Texas specimens show that the pegma of R. calcifera is almost horizontal. Ribeiria complanata Salter, 1866 Plate 9, figure 10 Discussion—This species is known only from the holotype (GB 12434) , which is not markedly attenu- ated posteriorly and which has a nearly straight dorsal margin. This specimen is from the lower Llandeilian (Middle Ordovician) of North Wales. Ribeiria compressa Whitfield, 1886 Plate 8, figures 1—5 Diagnosis.—Narrow1y convex Ribeiria with mar- kedly straight dorsal margin. Types and materials—The type material of R. compressa consists only of the holotype (pl. 8, figures 1, 2; AM 491). We place in synonymy with R. com- pressa, the name R. naculitiformis equilatera Cle- land. We choose as the lectotype of R. nuculitiformis equilatera the specimen figured by Cleland (1900, pl. 16, fig. 15), which is figured herein on plate 8, figures 3—5 (PR1 5081) . Stratigraphic distribution—The museum label lists the holotype of R. compressa as coming from the “Ft. Cassin bed, Fort Cassin, Vermont” (upper Canadian, upper Lower Ordovician). Cleland (1900, p. 22) listed R. nucalitiformis equilatera as coming from the “Calciferous [Beekmantown] at Fort Hunt- er, N.Y.” (Lower Ordovician). We have found R. compressa to be abundant in the “Fonda Limestone Member” of the Tribes Hill Formation (lower Ca- nadian; lower Lower Ordovician) of New York State. Ribeiria conformis Tromelin, 1877 Discussion—This name was first proposed as a nomen nudum by Salter, in Bigsby (1868, p. 141). The species has never been figured, and Tromelin (1877, p. 35) is the only author to have commented upon the form. He felt that it was probably synony- mous with R. pholadiformis Sharpe (R. bassacen- sis). R. conformis is from the Budleigh-Salterton Pebble Bed, a Triassic unit which contains Cara- docian (Middle-Upper Ordovician) fossils in the pebbles. Ribeiria crassa (Thoral), 1935 Discussion—We have seen no specimens of this species, and to our knowledge it is known only from the material figured by Thoral (1935, pl. 10, figs. 93, b). Thoral placed the species in the genus Ribeir— 52 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS ella Schubert and Waagen, which we regard as synonymous with Technophoras Miller. We do not regard Ribeiria crassa as belonging to Techhophoras because it lacks radial ornament. On his plate 10, Thoral listed the species as occurring in the “Trema- doc supérieur-Arenig inférieur” (Lower Ordo- vician) , St. Chinian, France. On his p. 209, figure 14, he gave the stratigraphic occurrence as “Tremadoc supérieur (?)”; also on p. 209, under Horizon and Locality he listed R. erassa as occurring in the “Arenig inférieur.” Ribeirin inflah Schubert and Waagen, 1904 Discussion—This species is much like R. apus- aides; we have seen no specimens of R. inflata. Schu- bert and Waagen (1904) felt that there might be intermediates between the two forms. It is probably best to consider R. inflata as a synonym of R. apas— aides; both forms are known only from the Cara- docian (Middle-Upper Ordovician) of Bohemia, Czechoslovakia. Ribeiria lucau (Waleott), 1924 Plate 8, figures 14—24 Diagnosis.—Ribeiria with pegma at high angle to dorsal margin and with side muscles bundled into discontinuous attachment areas. Types and materials.——This species is known only from the type suite, of which we choose the specimen herein figured on plate 8, figure 14, as the lectotype (USNM 209397). Paralectotypes are shown on plate 8, figures 15—24. Stratigraphic distribution—R. lawn is known only from the Mons Formation (Lower Ordovician) of Alberta, Canada. Discassion.—Walcott (1924) made this form the type species of his genus Ozomia, a name which may prove useful in the future. At present, however, Ribeiria lawn is well within the range of variation shown by species that we place in the genus Ribeim'a. Rilaeirin magnifica Tromelin, 1877 Discussion—This name was first proposed as a nomen nudum by Salter, in Bigsby (1868, p. 141). The species has never been figured, and Tromelin (1877, p. 36) is the only author to have commented on the form. R. magnifica is from the Budleigh— Salterton Pebble Bed, 3. Triassic unit which contains Caradocian (Middle-Upper Ordovician) fossils in the pebbles. Rlbelria mancllurlca Kobayashi, 1933 Plate 8, figures 6—11 Diagnosis.-—Tumid Ribeiria with small rounded pegma. Types and materials.———This species is known only from the material described by Kobayashi (1933, p. 291). We were able to obtain replicas of the holotype (pl. 8, figs. 6, 7; USNM 209394) and of a previously unfigured paratype (pl. 8, figs. 8—11; USNM 209395). In addition, we saw two poorly preserved specimens of the species which are not figured herein. Stratigraphic distribution—Kobayashi (1933, p. 291) listed the species as coming from the “Wan- wankou dolomite; Wan-wan-kou in the Niuhsintai Basin, South Manchuria.” He (1933, p. 259) indi— cated that the Wanwankou Dolomite is assigned to the Wanwanian Stage (Lower Ordovician). Ribeiria manchurica pennata Kobayashi, 1933 Discussion—We have seen no specimens of this form, and the only known figures are those of Koba- yashi (1933, pl. 9, figs. 4a, b). Kobayashi (1933, p. 292) noted that: “If complete specimens [of R. mancharica permata] be procured, it may not be possible to separate this specifically from the typical form [R. mahcharica].” Probably this form should be regarded as a synonym of R. mancharica. The stratigraphic occurrence and locality are the same as for R. manchurica. Ribeirin parva Collie, 1903 Plate 9, figures 7—9 Diagnosis.——Ribeiria with subcircular lateral out- line. Types and materials—R. parva is known only from the holotype (YU 7933) , which is figured here- in on plate 9, figures 7—9. Stratigraphic distribution—This species is known only from the Beekmantown Formation (Lower Or- dovician) , Bellefonte, Pennsylvania. Ribeiria personal: Thoral, 1935 Discussion—We have seen no specimens of this species. Thoral (1935, p. 201) gave the name of the species as R. personata; on the explanation for his plate 10, figures 6a, b, he gave the name as “R. per- sonata forme typica.” Thoral (p. 202) gave the oc- currence of the form as “Tremadoc supérieur ou de la base de l’Arenig des environs de Saint Chinian,” France. Ribeiria personnh lat: Thoral, 1935 Discussion—We have seen no specimens of this form. Thoral (1935, pl. 10, fig. 7) gave the occur- rence as: “Arenig inférieur, St. Chinian” (Lower Ordovician), France. SYSTEMATIC PALEONTOLOGY 53 Ribeiria personnta obsoleta Thoral, 1935 Discussion—We have seen no specimens of this form. Thoral (1935, p. 205, fig. 12) gave the strati- graphic occurrence as: “Trémadoc supérieur-Arenig inférieur” (Lower Ordovician), France. Ribeirin soleaeformis Thoral, 1935 Discussion.—We have seen no specimens of this form. Thoral (1935, pl. 10, fig 8) gave the occur- rence as: “Tremadoc supérieur-Arenig inférieur, St. Chinian” (Lower Ordovician), France. Comments about the species of Ribeiria proposed by Thoral (1935).—R. soleaeformis is much like R. pholadiformis Sharpe (1953) in shell shape, in hav- ing a projecting apical area in internal molds, and in having a prominent notch in the posterior dorsal margin of internal molds. R. soleaeformis is proba- bly a synonym of R. pholadiformis. The other spe- cies proposed by Thoral appear to be based on in- complete specimens which preserve only the anterior two-thirds of the valves. They are much like R. pho- ladiformis in shape and have a projecting apical area; it seems likely that they are synonyms of R. pholadiformis. Because we have not seen Thoral’s original material, it is difficult to synonymize his names with R. pholadiformis, however, on the basis of his figures, this synonymization seems likely. Forms similar to those figured by Thoral and to R. pholadiformis were figured by Gigout (1951, pl. 2, figs. 15-16) and Termier and Termier (1950, pl. 184, figs. 32—35) from the Llanvirnian and Llandeilian (Middle Ordovician) of Morocco. Ribeirin taylori n. sp. Plate 8, figures 12, 13 Description—Small Ribeiria with colinear parts of the hinge anterior and posterior to the beak. Beak projecting but little above the dorsal margin, not recumbent. Type—The holotype (USNM 209396) is shown on plate 8, figures 12, 13. It is an incomplete speci- men with the posterior end missing. We have pol- ished the posterior face in an effort to examine shell microstructure. The microstructure was not visible although shell layers were recognizable. Type locality—The holotype is from USGS 10c. 470B (old series), Hall Farm, 1 mile northeast of Whitehall, N.Y. (Taylor and Halley, 1974, p. 32). Stratigraphic distribution—The species is from the Whitehall Formation, Saukia Zone, probably Sau- kiella serotina (Trempaleauan, Upper Cambrian), of New York. According to Taylor and Halley (1974), the Cambrian-Ordovician boundary occurs in the Whitehall Formation. Etymology.—The species name is proposed for M. E. Taylor, U.S. Geological Survey, who brought the holotype to our attention. Ribeiria turgida Cleland, 1903 Plate 9, figure 1 Discussion—This form is known only from a specimen in the collections of the U.S. National Mu- seum (USNM 84630; pl. 9, fig. 1), which is labeled holotype. The specimen does not resemble either of the figures given by Cleland (1903, pl. 3 figs. 6, 7). We assume that the specimen is correctly labeled; however, it is incomplete, and it is thus difficult to determine the concept indicated by this name. The museum label gives the occurrence of the holotype as: “Tribes Hill ls., Canajoharie ?, New York” (Low— er Ordovician). Ribeiria spp. Plate 4, figure 25; plate 6, figure 13; plate 7, figures 17, 18; plate 9, figures 2—6 Discussion—The forms figured and discussed un- der this heading are based upon incomplete or poorly preserved specimens of Ribe’irz'a. They do add data to the stratigraphic and geographic distribution of the genus. The specimen shown on plate 4, figure 25, is from the Lower Ordovician part (Warendian) of the Ninmaroo Formation at northern peak of Digby Peaks, 60 miles north of Boulia, Queensland, Aus- tralia. The specimen shown on plate 6, figure 13, is a paralectotype of R. apusoz'des Schubert and Waagen from the Ashgillian (Upper Ordovician) rocks of Bohemia, Czechoslovakia; it differs from other speci— mens assigned to R. apusoides in its straight dorsal margin and in its length-height ratio. It is also the youngest known specimen of the genus Ribeiria. The specimens shown on plate 7, figures 17, 18, are from the Lower Ordovician of Utah; they are both incomplete posteriorly. The specimens shown on plate 9, figures 26, are from the Stonehenge Forma- tion (Lower Ordovician) of Maryland. Genus HERAULTIPEGMA new genus Plate 2 1920. [Non]Herault1‘a Villeneuve, Soc. Entomol. Belgique, Annales v. 60, p. 119. 1935. Heraultia Cobbold, Annales and Mag. Nat. History, ser. 10, v. 16, p. 37. 1974. Heraultia Cobbold, Runnegar and Pojeta, Science, v. 186, p. 315. 1974. Heraultia Cobbold, Pojeta and Runnegar, Am. Scientist, V. 62, p. 711. 1975. Heraultia Cobbold, Pojeta, Bulls. Am. Paleontology, v. 67, p. 375. 54 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS Type species.—-Heraaltia varensalensis Cobbold, 1935 (p. 38), is designated the type species of the genus H eraaltipegma. Diagnosis.——-Posteriorly elongate ribeiriids with a small pegma, wide anterior, posterior, and ventral shell gapes, and prominent rugose comarginal orna- ment. Stratigraphic distribution.—Upper Lower Cam- brian (Georgien). Geographic distribution—The genus is unequi- vocally known only from the St. Genies de Varensal area in the Hérault district of France. One speci- men, which was placed in Fordilla troyensis by Shaler and Foerste (1888) , from the Lower Cambri- an of Massachusetts has the shape of H eraulti- pegma; it may belong to that genus or to Watsonella, but nothing is known of its ornament or shell gapes. Discussion—The new generic name H eraultipeg- ma is proposed for those rostroconchs placed in Heraultia by Cobbold (1935). The name Heraultia Cobbold is a junior homonym of Heraaltia Villene- uve (1920) which was used for a dipteran insect. Etymology.———He’rault, a region of France; pegma, Greek, meaning fastened or fixed; also a structure in rostroconchs. Gender neuter. Heraullipegma varensalense (Cobbold), 1935 Plate 2, figures 1—13 Diagnosis—Small H eraaltipegma with about equally developed anterior, ventral, and posterior shell gapes. Types and materials—We have not seen Cobbold’s (1935) types of this species. However, S. C. Mat- thews of the University of Bristol, England, has gen- erously given us topotypes of the species; these topo- types are figured herein (USNM 209414—209417). Stratigraphic distribution—All specimens of H. varensalense are presently known only from the Georgien (Lower Cambrian) rocks near St. Geniés de Varensal, France. Genus PINNOCARIS Etheridge, 1878 Plates 9, 10 1878. Pinnocaris Etheridge, Royal Soc. Edinburgh Proc., v. 4, p. 167. 1880. Pirmocaris Etheridge, Nicholson and Etheridge, Mon. Silurian Fossils Girvan District in Ayrshire, v. 1, p. 207. 1892. Pinnocaris Etheridge, Jones, and Woodward, Mon. British Paleozoic Phyllopoda, pt. 2, p. 117. 1895. Pinnocaris Etheridge, Jones, and Woodward, Geol. Mag., Decade 4, V. 2, p. 542. 1907. Pinnocaris Etheridge, Reed, Geol. Mag., Decade 5, v. 4. p. 110. Type species.——Pirmocaris lapworthi Etheridge, 1878 (p. 169), by monotypy. Diagnosis.—Posteriorly elongated compressed ribeiriids with anterior clefts and with posterior end drawn out into a rostrum. Stratigraphic distribution—Lower Caradocian (Middle Ordovician)—Ashgillian (Upper Ordovici- an). Pinnocaris lapworthi Etheridge, 1878 Plate 9, figures 13—23 Diagnosis.—Pinnocaris with a nearly straight dorsal margin, a long straight rostrum, and rugose comarginal ornament. Types and materials—This is the type species of the genus. It is based upon a syntypic series, of which we choose as the lectotype the specimen fig- ured by Etheridge, 1878, on his plate 2, figure 5 (EM In 20367) ; this specimen is figured herein on plate 9, figure 21. The paralectotypes are figured herein on plate 9, figures 20, 22 (BM In 20366, In 20368). In addition, we figure four other specimens of the species. Stratigraphic distribution—This species is pres- ently known only from Scotland. It is from the Bal- clatchie Group which Whittington (1972) placed in the lower Caradocian (Middle Ordovician). Pinnocaris americana n. 31). Plate 9, figures 11, 12 Description.——Pinnocaris lacking rugose orna- ment, with a gently concave dorsal margin and a short posterior rostrum. The only known internal feature is the pegma. Types.—P. americana is presently known only from two specimens. The holotype is shown on plate 9, figure 12 (USNM 209393); it is about 20 mm long and 7.5 mm high. The paralectotype (USNM 209392) is shown on plate 9, figure 11; it is about 20 mm long and 8 mm high. Both specimens show the anterior clefts. Type locality—The museum label gives the locali- ty as: “Elkader, Iowa.” Stratigraphic distribution.—-The museum label gives the horizon as: “Prosser limestone’,(Middle Ordovician) . Etymology—The species name is derived from America. Pinnocaris curvata Reed, 1907 Plate 9, figures 24, 25; plate 10, figures 1—10 Diagnosis.—Pinnocaris with a markedly concave dorsal margin. SYSTEMATIC PALEONTOLOGY Materials.—We have not seen Reed’s type materi- al; we figure seven other specimens of the species. Stratigraphic distribution—P. curvata is present— ly known only from the Drummuck Group (Upper Ordovician) of Scotland. Genus RIBEIRINA Billings, 1865 Plate 3 1865. Ribeirina Billings [partim], Palaeozoic fossils, v. 1, Geol. Survey Canada, p. 340. 1934. [Non] Ribeirina Parker, Monog. frogs Microhylidae, p. 115. Type species.—Ribeiria? longiuscala Billings, 1865 (p. 341), designated herein. Diagnosis.—Posteriorly elongated ribeiriids with nearly parallel dorsal and ventral margins. Stratigraphic distribution.—Upper Canadian (up- per Lower Ordovician). Ribeirina longiuscula (Billings), 1865 Plate 3, figures 15-23 Diagnosis.—Ribeirina with small beak and with flaring shell margins to either side of the anterior gape. Types and materials—The holotype of the species (GSC 470) is shown on plate 3, figures 21—23; it is about 32 mm long and 15 mm high. The species is known from only two specimens besides the holo- type; these are shown on plate 3, figures 15—20. Stratigraphic distribution.—Two of the known specimens (GSC 470; ROM 26 cal.) of the species are labeled “Beekmantown” (Lower Ordovician), Ontario, Canada. The third specimen (USNM 209413) is from the Oxford Formation (GSC 10c. 89453), Ontario, Canada; recent work by Yochelson and Copeland (1974) suggests that this specimen is latest Canadian (late Early Ordovician) in age. Genus WANWANIA Kobayashi, 1933 Plate 3 1933. Wanwania Kobayashi Tokyo, Imp. Univ. Fac. Jour., v. 3, pt. 7, p. 282. 1957. Wanwam'a Kobayashi, Yang, Chung-kuo piao chun hua shih, p. 320. Type species—Wanwania cambrica Kobayashi, 1933 (p. 292), by original designation. Diagrzosis.—Dorsoventrally elongated ribeiriids in which the shell is higher than long or is subquadrate. Stratigraphic distribution—Upper Cambrian (Tsinania Zone, Yingtzu Series)-LOWer Ordovician (Wanwanian) of Manchuria. Sci. Wanwania cambrica Kobayashi, 1933 Plate 3, figures 5, 11—14 Diagnosis.—Subquadrate Wanwahia in which the length and height are subequal. l 55 M aterials.—We had three plastotypes of this species to examine. The plastoholotype is figured on plate 3, figures 11—14 (USNM 209412), and a plas- toparatype is shown on plate 3, figure 5 (USNM 209409). Stratigraphic distribution—According to Koba- yashi (1933, p. 283) , this species is known only from the: “Upper Cambrian, Tsiriania zone, Paichia—shan, in the northern part of the Wuhutsui Basin and at Hsishan in the southern part of the same basin, at the neck of the Liaotung Peninsula, Manchuria.” On p. 259, he showed the Tsinania zone as occurring in the lower Yingtzu Series. Jones, Shergold, and Druce (1971) show the Yingtzu Series as being late Late Cambrian in age. Wanwania compressa Kobayashi, 1933 Plate 3, figures 6-9 Diagnosis—Small Wanwania in which the height is greater than the length, and with an anteriorly sinuate pallial line. Materials.—This species is known only from the holotype, of which We had a replica (USNM 209410). The holotype is incomplete posteriorly, and the plastoholotype clearly shows the anterior gape (pl. 3, fig. 9), the pegma (pl. 3, fig. 7), and the pallial line (pl. 3, fig. 7). Stratigraphic distribution.——Wanwanian (Lower Ordovician). According to Kobayashi (1933, p. 284) , W. compressa is from the Wanwankou Dolomite, Wan-wan-kou in the Niuhsintai Basin, south Manchuria. Discussion—W. compressa is much like W. am- bonychiformis and may be a synonym of that species. Kobayashi (1933) separated them on slight differ- ences in shape. As we had only one plaster replica of each species to examine, we cannot be positive about the synonymy; however, it seems likely that the two names belong to the same form. Wanwania ambonychiformis Kobayashi, 1933 Plate 3, figure 10 Diagnosis—Large Wanwania in which the height is greater than the length. MateriaZ.-—A1though Kobayashi (1933, p. 284) noted that this was a common form, we had only a replica of the holotype to examine (USNM 209411) . Stratigraphic distribution.—Wanwanian (Lower Ordovician). Wanwankou Dolomite, Wan-wan-kou in the Niuhsintai Basin, south Manchuria. Discussion—As noted above under W. compressa, this species is probably synonymous with that 3 species. 56 PALEONTOLOGY OF ROSTROCONCH MOLLUSKS Genus WATSONELLA Grabau, 1900 Plate 3 Watsonella Grabau, Boston Soc. Nat. History Occas. Papers, v. 4, pt. 2, p. 631. [Non] Watsonella Thiele, Deutsch. Siidpolar Exped. 1901—03, v. 13, Zool. 5, p. 237. Watsonella Grabau, Cobbold, Ann. and Mag. Nat. His- tory, ser. 10, v. 16, p. 38. Stenotheca Salter [partim], Resser, Smithsonian Misc. Colln., v. 97, no. 10, p. 24. Type species.—-Watsonella crosbyi Grabau, 1900 (p. 631), by original designation and monotypy. Diagnosis—Small concentrically marked forms, probably with small anterior and posterior shell gapes. Stratigraphic distribution.—Lower Cambrian of eastern Massachusetts. Discussion.—Cobbold (1935) noted the similarity of Watsonella to Heraultipegma. His analysis was based on Grabau’s (1900) figures of Watsonella, which are stylized; the known specimens of Wat- sonella are not as well preserved as indicated by Grabau’s figures. It may be that Watsonella and H eraultipegma are synonymous, but the known ma- terial of Watsonella suggests that it had small an- terior and posterior gapes, unlike the large gapes of H eraaltipegma. On the basis of the known material of Watsonella, the genus is similar to such helcionel- lacean monoplacophorans as Anabarella. 1900. 1912. 1935. 1938. Watsonella crosbyi Grabau, 1900 Plate 3, figures 1—4 Discussion—The known material of this species is not well preserved. It does show that W. crosbyi is a laterally compressed form with comarginal or— nament. We choose as the lectotype the specimen figured by Grabau (1900) on his plate 31, figure 9b. This specimen is herein figured on plate 3, figure 1. Stratigraphic distribution.—Grabau’s specimens of W. crosbyi are all from Lower Cambrian boulders at Sandy Cove and Pleasant Beach, Cohasset, Mass. Family TECHNOPHORIDAE Miller, 1889 Diagnosis.—Ribeirioids with radial ornament, some with both divaricate and radial ornament. Stratigraphic distribution—Upper Cambrian (Id— amean)—Upper Ordovician (Richmondian) . The stratigraphic range of each species is shown in table 2. Discussion—This family contains five known gen- era and approximately 21 known species. It is known from all continents except Africa and Antarctica; more than half the known species occur in North America. Genus TECHNOPHORUS Miller, 1889 Plates 10—14 Technophorus Miller, North American Geology and Palaeontology, p. 514. Techrwphorus Miller, Ulrich, Lower Silurian Lamelli- branchiata of Minnesota, from Minnesota Geol. and Nat. History Survey Final Rept. v. 3, p. 612 [Pub lished under separate cover prior to entire v. 3.] Technophorus Miller, Ulrich, Minnesota Geol. and Nat. History Survey Final Rept., V. 3, pt. 2, p. 612 [Re- printing of 1894 paper.] Ribeirella Schubert and Waagen, K.K. Geol. Reichsanst Jahrb., V. 53, p. 45. Ribeirella Schubert and Waagen, Kobayashi, Tokyo Imp. Univ. Fac. Sci. Jour., Sec. 2, 3, pt. 7, p. 292, 316. Technophorus Miller, Kobayashi, Tokyo Imp. Univ. Fac. Sci., Sec. 2, v. 7, pt. 2, p. 299, 316. [Non] Ribeirella Thorral, Contr. étude Paléont. Ordo- vicien inférieur *** Montagne Noire, p. 208. Technophorus Mill-er, Kobayashi, Geol. Soc. Japan, Jour. v. 43, p. 350, [‘2] 352. Ribeirella Schubert and Waagen, Kobayashi, Geol. Soc. Japan, Jour. v. 43, p. 350. 1889. 1894. 1897. 1904. 1933. 1933. 1935. 1936. 1936. 1960. Technophorus Miller, Soot-Ryan, Norsk Geol. Tidsskr., v. 40, p. 125. 1965. [Non?] Technophorus Branisa, Bolivia Serv. Geol. BO]. 6, p. 76. Type species.—Techhophorus faberi Miller, 1889 (p. 514), by original designation and monotypy. Diagnosis.—Small, equivalved, posteriorly elon- gate technophorids, with a single pegma which is at or nearly at a right angle to the dorsal margin, and with well-developed posterior radial ribs. Stratigraphic distribution—Lower Ordovician (Wolungian)—Upper Ordovician (Richmondian). Discussion—As with Ribeiria, and for the same reasons, it is not our intention to describe each of the species we regard as belonging to Technophorus. In the following section, we list all the species known to us, figure and diagnose those for which we have been able to obtain material, make various com- ments, and provide descriptions only for those spe- cies that are newly named herein. We have not been able to locate the holotypes of T. divaricatus Ulrich and T. extenuatus Ulrich and regard them as lost, nor have we been able to locate the type material of T. otoviehsis Kobayashi. Of T. coreanica (Koba- yashi) , we have seen only a replica of the holotype. We have significant new material of Technophorus from Australia, Bohemia, Indiana, Kentucky, Min- nesota, Ohio, and Siberia. Technophorus faberi Miller, 1889 Plate 10, figures 16—21; plate 11, figures 1—6 Diagnosis—Techhophoras with two posterior 57 SYSTEMATIC PALEONTOLOGY 7///// // $33 w, .5 7///////////////////////////////// EE //////////// gmfisfiggufia S /////////// swig H 23:28:39 S ? Z ? III / 2%? 7 w3w395£3m E 7.///// msgsgfizfis 27/ m3u$§§®w8® .rH migtsawfi H fll llllllllll ll mgsiwmwzwe H ////A7//// e¥®§8®6ww .5 A 7/ 338328 w§so£ac§~umrm 3.29:3 mwssooxg fibwfifigwogb .& 7. 7//// 7. swam mgaofisgfifi z «£33 Se 33:: 1 w§s§§§§wgew§< Z defifitso Efizfiuo 7/A SPECIES TIME SCALE Cincinnatian (Caradocian-Ashgillian) Wildernessian-Shermanian (Caradocian) Marmorian-Porterfieldian (Llanvirnian-Caradocian) Canadnan-Whuterocklan (Tremadocian-Arenigian) Mindyallan-Trempealeauan Z<_O_>OQmO mP/j Z<.0_>ODmO m:.00:>_ Z<_O_>ODKO >JmODmO m._.<:_ Z<_O_>OQmO m.._n_0:>_ Z<_O_>ODN_O >4m PROFESSIONAL PAPER 968 PLATE 33 GEOLOGICAL SURVEY HIPPOCARDIAFigures PLATE 34 1-16. Mulceodens jaanussoni n. sp. (p. 73). I, 2. Anterior (X 7) and right-lateral (X 9) views of paratype. Silurian (Ludlovian, Eke, Marl), Gotland, Ronehamn, Sweden. SMNH Mo. 18563. 3. Interior of left valve of paratype showing marginal denticles (X 5). Silurian (Wenlockian, Mulde Marl), Gotland, Parish of Eksta, Djupvik, Sweden. SMNH Mo. 18302. 4. Interior of left valve of paratype showing marginal denticles (X 5). Horizon and locality the same as in figure 3 above. SMNH Mo. 18303. 5. Interior of right valve of paratype showing marginal denticles (X 5). Horizon and locality the same as in figure 3 above. SMNH Mo. 18304. 6. Dorsal view of paratype showing protoconch and rostral clefts (X 7). Silurian (Wenlockian, Slite Beds), Gotland, Parish of Othem, Samsugn, Sweden. SMNH Mo. 18336. 7. Dorsal view of paratype showing protoconch and rostral clefts (X 8). Horizon and locality the same as in figure 6 above. SMNH Mo. 18337. 8. Dorsal view of paratype showing protoconch and marginal denticles (X 8). Horizon and locality the same as in figure 6 above. SMNH Mo. 18338. 9. Anterior view of paratype showing marginal denticles (X 7). Silurian, Gotland, Sweden, SMNH Mo. 18464. 10. Anterior view of paratype showing marginal denticles (X 7). Horizon and locality the same as in figure 3 above. SMNH Mo. 18560. II. Anterior view of paratype showing marginal denticles (X 7). Silurian (Ludlovian, Hamra Beds), Gotland, Parish of Oja, Storviks Kanal, Sweden. SMNH Mo. 18322. 12,13. Left-lateral and anterior views (X 7). Silurian (Ludlovian, Hamra Beds), Gotland, Parish of Grotlingbo Uddvide, Sweden. SMNH Mo. 18546. 14. Anterior view of paratype (X 7). Horizon and locality the same as in figures 12-13 above. SMNH Mo. 18547. 15,16. Dorsal and right-lateral views of paratype (X 7). Horizon and locality the same as in figure 3 above. SMNH Mo. 151244.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 34 MULCEODENSPLATE 35 Figures 1-3,11,12. 4-7. 8-10. 13-17. Mulceodens jaanussoni n. sp. (p. 73). Left-lateral, dorsal, right-lateral, anteroventral, and posterior views of holotype (X 7). Horizon and locality the same as on plate 34, figure 3. SMNH Mo. 151245. Mulceodens eboraceus (Hall), 1860 (p. 74). Right-lateral, left-lateral, dorsal, and anterior views of lectotype (X 3). Hamilton Group (Middle Devonian), York, Livingston County, N.Y. AM5347/1. Mulceodens bifarius (Winchell), 1866 (p. 73). Dorsal, anterior, and right-lateral views (X 3). Traverse Group, upper Alpena Limestone, 4 Mile dam, Alpena County, Mich. (Middle Devonian). USNM 209306. Bigalea visbyensis n. sp. (p. 77). 13-15. Left-lateral, ventral, and right lateral views of paratype (X 6). Silurian (Wenlockian, formation unknown), Visby, Gotland, Sweden. SMNH Mo. 18554. 16,17. Right-lateral and ventral views of holotype (X 7). Horizon and locality the same as in figure 13-15 above. SMNH Mo. 18552.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 35 MULCEODENS AND BIGALEAPLATE 36 Figures 1-4. Bigalea visbyensis n. sp. (p. 77). 1,2. Left-lateral and dorsal views of holotype (X 7). Horizon and locality the same as on plate 35, figures 13-15. SMNH Mo. 18552. 3,4. Left-lateral and dorsal views of paratype (X 7). Horizon and locality the same as on plate 35, figures 13-15. SMNH Mo. 18553. 5-12. Bigalea ohioensis n. sp. (p. 77). 5-7. Oblique posterior (X 4), right-lateral, and ventroposterior views (X 6) of holotype. Horizon and locality uncertain, probably from Devonian rocks exposed at the Falls of the Ohio River. USNM 209302. 8, 9. Left-lateral and dorsal views of paratype ( X 6). Horizon and locality the same as in figures 5-7 above. USNM 209303. 10,11. Right-lateral and posterior views of paratype (X 6)- Horizon and locality the same as in figures 5-7 above. USNM 209304. 12. Highly oblique posteroventral view of paratype (X 6). Horizon and locality the same as in figures 5-7 above. USNM 209305. 13-16. Bigalea yangi n. sp. (p. 76). Oblique posterior view from left side, ventral, left-lateral, and anterior views of paratype (X 3.5). Traverse Group (Middle Devonian), Kegomic, Little Traverse Bay, Mich. FM 18331.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 36 BIGALEAPLATE 37 Figures 1-4. Bigalea yangi n. sp. (p. 76). 1-3. Dorsal, posterior, and left-lateral views of holotype (X 3.5). “Petoskey Limestone” in the Traverse Group (Middle Devonian) quarry at Mud Lake, about 1.5 miles northeast of Bay View, Emmet County, Mich. USNM 209301. 4. Posterior view of paratype (X 6). Traverse Group (Middle Devonian), Kegomic, Little Traverse Bay, Mich. FM 18332. 5-15. Bigalea clathra (d’Orbigny), 1850 (p. 76). 5-9. Left-lateral (X 3.5), dorsal (X 3), ventral (X 3), posterior (X 3.5), and anterior (X 3.5) views. Devonian, Pelm, Germany?. MCZ 15395. 10-12. Left-lateral, dorsal, and right-lateral views. (X 3.5). Devonian, Priim, Germany?. MCZ 15608. 13,14. Ventral and right-lateral views (X 3). Devonian?, Eifel, Germany. UM 1788. 15. Right-lateral view (X 3.5). Horizon, locality, and museum number the same as in figures 10-12 above. 16,17. Conocardium aff. C. elongatum (Sowerby), 1815 (p. 69). Left-lateral and dorsal views (X 2.5). Pennsylvanian, St. Joseph, Mo. USNM 100704.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 37 BIGALEA AND CONOCARDIUMFigures PLATE 38 1-7. Conocardium aff. C. elongatum (Sowerby), 1815 (p. 69). 1-3. Ventral, right-lateral, and posterior views (X 2.5). Horizon, locality, and museum number the same as on plate 37, figures 16, 17. 4-7. Anterior, dorsal, ventral, and left-lateral views (X 2.5). Carterville Formation (Upper Mis-sissippian), mine dump near Duenweg, Mo. USNM 209298. 8-24. Conocardium elongatum (Sowerby), 1815 (p. 69). 8. Left-lateral view of Hind (1900, pi. 51, fig. 8) hypotype (X 2). Carboniferous (Mississippian) Limestone of Settle, England. SM E.549. 9-14. Right-lateral, left-lateral, ventral, dorsal, anterior, and posterior views of holotype (X 2). Carboniferous (Mississippian), Derbyshire, England. Photographs courtesy British Museum (Natural History). BM PL 794. 15-20. Topotype. 15-18, Ventral, dorsal, left-lateral, and posterior views (x 2), 19, 20, Same figure of polished anterior end showing a longitudinal shelf on the right side (arrow fig. 20) (X 5). Carboniferous Limestone (Mississippian), Derbyshire, England. BM L 13496. 21. Right-lateral view internal mold showing some muscle scars (X 3). Four Laws Limestone (Vi-sean, Mississippian), Redesdale, Northumbeirland, England. BM PL 4431. 22-24. Right-lateral, dorsal, and oblique left-lateral views showing some muscle scars (X 4). Horizon and locality the same as on figure 21 above. BM PL 4432. 25,26. Conocardium pseudobellum n. sp. (p. 70). 25. Posterior view of holotype (X 2.5). Traverse Group, upper Alpena Limestone (Middle Devonian), Four Mile dam, Alpena County, Mich. USNM 209299. 26. Polished section of dorsal surface of paratype showing longitudinal shelf on right side (X 3). Horizon and locality the same as in figure 25 above. USNM 209300.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 38 CONOCARDIUMPLATE 39 Figures 1-3. Conocardium pseudobellmn n. sp. (p. 70). 1,2. Dorsal and left-lateral views of holotype (X 2.5). Rostrum rebuilt in plaster. Horizon, locality, and museum number the same as on plate 38, figure 25. 3. Left-lateral view of paratype showing elongate rostrum (X 2.5). Four Mile Dam Formation (Middle Devonian), Four Mile Dam on Thunder Bay River, 2 miles upstream from Alpena, Mich. UM 47287. 4-7. Conocardium normale Hall, 1883 (p. 70). Lectotype. Right-lateral, left-lateral, and dorsal views and latex replica of posterior end showing unusually thick longitudinal shelves (X 1.5). Hamilton Group (Middle Devonian), Cumberland, Md. AM 5349/1. 8-10. Conocardium aliforme (Sowerby), 1815 (p. 70). Posterior, left-lateral, and dorsal views (X 2) of Hind (1900, pi. 54, fig. 8) hypotype. Carboniferous Limestone (Mississippian) of Settle, England. SM E 561. 11-13. Conocardium attenuatum (Conrad), 1842 (p. 70). Dorsal, right-lateral, and ventral views (X 2) of a Hall hypotype. Schoharie Formation (Lower Devonian), Schoharie, N.Y. NYSM 2321.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 39 CONOCARDIUMPLATE 40 Figures 1,2. Conocardium attenuatum, (Conrad), 1842 (p. 70). 1. Polished section of anterior end showing longitudinal shelves (arrow) (X 2.5). Schoharie Formation (Lower Devonian), 1.75 miles north-northwest of Clarksville, N.Y. USNM 100705. 2. Right-lateral view (X 2) of Hall hypotype (1885, pi. 67, fig. 9). Schoharie Formation, Schohairie, N.Y. (Lower Devonian). AM 2850/3. 3-14. Pseudoconocardium lanterna (Branson), 1965 (p. 74). 3-8. Right-lateral, ventral, dorsal, anterior, posterior, and left-lateral views (X 2). Cisco Group (Pennsylvanian), Graham, Young County, Tex. USNM 209293. 9. Right-lateral view (X 2). Brad Formation, “Hog Greek Shale Member” of Caddo Creek Formation (Pennsylvanian), 6 miles west of Chico, 1.5 miles north of highway, Wire County, Tex. USNM 209294. 10. Right-lateral view (X 4). Horizon and locality unknown. UOK 800. 11,12. Anterior and right-lateral views (X 2). Pennsylvanian, Martin’s Lake, 3 miles southwest of Bridgeport, Tex. USNM 209295. 13. Dorsal view (X 2.5). Horizon and locality unknown. USNM 209296. 14. Right-lateral view (X 2). Brownwood Shale Member of the Graford Formation (Pennsylvanian), Signal Peak, Tex. USNM 209297.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 40 CONOCARDIUM AND PSEUDOCONOCARDIUMPLATE 41 Figures 1-5. Pseudoconocardium lanterna (Branson), 1965 (p. 74). SEM photographs of protoconch. 1, Posteroright-lateral view (X 26) ; protoconch is the bump at the junction of the linear ridges and the umbonal ridges. 2, Posterodorsal view (X 26); pirotoconch is the raised area between the umbonal ridges. 3, Left-lateral view of specimen (X 4). 4, Oblique posterior view (X 60) ; protoconch is the raised area dorsal to the projecting rostrum. 5, Dorsal view (X 28); protoconch is the dark area in the posterocentral part of the shell. Gaptank Formation (Pennsylvanian), 2 miles S. 17° E. of Gaptank, 23.5 miles northeast of Marathon, Tex. USNM 209292.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 41 PSEUDOCONOCARDIUMFigures PLATE 42 -7,12-14. Pseudoconocardium lanterna (Branson), 1965 (p. 74). 1,2. Posterior and ventral views (X 2). Graford Formation (Pennsylvanian), Martin’s Lake, 1.35 miles south of Bridgeport, Wise County, Tex. USNM 209287. 3,4. Right-lateral and dorsal views of an internal mold (X 2). Graham Formation (Pennsylvanian), 0.5 miles west of S. Bend, Young County, Tex. AM 28992. 5. Posteroventral view (X 2). Palo Pinto Limestone (Pennsylvanian), west side Martin’s Lake, 2 miles south of Bridgeport, Wise County, Tex. UOK 798. 6. Right-lateral view (X 4). Graford Formation (Pennsylvanian), west side of Martin’s Lake, 2 miles south of Bridgeport, Wise County, Tex. USNM 209288. 7. Right-lateral view (X 7). Horizon and locality the same as on plate 41, figures 1-5. USNM 209289. 12-14. Right-lateral, dorsal, and left-lateral views (X 2) of an internal mold that preserves remnants of muscle scars. Union Valley Formation (Pennsylvanian), 3 miles southeast of Ahloso, Okla. NEli, sec. 29, T. 3 N„ R. 7 E. UOK 794. 8-10. Arceodomus glabrata (Easton), 1962 (p. 71). Left-lateral, right-lateral, and dorsal views (X 2.5). Diamond Peak Formation (Mississippian), SW%, SE14, NEli, Sec. 28, T. 19 N„ R. 58 E. (USGS loc. 23837-PC), White Pine County, Nev. USNM 209290. 11. Arceodomus aff. A. glabrata (Easton), 1962 (p. 71). Ventral view (X 2). Pennsylvanian. Little Kickapoo Creek, 0.7 miles southeast of Kickapoo Falls^ Hood County, Tex. USNM 209291.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 42 PSEUDOCONOCARDIUM AND ARCEODOMUSPLATE 43 Figures 1-3,7-12. Arceodomus glabrata (Easton), 1962 (p. 71). 1-3. Right-lateral, dorsal, and anterior views of holotype (X 3). The anterior view shows the longitudinal shelves. Heath Formation (Mississippian), Stonehouse Canyon, Golden Valley County, Mont. USNM 118858. 7-12. Right-lateral, left-lateral, ventral, dorsal, posterior, and anterior views (X 2.5). Horizon and locality the same as on plate 42, figures 8-10. USNM 209286. 4—6. Arceodomus sp. (p. 71). Posterior, right-lateral, and dorsal views (X 2). Permian (Sakmarian) of the U.S.S.R. BM L 15570. 13-15. Arceodomus langenheimi (Wilson), 1970 (p. 71). Anterior (X 3), ventral, and left-lateral views ( X 1.5) of holotype. Figure 13 courtesy of E. C. Wilson. McCloud Limestone (Permian), Bollibokka Mountain, Shasta County, Calif. UCB 10589.PLATE 44 Figures 1-4. Arceodomus aff. A. glabrata (Easton), 1962 (p. 71). 1,2. External and internal views of right valve showing ornament and longitudinal shelves (X 2). “Dickerson Shale” (Pennsylvanian), just under Kickapoo Falls Limestone, 45 miles east northeast of Lipan, Hood County, Tex. USNM 209284. 3,4. Right-lateral and anterior views (X 2.5). “Dickerson Shale” (Pennsylvanian), southeast of Kickapoo Falls in creek, Hood County, Tex. U SNM 209285. 5-14. Hippocardia bella (Cooper and Cloud), 1938 (p. 75). 5. Right-lateral view of paratype (X 2). Devonian, first hollow south of Kritsville, Calhoun County, 111. USNM 95192b. 6,7. Right-lateral and posterior views of paratype (X 4). Horizon and locality the same as in figure 5 above. USNM 95192d. 8-10. Right-lateral, posterior, and dorsal views of paratype (X 2.5). Horizon and locality the same as in figure 5 above. USNM 95192c. 11-14. Right-lateral, posterior, ventral, and dorsal views of holotype (X 2.5). Horizon and locality the same as in figure 5 above. USNM 95192a. 15,16. Hippocardia monroica (Grabau), 1910 (p. 76). 15. Left-lateral view of La Rocque (1950) hypotype (X 2). Detroit River Group, Amherstburg Formation (Middle Devonian), Cummins’ quarry about 6 miles south and 1.75 miles east of Petersburg, Monroe County, Mich. UM 24524. 16. Dorsal view (X 2). Amherstburg Formation (Middle Devonian), Amherstburg, Ontario, Canada. USNM 60022.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 44 ■ , | j/Jfeyj i /l I 4 A*.^*-* ei lM| -i a ' *? ^ *;yrW „. vr'Jp^ f 11-' ®DD:- W a r ' «fc* ' .;- - v ?/ vHfe i M m Dim ,1 jt*D\ / li F jjatiM M? /S^attu- ^ W ARCEODOMUS AND HIPPOCARDIAPLATE 45 Figures 1-4. Hippocardia monroica (Grabau), 1910 (p. 76). 1,2. Right-lateral and anterior views of paratype (X 2). Horizon, locality, and museum number the same as on plate 44, figure 16. 3. Oblique dorsal view of latex replica of La Rocque (1950) hypotype (X 2). Amherstburg Formation (Middle Devonian). Livingstone Channel, Detroit River, Wayne County, Mich. UM 24529. Replica USNM 209283. 4, Left-lateral view (X 2). Lucas Formation (Middle Devonian), Patrick quarry, Grosse Isle, Mich. UM 24534. 5-9. Hippocardia cf. H. fusiformis (McCoy), 1844 (p. 75). 5,6. Interior and exterior views of left valve (X 1). Toumaisian (Mississippian), Toumai, Belgium. YU 28150. 7-9. Ventral, posterior, and right-lateral views (X 0.75). Carboniferous (Mississippian?), Tournai, Belgium. USNM 63372. 10-14. Hippocardia cooperi n. sp (p. 75). Dorsal (X 3), ventral (X 3), right-lateral (X 10), anterior (X 3), and left-lateral (X 10) views of holotype. Lower Chambersburg Limestone (Middle Ordovician), near Strasburg, Va., USNM loc. 600. USNM 162786. GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 45 'm&mmk VAV.W? HIPPOCARDIAFigures PLATE 46 1-12. Hippocardia hibernica (Sowerby), 1815 (p. 74). 1-3. Posterior, dorsal, and anterior views of polished surface showing longitudinal shelves (arrow) (X 1). Carboniferous (Mississippian), Tournai, Belgium. MCZ 433. 4. Posterior view (X 1). Carboniferous (Mississippian), Castle Cormell, County Limerick, Ireland. USNM 100712. 5-7. Posterior, right-lateral, and dorsal views (XI). Carboniferous (Mississippian), St. Doulagh’s, County Dublin, Ireland. SM E1169. 8,9. Left-lateral and posterior views (XI). Carboniferous (Mississippian), Ireland. SM E1176. 10-12. Ventral, right-lateral, and posterior views showing hood and elongate rostrum (X 1). Carboniferous (Mississippian), St. Doulagh’s, County Dublin, Ireland. SM E1185.Figures PLATE 47 1-7. Hippocardia bohemica (Barrande), 1881 (p. 75). 1,2. Posterior view showing hood on left side (X 2) and right-lateral view (X 1). Devonian, Konieprus, Bohemia, Czechoslovakia. USNM 100623. 3-7. Right-lateral, left-lateral, ventral, posterior, and dorsal views (X 2). Lower Devonian, Konieprus, Bohemia, Czechoslovakia. USNM 209282. 8-12. Hippocardia? zeileri (Beushausen), 1895 (p. 76). Left-lateral, right-lateral, posterior, anterior (X 3), and dorsal (X 3.5) views showing some muscle scars. Devonian, Germany. MCZ 18032. 13-15. Hippocardia'! (p. 4 ). Oblique left-lateral, posterior, and dorsal views (X 44) of a specimen showing a recumbent, snoutshaped protoconch. Shale below Lester Shale (Pennsylvanian), SW SW SW SW sec. 10, T. 6 S., R. 2 E., Love County, Okla. UOK 6083.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 47 HIPPOCARDIAFigures PLATE 48 1-15. Hippocardia cunea (Conrad), 1840 (p. 75). 1. Ventral view of Hall (1885, pi. 67, fig. 29) hypotype (X 1). Schoharie Grit (Lower Devonian), Schoharie, N.Y. NYSM 2313. 2. Ventral view of a specimen showing the entire hood (X 1). Schoharie Grit (Lower Devonian), Saugerties, N.Y. NYSM 6667. 3,4. Left-lateral and dorsal views of an internal mold showing the longitudinal shelves (X 2). Devonian, Columbus, Ohio. FM 59845 (Walker Mus. Coll.). 5. Dorsal view of Hall (1885, pi. 68, fig. 13) hypotype (X 2). Upper Helderberg Limestone (Lower Devonian), Columbus, Ohio. AM 2853a/3. 6, 7. Posterodorsal and ventral views showing the filling of the elongate ventral aperture developed in forms with a hood (X 1). Devonian (Middle Devonian?, Silver Creek Limestone?), near Louisville, Ky. USNM 33581. 8-11. Right-lateral (X 2), posterior (X 1), oblique posterior (X 1), and dorsal (X 2) views of Hall (1885, pi. 68, figs. 10, 11) hypotype. Onondaga Limestone (Jeffersonville?, Middle Devonian?), near Louisville, Ky. FM 12500 Walker Mus. Colin.). 12-14. Fragment of shell showing outer shell layer with ornament on one side and marginal denticles forming internal ribs on the other side. 12, Outside of shell showing ribs (X 2). 13, Inside of shell showing marginal denticles which form internal ribs as they grow (X 2). 14, Ventral edge of shell showing that marginal denticles and ornament are continuous at the shell margin (X 2). Horizon and locality uncertain, possible Middle Devonian (Beechwood Limestone?), at the Falls of the Ohio River. USNM 209280. 15. Oblique view looking into shell at ventral commissure; inner shell layer dissolved away and dorsal part of shell broken off. Hole at bottom leads into elongate ventral aperture running the length of the hood (x 2). Horizon and locality the same as in figures 12-14 above. USNM 209281.HIPPOCARDIA \Figures PLATE 49 1-15. Hippocardia cunea (Conrad), 1840 (p. 75). 1-5. Right-lateral (X 2), oblique dorsal (X 1,5), anterior (X 1.5), ventral (X 1.5), and posterior (X 2) views. The last view shows the lamellae that make up the hood in cross section. Horizon and locality the same as on plate 48, figures 12-14. USNM 209276. 6. Posterior view of edge of hood showing lamellae in cross section (X 3.5). Devonian (Jeffersonville Limestone), Falls of the Ohio River. USNM 51373. 7. Oblique view of broken right edge of hood showing length of lamellae which make up that hood (X 3). Devonian, 2 miles southwest of Sylvania, Lucas County, Ohio. MU 323. 8-10. Right-lateral, oblique dorsal, and ventral views (X 1.5). Horizon and locality the same as on plate 48, figures 12-14. USNM 209277. 11,12. Interior and exterior views of outer shell layer of posteroventral part of right valve with attached part of hood. Hood begins where radial ribbing stops (X 1). Horizons and locality the same as on plate 48, figures 12-14. USNM 209278. 13-15. Oblique dorsal, ventral, and right-lateral views (X 3) of Nettleroth hypotype (1889, pi. 5, figs. 16-18). Horizon and locality as in figure 6 above. USNM 209279.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 49 HIPPOCARDIAPLATE 50 Figures 1,2. 3. 5, 11-13. 6-10. 14. 15-19. 20-24. 25-27. 28-37. 38. Hippocardia cunea (Conrad), 1840 (p. 75). 1. Right-lateral view (X 2). Horizon and locality the same as on plate 48, figures 12-14. USNM 209267. 2. Inner surface of outer shell layer of right valve showing the internal ribs built by the marginal denticles and the tubular extension of the ventral aperture into the hood. Horizon and locality the same as on plate 48, figures 12-14. USNM 209268. Hippocardia pygmaea (Hisinger), 1837 (p. 76). Posterior view showing hood (X 3) of Branson hypotype (1942b, pi. 59, fig. 17). According to Valdar Jaanusson (written commun., 1975), the specimen is from Upper Ordovician rocks, Porkuni Stage, Porkuni “Borkholm”), Estonia. USNM 98871. Hippocardia sp. (p. 76). 4,5. Left-lateral and dorsal views (X 3). Silurian (Wenlockian, Slite Beds), Gotland, ditch between Angelbos and Norvarg, Parish of Larbro, Sweden. SMNH Mo. 18326. 11. Ventral view showing hood (X 3). Platteville Limestone and Briton Member of Mifflin Formation of Templeton and Willman 1952) (Middle Ordovician), Medusa Portland Cement Co., Lee County, near Dixon, 111. UI 5261. 12. Right-lateral view (X 2). Ordovician, quarry north of Church Stake Hall, Criburg district, west Shropshire, England. USNM 100707. 13. Ventral view showing hood (X 2.5). Middle Ordovician, Rich Valley, Porterfield quarry, 5 miles east of Saltsville, Va. USNM 206509. Hippocardia richmondensis (Foreste), 1910 (p. 76). 6-8. Right-lateral, dorsal, and left-lateral views of holotype (X 4). Elkhom Formation (Upper Ordovician), 3 miles south of Richmond, Ind. USNM 87041. 9,10. Right-lateral and left-lateral views (X 4). Richmondian (Upper Ordovician), Ohio. MU 209T. Hippocardia antiqua (Owen), 1852 (p. 75). Left-lateral view of holotype (X 3). Ordovician; the museum label reads: “Lower Fort Garry, Red River of the North, Manitoba,” Canada. USNM 17897. Hippocardia limatula (Bradley), 1930 (p. 75). 15. Left-lateral view of paralectotype (X 3). Kimmswick Limestone (Middle Ordovician), 1 mile north of Batchtown, 111. FM 29052 (Walker Mus. Colin.). 16. Right-lateral view of lectotype (X 3). Horizon, locality, and museum number the same as in figure 15 above. 17. Posterior view showing hood of paralectotype (X 3). Horizon, locality, and museum number the same as in figure 15 above. 18. Posterior view of paralectotype showing remnant of hood (X 3). Horizon, locality, and museum number the same as in figure 15 above. 19. Left-lateral view (X 3). Kimmswick Limestone (Middle Ordovician), New Hope, Mo. USNM 209273. Bransonia beecheri (Raymond), 1905 (p. 72). 20. Left-lateral view of lectotype (X 5). Chazyan (Middle Ordovician), Sloop Island, near Val-cour Island, N.Y. YU 15322C. 21. Left-lateral view of paralectotype (X 5). Horizon and locality the same as in figure 20 above. YU 15322B. 22. Left-lateral view (X 3). Row Park Limestone (Middle Ordovician), 200 feet above base section, 1.3 miles west of Marion, Pa. USNM 209274. 23. Left-lateral view (X 4). Mosheim Member of Lenoir Limestone (Middle Ordovician), Climer P.O., 7 miles east of Cleveland, Tenn. USNM 209275. 24. Left-lateral view (X 3). Chazvan (Middle Ordovician), Isle LaMotte, Vt. USNM 100709. Bransonia aff. B. paquettensis (Wilson), 1956 (p. 73). Left-lateral, right-lateral, and posterior views (X 3). Holston Formation (Middle Ordovician), Por terfield quarry, 5 miles east of Saltville, Va. USNM 144969. Bransonia alabamensis n. sp. (p. 72). 28. Right-lateral view of paratype (X 5). Little Oak Limestone (Middle Ordovician), crossroads 1.75 miles northeast of New Hope Church, Ala. USNM 209269. 29-31. Left-lateral, right-lateral, and dorsal views of paratype (X 5). Horizon and locality the same as in figure 28 above. USNM 209270. 32-34. Left-lateral, dorsal, and right-lateral views of holotype (X 5). Horizon and locality the same as in figure 28 above. USNM 209271. 35-37. Right-lateral, dorsal, and left-lateral views of paratype (X 5). Horizon and locality the same as in figure 28 above. USNM 209272. Bransonia paquettensis (Wilson), 1956 (p. 73). Right-lateral view of holotype (X 2). Leray-Rockland Beds (Middle Ordovician), Paquette Rapids, Ottawa River, Ottawa, Ontario, Canada. GSC 11585.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 50 mm ■ v£#] HIPPOCARDIA AND BRANSONIAPLATE 51 Figures 1-10,17. Bramonia wilsoni n. sp. (p. 72). 1-6. Left-lateiral, right-lateral, dorsal, ventral, anterior, and posterior views of holotype (X 1.5). Middle part of “Homevale beds,” lower Tiverton Formation (Permian), ridge southeast of Home-vale Homestead, Nebo District, Queensland, Australia. UNE F14789. 7-10. Posterior, anterior, right-lateral, and dorsal views of paratype (X 1.5). Horizon and locality the same as in figures 1-6 above. UNE F14790. 17. Oblique dorsal view of latex replica of paratype (X 1.5). Horizon and locality the same as in figures 1-6 above. UNE F14790. 11. Conocardium aliformet (Sowerby), 1815 (p. 70). Left-lateral view showing muscle impressions (X 3). Carboniferous limestone (Mississippian), Lowick, Northumberland, England. SM E1151. 12-16. Bransonia robustum (Fletcher), 1943 (p. 19). Dorsal, ventral, right-lateral, posterior, and anterior views (X 1.5). Permian (Wandrawandian Siltstone), Wyro, near Ulladulla, New South Wales, Australia. AMS F21930.PLATE 52 Figures 1-5,9. Bransonia wilsoni n. sp. (p. 72). 1. Dorsal view of paratype showing internal mold of protoconch (X 1.5). Horizon and locality the same as on plate 51, figures 1-6. UNE F14791. 2-5. Paratype. 2, Latex replica showing ornament and fenestellate bryozoan attached to left posterior face (X 1.5). 3—5, SEM photographs showing protoconch (X 25). Horizon and locality the same as on plate 51, figures 1-6. UNE F14792. 9. Posterior view of latex replica of paratype showing attached bryozoan (X 1.5). Horizon and locality the same as on plate 51, figures 1-6. UNE F14793. 1,7,10-14. Bransonia cressmani n. sp. (p. 72). 6,7. Left-lateral and dorsal views of paratype (X 5). Salvisa Bed, Perryville Limestone Member of the Lexington Limestone (Middle Ordovician), quarry on Mitchellsburg Road, 0.4 miles south of Perryville, Ky. (USGS loc. 5015-CO). USNM 209265. 10-14. Ventral, dorsal, posterior, right-lateral, and left-lateral views of holotype (X 5). Horizon and locality the same as in figures 6, 7 above. USNM 209266. 8. Bransonia? sp. (p. 19). Dorsal view of internal mold showing muscle scars (X 3). Windom Formation (Vitulina Zone, Hamiltonian, Middle Devonian), Tinkers Falls, Truxton, Cortland County, N.Y. USNM 100703.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 52 BRANSONIAPLATE 53 Figures 1-5. Bransonial sp. (p. 19). Right-lateral, anterior, oblique anterior, left-lateral, and posterior views of an internal mold showing muscle scars (X 3). Horizon, locality, and museum number the same as on plate 52, figure 8. 6-20. Bransonia cressmani n. sp. (p. 72). 6,7. Dorsal and right-lateral views of paratype (x 5). Perryville Limestone Member of the Lexington Limestone (Middle Ordovician), quarry on west side of U.S. Route 68, 1 mile north of junction with U.S. Route 150 in Perryville, Ky. (USGS loc. 6916-CO). USNM 209260. 8-10. Left-lateral, right-lateral, and dorsal views of paratype (X 5). Horizon and locality the same as in figures 6, 7 above. USNM 209261. 11-16. Right-lateral, ventral, dorsal, posterior, left-lateral, and anterior views of paratype (X 5). Horizon and locality the same as in figures 6, 7 above. USNM 209262. 17-19. Right-lateral, left-lateral, and dorsal views of paratype (X 5). Horizon and locality the same as on plate 52, figures 6, 7. USNM 209263. 20. Dorsal view of paratype (X 5). Horizon and locality the same as in figures 6, 7 above. USNM 209264. 21-23. Bransonia robustum (Fletcher), 1943 (p. 19). Oblique right-lateral, dorsal, and right-lateral views of paratype showing muscle scars (X 3). Horizon and locality the same as in figures 11-16, plate 51. AMS F. 21928.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 53 BRANSONIAPLATE 54 Bransonia cressmani n. sp. (p. 72). Stero triplet of internal mold showing a large muscle scar in the top center of the right side; anteriorly from this muscle scar passes a part of the pallial line (X 52). Tanglewood Limestone Member of the Lexington Limestone (Middle and Upper Ordovician), 81 feet above the Macedonia Bed of Grier Limestone Member of Lexington Limestone (Middle Ordovician) in the Frankfort East section on eastbound lanes of Interstate Highway 64, east side of Kentucky River crossing, Franklin County, Ky. USGS loc. D-1200-CO. USNM 209259.GEOLOGICAL SURVEY PROFESSIONAL PAPER 968 PLATE 54 BRANSONIA 7 DAYS :s\jMAll1 SOME ENGINEERING GEOLOGIC FACTORS CONTROLLING COAL MINE SUBSIDENCE IN UTAH AND COLORADO - 1SOME ENGINEERING GEOLOGIC FACTORS CONTROLLING COAL MINE SUBSIDENCE IN UTAH AND COLORADOFrontispiece .—Northward view across the North Fork Gunnison River showing the geologic and topographic setting in the Somerset coal mining district, Gunnison County, Colo. The rugged topography shown is characteristic of the Somerset mining district. The ridge in the left foreground overlies the mine workings that were selected for one of the subsidence studies. The Mesaverde Formation comprises the area from the light-tan ledges and cliffs in the foreground (Rollins Sandstone) to the ridge crests in the middleground. The coal-bearing portion of the Mesaverde is portrayed by the pinkish-red coloration in the left foreground, caused by the burning of the coal beds, which, in turn, has caused extensive subsidence. The Mancos Shale underlies the Mesaverde Formation. The light-tan ledges and cliffs above the Mesaverde (background) are the rocks of the Ohio Creek Formation. The subdued terrain in the background, with its extensive landslides and green aspen groves (right background), is developed on rocks of the Wasatch Formation, which overlies the Ohio Creek Formation. Some Middle Tertiary intrusives, which form a part of the West Elk Mountains, are on the right skyline.Some Engineering Geologic Factors Controlling Coal Mine Subsidence in Utah and Colorado By C. RICHARD DUNRUD GEOLOGICAL SURVEY PROFESSIONAL PAPER 969 A discussion of the processes of subsidence and their effects on mine safety, coal resource management, and the environment in two geologic settings UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON : 1976UNITED STATES DEPARTMENT OF THE INTERIOR THOMAS S. KLEPPE, Secretary GEOLOGICAL SURVEY V. E. McKelvey, Director Library of Congress Cataloging in Publication Data Dunrud, C. Richard Some engineering geologic factors controlling coal mine subsidence in Utah and Colorado. (Geological Survey Professional Paper 969) Bibliography: p. 1. Coal mines and mining—Utah. 2. Coal mines and mining—Colorado. 3. Mine subsidences—Utah. 4. Mine subsidences—Colorado. 5. Engineering geology—Utah. 6. Engineering geology—Colorado. I. Title. II. Series: United States Geological Survey Professional Paper 969. TN805.U8D86 622'.33*4 76-7452 For sale by the Superintendent of Documents, U.S. Government Printing Office Washington, D.C. 20402 Stock Number 024-001-02864-6CONTENTS Page Abstract.................................................... 1 Introduction................................................ 1 Coal as an energy contributor.......................... 2 Definition of subsidence............................... 2 Acknowledgments ....................................... 2 Previous studies............................................ 3 Current studies of subsidence processes..................... 8 Geneva mine area, Utah ................................ 8 Somerset mining district, Colorado..................... 9 Geometry of subsidence cracks...................... 13 Page Current studies of subsidence processes—Continued Somerset mining district, Colorado—Continued Subsidence measurements.......................... 17 Stress and deformation caused by subsidence .... 26 Case study of the relationship between a mine and the physical environment ...................... 30 Summary................................................. 34 Suggestions to mine planners............................ 36 Conclusions............................................. 38 References cited........................................ 38 ILLUSTRATIONS Frontispiece . View showing the geologic and topographic setting of the Somerset coal mining district, Colorado. Figure 1. Sections showing early concepts of surface and rock mass subsidence processes above mine workings.................... 3 2. Sections showing subsidence trough above a mined coal bed.......................................................... 4 3. Section showing author’s interpretation of how further mining would affect deformation of the surface.............. 5 4. Cross section showing general subsidence yield conditions for surface and stratified bedrock after migration of compression arch to surface..................................................................................... 6 5. Graph showing subsidence ratios for various ratios of mine panel width to mean overburden depth............. 7 6. Cross section perpendicular to mine panels, showing concurrent harmonic extraction of four flat-lying coal beds beneath overburden of constant thickness................................................................................ 8 7. Photographs showing Upper Cretaceous Mesaverde Group underlying the Book Cliffs near the Geneva Mine, Utah. 10 8. Cross section of the rocks of the Upper Cretaceous Mesaverde Group in the southern parts of the Book Cliffs and Geneva mines.................................................................................................... 11 9. Photographs showing subsidence damage in a massive sandstone...................................................... 12 10. Infrared Skylab 2 photograph of the Somerset mining district, Colorado ........................................... 14 11. Map showing plot of lineaments derived from the Skylab 2 infrared photograph of the Somerset mining district...... 15 12. Photograph of upper(?) Tertiary pluton that intrudes the Mesaverde and Ohio Creek Formations ..................... 16 13. Photographs of the Mesaverde Formation in the Somerset mining district............................................ 17 14. Map of the 3d South area in the Somerset mine..................................................................... 18 15. Photographs showing primary and subsidence fractures.............................................................. 20 16. Cross section A — A’through the 3d South area, Somerset mine...................................................... 22 17. Cross section B—B'through the 3d South area, Somerset mine........................................................ 24 18. Composite map of part of the Somerset mine workings and histograms of the daily seismic activity during 1973 and 1974, in the Somerset mining district.......................................................................... 28 19. Photographs showing stress effects from multiple-bed coal mining, Somerset district............................... 30 20. Composite aerial photograph and underground map of the Oliver No. 2 mine area..................................... 32 21. Photograph showing the rugged topography above the Oliver No. 2 mine and vegetation killed by escaping methane.... 33 22. Photograph and sections showing coal mine squeeze in “B” coal bed, Somerset district.............................. 36 TABLE Page Table 1. Relative displacement of U.S. Geological Survey subsidence bench marks located above the 3d South area of the Somerset mine, Delta and Gunnison Counties, Colo.............................................................. 21 vSOME ENGINEERING GEOLOGIC FACTORS CONTROLLING COAL MINE SUBSIDENCE IN UTAH AND COLORADO By C. Richard Dunrud ABSTRACT Subsidence plays a major role in coal mining activities and in the future use of the land surface above the mine workings. Stresses and deformations produced in mine workings, other coal beds, bedrock, and at the ground surface by the processes of subsidence significantly affect mine safety, extraction efficiency, and the surface environment. Basically, the subsidence process comprises two different stress-and-yield conditions in response to the excavation of mine workings. First, arcuate zones of compressive stress, called compression arches, tend to occur above and below the mine panels and transfer the overburden load in coal-extraction areas to adjacent solid-coal boundaries or barrier pillars. Second, caving and flexure of strata, within the destressed zone encompassed by the arches, into the mine cavities tends to increase the stresses again in the mine workings. Flexure of strata also produces tensile and compressive stresses within lithologic units and shear stresses across lithologic boundaries. With time, and as the mine voids are widened, the compression arches tend to migrate higher in the overburden strata and eventually may reach the surface. This migration continues to transfer overburden stresses back into the extraction area from the mine boundaries or barriers. The rate of migration of compression arches, and, consequently, the rate of stress transfer, depends on thickness and strength of overburden strata, duration and rate of mining, mine geometry, and mining sequence. The ground surface, other coal beds, ground-water aquifers, methane zones, and economic deposits above and below the mine workings can be damaged by the subsidence processes. Deposits in the mine overburden and the surface are subjected to stresses produced by compression arches and by flexure of strata. As a result of the subsidence processes, deposits are subjected to unloading followed by loading beneath mine openings and, in addition, are subjected to loading followed by unloading beneath remnant coal pillars and near solid-coal mine boundaries. In an area underlain by several thin beds or by one thick bed, mining sequentially from upper to lower beds, or, in the case of a thick bed, mining separate “benches” or “lifts” from top to bottom commonly is safest and most efficient. However, unless a uniform extraction plan is followed, mining hazards, such as roof falls, bumps, and squeezes caused by stress concentrations, may force premature abandonment of an area of the mine, with resultant loss of reserves. Knowledge of geologic, topographic, and socioeconomic conditions in prospective mining areas is vital to planning safe and efficient mining activities, particularly in areas underlain by thick coal beds or by more than one coal bed. INTRODUCTION Subsidence damage resulting from underground coal mining will become a greater problem in the United States as our population increases along with our demand for more energy. The clash between population growth and man’s demand for more energy is already very acute in many heavily populated industrial countries, such as Japan (Kaneshige, 1971), but also it is felt in the more heavily populated coal mining areas in the United States. Although lands underlain by coal reserves in the Western United States are still relatively free of industrial and urban development, many areas in the East support industry, housing, or farming. Many old established urban areas overlying coal mine workings, such as Scranton, Pa., have experienced severe and costly subsidence damage many years after the mines were abandoned. According to recent information (U.S. Bureau of Mines, written commun., Feb. 5 and April 10, 1973), $29 million worth of property, including 2,000 homes, 50 commercial and office buildings, 2 hospitals, and several schools are threatened by subsidence in the Hill section of Scranton—an area in which an estimated 10,000 people live and work. The total cost of surface stabilization of this area by hydraulic mine backfill is estimated to exceed $8 million. Subsidence damage to the rocks above underground coal mines also can reduce minable coal reserves, decrease mine safety, lower production efficiency, and possibly disrupt the hydrologic regimen. Coal deposits in parts of Wyoming, Colorado, and Utah locally comprise vertical sequences of several minable coal beds interbedded with thick to thin sandstones and mudstones. Deformation caused by mining one coal bed can damage other beds nearby unless the beds are extracted in proper sequence. Ruptures within the l2 ENGINEERING FACTORS, COAL MINE SUBSIDENCE, UTAH AND COLORADO rock mass and at the surface may tap and divert ground water or methane pockets associated with the coal beds or may even divert surface-water drainage to underground aquifers or mine workings, which in turn can upset established ground- and surface-water supplies as well as threaten the mine workings. Coal mine subsidence problems such as these point out that mining plans for new and existing coal mining areas must be based on all existing technology, with the goal of maximizing coal extraction percentages and minimizing the subsidence damage to the overburden and surface. To accomplish this we must know the processes of subsidence that occur both in the overburden and at the surface in various geologic environments and under various methods of mining. This is one of the major goals of coal mine deformation studies underway in the U.S. Geological Survey and is the central theme of this report. COAL AS AN ENERGY CONTRIBUTOR Coal is an important part of our current energy supply, particularly for use in generating electric power. The use of coal will no doubt continue to grow rapidly in response to increasing demands on domestic energy resources. In 1970, 17,000 trillion B.t.u.’s from various energy sources were required to produce some 1,260 billion kilowatt hours of electrical energy in the United States (Risser, 1973). Of this total, 1 percent was supplied by nuclear power, 16 percent by hydroelectric power, and 83 percent by the fossil fuels. Coal provided 55 percent of the power contributed by the fossil fuels. Although our total energy requirements are increasing rapidly with time, the demand for electrical energy nearly doubles every 10 years, and, assuming that coal continues to contribute about the same percentage of our electrical energy, the demand for coal will increase accordingly. The projected requirements for coal in the 1970’s alone may equal the total amount consumed in the previous seven decades. Many coal deposits are too high in sulfur, ash, or other constituents to be burned in power plants under existing or proposed clean-air standards. However, various techniques of obtaining synthetic crude oil, natural gas, char residue, sulfur, and fly ash from coal are in advanced stages of study or implementation under the auspices of U.S. Department of the Interior Office of Coal Research (W. A. Bear, oral commun., 1970, 1972). With these processes, the sulfur and ash are recoverable byproducts rather than pollutants. Another pollution-free method of producing electrical energy from coal involves the direct generation of electricity from coal and is called the magnetohydro- dynamic (MHD) process; with this process, electrical energy is produced by passing hot gases, derived from coal, through a stationary magnetic field. Should these new techniques eventually prove economical, coal would be an increasingly important source of electrical energy, as well as a source of other forms of energy and material byproducts, for years to come. It will thus be of continuing importance for us to study the effects, including subsidence, of mining needed supplies of coal. DEFINITION OF SUBSIDENCE Coal mine subsidence is defined in this report as all deformation within most of the overburden and at the surface that is caused by underground mining. It includes the local upward movement of strata that sometimes occurs above solid-coal mine boundaries or large barrier pillars, which is caused by down warping of overburden into mine cavities; it also includes the downwarping itself, the associated horizontal tensile and compressive strains produced by strata flexure, and the compressive strain induced by the compression arches. For the purposes of this report, the term “subsidence” applies to deformation or movement in the overburden two or more mine heights above the immediate mine roof; the term “roof fall” applies to the fall of roof rocks less than two mining heights above the mine roof. ACKNOWLEDGMENTS Officials of the U.S. Steel Corp., Bear Coal Co., and Western Slope Carbon, Inc., materially aided the studies undertaken for purposes of this report. They contributed maps of mine workings, offered helpful suggestions and information, and they allowed access to their properties and the use in this report of certain information and pictures taken in their mines. Their help was given in the spirit of making coal mining safer, more efficient, and less damaging to the environment. The author thanks them all for their help and cooperation, particularly Paul E. Watson, general superintendent, western district, U.S. Steel Corp.; J. Boyd McKean, district mining engineer, U.S. Steel Corp.; R. W. Ramey, superintendent, Somerset mine, U.S. Steel Corp.; W. A. Bear, president, Bear Coal Co.; and C. L. Heiner, Western Slope Carbon, Inc. Two part-time coworkers provided basic data for the preparation of this report. R. Edward McKinley, retired mining engineer, U.S. Steel Corp., made monthly subsidence measurements at Somerset, which were extremely helpful in determining thePREVIOUS STUDIES 3 processes of subsidence. Robert L. Rasmussen, mine engineer at the Somerset mine, who works part time on the U.S. Geological Survey seismic recording facility at Somerset, provided mine maps and helpful updates on mining operations. Two colleagues, R. L. Parker and R. B. Taylor, provided recent high-resolu-tion aerial photographs of the Somerset district that proved very useful to the mine deformation studies. PREVIOUS STUDIES Subsidence studies began in Europe before the turn of the century. Early concepts of subsidence included the so-called law of the normal and law of the vertical (fig. 1). According to the law of the normal (fig. L4), investigators believed that the overburden strata broke and subsided above and parallel to the boundaries of the mined-out areas along fractures perpendicular to the strata. The so-called law of the vertical was developed a few years later. According to this concept the overburden broke and subsided along vertical fractures above the boundaries of the mined-out areas (fig. LB). During the early 1900’s the concepts of limit (or draw) angle and break angle developed following further surface and underground measurements. Measurements in the Ruhr region of Germany and elsewhere in Europe and Russia revealed that subsidence or other deformation affected a surface area larger or smaller than the area mined (fig. 2A). The limit angle V, which is the acute angle defining the limit of subsidence above mine workings, is positive if the surface area affected by subsidence is greater than the mined-out area; it is negative if the deformed surface area is smaller than the mined-out area. The break angle 0, which defines the zone of maximum tensile stress caused by flexure of strata (fig. 2B), is steeper than the limit angle but may approach the limit angle when it is negative. The limit angle is measured from either the horizontal or from the vertical. The angle relative to the horizontal is chosen in this report because it is consistent with geologic measurement of bedrock attitudes. Limit angles commonly range from 45° to 65° in European coal fields, 55° to 65° in Great Britain, and 45° to 60° in the Ruhr region; they are about 55° in northern France and about 60° in the U.S.S.R.; and they range from 45° to 55° in the Netherlands (where angles are referenced from the horizontal rather than from the vertical) (Zwartendyk, 1971, p. 142-143). A troughlike subsidence geometry was observed within the limits of the draw in many European coal mining districts. This observation led Lehmann (1919, cited in Zwartendyk, 1971, p. 85-86) to propose his COAL BED — Mined where unshaded X / / z\ Surface subsidence ---------*- Fracture Figure 1.—Early concepts of surface and rock mass subsidence processes above mine workings. A, Law of the normal; B, law of the vertical. subsidence trough theory. This concept explains tensile and compressive strains in addition to the vertical and horizontal movement and tilting observed above coal mining areas. According to this concept, the rate of change of vertical settlement increases to a maximum from the limits of the draw inward, becomes constant at the point of inflection, and then decreases to zero in the center of the trough, provided the mining area is wide enough to prevent further differential settlement in this area (supercritical mining width) (fig. 2B). The mined-out area is said to be of critical extraction width if it is just wide enough to allow maximum subsidence (Smax) in the center of4 ENGINEERING FACTORS, COAL MINE SUBSIDENCE, UTAH AND COLORADO A Hi ■ COAL BED — Mined where-------------------Relative horizontal unshaded distance and slope |/ / A Surface subsidence -------Relative horizontal strain and curvature ---»- Direction of horizontal disp la cement Figure 2.—Subsidence trough above a mined coal bed; vertical scale of subsidence trough greatly exaggerated for clarity. D, overburden depth. Modified from Zwartendyk (1971, p. 132), Brauner (1973, p. 4), Mohr (1956, p. 141), and Warded (1971, p. 206). A, Subsidence trough with positive limit (or draw) angle (A and positive break angle ((3). B, Subsidence trough showing critical width of mining that will cause maximum surface subsidence (Smax), supercritical width of mining, and the general form of the curves depicting horizontal displacement, slope, strain, and curvature. the trough (fig. 2B). According to Wardell (1971, p. 205), this width varies, but in European coal fields it is between 1.0 and 1.4 times the average overburden depth, depending on the lithology and structure of the overburden and provided the panel length is more than 1.4 times the average overburden depth. Although most troughs do not precisely fit the model in figure 2JB, the figure illustrates the reasons for the vertical and horizontal strains that are commonly observed above mined-out areas. The ground surface is convex upward between the limits of the draw and the points of inflection; thus, the tensile strain increases to a maximum at the point of maximum positive curvature and then decreases to zero at the points of inflection, where ground tilt and horizontal extension are at a maximum. Between the points of inflection and the point of maximum subsidence (Smax, fig. 2B), the ground surface is concave upward, thus causing compressive strain. Between these points, the compressive stresses increase to a maximum and then decrease to zero, provided the mined-out area is of supercritical width. According to Wardell (1971, p. 209), maximum change in ground slope and length varies in relation to the maximum amount of subsidence (S max) divided by the average overburden depth CD), whereas the maximum horizontal curvature varies in relation to the maximum subsidence divided by the square of the average overburden depth (D2). The area subjected to maximum subsidence has reached a condition of zero strain. However, even this area of maximum subsidence and zero strain was initially subjected to strain during mining. Furthermore, if the mine opening is widened (fig. 2B), the surface adjacent to the area affected by subsidence will be subjected to a transient wave of first tensile, then compressive, and finally zero strain as the ground subsides above the enlarging mine opening. Ground-water- and methanebearing rocks, as well as subsurface water, gas, and sewer mains, and unyielding buildings and foundations, will be subjected to all these strains, and the damage will, in most cases, be cumulative. Mining of coal adjacent to the mined-out area, as shown in figure 2B, leaving a barrier pillar (fig. 3), can cause additional surface strains. Here, two adjacent subsidence troughs superposed over such pillars would interact to produce tensile strain approaching twice the strain produced above a solid-coal boundary. Both the rate and amount of surface settlement were studied by various foreign investigators (for example, Mohr, 1956, and research cited by Zwartendyk, 1971) in relation to various physical environments. Although much of this information is contradictory and somewhat confusing, they found that, in general, the amount of surface subsidence and deformation, although variable, depended on the strength and thickness of overburden, the width of the mined-out area, and the thickness of the coal bed. A wider mining area or a thicker coal bed increased the rate of surface settlement but, if the extraction wasPREVIOUS STUDIES 5 Barrier pillar Figure 3.—Author’s interpretation of how further mining of the coal bed shown in figure 2 would affect deformation of the surface. A barrier pillar now separates two mined-out areas; strain and horizontal displacement curves are shown above cross section. Note that the superposition of subsidence profiles greatly increases the tensile strain above the center of the barrier; also, the horizontal displacement is nearly at a maximum over the entire barrier. Surface damage is most accute above the middle of the barrier pillar; the barrier pillar must be twice as wide as the overburden depth (D) multiplied by the cotangent of the limit angle ( mining, Somerset district, Colorado. In A through C, the lower bed of coal was mined first; and in D through F, the upper bed was mined first. A, Damage to roof and rib in the “E” coal bed caused by mining “D” coal bed, 150 feet (45.8 m) below “E” bed, 25 years prior to mining “E” bed. Roof bolts and landing mats provided adequate support until the limit of mining in underlying “D” coal bed was traversed; timbering (background) was then required to prevent roof falls and possible coal bumps. B, Sites of roof falls and generally unstable roof conditions in a crosscut in the “C” bed that is located 45 feet (13.7 m) above workings in the “B” bed that were mined about 40 years before mining the “C” bed. C, Unstable roof conditions in the “C” bed along a slope haulageway above workings in the “B” bed that were mined about 40 years before mining the “C” bed. D, Damage to right rib and roof in the “B” coal bed beneath about 1,200 feet (366 m) of overburden caused by local stresses beneath an isolated pillar 60 feet by 60 feet (18.3X18.3 m) left in the “C” coal bed about 50 feet (15 m) above and collimated with the pillars in the “B” bed. Note the dramatic difference between the rib and roof conditions on the right side of the mine opening beneath the isolated pillar in the “C” bed and those of the left rib where the coal was mined out in the “C” bed before “B” bed was mined. E, High stress concentrations in an isolated pillar 60 feet by 60 feet (18.3X18.3 m) in the “B” coal bed beneath about 1,300 feet (397 m) of overburden. The pillar ribs in the left and right foreground show no signs of stress because the coal was completely mined out in “C” bed before “B” bed was mined, whereas the isolated pillar is yielding to stresses from an isolated pillar in “C” bed directly above the yielding pillar in “B” bed, even with massive props surrounding it. The author monitored about 100 coal bumps during a 3-hour period, as the yielding pillar was split by a mining machine. F, View of ideal rib and roof conditions in the “B” coal bed beneath about 1,300 feet (397 m) of overburden where the coal was mined out in the overlying “C” bed before “B” bed was mined. The overburden stress was reduced to essentially the 50 feet (15 m) of strata separating the two coal beds. is located in the “D” coal beds south of the North Fork Gunnison River, about 2 miles (3.2 km) east of the Somerset mine (fig. 20). The mine comprises a system both of raises on a 5- to 7-percent grade that roughly parallels the direction of the dip of the coal bed and of entries that parallel the strike of the coal bed. The mine portals are located at the outcrop of a “D” coal bed overlooking the North Fork Gunnison River. The surface area near the mine is sparsely settled and supports summer grazing for cattle and horses. The Oliver No. 2 mine, which was begun in the 1930’s and provided coal for the Oliver power plant, was closed in October 1953 after methane gas and water were encountered in quantities that were too costly to control. A four-entry raise was driven southward beneath overburden which increased in thickness from a few tens of feet near the outcrop to 1,250 feet (381 m) beneath a high ridge and thenCURRENT STUDIES OF SUBSIDENCE PROCESSES 31 decreased within a distance of about 1,500 feet (457 m) to 325 feet (99 m) beneath an east-trending, joint-controlled side canyon of Sylvester Gulch (fig. 20). At this point a four-entry system was driven eastward (7 East) directly beneath, and parallel to, the east side canyon. The 7 East panel was driven about 300 feet (91.4 m), then work began on driving 6 East. Suddenly large volumes of water and methane began to gush from the floor of the top entry of 6 East, forcing the evacuation and closure of the mine. The mine was sealed and has remained closed. The Oliver electric power plant also was subsequently closed, although other sources of coal supplied it for a32 ENGINEERING FACTORS, COAL MINE SUBSIDENCE, UTAH AND COLORADO Figure 20.—Composite aerial photograph and underground map of the Oliver No. 2 mine area, Colorado. This mine was closed in October 1953 because of a sudden intrusion of water and methane gas. The scrub oak and other woody plants that originally grew in the bare spot above the exhaust portal were killed when the few years. After the mine was sealed, the methane leaked out of the mine to the surface through fractures in the overburden in sufficient quantity to be detected with a miner’s lamp (C. L. Heiner, oral commun., 1974). The methane killed the scrub oak and all other woody plants in the area (fig. 21), leaving only the grasses unaffected. According to Garner (1974), the methane leaked through fractures to the surface after the portals were sealed. The beaver pond apparently dried up when subsidence fractures above 7 East tapped the ponds and the nearby spring source and diverted the water underground. Photograph by R. B. Taylor, U.S. Geological Survey, 1973. presence of methane in soil provides an environment in which certain bacteria utilize the methane and produce hydrogen sulfide and nitrous oxide in their life processes; the presence of the hydrogen sulfide and perhaps nitrous oxide disrupts root transpiration of woody plants and ultimately may kill the plants. Shortly after the mine was closed, the water in theCURRENT STUDIES OF SUBSIDENCE PROCESSES 33 Figure 21.—Southward view of the rugged topography above the Oliver No. 2 mine, Colorado. The exhaust portal (P, in left foreground) was sealed inthe late 1950’s. The scrub oak and other woody plants that originally grew in bare spot above exhaust portal apparently were killed when methane gas leaked through fractures from the mine to the surface. east-side canyon, which was fed by springs above the 7 East mine workings, ceased to flow (W. A. Bear, oral commun., 1972). This water, to which the Bear family had the rights, has not flowed with any regularity since the mine was closed. Investigation by the author revealed the presence of large dry beaver ponds, with dams as much as 6 feet (1.8 m) high, above the 7 East entries (fig. 20). These dams indicate that a perennial water supply was once available from springs nearby but that the springs have been dry or only flowed intermittently for many years. Although no surface cracks were positively identified in 1972,34 ENGINEERING FACTORS, COAL MINE SUBSIDENCE, UTAH AND COLORADO they could have been covered by erosion-deposition, mass wasting, and revegetation in only a few years. Local scarps in soil and colluvium noted in the area might be erosional remnants of subsidence scarps. Field evidence indicates that, after the mine was closed, subsidence fractures formed in the strata above the 7 East entries and faces of the raise and eventually migrated to the surface and drained the beaver ponds in the east-side canyon (fig. 20). Although coal pillars were not extracted in most of the Oliver No. 2 mine, the vertical and downdip component of stress produced by the weight of the high ridge above the mine workings, together with the reduction of frictional resistance at the top of the coal bed owing to mine development, apparently was sufficient to produce tension fractures along local joints in the thin overburden beneath, and parallel to, the side canyon. The fractures tapped the beaver ponds and nearby springs and diverted the surface flow underground. The mining reports indicate that the water- and methane-filled fracture encountered in 6 East might be a tension fracture produced by decollement-type movement of the strata above a “D” coal bed prior to mining. The “B” and “C” coal beds are present beneath these mine workings, and at least one “E” bed of minable thickness occurs above the mine workings. Subsidence may have significantly reduced the minable reserves represented by these beds because mines in the “B” and “C” beds may be threatened by intrusion of methane and water from the old “D” workings through subsidence cracks and the “E” bed(s) may be locally transected by subsidence fractures and stress concentrations above solid-coal boundaries. In summary, then, not only were wildlife, vegetation, and surface-water rights affected by surface-water diversion and methane leakage via subsidence fractures, but potential production from subjacent and superjacent coal reserves probably also was reduced considerably. SUMMARY Subsidence studies in Utah and Colorado show that the mode of subsidence depends upon: (1) the geometry of mine workings, (2) the lithology, structure, and thickness of the overburden, (3) direction of dip of the coal bed relative to its outcrop, and (4) proximity of mine workings to coal outcrop, unless an adequate coal barrier is left to support the overburden strata. In multiple-bed coal mining, the mining activities in one bed can cause stress problems in another bed during current or subsequent mining. Subsidence parameters, such as break line, limit angle, configuration of compression arch, and ratio of surface subsidence to coal extraction thickness, and how they are controlled by mine geometry and geology in two areas of Colorado and Utah are described in the following list: 1. A nearly vertical break line is caused by cantilever failure in strong overburden above a wide property barrier in the Geneva mine area and illustrates operation of the law of the normal or the law of the vertical in gently dipping strata that underlie overburden comprising strong, thick, jointed sandstones and interbedded mudstones (fig. 8). 2. Nearly vertical break lines also occurred in weak strata above a thin coal barrier between the 8th and 9th Left mining panels in 3d South, Somerset mine and likewise illustrate the law of the normal for tensile failure produced by positive flexure of strata above thin barriers separating adjacent mine cavities (figs. 3, 4, 17). 3. A break line, with a negative angle of 78° (0 = -78°), was observed in weak overburden above the solid-coal barrier north of 6th Left, Somerset mine only 4 months after mining was completed beneath the ruptured area. Strength of overburden appears to govern the rate of fracturing—a -73° (/3= -73°) break line was inferred in strong overburden above the Geneva mine workings 6-12 years after mining was completed (figs. 8, 17); in this case, however, pillars were not completely removed. 4. The limit angle in weak to moderately strong over- burden strata, which is 650 -900 feet (198-274 m) thick (fig. 16), measured 69° - 70° relative to the position of the room-and-pillar retreat line in the 6th Left mining panel in the Somerset mine. The limit angle appears to steepen to 75° or more in moderately strong overburden 900 -1,000 feet (274-305 m) thick, although bench mark control is not good beyond the 900-foot (274 m) overburden level (fig. 16). 5. The overburden above a 300-foot- (91.4 m) wide coal barrier north of 6th Left in the Someset mine was uplifted slightly through November 30, 1973, probably in response to overburden subsidence into the 6th Left mine void (fig. 17). If the uplifted area is included as part of the surface area affected by subsidence, the limit angle ranges from 35° to 65° in weak overburden 550-700 feet (168-214 m) thick; if the limit angle is measured relative to the subsiding zone, it ranges from 82° to 88°. A slight uniform subsidence occurred above the barrier in December,SUMMARY 35 thus precluding further determination of limit angle and suggesting that the limit angle is only a transient parameter above pillars between two mining panels. 6. Maximum subsidence to date (Jan. 1, 1974) in the B row of subsidence benchmarks above the 3d South area was measured above the barrier between 7th and 8th Left rather than above a mined-out area (fig. 17). This, in addition to subsidence-crack data, indicates that compression arches, perhaps about 50 feet (15 m) thick at the crest, are bridging the strata to the barriers and that the subsidence measured above roughly the 500-foot (152.4 m) overburden-thickness level actually results from yield or collapse of barriers pillars rather than from subsidence of strata into mine voids. The depth of the inverted arches that are believed to occur beneath the mine voids is not known but should be considerably less than that of the arches above the voids. The height-to-width ratio for the arches in the overburden appears to be about 1:1. The height of the arch parallel to the long direction of mining panels (fig. 16) is controlled by the position of the arch spanning the barrier pillars oriented perpendicular to the long direction of the panels (fig. 17). This produces abutment stresses on the pillar retreat line as well as on the barriers. A squeeze, such as the one shown in figure 22, can occur if rooms are developed too far ahead of extraction. 7. Stresses produced by flexure of strata within the destressed zone of the compression arches are oriented perpendicular and parallel to the long direction of the mining panels, as are the compression arches above and below the mine workings. Both flexure and caving of strata into mine workings or mined-out areas tend to transfer an increasing amount of the overburden stresses from solid coal boundaries and barrier pillars back into the mine workings or mined-out areas as the arches migrate higher into the overburden. This may explain why miners commonly encounter reduced stress levels on mining faces and along pillar lines after caving occurs in mined-out areas behind the pillar lines. 8. In multiple-bed coal mining, individual coal beds normally can be extracted more safely and completely if the beds are mined from top to bottom, provided the coal is uniformly extracted so that stresses cannot concentrate on isolated pillars and be thereby transmitted to underlying coal beds. However, it may be best to mine two beds concurrently using uniform methods if the beds are separated by only a thin (generally 25 - 50 ft (7.6 -15.2 m)) stratigraphic interval. It might also be best to reverse the sequence and mine coal beds uniformly from bottom to top in areas where the methane or water is abundant in nearby strata. If overburden rocks, for example, contained methane and water, large quantities of methane or water could accumulate in mined-out areas; if these accumulations were above current mine workings, they might be tapped by subsidence fractures and perhaps pose a greater threat to mine safety and coal reserves than would reversal of the recommended mining sequence. 9. The direction of dip of a coal bed relative to its location of outcrop can have a significant effect on mining safety and efficiency. If a coal bed is mined updip from where it crops out, even if the bedrock dips only a few degrees, a downdip decollement-type movement of the mine overburden is possible because the downdip component of gravitational force is unrestrained at the outcrop and the frictional resistance at the top of the coal bed is reduced due to mining. This could produce a severe hazard to life, coal reserves, and surface environment. For example, if the bedrock dipped westward instead of eastward at the same angle in the Geneva mine area (fig. 8), the entire block of overburden strata west of the fractures above the property barrier would be laterally unrestrained, and it could slide above mine workings, crushing pillars, men, and equipment beneath it. Needless to say, the effects on hydrologic regimen and on the surface environment also would be drastic. 10. The subsidence ratio for the ground surface above the 6th Left mining panel (table 1, figs. 14, 16) is Smax/t = 3.1 ft/10 ft = 0.31, where Smax = maximum surface subsidence, and t = mining height. Comparing subsidence ratio to the ratio of panel width (W) and average overburden depth (D) above the mined-out area (fig. 16) produces the following result: »yP=450ft/950ftt41Qft. =450 ft/680 ft= 0.66. The subsidence ratio is considerably lower than that determined by Warded (1971, p. 206; fig. 5) for the same W/D ratio in many longwall mines36 ENGINEERING FACTORS, COAL MINE SUBSIDENCE, UTAH AND COLORADO of the United Kingdom. This could be because longwall mining induces greater surface subsidence, because a compression arch is bridging the 6th Left mining panel, or because the overburden strata are stronger at Somerset than in the United Kingdom. Complete room-and-pillar extraction should induce at least as much subsidence as complete longwall extraction under similar conditions. Weak overburden strata and interbedded coal in the Somerset district probably is as weak as those of the United Kingdom. Therefore, the low subsidence ratio further indicates that compression arches bridge the overburden strata across mining panels. Should the compression arches fail with time, the subsidence ratio may approach Wardell’s (1971) curve. SUGGESTIONS TO MINE PLANNERS Previous and current subsidence studies indicate that a distinction must be made between subsidence and subsidence damage because most subsidence damage to the overburden and surface results from horizontal strain produced by differential vertical settlement of the mine overburden. Under geologic and mining conditions, such as those in the Geneva and Somerset mining areas, subsidence fractures can propagate through many hundreds of feet of strata. In order to estimate the probable effects of subsidence in areas of underground coal mining, mine planners should take into account such factors as (1) overburden thickness, lithology, and structure in the strata above and below the mine workings; (2) the geometry of mine workings; (3) the coal bed thickness; (4) the number of minable coal beds present; (5) the rate of mining; and (6) the natural and manmade environment at the surface. The first four factors vary with locality, and the last two factors often vary with time. The present and future conditions of an area, together with present and anticipated energy needs, determine the proper blend of extraction efficiency and preservation of the surface environment. Uniform coal extraction tends to cause the overburden to settle uniformly and, therefore minimizes subsidence damage. If, in addition, the coal bed could be completely extracted, this method would produce maximum yields. However, if complete extraction is planned by room-and-pillar methods, it may be impossible to mine uniformly during the development and extraction phases if the overburden is deep and stresses are high. Under these conditions, retreat lines must be close to solid coal or else a squeeze may develop (fig. 22), which would cause Cleat 0 50 FEET 15 METRES Figure 22.—Coal mine squeeze in “B” coal bed, Somerset district, Colorado. A, A mine opening that was closed by coal flowage in response to abutment stresses ahead of a pillar extraction area. The squeeze was caused by driving too many workings ahead of the area of pillar extraction. B, Cross section of mine workings and adjacent pillars before squeeze shown in photograph. C, Cross section of same area after squeeze. extensive loss of coal reserves. Therefore, in many instances uniform geometry can be attained only after mining is completed, provided that necessary coal barriers are thin enough to crush out so that stresses are relieved. The other alternative is to design for a uniform partial extraction procedure near solid coal, so that mining stresses are diffused and the possibility of a squeeze is minimized. The mine geometry,SUGGESTIONS TO MINE PLANNERS 37 therefore, is a very important factor to consider when planning a new mining operation. In shallow overburden, where stresses are low, certain compromises might be made in designing mine geometry that would preserve the integrity of the overburden and surface without increasing the stress problems underground, whereas such compromises might prove hazardous and nonproductive where overburden ranges in thickness from 500 to more than 1,000 feet (152-305 m). For example, room-and-pillar mine workings might be developed in a single coal bed beneath shallow overburden; then the pillar ribs could be uniformly sheared off in a multistage extraction sequence until a large percentage of the coal is extracted without damage to the overburden or surface. This is done locally in Japan, in areas where living space and energy are in an equally short supply. More than 90 percent of a coal bed was extracted by this method. The harmonic method of extraction also is used locally in multiple-bed mining (Kaneshige, 1971; fig. 6). Beneath deep overburden, however, the bump and roof-fall problems could be very severe if more than two or three rooms were developed ahead of extraction or if harmonic mining procedures deviated from a rigid time and tonnage schedule; consequently, uneven mining procedures locally may be inevitable during the extraction phase unless partial extraction is planned. A uniform partial extraction procedure with a secondary recovery procedure was recently successfully implemented in the Bear mine, Somerset district, according to W. A. Bear, Bear Coal Co. (oral commun., 1974). Mining officials oftheU.S. Geological Survey initially proposed that rooms and crosscuts be driven 16 feet (4.9 m) wide on 50-foot (15.2 m) and 60-foot (18.3 m) centers, respectively, in a block of coal in the “C” coal bed in order to minimize stress effects to the underlying “B” coal bed, which was controlled by another company. Scientific counsel for the company controlling the “B” bed proposed an alternative procedure wherein rooms 16 feet (4.9 m) wide would be developed on 84-foot (25.6 m) centers to prohibit the possibility of a squeeze. Then a 10-foot (3.05 m) slab of coal would be cut off two adjacent sides of each mine pillar on the retreat, thereby widening all mine openings to 26 feet (7.93 m) and, at the same time, reducing all pillars to 58 by 58 feet (17.7 x 17.7 m) (C. T. Holland, oral commun., 1971). This procedure, which was subsequently approved by mining officials of the U.S. Geological Survey, leaves uniform support pillars that diffuse overburden stresses and minimize stress effects to another coal bed, while at the same time yielding a reasonably high percentage (50 percent) for multiple-bed coal mining. A new modified longwall mining procedure is under development by Eastern Associated Coal Corp. (R. W. Thomas, vice president in charge of mine planning, oral commun., 1973) in one of their mines where changing geologic conditions make conventional longwall mining unsuitable. Basically, the procedure involves using a continuous miner to cut a longwall face 150-300 feet (45.7-91.5 m) long beneath self-advancing, cantilevered, hydraulic roof-support machines. This method takes advantage of both the safety, economy, and productivity of the longwall method and the ability of the continuous mining machine to adapt to varying coal thicknesses and other changing geologic conditions. Using this procedure, the chain pillars probably can be mined in sequence with the longwall face advance, making it possible to produce a uniform final mine geometry. Of course, a decision must be made as to whether complete extraction is feasible and is in accordance with applicable mining laws in a particular mining area before this procedure is implemented. Subsidence rupture can be damaging to mining operations as well as to the overburden and surface environment. In some cases, particularly in sparsely settled areas, coal mines may be threatened more than the surface environment by subsidence rupture because of the threat of the sudden intrusion of methane gas or large volumes of water. This is particularly true in geologic environments where water and methane are abundant. Mining companies should, therefore, balance production with ongoing research on subsidence and other types of mine deformation for their own benefit as well as in the interests of conserving coal and protecting the environment. Accurate structural and lithologic maps with overlays showing mine workings and overburden configuration relative to the coal bed or beds will prove very useful in planning new mines or expanding existing ones. Periodic subsidence measurements designed to determine subsidence parameters, such as limit angle, break angle, and transient and stable compression-arch configurations, can yield subsidence parameters, which in turn can help identify areas of potential stress concentrations and attendant deformations that might occur near or within the mine workings. This information, in turn, could produce tangible returns in extraction efficiency and mine safety. It is evident that mine safety and efficiency and the effects of mining on the physical environment are interlocked with subsidence and other facets of mine deformation, such as roof falls, bumps, and squeezes. It also follows that mine design should be based not38 ENGINEERING FACTORS, COAL MINE SUBSIDENCE, UTAH AND COLORADO only on expected subsidence damage above a retreating room-and-pillar line or longwall face, as Ropski and Lama (1973, p. 109- 118) recognized, but also on subsidence effects above barriers, solid-coal mine boundaries, and any other uneven geometry called for in the mining plans. In the case of multiple-bed coal mining, the overall success, in terms of mine safety, coal extraction ratios, and environmental protection, must be measured by the overall success of the multiple-bed mining operation, rather than by the overall successes of mining one bed. It also is evident that knowledge of overburden thickness, geology, and the environment are particularly critical to planning a multiple-bed mining operation that will produce a maximum amount of coal with a minimum risk to life and property. The sequence of mining coal beds also is very important because of the possibility of causing stress concentrations. Stresses produced in the strata by mining a coal bed will affect any coal beds present in overburden and, to a lesser but important extent, in the subjacent strata. A sequence of mining beds from top to bottom normally is safer and more efficient, paricularly if the final geometry is uniform so that stresses are not concentrated in isolated pillars and barriers and manifested downward to underlying coal beds; however, if the stratigraphic interval between two coal beds is generally 25 - 50 feet (7.62 - 15.24 m) thick, concurrent, partial, uniform extraction of both beds may be the safest and most efficient method. Also, a sequence of mining from bottom to top might be less hazardous and more productive in certain areas where water or methane is abundant because mine voids in upper beds may store large amounts of water or methane that could later be a hazard to life and property if tapped by subsidence fractures induced by mining in lower beds. In short, the safest and most efficient method and sequence of mining depends on all the geologic and environmental factors present in the area. CONCLUSIONS Analysis of subsidence processes and their effects upon underground coal mining lends a new perspective to the effects of mining on safety, conservation of coal resources, and protection of the ground surface above mine workings. The studies show that the mine workings, the strata above and beneath the mine workings, and the ground surface form a delicately balanced system of often irreversible causes and effects that can be seriously affected by subsidence. The excavation of underground mine workings can set in motion an interrelated chain of environmental and mining problems that can not only affect the manmade and natural elements of the environment but also threaten the mining operation. The studies clearly reveal that the geology of the coal beds and overburden strata in a proposed mining area should be accurately known in advance of mine planning so that the mine can be designed in harmony with the physical surroundings. Such an approach is highly desirable in planning mines in one coal bed, but it is vital in multiple-bed mine planning if hazards to life, property, and the environment are to be reduced to a minimum. One of the key aspects of mine planning is to weigh the various mining, geologic, and environmental factors in order to insure that the balance between coal recovery and subsidence effects is responsive to current and future needs for both coal resources and protection of the environment in the area. It should be recognized that a timelag of months or even years probably will occur between the following events: (1) recognizing the need for subsidence research, (2) implementing and performing subsidence research, and (3) applying the results of research studies to subsidence control and formulating subsidence-control legislation. Subsidence-control regulations should be flexible enough to change as national needs change. Historically, mining legislation has been inflexible to changing demands for coal production and environmental protection. For example, if the land surfaces were more valuable than underlying coal deposits at the time subsidence control legislation was enacted, the requirements for preservation of the land surface normally would be more stringent than requirements for protecting underlying coal resources. It is clear, from recent energy shortages, that provisions for flexibility should be written into future mining regulations so that we can maintain a judicious blend of coal-extraction efficiency, mine safety, and environmental protection, as the changing times require. REFERENCES CITED Brauner, Gerhard, 1973, Subsidence due to underground mining, pt. 2, Ground movements and mining damage: U.S. Bur. Mines Inf. Circ. 8572, 53 p. Briggs, H., 1929, Mining subsidence: London, E. Arnold and Co., 215 p. Dunrud, C. R., and Barnes, B. K., 1972, Engineering geologic map of the Geneva Mine area, Carbon and Emery Counties, Utah: U.S. Geol. Survey Misc. Geol. Inv. Map 1—704, 2 sheets. Garner, J. H. B., 1974, Death of woody ornamentals associated with leaking natural gas: Highway Research Abs., v. 44, no. 3, p. 5.REFERENCES CITED 39 Grard, C., 1969, Les Affaissements Miniers et les Moyens Permettant de Limiter Leurs Effets 3 la Surface du Sol: Paris, Rev. de l’lndustrie Minerale, v. 51, no. 1, p. 35- 70. Grond, G. J. A., 1947, Over Ontspanningsverschynselen in het Gebergte by Mynbouw, belicht uit het Standpunt van de Mynmeter [On stress phenomena in rock due to mining, as viewed by the surveyor]—A symposium in Heerlen [Holland], June 1947: Geologie en Mijnbouw, v. 10, no. 10, p. 240-291 [1948]. Kaneshige, Osamu, 1971, The underground excavation to avoid subsidence damage to existing structures in Japan, in Symposium [on] geological and geographical problems of areas of high population density, Washington, D.C., 1970, Proc.: Sacramento, Calif., Assoc. Eng. Geologists, p. 169—199. Lehmann, K., 1919, Bewegungsvorgange bei der Bildung von Pingen und Trogen: Gliickauf [Essen, Germany], v. 55, no. 48, p. 933-942. Lipman, P. W., Mutschler, F. E., Bryant, Bruce, and Steven, T. A., 1969, Similarity of Cenozoic igneous activity in the San Juan and Elk Mountains, Colorado, and its regional significance, in Geological Survey Research 1969: U.S. Geol. Survey Prof. Paper 650-D, p. D33-D42. Mohr, H. F., 1956, Influence of mining on strata: Mine and Quarry Eng., v. 22, no. 4, p. 140-152. Risser, Hubert, 1973, Fuel and energy dilemma: California Geology, v. 25, no. 12, p. 281- 283. Ropski, St., and Lama, R. D., 1973, Subsidence in the near-vicinity of a longwall face: Internat. Jour. Rock Mech. and Mining Sci., v. 10, no. 2, p. 105-118. Shoemaker, R. P., 1948, A review of rock pressure problems: Am. Inst. Mining and Metall. Engineers Tech. Pub. 2495, 14 p. Stemple, D. T., 1956, A study of problems encountered in multiple-seam coal mining in the eastern United States: Virginia Polytech. Inst., Eng. Expt. Sta. Ser. 107, v. 49, no. 5, 61 p. Wardell, K., 1971, The effects of mineral and other underground excavations on the overlying ground surface, in Symposium [on] geological and geographical problems of areas of high population density, Washington, D.C., 1970, Proc.: Sacramento, Calif., Assoc. Eng. Geologists, p. 201-217. Zwartendyk, Jan, 1971, Economic aspects of surface subsidence resulting from underground mineral exploitation: Pennsylvania State Univ. Ph. D. thesis, 411 p. . ' se* * w mmmm m •ff w » * V ' ^