7DAY ' Theory of Error in . Geochemlcal Data u: . GEOLOGICAL SURVEY PROFESSIONAL PAPER 574—A r K F t? ,. Theory of Error in Geochemical Data By A. T. MIESCH STATISTICAL STUDIES IN FIELD GEOCHEMISTRY GEOLOGICAL SURVEY PROFESSIONAL PAPER 574—A 14 classification of geocflemical errors am! discmsim of tfleir ejrects in data interpretation UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON : 1967 UNITED STATES DEPARTMENT OF THE INTERIOR STEWART L. UDALL, Secretary GEOLOGICAL SURVEY William T. Pecora, Director For sale by the Superintendent of Documents, US. Government Printing Office Washington, D.C. 20402 - Price 20 cents (paper cover) CONTENTS Page Page Abstract ___________________________________________ A1 Special models for geochemical sampling ............... A9 Introduction _______________________________________ 1 Analytical error ____________________________________ 10 Basic concepts ______________________________________ 2 Effects of error types in statistical analysis _____________ 13 The population _________________________________ 2 Analysis of variance _____________________________ 14 Restrictions due to limited access ______________ 2 Trend analysis _________________________________ 14 Other basic concepts ____________________________ 4 Summary __________________________________________ 15 Mathematical models and sampling error ______________ 5 Literature cited _____________________________________ 16 ILLUSTRATIONS ‘ Page FIGURE 1. Diagram showing directions of stepwise inferences from Specimens to the target population ___________________ A3 2. Diagram showing directions of stepwise inferences from the specimens to the target population where sampling localities comprise subpopulations .................................................................. 6 3. Graphs of analytical determinations on prepared standards of detergent in water ............................ 12 111 28055 4r-r—‘v 44‘ ‘ ' STATISTICAL STUDIES IN FIELD GEOCHEMISTRY THEORY OF ERROR IN GEOCHEMICAL DATA By A. T. MIESCH ABSTRACT All geochemical data accumulated in field studies directed at the determination of spatial variation in rock bodies contain error due to both sampling and analysis. Sampling error asso- ciated with a rock specimen is the compositional difference between the specimen and the part of the rock body that the specimen is intended to represent. Analytical error is the difference between the analysis and the concentration of the constituent in the total specimen. Error that is unbiased and of about equal variance from one sampling locality to another may not greatly hinder attempts at data interpretation; addi- tional data will always serve to reduce the error efi’ects. How- ever, either sampling or analytical error may contain an overall bias (when the mean error does not tend toward zero), a variable bias (when bias is variable from one sampling locality to another), or variable precision (when the variance of the error differs from one locality to another). Errors of these types may significantly affect interpretation, and their relative importance is about the same whether the methods of interpretation are based on formal statistical procedures or on intuitive judgment. Some types of error can commonly be avoided or controlled by taking suitable precautions in the field and laboratory. But where errors result from conditions of the rock body and its outcrop or from an inherent shortcoming of an analytical tech- nique that must be used, they are unavoidable. Detrimental effects of some error types may be overcome by suitable data transformation prior to interpretation. Other types of error, if of sufficient magnitude, may invalidate any attempt at inter- pretation. INTRODUCTION Geochemical field studies are highly varied in scope and objectives, but they are most commonly designed to describe and interpret the compositional variation within bodies of rock, soil, or alluvium, or the composi- tional variation among different rock bodies. The distinction between these two types of problems is not necessary to a discussion of error, and here we need consider only the variation within a single rock body. Depending on how it is defined, this rock body may or may not contain sharp contacts or discontinuities that would allow its division into multiple rock bodies. Any geochemical field study of this kind involves the problems of obtaining specimens to represent larger bodies of rock and of obtaining analyses to represent the specimens. Thus, two main kinds of error—that due to sampling and that due to analysis—are always present, even though the importance of each may vary greatly from one study to another. Both the type and the magnitude of error should be considered; some types of error need not seriously affect data interpreta— tion even if they are large. Other types can never be overcome and, if of sufficient magnitude, have serious effects on data interpretation. A classification of types of error important in geo- chemical prospecting was described briefly in a previous paper (Miesch, 1964). The same classification as it applies to a broader class of geochemical field problems is presented here, along with some basic concepts that may be essential to understanding the nature of error. Geologists have always employed most of the concepts, but some geologists express them in terms of a rigid statistical framework within which experiments in field geochemistry are carried out, whereas other geologists View the statistical framework informally and conduct their experiments according to intuitive judgment. Actually, the concepts employed by the two groups are virtually the same. Moreover, there are few require- ments demanded of data, and of data errors, for rigid statistical analysis that are not demanded otherwise, and it can be shown that sampling and analytical errors similarly affect statistical and nonstatistical interpreta- tions of the data. The concepts and mathematical frameworks used in statistical experiments, and the requirements and assumptions regarding data errors, are not so different from those most geologists have been employing for many years Without formal statistical analysis. The principal theme of this report is in support of the notions presented in the preceding paragraph. The report aims to show that adoption of rigid ex— perimental design techniques in field geochemistry does not necessarily require radical departure from methods and concepts now being used, but, instead, requires mainly a greater awareness of things we already know. Statistical principles are commonly stated for A1 A2 the general case, and their application to individual problems is not always obvious. It is hoped that this discussion will be of some help in interpreting a few basic statistical principles in terms of the particular situations in field geochemistry. To avoid making a subject that is already seemingly complex even more so, the geologic and analytical problems used as examples will be kept simple, perhaps even trivial. The examples, however, will be ones that are familiar to nearly every field geochemist or petrologist. Geochemical errors in general are departures of determined values from true or expected values or from values one is attempting to estimate; but such a departure does not necessarily imply that a mistake has been made in sampling or analysis. According to definition (Webster’s Seventh New Collegiate Dic- tionary, 1963, p. 282), the term “error” “* * * may suggest an inaccuracy where accuracy is impossible.” Error may result from mistakes or blunders in relatively rare cases. Thus, the degree or error in a particular geochemical study may be controlled by the nature of the study itself, and not necessarily by the care with which the study is conducted. Two geochemical studies that are identical in objectives and are con- ducted with identical degree of precaution in sampling and analysis may produce difierent types and magni- tudes of error, depending on the natures of the two areas or rock bodies involved or on the methods of analysis used. In studies directed at rock bodies that are highly variable in composition, more sampling error must be expected than would be met in studies of relatively homogeneous rocks. Where rapid low- cost analytical methods are used, the errors from analysis are commonly greater than Where more elaborate methods are used, but the high precision of the elaborate methods may be unnecessary, and of little advantage, in studies Where sampling error is large. Acknowledgments—I am grateful to Dr. W. C. Krumbein, Department of Geology, Northwestern University, and to Dr. R. F. Link, Department of Mathematics, Princeton University, for valuable criti- cism of an early draft of the manuscript. BASIC CONCEPTS THE POPULATION Geochemical sampling is normally directed at rock bodies, or parts of them, whether they are granitic plutons, selected stratified layers, covers of soil or alluvium, or whatever is of geologic interest. The boundaries of some rock bodies are distinct and well defined. The boundaries of others are arbitrary and may or may not be well defined. Nevertheless, the rock body is a single unit and cannot, while considered STATISTICAL STUDIES IN FIELD GEOCHEMISTRY in this sense, be regarded as a geologic population; moreover, a rock body in this sense has a bulk com- position but has no mean composition or composi- tional variance until it is conceptually divided into parts. In studying a rock body it is convenient for the geologist to collect specimens or samples, and the population is frequently regarded as all the potential specimens that the rock body contains. The fact that some potential specimens cannot be obtained has led to the concepts of target and sampled populations introduced by Cochran, Mosteller, and Tukey (1954) and used by Krumbein (1960, p. 351). Characteristics of the population of specimens, such as the variance, may largely depend on specimen size. In a study of uranium ores from the Colorado Plateau, Shoemaker, Miesch, Newman, and Riley (1959) used samples split from crushed and homogenized ore ship- ments aggregating at least several tons. The compo- sitional variation among these samples was less than might have been found in samples cut from only a few pounds. If the samples were so small that each in- cluded only a few mineral grains, the amount and nature of the variation from one sample to another, again, would be entirely different. Laffitte (1957, p. 101—105) discussed efliciency of sample size in rela- tion to size of constituent mineral grains. The geologic population in geochemical sampling, then, is determined by the geologist when he decides on the target of his investigation. Commonly the population consists of all possible specimens of about a few pounds’ size that occur within a rock body. Be- cause most rock bodies under study are large, and because possible specimens can overlap, the population is, for all practical purposes, infinite in size. RESTRICTIONS DUE TO LIMITED ACCESS Although the geologic population of interest in most geochemical field studies comprises all possible speci- mens occurring within a rock body, sampling is generally limited to parts of the rock body that are exposed at the surface. In some work, mostly that bearing on resource appraisal, deep drilling can be conducted in such a way that the entire rock body is theoretically accessible for sampling; in other work, shallow drilling or trenching may furnish samples from within a few feet of the surface. Except where drilling or trenching makes the entire rock body accessible for sampling, there may be a large difference between the composition of the target population (the population of interest) and the sampled population (the population that can be sampled). The difference is due to both original compositional variation in the rock body and secondary variation caused by weathering and other related processes. THEORY OF ERROR IN GEOCHEMICAL DATA A3 Sampling limitations frequently necessitate that one or a series of indirect inferences about the target pop- ulation be made from the collected specimens. The inferences may proceed in stepwise fashion, as illus- trated in figure 1. The correctness of the inferences may partly depend on the degree to which the sampled population corresponds to the target population; the sampled population, in turn, may be determined by the facilities available for sampling. If facilities such as deep—drilling equipment are avail- able, making the entire rock body accessible for sam- pling, the inference from the specimens to the target population may be direct and straightforward. Objec- tive schemes of drilling and of sampling the drill core or cuttings would be especially helpful in validating the inference. Where shallow-drilling devices or trenching equip- ment is available, the sampled population may consist of all possible specimens from within a few feet of the surface (A, fig. 1). Inferences from the specimens to this sampled population may be direct and straightfor- ward, but the additional inference from the sampled population to the target population is more difficult. The validity of this second inference depends on the nature of the rock body being studied, especially on such factors as compositional zoning and depth of weathering. If the sampling devices are used only to obtain speci- mens from the outcropping parts of the rock body (such devices are necessary, for example, in sampling a smooth face of dolomite or ganite that cannot be sampled sat- isfactorily with a hammer and chisel; see Baird and others, 1964, p. 258), an additional indirect inference may be required to interpret the nature of the rock body from the nature of the specimens (B, fig. 1). One must infer that the outcropping part of the rock body adequately represents the part of the rock body Target population: near the surface, and that this part, in turn, adequately represents the rock body as a whole. In most geochemical field problems the only available sampling devices are a hammer and chisel or similar tools. (Most rock specimens are collected by means of only a geologic pick and hammer.. However, the portion of an outcrop that can be sampled with an ordinary pick and hammer is commonly greatly limited in comparison with that which can be sampled using a 2-pound sledge hammer and a heavy steel cold chisel. Thus, bias in sampling may be significantly reduced by using a chisel for sampling many types of rock.) Consequently, depending on the type of rock being studied, not all parts of the outcrop can be sampled. Specimens are generally taken only where the rock protrudes, as along bedding planes or fractures. This is especially true where the rock is hard and dense, as are many quartzites, dolomites, and granites. Where a hammer and chisel are used as sampling devices, the sampled population consists only of specimens that can be collected with this equipment (0, fig. 1), and the inferences to be made about the target population from the collected specimens, with few exceptions, are subject to severe limitations. These few examples illustrate the kinds of problems met in inferring the composition and compositional variation of an entire rock body from specimens collected with limited sampling access, but the problems are similar to those commonly encountered. The difficulty of inferring from the sampled populations (A, B, and 0, fig. 1) to the target population, as defined in figure 1, may discourage one from such attempts. In some specialized geochemical problems the target population is not the potential specimens from the entire rock body but only those near or at the surface. This is true, for example, in studies directed at envi- ronmental health problems, because it is primarily the Sampled populations: A B C All possible specimens from entire rock body, or from se- lected part of the rock body All possible specimens from within several feet of the surface All possible specimens from outcropping parts All possible specimens that are obtainable with a hammer and chisel T Total access for sam- pling (by deep-drill- ing equipment) Limited access for sampling(by shallow- drilling or trenching equipment) L Limited access for sampling Target subpopulation 2 All possible specimens from sampling locality 2 Sampled subpopulation 1 All specimens that are acces- sible within locality 1 Sampled subpopulation 2 All specimens that are acces- sible within locality 2 l Limited access for sampling l Limited access for sampling Total access for sampling Total access for sampling I Specimens from locality 1 I I Specimens from locality 2 J FIGURE 2.——Directions of stepwise inferences from the specimens to the target population where sampling localities comprise subpopulations. THEORY OF ERROR IN GEOCHEMICAL DATA mens in the total rock body. The U, component in equation 3 is now represented by two components, 6, and am, Where )8, is the deviation of the ith locality mean from the grand mean, p, and w“ is the deviation of the jth sample from the ith locality mean. The value 11+ )6, is the true mean concentration of a constituent in the ith locality (the ith subpopulation), and the quantity u—I—Bi—l—w” is the true concentration in the jth specimen from the ith locality. The w” component is the difference between the concentration in the specimen (u-l-firfwu) and the true mean for the locality (n+3) As the purpose of the jth specimen from the ith locality is to represent the mean of the ith locality (u—Hii), the component w“ may be regarded as the sampling error associated with the specimen, and the mean of all the w” components associated with speci- mens from the ith locality is the sampling error asso- ciated with the locality. Thus, sampling error is caused by local variation within the rock body. Rela- tively homogeneous rock bodies can be sparsely sampled with little chance of large sampling error, whereas highly variable rock bodies require taking a large number of specimens to represent them adequately. The w“ component associated with the specimen or the mean 0.3,, component associated with a sampling locality is never known in practice. Its computation would require prior knowledge of p—l—Bi, the true mean for the locality, and this, in turn, would require com- plete sampling of the locality and analysis of the specimens (determination of X”) by a perfectly accurate and precise technique. Nevertheless, we must make judgments about on, prior to any sort of interpretation of geochemical data. Four aspects of the a)” components are discussed next. Three are of particular significance, regardless of whether the interpretation is statistical or intuitive. The fourth is important mainly where the statistical methods used involve certain kinds of probability theory. First aspect The grand mean of the sampling error components (am) over all specimens and localities should tend to- ward zero as the number of localities, m, and of speci- mens per locality, n, is increased. That is, (5) 1 m n _ Z 2 “1.1%0; as my ”600' mn 5:1 1:1 If equation 5 is not true, an overall bias is present in the data. Such bias may occur when the total rock body (the target population), or the total sampling locality (the target subpopulation), is not accessible for sampling, or when sampling is not done according to some objective operational procedure (Krumbein, A7 1960, p. 351). In the latter instance the selection of samples may involve personal bias on the part of the sampler. Personal bias may be easily avoided, but bias caused by limited accessibility of the rock body—— such as lack of outcrop—is more difficult to overcome. For example, parts of the rock body rich or deficient in some constituent may tend to be either covered or to crop out as smooth surfaces that are difficult to sample with a hammer and chisel, as previously discussed. In this circumstance the geologist may resort to another sampling device, such as a portable drill. If drilling or other sampling devices are not practical, and the out- cropping parts of the rock body are unlike the total sampling locality, the inference from the sampled sub- population to the target subpopulation (fig. 2) may be difficult. Nevertheless, this is the common situation, and consideration of possible differences between the sampled and target populations is required in most field studies. Where the target subpopulation, however defined, is completely accessible for sampling, objective sampling procedures (commonly involving some randomization in sample selection) can reduce or eliminate the pos- sibility of personal bias. Overall bias in sampling, if of sufficient magnitude, may seriously affect the estimation of means, but it will not hamper attempts to describe variation among sampling localities. Unless the bias differs from one locality to another, as discussed under aspect 2, the errors in locality means tend to be constant, and esti- mates of differences among means tend to be correct. Second aspect The mean of the sampling—error components should tend toward the same value for each locality as the number of specimens per locality is increased. That is, 1 m 1 m 1 m — Zwu+ 2w21—’ - - - —‘ 260112315711, "12w” ”1"“, n1 j=1 n2 j= 7% i=1 (6) where n, is the number of specimens from the ith locality. Where equation 6 is not true, variable bias is present. If each term in equation 6 tends toward zero, neither variable bias nor overall bias is present. Variable bias in sampling, or bias that varies from one sampling locality to another, may result from differences in the circumstances under which the sampling is done. This can be caused, for example, when localities are sampled by different individuals having different personal biases, or when the outcrop characteristics vary among localities. The problem of personal bias is easy to resolve, but the problem caused by differences in outcrop characteristics is more difficult. For example, all localities within the rock body may actually have the same total composi- A8 tion, but parts rich in, say, calcium carbonate may tend to crop out at one locality and to be covered at another. Even objective sampling of the exposures at each local- ity, then, may result in maps showing regional varia- tion caused perhaps by climatic differences. Variable bias is often unavoidable, and it is therefore necessary to consider all the factors (even those which are nongeo- logical) that may be important in the interpretation of the experimental results. Variable sampling bias, however, can never be completely overcome in data interpretation, and if it is of sufficient magnitude it may seriously hamper attempts to describe regional varia- tion in the rock body. Variable bias, however caused, may be the most serious type of error that can be present in geochemical data. Third aspect The variation, or precision, of the sampling error, conveniently expressed as the variance, should tend to be the same for each locality as the number of specimens per locality is increased. That is, 1 IV: (w —' ra—l— i (w -—' )2» l 7“ _ —>— Z (mu—my, as n1, n2, . . . raj—>00, (7) 7111—1 j:1 where E, is the mean sampling error component of the ith locality. Using 83,, for the variance of a)” at the ith locality, equation 7 may be written as: 82198E,Z—> . . . 31,, as n1, n2, . . . 12,—)00. (8) If equation 8 is not true, variable precision in sampling is present. The conditions defined by equations 7 and 8, unlike those defined by equations 5 and 6, may be tested from the sampling results if analytical errors are sufficiently small to insure that sampling precision, rather than analytical precision, is being observed. The ability to test for the relation in equations 7 and 8 results from the fact that the quantity cow—51 for the ith locality is equal to XU—Ei. Therefore, equa- tions 7 and 8 are identical with: (9) 2 2 2 8x,—>sX2—> . . . 8,5,, as nhnz, . . .n,——>oo, where s?“ is the variance of a constituent in samples from the 13th locality. If all samples from a particular locality have about the same composition, those locality means estimated from repeated sampling experiments will also be closely similar, or reproducible. If, on the other hand, the samples differ widely in composition, the means from repeated experiments, each involving a small number of samples, will also differ widely. The expected variance STATISTICAL STUDIES IN FIELD GEOCHEMISTRY of means may be estimated for a single set of samples by use of the statistical relation: 2. =3; (.0) X1 n, That is, the expected variance of means for the 73th locality (the degree to which a locality mean may be expected to vary in repeated experiments) is 1/72,,- of the variance among the samples from the locality. If objective sampling is employed at all the sam- pling localities, the variance among collected specimens is controlled by the nature of the rock body (its local variability) and its outcrop characteristics. Vari- able precision, then, is caused by variation of these factors from one locality to another. For example, samples from one locality may be widely different in composition because the rock body there is highly variable and completely exposed for sampling, whereas samples from another locality may be virtually identi- cal because the rock body is relatively homogeneous or only a relatively homogeneous part of it is exposed. Variable sampling precision may commonly occur in minor-element data because the variation of minor- element concentrations in rocks tends to be propor- tional to their mean concentration. Where the means vary widely among localities, the variation, and there— fore the sampling precision, may also vary widely. The significance of deviations from the mean is difficult to evaluate where sampling precision is vari- able. Locality means are estimated with varying degrees of confidence or reliability, and a minor devia- tion of one locality mean from the grand mean for the rock body may be geologically significant, Whereas a major deviation for another locality mean may result only from sampling error. The relative statis- tical significance and possible geologic implications of deviations are not easily determined from their magnitude. Variable precision may be overcome, in some exper- iments, by using suitable data transformations. For example, if variation within localities tends to be proportional to the locality means, the variation of the logarithms of the concentrations tends to be constant among the localities. Transformation of the data to logarithms, in this circumstance, would be a valid and effective means of overcoming variable precision. Transformations must be used cautiously, however, as complicating factors may thereby be introduced into the experiment. Link, Koch, and Gladfelter (1964, p. 35) referred to the difficulties that had to be overcome by Sichel (1952) and Krige (1960) in estimation problems where the data required loga- rithmic transformation. THEORY OF ERROR IN Variable precision of means among localities can be overcome by collecting different numbers of speci- mens at the several localities. This is equivalent to adjusting the values of n, in equatiOn 10 so that the value of 8;, tends to be the same across localities. The adjustment of n, values would have to be made on the basis of preliminary estimates of 8,”, obtained from a pilot sampling survey (Miesch and Connor, 1964). , Fourth aspect The sampling errors (the w” components of equation 4) for any particular sampling locality should tend to be normally distributed; that is, the frequency dis- tribution of a)” should approximate the bell-shaped normal, or Gaussian, curve. This requirement, as that in aspect 3, can be tested by using the observed data, X”, if analytical error is relatively small. The frequency distribution of the data, X”, for the 71th locality will have the same form as that of the theoret- ical component, 0.1”. Frequency distributions that are highly asymmetri- cal are less easily interpreted than symmetrical dis— tributions, regardless of the method of interpretation; but only in methods based on certain probability theory must the frequency distributions be normal (Gaussian). hdost of these methods, however, allow at least a moderate, and commonly a large, departure from normalcy. SPECIAL MODELS FOR GEOCHEMICAL SAMPLING The geochemical sampling model given in equation 4 is applicable to the general case where the objective of the field study is to detect and describe the variation among parts of the rock body larger than a single sample—in other words, among sampling localities. These localities are commonly about evenly spaced over the rock body, or over the part of the rock body that is of interest, and are referred to as the major sampling localities. More specialized models have been employed in geochemical sampling by Krumbein and Slack (1956) and by Baird, McIntyre, Welday, and Madlem (1964), and models similar to theirs are currently being used in several geochemical investigations by the US Geological Survey. The purpose of each such investi- gation is to determine the type of variation present in a rock body and to devise an efficient sampling plan based on the type of variation found. Krumbein and Slack, for example, measured radioactivity in a shale bed associated with a cyclothem deposit over a large part of southwestern Illinois. In the regional phase of their study they collected and grouped samples by source as within mines, selected mines within GEOCHEMICAL DATA A9 townships, and townships within supertownships. Their experimental model (Krumbein and Slack, 1956, p. 754) is given as: Xijkm= Il+011+l3u+7uk+5ukm (11) where Xmm is the radioactivity of the mth sample (1 Smgd) from the kth mine (lgkgc) in the jth township (1 oo. 19m {:1 j=1 (17) If equation 17 is not true, overall analytical bias is present. This may result from a large number of causes (including sample contamination) that may occur at any time from when the sample was collected to when the analytical determination is received by the geologist. (This all—inclusive notion of analytical error is not fair to the analyst, but errors due to sample handling and preparation must be accounted for some— where. The part of the analytical error that is the responsibility of the analyst may be referred to as laboratory error when such subdivision is necessary.) There are numerous ways in which overall bias and other types of error could occur in analysis. Many are well known to chemists and other analysts, and examples could be given with reference to any par- ticular analytical method. However, it may be useful here to refer to the well-known method of estimating uranium content by measurement of radioactivity. In the commonly used laboratory method a small pow- dered rock specimen is placed in a lead chamber con- taining radioactivity counters of various types; the lead chamber eliminates nearly all radioactivity other than that from the specimen itself. Results of the radioactivity measurements are commonly expressed in units of equivalent uranium (percentage eU), where equivalent uranium is defined as the ratio of the net counting rate of a specimen to the net counting rate per percentage of a uranium standard in equilibrium with all of its disintegration products, both measured under similar geometry (Rosholt, 1954, p. 1307). Viewed as a measurement of radioactivity, the eU analysis is generally unbiased. As a method of uranium analysis, its accuracy depends on two main assumptions: 1. That all radioactivity in the specimen is from ura- nium and its radioactive decay products. 2. That uranium and all its radioactive decay prod- ucts are in equilibrium within the specimen; that is, each radioactive member of the uranium series is disintegrating within the specimen at the same rate at which it is being formed. The first assumption may be invalid when, for example, part of the radioactivity from the specimen arises from thorium or potassium, rather than from the A11 uranium series. The second assumption may be in— valid when radioactive decay products of the uranium series have been removed from the specimen or where they are present in excessive amounts. Either situ- ation can give rise to overall bias in uranium assays by the eU method. The mean of the assay error may tend toward some positive or negative value, rather than toward zero. Rosholt (1959) discussed this subject thoroughly. Overall bias in low-level radiometric uranium de- terminations was illustrated by Newman (1962, table 33). The data consist of fluorimetric uranium and eU analyses of 95 mudstone specimens. The fluorimetric analyses indicate that most of the specimens contain less than 10 parts per million uranium; the radiometric analyses are somewhat higher and, if the fluorimetric analyses are accepted as correct, have an average error of plus 13 parts per million. The bias is probably due to the presence of potassium (containing K“) in the analyzed specimens (Newman, 1962, table 33). Variable bias Variable bias is absent when the mean analytical error tends to be the same from one locality to another as the number of specimens per locality is increased. That is, 1 m 1 m 1 ’m —29u—>—‘ 2921‘—> - - - ‘20:» as ”1,712, - - Uni—9°“ n1 j=1 71/2 i=1 n: j=1 (18) If equation 18 is not true, variable analytical bias is present. Analytical bias that is variable from one sampling locality to another may result when the speci- mens from different localities are treated diflerently or when the localities’ characteristics that ‘afl’ect the analytical procedure vary. For example, if the analytical work extends over a long period of time, improvements in technique or analysts are likely to occur. If the specimens are treated locality by locality rather than in randomized sequence, the analytical bias associated with localities can easily vary. Variation in the chemi— cal or physical characteristics of sampling localities can also cause variable bias. Where the rock at some sampling localities contains large amounts of one con- stituent that interferes with the determination of another constituent, variable bias could occur if suit- able precautions were not taken. In some spectro- graphic procedures, for example, high concentrations of manganese may interfere with the determination of silver (Bastron and others, 1960, p. 179). If steps had not been taken to allow for the interference, biased determinations of silver would result only on specimens from localities where manganese concentrations are high. A12 Finnell, Franks, and Hubbard (1963, p. 46—49) discussed an excellent example of variable bias in eU analyses among sampling localities within uranium- copper deposits in San Juan County, Utah. Because of differential uranium leaching, specimens from weathered parts of the outcrop contain less uranium than the eU analysis would indicate, whereas un- weathered specimens from underground contain about the same amounts of uranium as indicated by the eU analyses. If the eU analyses had been viewed as ura- nium determinations rather than as measurements of radioactivity, the bias would have ranged from some high positive value at localities where weathered out- crop was present to some value near zero at unweathered localities. Variable bias from one sampling locality to another may also result when the amount of bias is related to the amount of the constituent present. In some types of colorimetric methods an analyst may tend to over- estimate low concentrations and underestimate high concentrations. As the amount of the constituent present 'can be expected to vary from one sampling locality to another, the bias related to the amount present also varies among localities. Bias related to the amount present may also result from incorrect standards or from the chemistry involved in the analysis itself. Examples of both overall bias and bias that varies with amount of constituent present were given by Wayman and Miesch (1965), with regard to field methods for determining the amounts of detergent in 1.0 — X; 0.8 - . /? ° / 0.6 — 3. ,/ 0.4 — /: :r’ ./ 0.2 — )x o 0 d2 014 d5 01.8 11.0 A. LABORATORY METHOD FIGURE 3.—Analytical determinations on prepared standards of detergent in water. STATISTICAL STUDIES IN FIELD GEOCHEMISTRY water. Analyses of prepared standards containing 0—1 part per million detergent (alkylbenzenesulfonate) were made by two analysts using three methods. Some of the results are given graphically in figure 3. Example A is from a standard laboratory technique; the mean of the deviations from the accepted values of the prepared standards is about zero at each level of concentration. Neither overall nor variable bias can be demonstrated. In example B, based on a rapid field method, the bias present is clearly related to concentration; it is small or absent at low and high concentrations but large (and negative) at intermediate concentrations. As con- centrations at different sampling localities may certainly vary when the technique is applied in a field problem, variable bias among the localities is certain to result. As the biases are not compensating, an overall negative bias will also be present. Varlable precision Variable precision in analysis is absent when the variance of the analytical error associated with speci- mens, 6”, tends to be the same from one sampling locality to another as the number of specimens per locality is increased. That is, silasge . . . 83,, as n1, n2, . . .niaw. (19) If equation 19 is not true, variable analytical precision is present in the data. This type of error may be expected when the samples from various localities are treated difierently or when characteristics of the localities that affect the analytical precision vary. ), 0.8- /i E / O / 06— / x // !° _ / 04 / : t a// .0. I 0-2— I? s / . 4. . o .. I J. I I I o 0.2 0.4 0.6 0.8 1.0 X B. FIELD METHOD X = concentration in standard; Y= measured concentration. Concentrations are given in parts per million. From Wayman and Miesch (1965). THEORY OF ERROR IN GEOCHEMICAL DATA The conditions under which variable precision is likely to occur are about the same as those that might lead to variable analytical bias. It seems, however, that variable analytical precision is much more common than variable bias, especially in studies directed at minor elements. The precision of minor-element determinations, as is well known, nearly always varies with the concentration. Fortunately, however, vari— able analytical precision is an easily recognized type of error and, because it is frequently systematic, can often be corrected by suitable data transformations (Bennett and Franklin, 1954, p. 356). Variable pre- cision may also be corrected by making a number of replicate analyses of each sample proportional to their variance—by adjusting p in equation 13 propor— tional to the variance of am. Variable precision of the analytical error, am in equation 12, from one specimen to another may be detected by computing the variance of analytical values, X,,,,, for each specimen and applying several statistical tests for variance homogeneity (see Bennett and Franklin, 1954, p. 196—200). The same tests may be used to detect variable precision of the total sampling and analytical error, e” in equation 16, among sampling localities. Frequency distribution The components of analytical error, 6,), should tend to have a normal distribution. The frequency distri- bution of the total experimental error for sampling localities, e” in equation 16, may be examined by constructing the frequency distribution of analytical values, Xu, representing the locality. If the distribu- tion departs Widely from the normal, it can commonly be corrected by suitable data transformations (Bennett and Franklin, 1954, p. 91). The requirement of a normal distribution is necessary only when probability theory is to be employed in data interpretation, although symmetrical frequency distributions are easier to interpret by any method. The frequency distribution of the analytical error alone is best determined from a large number of replicate analyses of the same specimen, or from a large number of estimates of am obtained by comparing analytical values With corresponding values known to be more correct. EFFECTS OF ERROR TYPES IN STATISTICAL ANALYSIS In the preceding discussion several types of sampling and analytical error have been described, and the effects of each error type in data interpretation are briefly discussed. The effects are the same whether the error is due to sampling or to analysis. Overall bias is present When the mean of the error components does not tend toward zero, and its effect is A13 merely to distort the estimate of the mean composition of the rock body. Unless the bias is variable from one sampling locality to another, the estimated locality means tend to be wrong by a constant amount, and estimates of differences among locality means tend to be correct. Bias that is variable from one sampling locality to another may have disastrous effects on data interpre- tation. As the bias for each locality varies and is unknown, determining the differences among localities—— the primary objective of most geochemical field sampling—may be impossible. Variable precision of sampling and analysis from one locality to another presents difficulties in judging both the statistical and geologic significance of differences but does not present as severe difficulties as does variable bias. The type of frequency distribution of errors is of even less concern in data interpretation, though symmetri- cally distributed error is easier to evaluate by both formal statistical or other methods. When some types of probability theory are used in data interpretation, however, the type of frequency distribution displayed by the error components should be considered. In formal statistical methods of data analysis, certain assumptions pertaining to the kinds of sampling and analytical error are required to make valid use of statistical models as given in previous parts of this paper. These models are the bases for formal statistical analysis of the data. However, it is a reasonable circumstance that the required assumptions regarding data errors are much the same for formal statistical analysis as for other methods of data analysis. More— over, the relative importance of error assumptions required for formal statistical methods is about the same as for other methods. The lone exception here is the importance of the normal distribution in statistical methods based on normal probability theory. The required assumptions regarding data errors are about the same for all statistical methods, even though the requirements of the data themselves, and the man- ner in which the data are collected, may differ. Multi- variate statistical methods involve some assumptions regarding data errors, such as uncorrelated errors among variables, that are not considered here; but the requirements of most statistical methods that deal with one estimated variable at a time can be illustrated by reviewing two methods that have been shown to be useful in geochemical problems. These methods are analysis of variance and trend-surface analysis (poly- nomial multiple regression). Applications of these methods to geochemical and petrologic problems are so numerous that citation of individual papers is unnecessary. A14 ~ STATISTICAL ANALYSIS OF VARIANCE Analysis of variance has been applied in geochemical problems for testing compositional diflerences for statistical significance and for partitioning the total variability of a constituent among several factors. Applications of the second type are efforts to ascribe variability to geologic or spatial factors in the correct proportions. The assumptions that are required for analysis of variance techniques in general were sum- marized by Eisenhart (1947). These assumptions, as stated for the general case, appear somewhat for- midable; viewed in terms of a particular problem, they are more easily understood. In a companion paper to Eisenhart’s, Cochran (1947) reviewed the effects of failure of the data to meet the requirements set forth by Eisenhart. Insofar as experimental error is concerned, Cochran assumed that all the errors have zero means, and he tabulated the other assumptions required as follows: 1. Errors must be independent. 2. Errors must have a common variance. 3. Errors should be normally distributed. If the assumption of zero means among the errors fails (overall bias is present), the analysis of variance may still be valid so long as the errors are independent, have a common variance, and are normally distributed. The grand mean and individual means for sampling localities will be biased, but the analysis of variance method is still valid for testing the significance of differences among them. The three error requirements will be discussed as applicable to analysis of variance based on the model given in equation 16. Errors lack independence when the amount of one error is related to that of another. This may occur in geochemical field sampling when the errors associated with specimens from one locality tend to be distributed about one value and those associated with specimens from another locality tend to be distributed about a different value. The errors, then, are related to the localities and lack independence. The errors associated with at least one of the localities, moreover, are biased, and the bias is variable among localities. The effect of variable bias, or dependent errors, in analysis of variance is that the additivity of variances is destroyed, as shown below in an example analogous to one given previously: 7: .7 Xi} = (“+30 + 6:7 1 1 1 2 -—-1 1 2 4 2 +2 2 1 6 4 ' +2 2 2 1 4 —3 Here the mean of the 6” values (—1 and +2) added STUDIES IN FIELD GEOCHEMISTRY to p+fii=2 is different from that of the eu values (+2 and ——3) added to u+fi,=4. The variance of p+B¢ (= 1%) plus the variance of e” (= 6) is equal to 7%, whereas that of X1, is 6. Because the additive character of variance is destroyed by variable bias, the analysis of variance method, based on models given in this paper, is invalid for estimating variance components and testing the significance of differences among sampling locality means. Mean square esi- mates derived in the analysis of variance procedure tend to be incorrect. Common variance among the errors is absent when the errors in one category of the experimental design have a different variance than those in another category. This occurs in geochemical field problems when the precision of sampling or analysis, measured by the variance, varies from one sampling locality to another. Variable precision prohibits valid estimation of mean squares in analysis of variance because the true com— ponent of error variance is not a single value, but a range of values. Data transformations that may be suitable for obtaining a common, or homogeneous, variance were reviewed by Bartlett (1947). The requirement of a normal distribution for ex- perimental errors does not affect the validity of the analysis of variance method for estimation of variance components, but it does affect tests for significance of differences among sampling locality means. The effects, however, are not severe, and non-normality is not regarded as a rigid requirement when F—tests are applied (Cochran, 1947, p. 24). The F—tests, because of this lack of high dependence on a normal distribution, are said to be “robust.” When the normal distribution assumption cannot be satisfied, nonparametric methods of statistical analysis may be useful (Siegel, 1956). Two-tailed t—tests for estimating the significance of difference between two means—a special application of analysis of variance—and methods for estimating the confidence interval of a mean based on the t- distribution are also “robust”; the normal-distribution requirement is not severe. On the other hand, methods based on the one-tailed t-distribution, as for comparing a mean with some critical value, lack “robustness,” and the normal-distribution requirement is more critical (Cochran, 1947, p. 24—25). The lack of “robust- ness” of Bartlett’s test for common, or homogeneous, variance is well known (Box, 1953). TREND ANALYSIS The method of trend analysis, as currently applied to geologic problems, is a form of multiple regression wherein polynomial or other mathematical surfaces are fitted to map data by least-squares techniques. Several THEORY OF ERROR IN GEOCHEMICAL DATA objectives can be achieved by this technique, but chief among these is the separation of total map variability into regional and local components. The essential fea- tures of the method, as applied in geology, were sum- marized by Krumbein (1959). Trend analysis may be viewed as an empirical tool or as a descriptive technique in geologic data analysis; as such, the data errors in- volved are not subject to unusual requirements. Over- all bias in the data merely affects the estimation of the constant term in the computed regression equation and the absolute values on the contours of the regression surface. Variable bias affects the regression coeffi— cients and the form of the fitted surface. Variable precision results in unequal reliability of the locality means described by the fitted surface. These effects are the same as would be present if the data were merely contoured. The frequency distribution of the error has no unique effect on either trend analysis or con— touring procedures. Testing the statistical significance of fitted trend surfaces, by F—tests as outlined in Bennett and Franklin (1954, p. 431—436), requires the same assumptions regarding data errors as listed previously. Other as— sumptions regarding the data and the mathematical function fitted to the data in trend analysis are beyond the scope of this paper. (See Link and others, 1964.) From the above examples it is apparent that data containing errors causing the data to be unsuitable for statistical analysis are probably unsuitable for any other rigid type of interpretation. When neither over- all bias, variable bias, nor variable precision of the errors is of large magnitude, the data, insofar as the error is concerned, are suitable for interpretation on either statistical or substantive grounds, although sta- tistical methods are more likely to overcome the effects of unbiased error. However, statistical methods have no greater power to overcome effects of overall and variable bias than have other methods. SUMMARY The entire purpose of statistical analysis in geochem— ical field problems is to evaluate and allow for the effects of error in the data in drawing geochemical conclusions. However, neither statistical nor other types of data analysis can be effective when certain kinds or error are present. Some kinds of error can be effectively reduced by collecting more data or by making data transforma- tions; but other kinds, if of sufficient magnitude, may ‘ persist and invalidate any type of data interpretation. The individual components of sampling and analyt- ical error are never known in a real field problem, though the variance of total experimental error and the form of its frequency distribution can be deter- mined. In spite of the fact that the error components A15 are unknown, we are faced with the necessity of making judgments about them prior to any sort of data inter- pretation. These judgments are necessarily based on knowledge of the sampling and analytical procedures and the circumstances under which they were carried out. No sampling or analytical procedure gives results that are perfectly precise, or reproducible; some imprecision is present in all observational data. Over- all bias, variable bias, and variable precision, however, may or may not be present in important amounts, and the presence of one is not necessarily related to the presence of another. In some studies the occurrence of these error types can be reduced or controlled by suitable precautions in the field and laboratory. In other studies, owing to the nature of the rock body and its outcrop characteristics or the nature of an analytical method that must be used, the occurrence of these error types in important amounts is beyond control. Overall bias may be controlled when it is due to misjudgment on the part of the sampler or analyst. The sampler may be persuaded to use an objective sampling technique, and the analyst may change his analytical methods. Overall bias may also result, however, under circumstances beyond the sampler’s or analyst’s control, as when the sampler is restricted to outcropping parts of the rock body and the outcrop is not representative of the rock body as a whole. Overall bias in analysis may be beyond control, for practical purposes, when a method that must be used, owing to cost or other factors, involves such short- comings as inadequate sample digestion or inadequate complexing or precipitation. Overall bias in analysis is subject to tests by analysis of prepared standards or by comparison with other methods accepted as more precise and accurate, but bias in sampling can only be inferred from knowledge of the field procedures and the nature of the rock body. Variable bias in sampling and analysis, where it is due to variation in the field and laboratory procedures, may be reduced by exercising precautions to keep the procedures uniform. However, variable bias may be caused by circumstances that are beyond control, as when outcrop characteristics differ from one sampling locality to another and when attributes of the rock that afiect analytical bias vary among localities. Variable sampling bias may only be inferred from knowledge of the field conditions and procedures, but variable analytical bias sometimes may be detected by analysis of prepared standards or by comparison with methods known to be more precise and accurate. Variable precision in sampling, like overall and vari- able sampling bias, will rarely be introduced by the sampler who uses some objective method of sample A16 selection. Variable precision resulting, however, from natural differences in the amount of local variation over different parts of the rock body is unavoidable. Variable sampling precision is also unavoidable when caused by the variable nature of the outcrop from one sampling locality to another, if only outcrops can be sampled. Analytical precision that varies with the amount of the constituent present is an inherent feature of most analytical methods for minor-element deter- minations and is generally unavoidable. Variable precision in total experimental error, that due to both sampling and analysis, is comparatively easy to detect in the analytical data. Experimental results which are affected by bias in sampling cannot be corrected; effects of analytical bias may be corrected in certain restricted instances where it is either constant or systematically variable (Youden, 1962). The effects of variable precision commonly may be overcome by suitable data transformations; these same transformations may also bring the frequency distribution closer to normal (Bartlett, 1947, p. 40). Variable precision among sampling locality means may be overcome by collecting additional specimens at the more variable localities. The efl’ects of overall bias in geochemical errors are usually not severe, unless the purpose of the investiga- tion is to estimate absolute mean concentrations of constituents to several significant figures. Variable bias, however, may seriously affect both statistical or intuitive interpretation of the data. Variable pre- cision or a non—normal frequency distribution of error components renders judgments of the relative signifi- cance of differences difficult and may prevent use of some useful statistical methods based on probability theory. Despite the difficulties that can and do arise in the field and in the laboratory, sound geological judgment in designing a study and use of objective sampling plans can do much to reduce their effects. The many valid inferences made in geochemical studies attest to the fact that in many instances the errors are not great enough to invalidate the findings of the study. N ever- theless, we cannot afford to ignore the implications of error. The different types of error that may be present must be recognized and the kinds of effects each may have on data interpretation should be known. Only through a clear understanding of these factors can we arrive at sampling designs and methods of data inter- pretation that are both consistent with the geologic problem under consideration and efficient with respect to the available field and laboratory resources. STATISTICAL STUDIES IN FIELD GEOCHEMISTRY LITERATURE CITED Baird, A. D., McIntyre, D. B., Welday, E. E., and Madlem, K. W., 1964, Chemical variations in a granitic pluton and its surrounding rocks: Science, v. 146, no. 3641, p. 258—259. Bartlett, M. S., 1947, The use of transformations: Biometrics, v. 3, no. 1, p. 39—52. Bastron, Harry, Barnett, P. R., and Murata, K. J., 1960, Method for the quantitative spectrochemical analysis of rocks, minerals, ores, and other materials by a powder D—C arc technique: U.S. Geo]. Survey Bull. 1084—G, p. 165—182. Bennett, C. A., and Franklin, N. L., 1954, Statistical analysis in ’chemistry and the chemical industry: New York, John Wiley & Sons, 724 p. Box, G. E. P., 1953, Non-normality and tests on variances: Biometrika, v. 40, p. 318—335. Cochran, W. G., 1947, Some consequences when the assumptions for the analysis of variance are not satisfied: Biometrics, v. 3, no. 1, p. 22—38. Cochran, W. G., Mosteller, F., and Tukey, J. W., 1954, Princi- ples of sampling: Am. Statistical Assoc. Jour., v. 49, p. 13—35. Eisenhart, Churchill, 1947, The assumptions underlying the analysis of variance: Biometrics, v. 3, no. 1, p. 1—21. Finnell, T. L., Franks, P. C., and Hubbard, H. A., 1963, Geology, ore deposits, and exploratory drilling in the Deer Flat area, White Canyon district, San Juan County, Utah: U.S. Geol. Survey Bull. 1132, 114 p. Krige, D. G., 1960, On the departure of ore value distributions from the lognormal model in South African gold mines: South African Inst. Mining and Metallurgy Jour., v. 61, no. 4, p. 231—244. Krumbein, W. C., 1959, Trend surface analysis of contour-type maps with irregular control-point spacings: Jour. Geophys. Research, v. 64, no. 7, p. 823—834. 1960, The “geological population” as a framework for analysing numerical data in geology: Liverpool and Man- chester Geol. Jour., v. 2, pt. 3, p. 341—368. Krumbein, W. G., and Slack, H. A., 1956, Statistical analysis of low-level radioactivity of Pennsylvania black fissile shale in Illinois: Geol. Soc. America Bu11., v. 67, no. 6, p. 739—762. Lafiitte, Pierre, 1957, Introduction a l’étude des roches méta- morphiques et des gites métalliféres: Paris, Masson et Compagnie, 343 p. Link, R. F., Koch, G. 8., Jr., and Gladfelter, G. W., 1964, Computer methods of fitting surfaces to assay and other data by regression analysis: U.S. Bur. Mines Rept. Inv. 6508, 69 p. Miesch, A. T., 1964, Effects of sampling and analytical error in geochemical prospecting, in Parks, G. A., ed., Computers in the mineral industries, pt. 1: Stanford Univ. Pubs. Geol. Sci., v. 9, no. 1, p. 156—170. Miesch, A. T., and Connor, J. J., 1964, Investigation of sa mpling- error effects in geochemical prospecting, in: Short papers in geology and hydrology: U.S. Geol. Survey Prof. Paper 475—D, p. D84—D88. Newman, W. L., 1962, Distribution of elements in sedimentary rocks of the Colorado Plateau—A preliminary report: U.S. Geol. Survey Bull. 1107—E, p. 337—445. Ostle, Bernard, 1963, Statistics in research: 2d ed., Ames, Iowa, Iowa State College Press, 585 p. THEORY OF ERROR IN GEOCHEMICAL DATA Rosholt, J. N., 1954, Quantitative radiochemical method for determination of major sources of natural radioactivity in ores and minerals: Anal. Chemistry, v. 26, no. 8, p. 1307— 1311. 1959, Natural radioactive disequilibrium of the uranium series: U.S. Geol. Survey Bull. 1084—A, p. 1—30. Shaw, D. M., 1961, Manipulation errors in geochemistry: A preliminary study: Royal Soc. Canada Trans, v. 55, ser. 3, sec. 4, p. 41—55. ' Shoemaker, E. M., Miesch, A. T., Newman, W. L., and Riley, L. B., 1959, Elemental composition of the sandstone-type deposits, in Garrels, R. M., and Larsen, E. 8., 3d, compilers, Geochemistry and mineralogy of the Colorado Plateau uranium ores: U.S. Geol. Survey Prof. Paper 320, p. 25—54. 0 A17 Sichel, H. S., 1952, New methods in the statistical evaluation of mine sampling data: London, Inst. Mining Metallurgy Trans., v. 61, p. 261—288. Siege], Sidney, 1956, Nonparametric statistics for the behavioral sciences: New York, McGraw—Hill Book Co., 312 p. Tippett, L. H. C., 1952, The methods of statistics: 4th ed., New York, Dover Pub. Inc., 395 p. Wayman, C. H., and Miesch, A. T., 1965, Accuracy and pre- cision of laboratory and field methods for the determination of detergents in water: Water Resources Research, v. 1, no. 4, p. 471—476. Youden, W. H., 1962, Accuracy of analytical procedures: Assoc, Official Agr. Chemists Jour., v. 45, no. 1, p. 169—173. E 75’ 7 DAY ’é 57#—B ethods of Computation for Estimating Geochemical Abundance GEOLOGICAL SURVEY PROFESSIONAL PAPER 574—B Methods of Computation for Estimating Geochemical Abundance By A. T._ MIESCH STATISTICAL STUDIES IN FIELD GEOCHEMISTRY GEOLOGICAL SURVEY PROFESSIONAL PAPER 574—B A review of statistica/[y ejicient procedures far estimating tne population aritnmetie mean w/zere t/ze data are eensorea' at tne lower limit ofanaiytica/ sensitivity or positive/y séeweaI UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON : 1967 UNITED STATES DEPARTMENT OF THE INTERIOR STEWART L. UDALL, Secretary GEOLOGICAL SURVEY William T. Pecora, Director For sale by the Superintendent of Documents, US. Government Printing Office Washington, D.C. 20402 - Price 20 cents (paper cover) CONTENTS Page Page Abstract ........................................... Bl Recommended techniques—Continued Introduction _______________________________________ 1 Confidence intervals _____________________________ B8 Problems in estimation of geochemical abundance _______ 2 Summary of methods ________________________________ 8 Analytical sensitivity ____________________________ 3 Examples __________________________________________ 10 Small groups of data ____________________________ 3 Molybdenum sulfide in drill core _________________ 10 Geometric classes ............................... 3 Iron in sandstone- ___~ ___________________________ 11 Analytical discrimination ________________________ 4 Uranium in granite _____________________________ 12 Recommended techniques ............................ 4 Arsenic in basalts and diabases ___________________ 13 Transformations ________________________________ 5 Conclusions ________________________________________ 14 Cohen’ s method for censored distributions _________ 5 Literature cited _____________________________________ 14 t and ta. estimators of Sichel and Krige ____________ 7 ILLUSTRATIONS Page FIGURE 1. Histograms showing frequency distributions _____________________________________________________________ B6 2. Curves for estimating x for equations 5 and 6 ___________________________________________________________ 7 3. Graphs of 7 as a function of the number of analyses, n, and the antilog of s or 3 _____________________________ 8 4. Flow chart showing methods of computation for estimating geochemical abundance __________________________ 9 5. Probability graphs of frequency distributions A—G’ in figure 1 ______________________________________________ 11 TABLE Page TABLE 1. Summary of computations f or estimating geochemical abundance from data represented in histograms in figure 1 .- _ B12 111 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY METHODS OF COMPUTATION FOR ESTIMATING GEOCHEMICAL ABUNDANCE By A. T. MIESCH ABSTRACT Geochemical abundance of an element is regarded as the pro- portion of a total rock body, or group of, rock bodies, that is made up of the element and is equivalent to the population arithmetic mean of sample analyses, generally in units of per- centage or parts per million. Computational problems encoun- tered in estimation of abundances can arise from having lim- ited ranges of analytical sensitivity, from having only small groups of data, and from reporting of data in broad geometric classes. These problems can be partly resolved by use of a com- bination of techniques described by Sichel (1952), Cohen (1959, 1961), and Krige (1960).. The combination of techniques allows eflicient estimation of geochemical abundances from a wide variety of frequency distribution types if the analytical discrimination is suflicient to allow effective use of data transformations. Examples of data from the literature were used to demon- strate the techniques; up to 76 percent of the data values were arbitrarily placed in the “not detected” class, and estimated arithmetic means agreed with those estimated from the com- plete data set to two significant figures. The types of frequency distributions displayed by the data examples are generally representative of those commonly encountered in geochemical problems. IN TRO DUCTI ON The estimation of abundances of constituents in rock bodies, or in groups of rock bodies, is necessary in a broad range of geologic problems. In mining and ore reserve estimation, for example, unbiased and precise estimates of abundance of ore constituents are prereq- uisite to efficient operational planning, and become man- datory as the grade of the ore declines toward the minimum grade that can be minedprofitably. As lower grade deposits of broad extent are being appraised as potential ores of the future, the need for accurate and precise abundance estimates will increase. Aside from these economic problems, the estimation of element abun- dances is necessary in geochemical problems related to the origin of rock bodies and in both small- and large- scale assessments of geochemical balance among different rock units and other material at the earth’s surface. Estimates of abundance are generally expressed in terms of weight percent or parts per milliOn and are regarded as estimates of the proportion of the rock body that is made up of the constituent of concern. An abundance of 2 percent iron in a 500—million-ton rock body, for example, implies that 10 million tons of iron are present. . The estimation of reliable geochemical abundances is met with serious difficulties in both sampling and in computation. Errors due to sampling are undoubtedly . the more serious in a majority of problems, but where ' sampling has been well planned and successfully exe- cuted, correct computational procedures are increasingly desirable. Where sampling has been inadequate (often unavoidably so owing to limited outcrop, for example), reliable abundance estimates are difi’icult to obtain. Where the sampling has been unbiased but of limited extent, the problem faced is to obtain the best estimate of abundance possible; computational methods _ are ‘ important for this purpose. Many types of averages (such as medians, modes, geometric means, mean logs, and arithmetic means) are suitable for specific geochemical problems, but the arithmetic mean is the only one of these that is a correct expression of abundance—or an unbiased estimate of abundance—under all circumstances. This was dis- cussed and verified by Sichel (1947, 1952).- If the frequency distribution of values used in computing the average is unimodal and symmetrical, the median, mode, and arithmetic mean will be the same; but if the distri- bution is skewed (asymmetrical), they may differ Widely. The geometric mean will always be less than the arithmetic mean regardless of the type of frequency distribution, except Where they are equal owing to zero variance in the data. Any type of average may be appropriate for a specific geochemical problem, depend- ing on the purpose for which the average is estimated. The geometric mean, for example, is useful for repre- senting the typical concentration of an element in a B1 B2 group of rock specimens where the concentrations ex- hibit a symmetrical frequency distribution on a log scale. The median is a useful expression of average concentration among specimens regardless of the fre- quency distribution and will be the same as the geo- metric mean where the distribution is symmetrical on a log scale. The equivalence of the arithmetic mean of the total population of samples in an entire rock body to geo- chemical abundance follows from the definition of abundance, A (in weight percent), as: _W A_—T><100, Where Wis the mass of the constituent in the rock body, and T is the mass of the rock body. If there are N potential discrete samples in the total rock body, the mass, W, of the constituent is given by: 1 N W=m6 Z 35131. J=1 Where x,- is the concentration (in weight percent) of the constituent in the jth sample and S,- is the weight of the sample. If all samples are of the same weight, then T N W—100N E 20,, and N A: N - The right side of this equation is the population arith- metic mean which can be estimated from analytical data by a number of difierent techniques. In some problems involving estimation of geo- chemical abundance, the weighted arithmetic mean is necessary to estimate the population arithmetic mean and has been known to provide accurate abundance estimates. This is well illustrated in many papers on, for example, polygonal methods of ore-reserve esti- mation. In some other types of problems, weighted arithmetic means are required to correct for highly uneven (biased) sampling, such as in estimation of the abundance of elements in large parts of the earth’s crust. Some of the problems met in this work were reviewed by Fleischer and Chao (1960). The methods described in this paper are not intended for problems where weighted arithmetic means are more suitably used. I am grateful to N. C. Matalas and L. B. Riley of the U.S. Geological Survey for technical criticism and much helpful discussion. STATISTICAL STUDIES IN FIELD GEOCHEMISTRY PROBLEMS IN ESTIMATION OF GEOCHEMICAL ABUNDANCE Problems in the estimation of abundance (population arithmetic means) arise in both sampling and computa- tion, but only the computational problems are consid- ered in this paper. The more common circumstances leading to computational problems result from (1) lim- ited ranges of analytical sensitivity, (2) small groups of data (small 72.), and (3) data reported in geometric classes. Where none of these circumstances cause diffi- culty, abundance estimates derived from the ordinary expression for the arithmetic mean, _. 1 x=7—b 2x, (1) will tend to be correct and will be at least relatively eflicient. An efficient estimate, in the statistical sense, is one that has a small variance (Fisher, 1950, p. 12), or one that would be expected to change little with the addition of new data. Concern for the efliciency of sta- tistical estimates gave rise to the “maximum-likelihood method” used in devising estimation methods (Fisher, 1950, p. 14). Estimators derived in this manner are said to yield statistics with minimum variance (maxi- mum efliciency). The expression for the arithmetic mean in equation 1 is a. maximum-likelihood estimator where the data are derived from a normally distributed population; it is not a maximum-likelihood estimator for many of the problems in geochemical studies. In problems where data from a normally distributed population are censored (some concentration values fall beyond the range of sensitivity for the analytical method), a maximum-likelihood method described by Cohen (1959, 1961) may be used. Where the data are from a lognormal population, a maximum-likelihood method given by Sichel (1952) will be applicable. A modification of the method given by Sichel will be use- ful where the data indicate certain kinds of departure from the lognormal (Krige, 1960). Some estimation methods derived by the method of maximum likelihood are slightly biased for small In. Unbiased likelihood estimators are those that have been corrected for the bias, where such correction is needed. Abundance estimates derived by maximum-likelihood techniques will not differ greatly, in many instances, from estimates derived by other reasonable though less precise methods. In much geochemical work, especially that conducted on a large scale, errors due to sampling and analysis are overwhelmingly dominant in deter- mination of the total estimation error; computational errors are relatively small. Nevertheless, the possibility of large errors from other sources cannot serve to justify additional errors where they can be easily avoided. METHODS OF COMPUTATION FOR ESTIMATING GEOCI—IEMICAL ABUNDANCE ANALYTICAL SENSITIVITY Every analytical method has limits of sensitivity be- yond which it is ineffective for determination of con- centration valves. These limits may occur at both the lower and the upper bounds for a concentration range. The spectrographic method described by Myers, Havens, and Dunton (1961), for example, may be used to determine silicon within the range from 0.002 to 10 percent. Concentrations judged to be lower or higher than this range are reported as <0.002 percent or 8> 10 percent, respectively. Thus, the spectrographic data may be either left or right censored. The term “censored” is applicable here because the number of values beyond each sensitivity limit is known for any set of analyzed rock samples. In other types of prob- lems, where the number of values beyond certain limits is unknown, the data are said to be truncated (Cohen, 1959, p. 217). A number of methods have been used by individual geologists for computing means from censored data. (See, for example, Miesch, 1963, p. 21—23). Other methods that are available, however, are a great deal more satisfactory (Cohen, 1959, 1961; Hald, 1952, p. 30; Hubaux and Smiriga-Snoeck, 1964). The maxi- mum—likelihood method described by Cohen will be applied to abundance estimation problems in later sections. Abundance estimates can be obtained directly from censored normal distributions, using Cohen’s method, but not from other types of censored distributions. Where the analytical data do not indicate a normal dis- tribution, some transformation of the data can be used, as will be demonstrated in a following section. The mean and variance of the transformed values, in many < problems, can then be used to derive abundance estimates. SMALL GROUPS OF DATA The precision, or reproducibility, of any statistical estimate is largely dependent on the number of values on which the estimate is based. Those estimates derived from large data sets will be relatively precise, even where the statistical method is not the most efficient one that could be used (where the statistical method has not been derived by the method of maximum likelihood). Where the data set is small, however, an unbiased max- imum-likelihood technique should be used wherever pos— sible if the precision of the statistical estimate is important. It is not possible here to state a value of n which will serve to distinguish between large and small data sets because this value will vary according to the variation in the data. A better apprOach may be to accept the maximum-likelihood estimate, if obtainable, B3 wherever it differs from an estimate derived by other techniques. - Where the data are derived from a normal distribu- tion, an estimate of abundance derived from equation 1 is based on the method of maximum likelihood and is the most eflicient estimate of abundance possible for a given number of samples. However, most underlying frequency distributions in geochemical studies are not normal; more commonly they are ( 1) asymmetrical with a long tail toward high values (positive skewness), (2) asymmetrical with a long tail toward low values (negative skewness), or (3) multimodal with more than one peak in the frequency distribution curve. Where skewness or the presence of more than one mode is suffi- cient to indicate a departure of the underlying fre- quency distribution from the normal form, abundance estimates derived from equation 1 will not be as efficient as maximum-likelihood estimates, if the latter are avail— able. The difl’erence in efficiency may be large if n is small. The most common departure of geochemical data from the normal distribution is a positive skewness. In many frequency distributions the skewness, along with other properties of the distribution, may reflect an underlying lognomnal distribution. Where this is true, an unbiased maximum-likelihood method for deriving efiicient estimates of abundance (the population arith- metic mean) can be applied (Sichel, 1952). Where the positive skewness in a particular data set is less or greater than that which can be ascribed to a lognormal distribution, a modification of this method may be used (Krige, 1960). The methods of Sichel and Krige involve data transformations. Unbiased maximum-likelihood methods of abundance estimation applicable to frequency distributions which are negatively skewed or multimodal are unknown to the writer. GEOMETRIC CLASSES Any quantitative analytical determination may be regarded as a reported range or class; a concentration reported as 78.62 percent, for example, signifies the range from 78.615 to 78.625 percent as the analyst’s best estimate of the true value. The classes are broader where fewer significant figures are reported. Where the values are reported to an equal number of decimal places, the class widths are equal; the class boundaries are arithmetic, because they increase by a constant in- crement. In much spectrographic and colorimetric work the analyst reports concentration values in broad classes, without specifying single values (except where they are meant only to identify a class). Classes are used for B4 reporting because of relatively poor discriminatory capacity of the analytical methods. Generally, the boundaries of the classes increase geometrically, each being higher than the previous boundary by a constant multiplier. Geometric classes are used because the error variance is at least approximately proportional to the amount of the constituent present. An example of the geometric classes used in spectro- graphic work was given by Myers, Havens, and Dunton (1961, p. 217), who formerly reported the concentra- tions of 68 elements in classes having the boundaries—— 0.00010, 0.00022, 0.00046, 0.0010, 0.0022, . . .—increas- ing by a factor equal to the cube root of 10, up to 10 percent. Myers and other US. Geological Survey spectrographers currently report values in classes with boundaries increasing by a factor equal to the 6th root of 10 (0.00012, 0.00018, 0.00026, 0.00038, 0.00056, 0.00083, 0.0012, . . .). Other similar systems of reporting, using various other factors for the generation of class boundaries, were described by Barnett (1961, p. 184). Although the reporting of spectrographic analyses in geometric classes has allowed spectrographers to pro- duce a vast amount of data useful in geochemical prob- lems, the practice has presented some difliculties in sta- tistical analysis that have not, so far, been satisfactorily resolved. Because the class boundaries form a geo- metric progression, the class widths are unequal in size, and conventional methods of treating grouped data are awkward and less eflicient than other methods that might be used. There is also the problem of choosing the best midpoint to represent a frequency class in the grouped-data computations. Where a number of con- centration values are reported to occur within the range from 0.0046 to 0.010 percent, for example, it would seem proper to use the arithmetic midpoint of 0.0073, rather than the geometric midpoint of 0.0068, for computing the sample arithmetic mean for the whole data set. The use of the geometric midpoint, however, nearly always gives better answers. (An example is given later in this paper.) The reason for this is that the use of the lower value partly compensates for a positive bias caused by the grouping technique, but the compensation is with- out any sound theoretical basis and should not be re— lied on. When a logarithmic transformation of the class boundaries is made, the class widths (in log units) be- come equal, thus allowing straightforward use of grouped-data computational methods for deriving the mean logarithm and standard deviation of the logs. By using these values, the abundance (population arith- metic mean) can be estimated according to the maxi- mum-likelihood method of Sichel (1952), if the form STATISTICAL STUDIES IN FIELD GEOCHEMISTRY of the underlying frequency distribution is at least ap- proximately lognormal. ANALYTICAL DISCRIMINATION For many types of spectrographic or colorimetric analytical techniques the analyst is unable to estimate concentrations to more than one significant figure. If a number of concentrations of the constituent sought are similar among the samples analyzed, the suitability of the technique for discriminating among the samples is poor. Commonly, a large proportion of the deter- minations are reported as the same value. For example, Hufi' (1955, p. 111) gave copper determinations from a chromatographic field technique on 23 samples of Tapeats Sandstone (Cambrian) from near Jerome, Ariz. Seventeen of the 23 determinations are given as 10 ppm (parts per million) ; the remaining 6 are either 50, 100, or 150 ppm. Many other examples of poor analytical discrimination could be given involving data reported in geometric classes. In many instances only three or four adjacent classes are used in reporting, and commonly 20-50 percent of the concentrations occur within one class. Discrimination among values in the same class, of course, is impossible. The proportion of values reported in a single class is dependent on the class widths, the variation in the concentrations, and the form of the frequency distribution. As has been pointed out, some of the computational problems encountered in abundance estimation can be wholly or partly resolved by means of data transforma’ tions. Where analytical discrimination is poor, how- ever, some of the more useful transformations are impossible or ineffective. No transformation would be useful, for example, in normalizing Huff’s data referred to above, and transformations other than the logarith-' mic transformation would be ineffective in treating most data reported in broad geometric classes. RECOMMENDED TECHNIQUES Where large sets of uncensored data are available, abundances may be estimated directly by using the con- ventional formula for arithmetic means in equation 1. If the data are part of an underlying normal distribu— tion of values, the abundance estimate will be the most efficient possible. If the underlying distribution is not normal, it may be possible to obtain a more efficient estimate by other techniques. However, if n, the num- ber of values, is large, the increase in efficiency may be small. Where small sets of uncensored data are used, abun- dance estimates derived from the conventional formula for the arithmetic mean will be as eflicient as possible if the underlying frequency distribution is normal. METHODS OF CONEPUTATION FOR ESTIMATING GEOCHEMICAL ABUNDANCE Where the distribution is positively skewed, abundance estimates that are significantly more eflicient may be obtained by other methods. These methods begin with data transformations. Where the data are censored, abundances may be esti- mated using the techniques available for estimating the population arithmetic mean, if the total underlying dis— tribution is normal. Where it is not normal, data transformations may be employed. The mean and standard deviation of the transformed values may then be used to derive abundance estimates. TRANSFORMATIONS . Many types of data transformations have been used in geologic and geochemical problems to normalize ob- served frequency distributions, but the types that ap- pear more useful in problems of estimating geochemical abundance are the log 3/ and log (y+a) transforma- tions, where 3/ is the concentration of the element, in percent or in parts per million, and o: is a constant. (All logarithms used in this paper are to the base 10.) Data which can be transformed to the normal form by these functions are referred to as 2- and 3-parameter lognormal, respectively (Aitchison and Brown, 1957, p. 7, 14). These two transformations generally appear to be effective where the distribution of the original analyti— cal data is unimodal and positively skewed. The sim- ple log transformation of some positively skewed data will lead to a log distribution with negative skewness. For other data (fig. 1E) the log transformation will lead to a distribution with some degree of positive skew— ness remaining (fig. 1F). In either case the constant, oz, can be estimated using techniques described by Krige (1960, p. 236) and used in the transformation log (y+a). Where a negative skewness of logs is to be corrected, 0: will be a positive value; where a positive skewness in the logs is to be corrected, or will be negative. In some cases the absolute quantity of the derived negative value of a will equal or exceed some of the concentration values, 3/, so that the quantity (y+a) is zero or negative and log (y+a) is undefined for some analytical values. Transformed distributions that re- sult are, in elfect, censored (fig. 161), and abundance estimates may be obtained by using techniques for censored distributions. The choice of a transformation that may be effective in producing a normal, or at least symmetrical, fre- quency distribution can be governed by examination and testing of the data available or by previous experience with similar data. Selections based on the data avail- 240—308—67———2 B5 able are generally preferred, though this is not always possible when a group of data is small or when a large proportion of the data is censored. Many data pertain- ing to the concentration of minor elements in rock sam- ples collected over large areas correspond more closely to the lognormal form than to the normal, and the log transformation is appropriate in a large number of problems. The log (y+a) transformation can fre- quently be used to improve on the simple log transforma- tion if a satisfactory estimate of a can be obtained. When the data have been transformed by w=log y or w=log (y+a), the mean of a: is estimated from equa- tion 1 and the standard deviation is estimated from equation 2. The mean and standard deviation of these transformed values are of little value themselves in problems of abundance estimation, but they may be used to derive estimates of arithmetic means by tech- niques developed and described by Sichel (1952) and Krige (1960). 2332—7152 1/2 s7] Equation 2 is a biased estimator of the population stand- ard deviation but is used in this form in computations described later in the paper. Where unbiased estimates of standard deviation are needed, n in the denominator of equation 2 is replaced by n—l. COHEN’S METHOD FOR CENSORED DISTRIBUTIONS Cohen (1959, 1961) presented maximum-likelihood techniques for estimation of the mean and standard deviation from either censored or truncated normal dis- tributions. These techniques may be used whether the unlmown part of the distribution is in the low- or the high-value region (left- or right-censored, respec- tively). Because left—censored distributions are by far the more common in geochemical problems, only the part of Cohen’s techniques applicable to these distribu- tions will be discussed here. Although Cohen’s methods are strictly applicable to normal distributions only, it can easily be demonstrated that the method for left- censored distributions provides satisfactory estimates of arithmetic means wherever the total distributions are symmetrical about one mode. The method given by Cohen has not been corrected for a small bias, and other methods for treating censored distributions may be more accurate when n is less than about 10 (Cohen, 1959, p. 218). The methods given by Cohen (1959, 1961) are pre- ferred to others given by Hald (1952) and Hubaux and Smiriga-Snoeck (1964) only because of the simplicity of computation. B6 30 N O FREQUENCY H o 0 FREQUENCY 8 8 ‘5 ‘6‘ ._I O O 30 N O FREQUENCY 5 STATISTICAL STUDIES IN FIELD GEOCIIEIMISTRY 40 : L L A : 7 B 30 — L n=101 : 7 [2:85 L L L 20 _— 77 7 m : 7 7/1, @777, - V % . 0.1 0.2 0.3 0.4 0.5 0.6 0.7 —1 1/3 —2/3 0 2/3 L L M082, IN PERCENT LOG IRON, IN PERCENT 7 E 7 L ‘c D E L r} ‘=185 n =135 77/ W, M A W W WV W 'o 2 4 6 s 10 12 14 —0.2 o, 0.2 0.4 0.6 0.8 1.0 1.2 URANIUM, IN PARTS PER MILLION ' LOG URANIUM, IN PARTS PER MILLION :Ili Z7 E i i i F L . G 2 - I 7 ”"58 [7—58 E??? ”:58 % 7 7 *3 o I IV /]/ J r If /I // Wm 0' A o 2 ' 4 6 s 10 ~04 o 0.4 0.8 —1.0 —0.2 0.6 ARSENIC, IN PARTS PER MILLION LOG ARSENIC, IN PARTS LOG (ARSENIC, IN PARTS PER MILLION PER MILLION—0.6) FIGURE 1.—Frequency distributions. A, M082 values for 300—degree samples at 2-foot intervals in drill core from Climax, 0010. (from Hazen and Berkenhotter, 1962, p. 84—86). B, Iron in sandstones (from Miesch, 1963, pl. 3). 0 and D, “Uranium in granite a deux micas de Pontivy” (from Hubaux and Smiriga-Snoeck, 1964, p. 1207). E, F, and G, Arsenic in basalts and diabases (from Onishi and Sandell, 1955, tables 5, 10). NIETHODS 0F COMPUTATION FOR ESTIMATING GEOCHEMICAL ABUNDANCE In Cohen’s technique 2:, is taken as either the trans- formed value of the lower limit of analytical sensitivity or as the limit of sensitivity (in percent or in parts per million) if the data transformation is not required. The quantity n’ is the number of concentration values that are below the limit of sensitivity; such concentra- tions are generally reported by the analyst as “not detected” or zero. The ratio h=n’/n is the fraction of the total number of analyzed specimens in which the element was not detected. The mean and standard deviation of the analytical values above the limit of sensitivity are computed as: _,_ Z a; x _n—n' (3) and 2 1/2 s'= Ezrefl - <4) The mean and standard deviation of the entire distri- bution are then estimated from: B7 The value x in equations 5 and 6 is a function of hand the quantity (s’)2/(aE’—x.,)2. Values of )1 for 0.013h $0.90 and 0.003(8’)2/(:i:’—x.,)2S 1.00 were tabulated by Cohen (1961, p. 538), and graphs of A were given in an earlier paper (Cohen, 1959, p. 231). The graphs are reproduced here in figure 2, with Dr. Cohen’s permission. t AND ta ESTIMATORS OF SICHEL AND KRIGE Where the original analytical data have been used in computation without transformations, the derived val- ues of m (eq 1) or 1'; (eq 5) may be taken directly as estimates of geochemical abundance. However, when E or ,u are derived for m=log y or w=log (g/+a, abun- dances can be estimated using methods described by Sichel (1952) and Krige (1960). The method by Sichel is an unbiased maximum-likelihood technique; that by Krige is a modification of Sichel’s method, to be used where the log (y+a) transformation is required. The estimators, Where the w=log 3/ and m=log (y+a) fi=—’—->\(5’—$o) (5) transformations are used, are referred to as t and ta, and respectively, and should not be confused with Student’s 3=[(s')2+)\(5:"——x,,)2]1/2. (6) t, which has wide application in statistical procedures. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 2‘0 '11:: 'I”"|””I”'ri””l'"'1'”'l"”l""l”"l""l”"l‘”‘|""1"""” |""|””—2-° 1.9 f— h‘0'75 —_ 1.9 1.8 :— —E 1.8 1.7 E— —f 1.7 1.5 f— —E 1.6 1.5 :— ": 1-5 1.4 f—' 3 1.4 1.3 :— —: 1.3 1.2:— —f 1.2 1.1 :— —‘ 1.1 A 1.0f— '— 1.0 >\ 0.9 :— —f 0.9 0.8 g- —j 0.8 0.7 — 0.7 0.6 : 0.3o__________._._—; 0.6 0'5 0.25 f 0-5 0.4E 0.20 E 0.4 0‘3 :— 0.15 “3 0-3 0-2 E— 010 -E 0.2 0-1 :— h=0.05 —: 0.1 o a...l....1....1.l..1....1....1....1....I....I....1....|....1....I....I....1....1....1....1....I....1....|..9.1....1....1....1...: 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1,2 1.3 (s’)2/(3(‘-'Xo)2 FIGpRE 2.—Curves for estimating A for equations 5 and 6 (from Cohen, 1959, p. 231; reproduced with author’s permission). B8 The t and ta statistics provide useful estimates of the arithmetic mean and geochemical abundance when the distribution of log y or log (y+a), respectively, ap— proximates the normal distribution form. Where log y is normally distributed: t=1x1fi or , (7) t=Tx1m where 5 and ii are the arithmetic means of log y esti— mated from complete (eq 1) and censored (eq 5) distributions, respectively. The factor 1- is a function of n, the total number of specimens analyzed and either the antilog of 8 (eq 2) or of 3 (eq 6), depending on whether the distribution is complete or censored. Values of 1- may be read from graphs in figure 3; for more exact work the reader is referred to the original equation and tables (Sichel, 1952, p. 275, 284—288) from which equation 7 and the graphs in figure 3 were derived. Simplified equations and tables were given recently by Sichel (1966). Other equations and tables that are mathematically equivalent to those of Sichel were given by Aitchison and Brown (1957, p. 45, 156—158) and were based on the work of Finney (1941). Where log (y+a) is normally distributed: ta: (TX 10;)—a or A , (8) ta: (1X10")—oz 7» ........ I ......... , .................. ,........., ......... _ n: number of analyses _ 5 :standard deviation (equation 2) 6 ; g:standard deviation (equation -6) r4:— I|Illlllllllll|l||lI|l|llllll1l1lll 1 2 3 4 5 e 7 io‘or 10¢9 FIGURE 3.—Graphs 0f 1' as a function of the number of analyses, n, and the antilog of s or 6' (based on tables by Sichel, 1952). STATISTICAL STUDIES IN FIELD GEOCHEMSTRY where 9's and ii are the arithmetic means of log (y+ as) estimated from complete (eq 1) and censored (eq 5) distributions, respectively. The equations in 8 are modified forms of one given by Krige (1960, p. 239). The factor -r is a function of n, the total number of specimens analyzed, and of the standard deviation of the transformed values, log (y+a). If complete distributions are used, the standard deviation is esti- mated by 8 (eq 2); if censored distributions are used, it is estimated by 3 (eq 6). The value of 1' can be read from figure 3 by using 72. and the antilog of either 8 or c? for the transformed values. For more exact computation the reader is referred to Krige (1960, p. 239) and tables given by Sichel (1952, p. 284—288), or to Sichel (1966). CONFIDENCE INTERVALS Estimates of geochemical abundance should be accom- panied by estimates of the confidence intervals wherever possible. The method for estimating this interval about the arithmetic mean derived from equation 1 is well known (see Dixon and Massey, 1957, p. 128) and leads to approximately correct intervals even where the dis- tributions are not normal. Cohen (1959, p. 231—232) gave methods for estimating confidence intervals of It derived from censored distributions. Approximate confidence intervals for the t and ta, may be estimated by using equations given by Aitchison and Brown ( 1957, p. 46) ; more exact confidence intervals can be obtained by using methods given recently by Sichel (1966). SUMMARY OF METHODS A flow chart indicating recommended computational procedures for estimating geochemical abundances is given in figure 4. With large groups of uncensored data, abundances may be estimated as arithmetic means in the conventional manner, using equation 1. How- ever, where means obtained from equation 1 differ from those obtained from procedures indicated in figure 4, the means obtained from the latter procedures should be preferred. The higher precision of the t and ta esti- mators have been adequately demonstrated by Sichel (1952, p. 276—278) and Krige (1960, p. 242—244), respec- tively, and the advantages of treating censored distribu- tions by the statistical method of Cohen (1959, 1961) over the approximate methods used in the past will be obvious. The methods of Sichel (1952), Krige (1960), and Cohen (1959, 1961) are individually inadequate for a large number of computational problems met in analyz- ing geochemical data. Sichel’s t estimator is useful where the data correspond to the lognormal frequency distribution and are uncensored. Krige’s ta estimator B9 METHODS OF COlVIPUTATION FOR ESTIMATING GEOCHEMICAL ABUNDANCE mocmucanm ho camEme uwmmfics cm 28 cozmscové NUS: mcofimsaw Eo‘c A8 + x; mo. *0 m .25 k wasano .33. no 359% :35 Sony 85655: Eggnooow wqfldfiuwa .8“ now—559:8 yo mcofiwzllé unborn n. Ohm wumEzmw uwmmficz cm m. 3 cozmsumvk mucmucann *0 wamEmefiwmflnca cu m_ 3 cozmsaovu wocwucsna ho Suzanna 30.an0 14 mOPmtlllnl .mEacco: m_ ”30 corantumfi sconce“. All cozzntuflu wuflano ovcmm mcoZmacw E0: Au + \v mo. *0 .« uca a 3:an0 5505. x. we mw:_n> .3 L8 3E: -ou m_ R van .EES: All 2355 _xo‘_aam m_ x we :25“...ng >ocoacw£ .mELo: >_wumE_xoLnam m_ corantami saws—ugh. E II \Hkuoq nonomcwo .mELocco: m_ cozantumfi 35:59.“. £03.43me ‘I m Ohm mOhw Uwswxm 225%? m_ cozantumfi >ocwceu .mo_boEE>m m_ :0: -sntumfi Sconce... :ozaehmfi umromcmo \ we mm: uwaxm boztmoa ucm .muoEEsE m. cozantums >055qu A583: .29: yo = o :3 n I a wucmucsna ho wumEme uwmmfica cm mZN £032.5wa N ucm a mac: -mscw Eot \ mo. “6 m tank ouanEoo -_m> mEom he uvccou -c: m_ Emu—«Sienna; m_ x we cozantums >ocoacwi R «an .553: Al. cot—5:55 oumEEOO Au+5 mo_Hk Enemcmfi. 8H » .mELo::o: m_ x we :ozantumzu 3:030?“— .w ucm m mcoznzuo E o: A. no. ho « ucm q 3:an0 .9: ice >_2mE_xo\_Q -Qm m_ x *0 50339 -tumfisnvcwzcmi 3H cozaezmfi “.990.ch a Ohm Ectocco: m_ x *0 .5339me Seance“. L -mB >ocmsamt goon m_ we cozhEmem HS :oszEtowG .moflb 10.5 -mcm unmimxm >63: All men Ucn .muoEE: m_ .5532wa >occacmi BouoameNm m_ :0: -mEEtomE .mofibmcm momma o o. mE Em \ me." x fiwimxm >_w>_:moa AI EWEWEM . EgowmcmF 5: van _m u o E _: a m_ . cozzntuflu >ocwsaofi :oAMWWrmmvaWWL mmwmflo 0505 53.5 E zocmaumt HOE "D -om E 9.x Mano B10 is applicable to a Wider variety of distribution types, but, as with the t estimator, the data must be uncensored. Moreover, with highly skewed frequency distributions, the log (y+a) transformation may impose censoring of data which are completely above the lower sensitivity limit of the analytical method. Cohen’s method is directly applicable to data which approximate a normal distribution, but data of this type are not common in minor-element studies pertaining to areas larger than a single mine or small outcrop. Generally the required normal distribution can be achieved only through a data transformation; estimates of the mean of the trans- formed variate are not valid estimates of abundance but may be transformed to valid abundance estimates using the methods of Sichel or Krige, if the necessary data transformation is accomplished by log y or log (y+a). The transformations are, indeed, sufficient in a wide variety of geochemical studies. Some properties of small data sets prohibit the esti- mation of precise arithmetic means, or abundances. These are indicated by STOP signs on the flow chart in figure 4, and are as follows: 1. A censored frequency distribution which, though believed to be part of a symmetrical distribution, departs widely from the normal form (for exam- ple, some censored multimodal distributions). 2. A frequency distribution which is markedly asym- metrical and cannot be transformed to an approx- imate normal distribution by log 3/ or log (y+a). This particularly includes multimodal skewed dis— tributions and all negatively skewed distributions. 3. A frequency distribution that is unimodal and posi- tively skewed but is based on data that cannot be normalized by the log y or log (y+a) transforma- tions owing to poor analytical discrimination or other factors. 4. Data reported in broad geometric classes, but not approximately normal on a log scale. Where these properties are present, the conventional method of estimating the population arithmetic mean (eq 1) may be the only way readily available to the geologist for estimating geochemical abundance. The use of the flow chart, figure 4, is demonstrated in the following section by employing examples of data from the literature. EXAMPLES Four data sets were selected from the literature to illustrate use of the recommended methods for widely differing types of data. One data set is approximately normally distributed, two are approximately lognormal (one of these is reported in geometric classes), and a fourth set is approximately lognormal after adjustment STATISTICAL STUDIES IN FIELD GEOCHEMISTRY by a constant, 0;. However, except for the data in geo- metric classes, none of the data sets indicate a close correspondence to the normal or lognormal form; this selection has been intentional to demonstrate that the distribution requirement is not rigid. As shown by these and other examples, the principal requirement is that the distributions be unimodal and approximately symmetrical on any of the three scales— the scale of original measurement, 3/, or one of the two transformed scales, log y or log (3/+a) . The number of values in each of the four data sets used as examples (fig. 1) is large compared with the number available in many geochemical problems. Most of the estimated abundances, therefore, agree fairly well with the abundances derived using the conventional method for estimating the population arithmetic mean. Large data sets were used so that this comparison could be made to verify the accuracy of the techniques for use in actual abundance estimation problems where n, is small or where the data are censored. MOLYBDENUM SULFIDE IN DRILL CORE The first set of data is from Hazen and Berkenhotter (1962, p. 84—86). The assays, of percent M082, were made on samples of drill core (300-degree segments) from the Climax Molybdenum mine, Lake County, Colo. We shall assume, for purposes of illustration, that the assays are an objective and unbiased sample of those that might have been obtained from the block of ore penetrated by the drill hole (that is, there is no sam- pling problem). The frequency distribution of the original assays is represented by histogram A in figure 1 and by the probability graph for distribution A in figure 5. Both illustrations indicate that the frequency distribution of assays (at least the central part) is ap- proximately symmetrical. The distribution was arbi- trarily censored successively at three points indicated by the small arrows (fig. 1A), and four abundance esti- mates were made using the complete and the partial data sets. The estimates, With intermediate computa- tional values, are given in table 1. In reference to figure 4: [1] The M082 assays are in arithmetic classes. ceed to [2]. [2] The frequency distribution of M082 assays, y is approximately symmetrical (figs. 1A, 5A). Proceed to [3]. [3] If the complete distribution is used, the abundance of M082 in the ore block is estimated, by the con- ventional method in equation 1, to be 0.359 per- cent. If only assays equal to or greater than the arbitrary (or hypothetical) analytical cutoffs—0.25, 0.35, Pro- METHODS OF COMPUTATION FOR ESTIMATING GEOCHEMICAL ABUNDANCE and 0.40 percent—are used (table 1, col. 2), the quantities indicated in columns 6—14 (table 1) are derived from the indicated equations, and the successive abundance estimates are derived as shown in column 17. The abundance esti- mates are in agreement to two figures, even though as much as 63 percent of the data was arbitrarily censored (table 1, col. 8). CUMULATIVE FREQUENCY. IN PERCENT A, M082. IN PERCENT 0123456 78910I1.121} C, URANIUM. ‘AND EMARSENIC, IN PARTS PER MILLION 99 98 95 90 so 70 so so 40 30 20 10 CUMULATIVE FREQUENCY, IN PERCENT 5 2 l D, LOG URANIUM.IN PARTS PER MILLION .-1 -2/3 -1/3 0 1/3 '8. LOG IRON. IN PERCENT -1.0 -O.6 -O.2 0.2 0.6 1.0 F, LOG ARSENIC. IN PARTS PER MILLION, AND G, LOG (ARSENIC, IN PARTS PER MILLION-0.6) FIGURE 5.—Probabi1ity graphs of frequency distributions A—G in figure 1. B11 Had the data, in fact, been censored at the arbitrary cutoff values, it would have been necessary to judge the nature of the total frequency distribution from as little as 37 percent of the data occurring above the cutoffs. In this extreme example of data censoring, the judg- ment regarding the form of the frequency distribution could be made only on the basis of prior experience with similar data. Where the censored part of the distribu- tion is minor (less than about one-third of the total dis- tribution), the probability graphs may still be useful for this purpose. IRON IN SANDSTONE The iron concentration in 85 drill-core samples of quartzose sandstone from the Salt Wash Member of the Morrison Formation (Jurassic) in San Miguel County, 0010., are represented by histogram B in figure 1. The data are from Miesch (1963, pl. 3) and were obtained by means of a spectrographic method (Myers and oth- ers, 1961) wherein the results are reported in classes having the boundaries 0.046, 0.10, 0.22, 0.46, . . . per- cent. The corresponding logarithms of the boundaries are —1%, —1, —2/3, "V3, . . . . The boundaries, in percentage concentration values, are at geometric inter— vals and increase by a factor of 2.15. The boundaries, in log values, increase by an increment of one-third. We shall assume that the analyzed samples are an unbiased representation of the body of rock for which the geochemical abundance is to be estimated. The estimate of iron abundance in the sandstone, derived from the grouped data form of equation 1, is 0.38 percent. The grouped—data form of equation 1 is: (9) where f. is the number of values in the ith class and x; is the class midpoint. The midpoints used were the geo- metric centers of the classes—the values 0.07, 0.15, 0.32, 0.68, and 1.46 percent. There is little justification for accepting the value of 0.38 percent as a valid estimate of abundance other than the fact that it agrees closely with the abundance of 0.37 percent derived with the theoreti- cally justified t estimator. It is not expected that equa- tion 9, used in the manner described here (with geometric midpoints) , will consistently lead to unbiased and efficient abundance estimates. The use of arithmetic midpoints (0.07, 0.16, 0.34, 0.73, and 1.58) leads to an abundance estimate of 0.40 percent. Although the esti- mate of 0.40 percent is not greatly different from that of 0.37, it does demonstrate the positive bias in the technique by which it was obtained. The positive bias is commonly much larger. _ 1 F—fl-Efmi, B12 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY TABLE 1.—Sammarg of amputations for estimating geochemical abundance from data represented in histograms in figure 1 l 2 3 4 5 6 7 8 9 10 ll 12 13 14 15 16 17 i Hypo- _ nr (eq 1) 8 (eq 2) (3’) 3 A A Esti- Distribution 1 thetical Transformatlon a I, n’ n h=—— or x" or 3’ x (fig. 2) u (eq 5) a (eq 6) t (eq 7) ta (eq 8) mated analytical 7:. (eq 3) (eq 4) (z’—z.,) 2 abun- cutofl dance Molybdenum sulfide (percent) 0 101 O 0. 359 0. 110 O. 359 20 101 . 20 . 396 . 091 . 356 48 101 . 48 . 445 . 072 . 356 64 101 . 63 . 475 . 066 . 359 Iron (percent) 0 —. 543 . 06 . 35 . 76 . 314 0 185 0 4. 53 23 185 . 12 4. 89 80 185 .43 5. 78 0 185 0 . 612 23 185 . l2 . 660 80 185 . 43 . 742 135 185 . 73 . 845 Arsenic (ppm) 0 58 0 1. 69 l. 69 0 58 0 . 096 l. 58 8 58 . 14 . 152 1. 58 22 58 .38 . 257 1. 59 42 58 . 72 . 488 1. 51 9 58 . 14 —. 208 1. 77 22 58 . 38 . 032 1. 68 42 58 . 72 . 375 1. 75 I The letters in this column refer to frequency distributions shown in figs. 1 and 5. In reference to figure 4 : [1] The data are grouped in geometric classes. ceed to [7]. [7] The frequency distribution, on a log scale, is shown in figure 13. The probability graph for this dis— tribution is given in figure 5. No important departure from the normal form is indicated, if the log scale is used. Proceed to [5]. [5] The abundance of iron in the sandstone, using Sichel’s t estimator, is estimated to be 0.37 per- cent. If the data are censored at the arbitrary points, 0.10, 0.22, and 0.46 percent, and 6, 35, and 76 percent of the data is effectively lost, the abundance estimates are 0.37, 0.38, and 0.37 per- cent, respectively (table 1, col. 17) . If only data values above the 0.46—percent class boun- dary (20 of the 85 values) are used, a large positive skewness is apparent from the fact that the median is between 0 and 0.46 percent—far below the central part of the total range of concentrations (0—2.2 percent). Therefore, a log transformation could be judged to be an appropriate step toward achieving a distribution closer to the normal form. If the data had shown a significant departure from the lognormal form, a “STOP” sign would have been Pro- encountered in figure 4. Because the data are in broad geometric classes, the log (y +a) transformation would have been awkward, and no method is available that could have been used to obtain a precise estimate of abundance. URANIUM IN GRANITE The data represented in histogram 0 (fig. 1) were originally from Coulomb (1959), but were repro- duced by Hubaux and Smiriga—Snoeck (1964, p. 1207) and used in testing a digital—computer method for de- riving means and standard deviations of censored dis- tributions. Hubaux and Smiriga—Snoeck derived mean logarithms rather than arithmetic means or abundance estimates. The data represent uranium concentrations in samples from a homogeneous granitic massif; we shall assume that the sampling was unbiased. The frequency distribution of the original data (fig. 10) exhibits a clear positive skewness, and the data, therefore, were transformed logarithmically (fig. 1D). The frequency distribution, as noted by Hubaux and Smiriga-Snoeck (1964, p. 1206), is only. imperfectly lognormal. The effectiveness of the log transformation in bringing the data closer to the normal form can be METHODS OF COMPUTATION FOR ESTIMATING GEOCHEMICAL ABUNDANCE seen by comparing the probability graph for distribu— tion 0 with that of distribution D (fig. 5). If all the data (n=185) are used, the geochemical abundance of uranium in the granitic massif is esti- mated, as the arithmetic mean (eq. 1), to be 4.53 ppm (table 1, col. 17). In reference to figure 4 : [1] The uranium data are in arithmetic classes. Pro- ceed to [2] . [2] The frequency distribution of uranium analyses is unimodal and positively skewed; analytical discrimination is satisfactory (figs. 10 and 50). Proceed to [4]. [4] The frequency distribution of the logarithms of the uranium analyses is approximately normal (distribution D; in figs. 1 and 5). Proceed to [5]. [5] If the complete distribution is used (n=185), the abundance, derived with the t estimator of Sichel (1952), is 4.54 ppm, virtually the same as the value, 4.53, derived by the conventional procedure. After censoring the data at the arbitrary analytical cutoffs 2.6, 4.0, and 5.1 ppm and effectively dis- carding 12, 43, and 73 percent of the data, respec— tively, the abundance estimates, as derived with the t estimator, are 4.55, 4.54, and 4.24 ppm (table 1, col. 17) . Had the decision been made to use the original data rather than to transform them logarithmically (that is, had the distribution represented by histogram 0 in, figure 1 been judged approximately symmetrical), the computation method would have proceeded as in the previous example for MOS2 assays. Censoring the data, then, at 2.6 and 4.0 ppm would result in abundance esti- mates of 4.48 and 4.11 ppm, respectively (table 1, col. 17) . These are somewhat poorer than the correspond- ing values of 4.55 and 4.54 obtained by way of the t estimator, and the importance of the log transformation is apparent. The problem of judging, from censored data, whether or not the total concentration values (fig. 10) exhibit a symmetrical frequency distribution is not as difficult as for MOS2 assays in the previous example. Because only 27 percent of the uranium values are above 5.1 ppm, for example, the median of all the data is known to occur between 0 and 5.1 ppm. As the data extend to more than 14 ppm some positive skewness is evident. However, judging the degree to which the log transfor- mation corrects the skewness is more difficult. Where a large proportion of the data is censored but the re- mainder is sufficient to indicate definite positive skew- ness in the original data, the log transformation might B13 be used as an approximation. Where a lesser propor- tion of the data is censored, the transformation using log (y+a) might be used if needed, and thereby offer a means for improving the accuracy of abundance estimates. ARSENIC IN BASALTS AND DIABASES The data with extreme positive skewness, represented by histogram E in figure 1, are from Onishi and Sandell (1955, tables 5, 10) and were used by Ahrens (1957, p. 207—209) in a discussion of frequency distributions of minor elements in igneous rocks. The data—arsenic determinations, in parts per million—were obtained on 58 samples of basalt and diabase from widely separated localities in the United States and from localities in Japan and Sicily. For purposes of illustrating the computational techniques we shall assume that the samples adequately represent the rock bodies from which they were taken. The geochemical abundance of arsenic in the basalts and diabases, estimated by the conventional method (eq 1), is 1.69 ppm (table 1, col. 17). In reference to figure 4: [1] The arsenic data are in arithmetic classes. ceed to [2]. [2] The frequency distribution is unimodal and posi- tively skewed; analytical discrimination is satis- factory. Proceed to [4]. [4] The frequency distribution of the logarithms of the arsenic analyses (distribution F in figs. 1, 5) retains a definite positive skewness. Proceed to [6]. [6] If the methods described by Krige (1960, p. 236) are used, the appropriate constant, a, is estimated to be —0.6. (Had the skewness of distribution F in fig. 1 been negative, the estimated constant would have been a positive value.) Eight of the analytical values are equal to or less than 0.6, and as a is negative, the quantity log (3/+a) for these values is undefined; the transformed data (distribution 0' in fig. 1) are censored with h=%8=0.l4. The probability graph for the newly transformed data, drawn using log (y—O.6), is shown in figure 5 (distribution 61). Neither the histogram nor the probability graph indicates a large departure of the transformed data from the censored normal form. If Cohen’s technique for censored distributions (eq 5, 6) and Krige’s ta estimator are used, the abundance estimate for arsenic in the basalts and diabases is 1.77 ppm (table 1, col. 17). Because the number of analyses in this example is small (n=58) and the data are highly skewed, it may Pro- 314 be argued that the estimate of 1.77 ppm is better than the estimate of 1.69 derived from the con— ventional procedure. However, the two esti- mates are in at least fair agreement. When censoring the data at the arbitrary cutoffs, 1.0 and 1.6 ppm, and eflectively losing 38 and 72 percent of the data, respectively (table 1, col. 8), the abundance estimates derived with the ta estimator are 1.68 and 1.75 ppm (table 1, col. 17). Had the data actually been censored below 1.6 ppm, estimation of the constant a may have been diflicult, but some useful estimate might have been made unless the point of censoring was lower than about 1 ppm. Whether the point of censoring had been at either 0.7, 1.0, or 1.6 ppm, a high positive skewness would have been apparent from the fact that the median lies well below the central part of the known range of the data— 0—10 ppm. If the log transformation is used (without adjusting the data by the constant a) and all the data are used, the abundance estimate is 1.58 ppm. If only data equal to or greater than 0.7, 1.0, and 1.6 ppm are used, the abundance estimates are 1.58, 1.59, and 1.51 ppm, respectively (table 1, col. 17). These estimates are notably poorer than those derived using the ta (rather than t), estimator, but they are, nevertheless, sufficiently good for many types of geochemical studies. CONCLUSIONS Computational methods given by Cohen (1959, 1961), Sichel (1952, 1966), and Krige (1960) are useful in providing accurate and eflicient estimates of geochemi- cal abundance in many problems Where the conventional method for estimating population arithmetic means is not applicable (owing to censored data) or is ineffi- cient (because of small data sets from nonnormal dis- tributions). A combination of the methods may be useful where censored data are from an underlying frequency distribution that is nonnormal. Each of the methods requires that the data, or transformations of the data, be normally distributed. Two transforma— tions that have proven satisfactory in much geochemical work are log y and log (3/ +41). Neither transformation is effective, however, where analytical discrimination has been poor. Moreover, selection of the appropriate transformation may be difficult where a large propor— tion of the distribution has been censored. The methods discussed are recommended for esti- mating geochemical abundance primarily because they are the most efficient methods available. Indeed, the methods developed by Sichel and Cohen are the most efficient possible where the frequency distribution re- quirements are satisfied. Those developed by Sichel STATISTICAL STUDIES IN FIELD GEOCHEMISTRY and Krige, moreover, have been corrected for a small bias that exists where n is small. The method of Cohen does contain some bias where n is small, but the bias is probably much less than that introduced by most arbitrary methods used in handling the censored- data problem. LITERATURE CITED Ahrens, L. H., 1957, Lognormal-type distributions, [Pt.] 3 of The lognormal distribution of the elements—a fundamental law of geochemistry and its subsidiary: Geochim. et Cosmochim. Acta, v. 11, no. 4, p. 205—212. Aitchison, John, and Brown, J. A. 0., 1957, The lognormal dis- tribution, with special reference to its uses in economics: Cambridge Univ. Press, 176 p. Barnett, P. R., 1961, An evaluation of whole-order, l/g-order, and 1Ag-order reporting in semiquantitative spectrochemical analysis: U.S. Geol. Survey Bull. 1084—H, p. 183-206. Cohen, A. 0., J r., 1959, Simplified estimators for the normal dis- tribution when samples are singly censored or truncated: Technometrics, v. 1, no. 3, p. 217—237. 1961, Tables for maximum likelihood estimates; singly truncated and singly censored samples: Technometrics, v. 3, no. 4. p. 535—541. Coulomb, René, 1959, Contributions a la géochimie de l’uranium dans les granites intrusifs: France, Centre de’Etudes Nu- cléaires de Saclay, rap. 0.E.A. 1173. Dixon, W. J ., and Massey, F. J ., J r., 1957, Introduction to statis- tical analysis: 2d ed., New York, McGraw-Hill Book 00., 488 p. Finney, D. J ., 1941, On the distribution of a variate whose loga- rithm is normally distributed: J our. Royal Statistical Soc., Suppl. 7, no. 2, p. 155461. Fisher, R. A., 1950, Statistical methods for research workers: 11th ed., New York, Hafner Publishing 00., 354 p. Fleischer, Michael, and Chao, E. 0. T., 1960, Some problems in the estimation of abundances of elements in the earth’s crust: Internat. Geol. 00ng., 21st, Copenhagen 1960, Rept., pt. 1, p. 141—148. Hald, A., 1952, Statistical tables and formulas: New York, John Wiley & Sons, Inc., 97 p. Hazen, S. W., Jr., and Berkenhotter, R. D., 1962, An experi- mental mine-sampling project designed for statistical analy- sis: U.S. Bur. Mines Rept. Inv. 6019, 111 p. Hubaux, A., and Smiriga-Snoeck, N., 1964, On the limit of sen- sitivity and the analytical error: Geochim. et Cosmochim. Acta, v. 28, no. 7, p. 1199—1216. Huff, L. 0., 1955, A Paleozoic geochemical anomaly near Jerome, Arizona: U.S. Geo]. Survey Bull. 1000—0, p. 105—118. Krige, D. 0., 1960, On the departure of ore value distributions from the lognormal model in South African gold mines: South African Inst. Mining Metallurgy Jour., v. 61, no. 4, p. 231—244. Miesch, A. T., 1963, Distribution of elements in Colorado Plateau uranium deposits—a preliminary report: U.S. Geol. Survey Bull. 1147-E, p. E1—E57. Myers, A. T., Havens, R. 0., and Dunton, P. J ., 1961, A spectro- chemical method for the semiquantitative analysis of rocks, minerals, and ores: U.S. Geo]. Survey Bull. 1081—1, p. 207— 229. METHODS OF COMPUTATION FOR ESTIMATING GEOCHEMICAL ABUNDANCE B15 Onishi, Hiroshi, and Sandell, E. B., 1955, Geochemistry of ar- Sichel, H. S., 1952, New methods in the statistical evaluation of senic: Geochim. et Cosmochim. Acta, v. 7, nos. 1 and 2, p. mine sampling data: London, Inst. Mining Metallurgy 1—33. Trans, v. 61, p. 261—288. Sichel, H. S., 1947, An experimental and theoretical investiga- 1966, The estimation of means and associated confidence tion of bias error in mine sampling, with special reference limits for small samples from lognormal populations: South to narrow gold reefs: London, Inst. Mining Metallurgy African Inst. Mining and Metall. J0ur., preprint no. 4, 17 p. Trans, v. 56, p. 403-474. n7 1% .7: DAY no: 574 ‘C Geochemical Environments and Cardiovascular Mortality Rates in Georgia GEOLOGICAL SURVEY PROFESSIONAL PAPER 574—C Prepared in cooperation wit/z t/ze U.S. Puo/z'c Hea/t/z Service Geochemical Environments and Cardiovascular Mortality Rates 1n Georgla By HANSFORD T. SHACKLETTE, HERBERT I. SAUER, and A. T. MIESCH STATISTICAL STUDIES IN FIELD GEOCHEMISTRY GEOLOGICAL SURVEY PROFESSIONAL PAPER 574—C PreparedI in coaperatz’on wit/z t/ze U.S. Pun/2'6 Hen/tn Service A stnq’y of t/ze cfienzz'ca/ elements in soils and plants from counties t/zat nave contrasting rates for diseases of t/ze neart and Mood vessels UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON : 1970 UNITED STATES DEPARTMENT OF THE INTERIOR WALTER J. HICKEL, Secretary GEOLOGICAL SURVEY William T. Pecora, Director For sale by the Superintendent of Documents, U.S. Govermnent Printing Office Washington, D.C. 20402 - Price 50 cents (paper cover) CONTENTS Page Abstract ___________________________________________ Cl Results of analyses—Continued Introduction _______________________________________ 1 Compositions of trees and uncultivated soils ______ Cardiovascular mortality rates in Georgia ______________ 2 Correspondence between the compositions of plants Physiography, geology, and soils of Georgia ____________ 4 and soils _____________________________________ Collection and analysis of geochemical data ____________ 6 Vegetables and garden soils __________________ Sample design __________________________________ 6 Trees and uncultivated soils __________________ Sampling media ________________________________ 9 Correspondence among the compositions of soil Plants _____________________________________ 9 horizons _____________________________________ Soils ______________________________________ 9 Correspondence between the compositions of stems Procedures used in chemical analysis ______________ 10 and leaves of trees ____________________________ Methods of statistical analysis ____________________ 10 Summary __________________________________________ Results of analyses __________________________________ 13 Conclusions ________________________________________ Compositions of vegetables and garden soils ______ 13 Literature cited ____________________________________ ILLUSTRATIONS FIGURE 1. Map showing physiographic regions and provinces in Georgia, and locations of counties with high and low death rates per 100,000 population for cardiovascular diseases ________________________________________ 2. Map showing Great Soil Groups of Georgia, and the locations of counties that have high and low cardiovascular mortality rates _______________________________________________________________________________ 3—7. Diagrams showing: 3. Correlations between the amounts of elements in garden soils and the amounts in vegetables ________ 4. Correlations between the amounts of elements in the C horizon of uncultivated soils and the amounts in the A and B soil horizons and in the stems and leaves of trees _____________________________ 5. Correlations between the amounts of elements in the B horizon of uncultivated soils and the amounts in the A horizon of soils and in the stems and leaves of trees _________________________________ 6. Correlations between the amounts of elements in the A horizon of uncultivated soils and the amounts in stems and leaves of trees ______________________________________________________________ 7. Correlations between the amounts of elements in stems and in leaves of trees ______________________ TABLE TABLES ._. 0‘9““?‘9‘5‘9‘359? Death rates for selected causes, white males age 35—74, selected counties of Georgia, during the period 1950—59- Localities, kinds, and numbers of samples that were collected in Georgia _________________________________ Analytical limits of detection _______________________________________________________________________ Components of logarithmic variance between samples, sites, and areas ___________________________________ Ash contents of vegetables and element contents of vegetable ashes and garden soils ______________________ Ash contents of trees and element contents of tree ashes _______________________________________________ . Element contents of uncultivated soils ______________________________________________________________ . Correlations between the amounts of elements in vegetables and garden soils _____________________________ . Correlations between the amounts of elements in trees and in the C horizon of uncultivated soils ___________ . Correlations between the amounts of elements in trees and in the B horizon of uncultivated soils ___________ 11. 12. 13. Correlations between the amounts of elements in trees and in the A horizon of uncultivated soils ___________ Correlations between the amounts of elements in horizons of uncultivated soils ___________________________ Correlations between the amounts of elements in stems and leaves of trees _______________________________ III Page 013 13 19 25 31 34 34 38 39 Page C5 7 20 22 28 32 36 Page C3 10 12 16 18 19 26 26 34 34 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY GEOCHEMICAL ENVIRONMENTS AND CARDIOVASCULAR MORTALITY RATES IN GEORGIA By HANSFORD T. SHACKLETI‘E, HERBERT I. SAU’ER, and A. T. MIESCH ABSTRACT Nine contiguous counties of northern Georgia that have a low cardiovascular mortality rate lie in an area where the rocks, by weathering. continuously contribute a supply of minor elements to the soil that is formed. Nine counties of central and south- central Georgia, in contrast, have a high cardiovascular mortality rate, and the soil is largely composed of materials that generally are deficient in weatherable minerals. The concentrations of the elements that were studied in both garden soils and uncultivated soils from the two areas tend to be significantly diiferent and are generally higher in the low-death-rate area. The difference in element composition of soils from the two areas is not apparent in analyses of most vegetables; only cabbage and green beans tend to reflect, on a broad scale, the compositional differences in amounts of certain elements in soils. The concentrations of most elements in tree samples tend to be higher in the low-rate area ;‘ therefore, trees appear to be more sensitive than vegetables to the variation in element content of soils. The difference between the two areas in the levels of various chemical elements present in vegetables appears to be so slight as to be unlikely to contribute appreciably to the observed diiferences in death rates in middle- aged humans. The chemicals in soils, however, may enter the human food chain in water, in food plants that were not sampled in this study, and in meat and milk from animals that have fed on cultivated forage plants and native vegetation. If geochemi- cal difl’erences between the two areas do, in fact, have a causal relationship to death from cardiovascular diseases, the cause would appear to be a dietary deficiency, rather than an excess, of the elements that were studied. IN TRODUCI‘ION The counties of Georgia that have low rates of cardio- vascular mortality are in the northern part of the State, whereas those that have high rates are in the southern and central parts. Northern and southern counties are very different in their characteristic geological forma— tions and soil types. The fact that the two areas of con- trasting mortality rates coincide with two areas of con- trasting physiography gave rise to the hypothesis that cardiovascular mortality might be causally related to physical factors of the environment. Of these factors, the chemical composition of the surficial deposits and the vegetation appeared likely to have the greatest effect on the human inhabitants of a region. At the beginning of this study, the patterns of heart-disease mortality rates in this State were well known; the chemical ‘ele- ment contents of soils and plants, on the other hand, were but poorly known. Therefore, the Heart Disease and Stroke Control Program of the US. Public Health Service and the US. Geological Survey undertook a co- operative study of the geochemical properties of soils and plants as they relate to the occurrence of cardio- vascular mortality in Georgia. A preliminary survey was made in the summer of 1963 in which rocks, uncultivated soils, and native plants were sampled in counties that have low, intermediate, and high rates of heart-disease mortality. These samples were analyzed chemically, and the results of the anal- yses indicated that the areas of high and low mortality rates could be identified by the amounts of certain ele- ments that were in samples from the areas (Shacklette and Sauer, 1964). Following this study, a more rigorous design for sampling, which included sampling of garden soils and vegetables as well as uncultivated soils and native plants, was devised for the areas under study, and fieldwork was completed in the summer of 1965. The re— sults obtained from chemical analyses of the samples, and the abundance of the elements in the areas of high and 10w mortality rates, are presented in this report. Interest in the relation of the environments in Georgia to human ‘health began early in the history of the State and continued up to the time of the Civil War. Cramer (1967) gave an account of the papers that reflect this interest. Most early reports were prepared by physicians who were careful to note the effects of physiography on the prevalence of diseases and the occurrence of epi- demics. Yellow fever and malaria were the two diseases most commonly reported to have been associated with topographic features because they were more prevalent in areas of low-lying, swampy ground than in well- drained upland areas. Methods of clearing the land of Cl CZ trees and cultivating the soil that resulted in impeding the flow of streams and increasing the amount of stand- ing water caused an accompanying increase in incidence of these diseases. Putrefaction of vegetation in wet lo— cations was thought to produce a noxious effluent that caused the diseases; the fact that mosquitoes, which breed in standing water, are the vectors in the trans- mission of these diseases was not known at the time. Speculation on the association of health problems with geological features, however, was not limited to specific diseases. Cramer (1967, p. 216) quoted a report that was read to the Georgia Medical Society in 1806 by Dr. Joshua E. White, who said, To obtain a correct knowledge of medical topography, we should possess some information respecting the face of the country; its elevations and depressions, the quality and proportion of low and high grounds; the number and magnitude of the streams; and various other local circumstances. In relating facts, and detailing alterations which have been produced by the art and industry of man or by fortuitous causes, on the surface of the country, we should not confine our views solely to the present moment. To those who succeed us, it may be satisfactory to be informed what was, and thus they may be enabled to calculate what may be. Cramer (1967, p. 217—218) reported further that as early as 1837 physicians had recognized the difference in disease rates between the Coastal Plain and the Pied- mont regions and that the healthful environment of northern Georgia counties had been noted in 1853 by Dr. Robert C. Ward, who related the effects of lithology and topography on the properties of soil that was formed. Geochemical characteristics of regions are, in the final analysis, expressions of bedrock geology and the con- comitant changes produced by pedological and biologi- cal processes that relate to climate and the passing of time. An understanding of these characteristics is nec- essary in assessing the factors that control the chemical environment of a region. The data included in this re- port illustrate some chemical relationships of soils to their geologic parent materials and of plants to the soils that support them. The distribution and chemical be- havior of elements in the system “rock-soils-plants” is a subject of increasing concern in several other types of geological investigations. For these reasons, these data are of geochemical and biogeochemical interest, inde- pendent of their possible usefulness in interpreting pat- terns of human disease mortality. Shacklette, a botanist of the US. Geological Survey, directed this project and made the field studies; Miesch. a geologist of the same organization, devised and con- ducted statistical studies of the data; and Sauer, a statistician of the US. Public Health Service, provided all data on disease mortality rates and assisted in con- STATISTICAL STUDIES IN FIELD GEOCHEMISTRY structing the sampling plan and in interpreting the data. We express our appreciation to Messrs. David F. Davidson and James A. Banta for their assistance in initiating and planning the study, to Messrs. Todd Church and John R. Keith for their help in collecting and preparing samples, to Messrs. Gary L. Selner, Rob- ert Terrazas, and George Van Trump for assistance in computer programming, and to Mmes. Josephine G. Boerngen, Jessie M. Bowles, and Dorothy Moore and Mr. Darrel Parke for their contributions to data proc- essing and analysis. The following chemists of the Geo- logical Survey analyzed the samples: J. C. Hamilton, Thelma F. Harms, J. B. McHugh, Harriet G. Neiman, Clara S. E. Papp, A. L. Sutton, J r., Barbara Tobin, and J. H. Turner. We also acknowledge the cooperation of the many property owners in Georgia who freely gave us permis- sion to obtain samples on their land; without their gen— erosity this study could not have been made. CARDIOVASCULAR MORTALITY RATES IN GEORGIA Death rates for the cardiovascular (CV) diseases in- clude those for coronary heart disease (which accounts for more than half of all CV deaths), stroke, hyper- tensive diseases, and other diseases of the heart and blood vessels. Rates have been computed for the State of Georgia by usual county of residence, utilizing all deaths for the period 1950—59, inclusive (Sauer and others, 1966). Standard vital statistics techniques have been used in computing rates, which are expressed as average annual death rates per 100,000 population at risk, along with several refinements to meet the specific needs of this type of ecological study. The age-sex group of greatest epidemiological concern is middle—aged White males, which may be defined broadly as white males age 35—74. Rates have been computed for this age group, using the average of the 1950 and 1960 censuses as the population at risk, and have been ad- justed by the direct method by 10-year age groups to the total United States population of that age in 1950 (Linder and Grove, 1943, p. 66—68). Differences in rates are, therefore, not due to differences in the age, sex, or race composition of the population of the various counties. The high-rate counties have mortality rates for the cardiovascular diseases roughly double, and in some in- stances more than double, the rates for the low-rate counties. In this one State, contrasts are found that are almost as great as those for the entire United States (Sauer, 1962). Rates for 1949—51 by economic subre- gions show northern Georgia in a very low death-rate area and south-central Georgia in a very high rate area (Enterline and other, 1960, p. 762) . GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA C3 The nine contiguous low-rate counties chosen for this study are listed in table 1. The cardiovascular diseases death rate for white males age 35—74 ranged from 560 to 682. Only three other counties of the 159 counties in Georgia had rates equally low—Irwin, Liberty, and Miller; each of these three has very small populations of middle-aged white males, and these counties are iso- lated from other low-rate counties, therefore the pos— sibility is strongly suggested that the low rates for these three counties are phenomena of chance fluctuation or random error. Eighteen counties are listed as “high-rate,” with death rates from 1,151 to 1,446; the first nine counties in this list were chosen for the study because they are rural counties in the section of the Sate which consist- ently had high rates. Two metropolitan counties, Chat- ham (including the city of Savannah) and Richmond (including Augusta), were not included because soils and locally grown plants are presumed to have a lesser effect upon the risk of dying in urban areas than in rural areas. Five scattered counties were not included because they are surrounded by counties that have more mod- erate rates, which again suggests the possibility of chance fluctuation of rates. The determination of the underlying cause of death frequently may be difficult, and the available diagnostic information may not always be adequate. For a broad group of diseases, such as the CV diseases, the available evidence suggests that this potential shortcoming does not invalidate the comparison of the counties that we selected (Sauer and Enterline, 1959). The rates for “all non-CV” causes follow a pattern similar to that for CV diseases (table 1). Of the 'nine counties with low CV rates, all but one had “all non-CV causes” rates below the State average, and of the nine counties selected with high CV rates, all but two had rates for all non-CV causes above the State rate. The factors which are causing these differences in cardiovascular disease rates are presently unknown, but whatever they are, they (or factors associated with them) tend to produce a somewhat similar pattern of rates for non- CV causes. Similar findings are presented in greater detail for the nation as a whole by Sauer and Enterline (1959) and Sauer and Brand (unpub. paper by H. I. Sauer and F. R. Brand, “Geographic differences in cause-specific death rates,” presented to the Society for Epidemiologic Research, VV'ashington, D.C., May 10, 1968). The all—causes death rates, therefore, show almost as much contrast as do the rates for the cardiovascular diseases. Several counties, in fact, in the high-death- rate area have all-causes rates approximately double those for several counties in the low-death-rate area (table 1). In general, the all-causes death rates for the TABLE 1.——Death rates for selected causes, white males age 35—74, selected counties of Georgia, during the period 1950—69 [Average annual death rates per 100,000 population, age adjusted by 10-year age groups by the direct method to the entire United States population age 35-74 in 1950. Deaéms t]abu1ated by Georgia State Department of Health, by county of usual resx ence Caigiovascular diseases 30-334, 400—468 1) All non- County Coronary Other cardio- All Total heart cardio- vasCular causes disease vascular causes of death (420 1) diseases of death Low-rate counties: Cherokee ........... 671. 0 344. 1 326. 9 525. 4 1, 196. 4 Fannin ______ 655. 2 262. 2 393. 0 607. 9 l, 263. l Forsyth _____ . . 635. 7 275. 1 360. 6 529. 6 1, 165. 3 Gilmer ...... 647. 5 296. 9 350. 6 485. 4 1, 132. 9 667. 7 328. 6 339. 1 573. 3 1, 241. 0 680. 8 274. 8 406. 0 593. 6 1, 274. 4 681. 7 332. 2 349. 5 735. 5 1, 417. 2 587. 6 201. 2 386. 4 480. 0 1, 067. 6 560. 1 287. l 273. 0 477. 5 1, 037. 6 1,302. 2 743. l 559. 1 789. 9 2,092. l l, 205. 5 650. 6 554. 9 546. 0 1, 751. 5 1, 167.0 603. 7 563. 3 834. 3 2, 001. 3 1, 206. 3 569. 4 636. 9 736. 1 1, 942. 4 1,167. 8 467. 4 700. 4 731. 5 1, 899. 3 , 446. 701. 1 745. 3 568. 6 2, 015. 0 1, 151. 2 679. 3 471. 9 677. 9 l, 829. l 1, 330. 8 530. 1 800. 7 802. 5 2, 133. 3 1,321.7 915. 6 406. 1 760. 2 2, 081. 9 Other high-rate counties: Baldwin 1 ___________ 1, 218. 8 781. 5 437. 3 744. 7 1, 963. 5 1, 177. 2 753. 7 423. 5 757. 2 1, 934. 4 1,170. 3 644. 1 526. 2 700. 1 1, 870. 4 1, 170. 6 754. 5 416. 1 927. 8 2,098. 4 l, 162. 3 614. 6 547. 7 638. 1 1, 800. 4 1, 155. 8 667. 2 488. 6 686. 6 l, 842. 4 1,252. 1 572. 2 679. 9 618. 1 1,870. 2 1, 203.0 773. 5 429. 5 771. 3 1, 974. 3 1,171.7 533. 2 638. 5 766. 2 1,937.9 All counties of Georgia ......... 925. 2 520. 4 404. 8 649. 0 1, 574. 2 ‘ Code number of International Statistical Classification of Diseases, 7th Revision (World Health Organization, 1957). 3 With adjustments for resident institution populations. low-rate counties are similar to or lower than the CV diseases rates alone for the high-rate counties. Calculation of death rates for the CV diseases for 1959—61 indicates that all nine high-rate counties have rates which are above the United States rate and which are roughly 7 0 percent higher than the rates for the low- rate counties. For 1950—59, the pattern of rates for white females age 35—74 tended to be similar to that for males. Since there are fewer deaths among white females, however, chance fluctuation will tend to be higher (Lilienfeld and others, 1967, p. 121—126). Of the nine counties selected because of their low male death rates, seven had rates for white females below the State average, several of the rates being among the lowest in the State. Of the nine counties selected because of their high male death rates, seven also had white female rates higher than the State average. Although nonwhite rates tend to a considerable extent to present patterns parallel to those for whites, the number of nonwhites in the north- ern Georgia counties is generally too small to provide a basis for calculating meaningful rates. C4 These contrasts in rates are in general not due to random error, and thus far show no evidence of being due to methods of data collection or classificationfland we have made extensive search for such evidence. There is an incentive, therefore, to search for and test hypothe- ses regarding factors which may be responsible for these differences. The death rates tell us that the risk of dying of CV diseases in middle age is twice as great in the high-rate area as in the low-rate area. The rates do not, however, tell us whether the attack rate is higher in the high-rate area, or whether these diseases are more lethal in this area, or both. It is possible that the attack rate or in- cidence rate in the high-death-rate area is higher than in the low-death—rate area, or it may be that the time period between onset of clinical disease and death is shorter—that is, that the lethality or fatality rate is higher; it is also possible that both the attack rate and the fatality rate are somewhat higher in the high-rate areas than in the low. The nine counties in the northern part of the State that have very low death rates due to CV causes are identified in figure 1. They are part of a group of closely associated or contiguous low-rate counties in the Blue Ridge, the Valley and Ridge, and the upper part of the Piedmont provinces. The nine counties with very high death rates are likewise part of a group of contiguous or closely associated high—rate counties in the east-central part of the State; all counties but one lie entirely within the Atlantic Coastal Plain region. The counties that were chosen for this study give substantial evidence that they are properly classified. As a further illustration: Of the 159 counties in Georgia the low-rate counties are all among the 20 lowest for all causes of death as well as among the very lowest for the cardiovascular diseases generally, and coronary heart disease specifically. Similarly, the high-rate counties are all among the 20 highest for all causes of death. PHYSIOGRAPHY, GEOLOGY, AND SOILS OF GEORGIA The northern part of the State lies Within four prov- inces of the Appalachian Highlands region, and the southern part is in the Atlantic Coastal Plain region (US. Geological Survey, 1968), as is shown in figure 1. These regions are characterized by features of relief and geologic formations (Georgia Division of Mines, Min- ing and Geology, 1939) that have. been largely respon- sible for the differences in soils within the State. The Southern Appalachian region, according to Laurence (1968, p. 156), comprises four major geologic provinces whose boundaries are almost identical with those of the four physiographic provinces, therefore the same names may be used for both kinds of provinces. The Appala- STATISTICAL STUDIES IN FIELD GEOCHEMISTRY chian Plateaus province barely extends into the extreme northwestern part of the State, where high, narrow ridges of sandstone occur interbedded with layers of shale. This province was not sampled in this study. The Valley and Ridge province is characterized by long, narrow ridges that parallel broad valleys, which have a northeast-southwest trend. The folded and faulted sedimentary rocks range in age from Early Cam- brian to Pennsylvanian. The Cambrian, Ordovician, and Mississippian Systems include thick and extensive deposits of carbonate rocks. Clastic rocks predominate in the Lower Cambrian, Middle Ordovician, Silurian, Devonian, Lower Mississippian, and Pennsylvanian (Laurence, 1968, p. 158). Soils derived from limestones in the valleys generally are rich loams that are suitable for agriculture. The noncalcareous shales produce a soil that is less fertile than that derived from limestone, and the soils that have sandstone parent materials are rela- tively infertile (Carter and Giddens, 1954). The Blue Ridge province in the northeastern part of the State is characterized by high mountains, narrow valleys, and rapid streams. The rocks range from crys- talline gneisses and schists to unmetamorphosed elastic sedimentary rocks of late Precambrian and Early Cam- brian age. Igneous rocks of Paleozoic age, from perido— tite to granite, are widespread in this province (Lau- rence, 1968, p. 156—157). The cool climate of the high elevations has favored the retention of organic matter in the generally acid soils, but the great relief of the region does not promote large-scale agriculture, there- fore subsistence farms are prevalent. The Piedmont province extends from the foot of the Appalachian Mountains to the fall line at the coastal plain and constitutes about one-third of the State. This province includes a wide variety of Precambrian and Paleozoic metamorphic rocks, and intrusive rocks of several different ages. Much of the area has been deeply weathered, and saprolite conceals the unweathered bed- rock at places (Laurence, 1968, p. 156). The soils of the part of the province that was sampled in this study commonly are sandy loams, but at places erosion has exposed clay subsoils. Southern Georgia and parts of central Georgia lie within the Atlantic Coastal Plain region. The landscape has but little relief, and swamps and sluggish streams are common. The typically sandy soils overlie marine rocks of Cretaceous and Tertiary age. These rocks, especially where exposed, have been subjected to inten- sive weathering. The counties that were studied, if low lying, were largely covered with swamp forests; those with greater relief and better drainage were devoted to small-scale diversified agriculture. GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA C5 85° 84" EXPLANATION Appalachian Plateaus Blue Ridge province 83° . provmce Valley and Ridge Piedmont province province Atlantic Coastal Plain region 82' \\ Low—death-rate | counties O 50 IOOMILES Ii...|....4 ‘_ FIGURE 1.———Physiographic regions and provinces in Georgia, and locations of counties with high and low death rates per 100,000 population for cardiovascular diseases among white males of ages 35—74 during the period 1950-59. 2 368—640 0—70 C6 The kinds of soils that occur in Georgia were shown on a map prepared by the US. Soil Conservation Serv- ice (1969), which utilized the new soil classification system that was established by the Soil Survey Staff (1960, 1967). This map classified soils into the four highest categories (Order, Suborder, Great Group, and Subgroup) of the hierarchical system. The lowest of these categories, Subgroup, is distinguished principally by degree of slope and combinations of other Great Groups that are given for the map unit. For simplicity of presentation in this report, Subgroups have been ignored in preparing the classification outline and the soils map (fig. 2) that follow. ORDER Entisols—Soils that have no pedogenic horizons. Suborder Psamments—Entisols that have textures of loamy fine sand or coarser. Great Group Quartzipsamments—Psamments that con- sist almost entirely of minerals highly resistant to weathering, mainly quartz. ORDER Histosols—Wet organic (peat and muck) soils. Suborder (not named)—Histosols of warm regions. Great Group (not named)—Plant residue highly decom- posed. _ ORDER Inceptisols—Soils with weakly differentiated horizons showing alteration of parent materials Suborder Aquepts—Seasonally wet Inceptisols that have an organic surface horizon and a sodium saturation, mot- tles, or gray colors. Great Group Humaquepts—Aquepts that have an acid dark surface horizon (in Georgia, tidal marshes). Suborder Ochrepts—Inceptisols that have formed in mate- rials with crystalline clay minerals, have light—colored surface horizons, and have altered subsurface hori- zons that have lost mineral matter. Great Group Dystrochrepts—Ochrepts that are usually moist and low in bases and have no free carbonates in the subsurface horizons. ORDER Spodosols—Soils with low base supply that have in subsurface horizons an accumulation of amorphous mate- rials consisting of organic matter plus compounds of alu- minum and usually iron. Suborder Aquods—Seasonally wet Spodosols. Great Group Haplaquods—Aquods that have a subsur- face horizon that contains dispersed aluminum and organic matter, but only small amounts of free iron oxides. ORDER Ultisols—Soils with horizons of clay accumulation and low [base supply. Suborder Aquults—Seasonally wet Ultisols that have mot- tles, iron-manganese concretions, or gray colors. Great Group Ochraquults—Aquults that have a light- colored or thin black surface horizon. Suborder Udults—Ultisols that are usually moist, relatively low in organic matter in the subsurface horizons; formed in humid climates that have short or no an- nual dry period. Great Group Hapludults—Udults that have a thin sub- surface horizon of clay or appreciable weatherable minerals, or both. Great Group Paleudults—Udults that have a thick horizon of clay accumulation without appreciable weatherable minerals. STATISTICAL STUDIES IN FIELD GEOCHEMISTRY The counties that have a low cardiovascular mortality rate are in an area where Hapludults predominate (fig. 2). The principal characteristic, from the geochemical standpoint, of soils in this Great Group is their con- tent of weatherable minerals which serve as a constant source of element enrichment in the process of soil devel- opment. The counties that have a high mortality rate for these diseases are located in areas where one of the three soils, Paleudults, Quartzipsamments, and Ochra- quults, predominates. Paleudults are characterized by the absence of appreciable amounts of weatherable minerals, Quartzipsamments consists mostly of minerals that are highly resistant to weathering, and Ochraquults are deficient in available elements because of their high seasonal water content which causes ex- cessive leaching. Soils in the low—death—rate counties, therefore, were expected to contain certain elements in greater abundance than is found in soils from the high- death-rate counties. The analyses given in this report support this hypothesis. Most soils of Georgia are not very fertile as judged by agricultural standards in their natural condition. This infertility was attributed by Carter and Giddens (1954) to climatic conditions and soil development processes. The soils generally are deficient in the major elements for plant nutrition, as well as in magnesium, boron, and zinc; therefore, trace elements commonly are added to the fertilizers used on certain crops. The relation of the two areas of contrasting rates of heart disease mortality to the occurrence of soil types can be summarized by stating that the high-mortality counties have soils that are derived from the highly weathered marine sediments of the coastal plains, whereas the low-mortality counties have soils that originated from a great variety of rock types from which weathering is continuously providing a fresh supply of chemical elements to the soils. COLLECTION AND ANALYSIS OF GEOCHEMICAL DATA SAMPLE DESIGN The counties of Georgia to be sampled were divided into two groups of nine counties each, according to their classification by rates of death due to heart disease. The counties that have high death rates are Bacon, Bleck- ley, Burke, Dodge, Emanuel, J efi' Davis, Jefferson, Jenkins, and Warren; those with a low death rate are Cherokee, Fannin, Forsyth, Gilmer, Hall, Murray, Pickens, Towns, and Union. Within each of the two groups, 30 sites were selected for sampling native plants (trees) and uncultivated soils, and 30 sites were chosen for sampling vegetables and garden soils. This made a total of 120 sample sites in the two areas. A sample site for trees and uncultivated soils was defined as an area in which the required species of trees GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA C7 EXPLANATION 85' a4- Quar‘tZIpsamments Humaquepts Halplaquods \ 1‘ 83" Plant residue highly Dystrochrepts Ochraquults I \ decomposed | ,_[_ srsz'NS P Mm” a“ I ’4» f Y_:‘:T/fl;w)\\\— Hapludults Paleudults \ 761‘“ ‘T/cg‘m‘r- §\\\ \ 53 V Low-death-rate \ counties \ H igh-d eath-rate counties 100 MILES I FIGURE 2,—Great Soil Groups of Georgia, and the locations of counties that have high and low cardiovascular mortality rates. C8 could be found rooted in a soil unit which was judged to be relatively homogeneous. Ordinarily, the sample site was considered to be within a radius of 20—60 feet of the point that was selected for the soil-sample hole; however, this radius was extended at places in order to include a required species of tree, if the same soil unit continued into this enlarged area. A sample site for veg- etables and garden soils was defined as a single home garden (not a commercial truck farm). The size of these gardens ranged from about 400 square feet to a quarter acre (10,890 sq. ft.). The distribution of sample sites by counties within each area (the high and low-death—rate areas) was arbi— trarily set at three native plant and soil sites and three garden sites for each of six counties of an area, and four native plant and soil sites and four garden sites for each of three counties (table 2). Within a county the exact location of sample sites was dictated, to a large extent, by the restrictions of the sampling plan—that is, the locations of the sites often were determined by the lim- ited occurrence of the required sampling media. The selection of sites for sampling native plants and uncul- tivated soils first required that relatively undisturbed forests be found in which the selected species of trees and a reasonably homogeneous soil unit occurred. The selection of garden sites was controlled largely by the requirements that the kinds of vegetables that were STATISTICAL STUDIES IN FIELD GEOCIEUEMISTRY wanted occurred in each garden and that permission of the owner to sample the garden was obtained. The former requirement was at places difficult or impossible to meet, therefore complete suites of samples were not obtained at all sites. Permission to sample gardens was almost invariably given by the owners, many of whom assisted us in gathering the vegetable samples. Within the limitations imposed on selection of sites as described above, We attempted to obtain samples that were representative of a county. In effect, the unpredict- able occurrence of suitable sites contributed a degree of randomness to the sampling plan. The two sampling areas are disposed in a north- south orientation and, therefore, encompass a wide range of climates, soils, forest types, plant species, and cultural preferences. In the most southern counties, where the fieldwork began, we collected samples of per- simmon, blackgum, black cherry, red maple, blackjack oak, smooth sumac, sassafras, and sweetgum. Of these eight species, only five (blackgum, persimmon, red maple, smooth sumac, and black cherry) could be found at all sites throughout the State (table 2). Blackjack oak was found at all sites in the high—death-rate area, but did not occur at all sites in the low-death-rate area; therefore, two other species of oak (black oak and post oak) that are closely related morphologically were sub- stituted in the latter area. Sassafras was more abundant TABLE 2.——Loealt'ties, kinds, and numbers of samples that were collected in Georgia [_.._, no samples collected] Kinds and numbers of samples Native trees 1 Vegetables Uncultivated soils Gar- Soils, Counties den compo- Black- Per- Red Sassa- Smooth Sweet Black Black- Cab— Green Lima Toma- A B C soils, nents of gum Oak 2 sim— maple fras sumac gum cherry eyed bage 3 Corn beans beans toes hori- hori- hori- plow variance mon peas zon zon zon zone study High-death-rate area Bacon_._......._ 3 3 3 3 1 3 2 3 3 2 3 3 3 3 3 3 3 Bleckley ........ 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 Burke ........... 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 Dodge .......... 3 3 3 3 3 3 2 3 3 3 3 3 3 3 3 3 3 Emanuel ........ 4 4 4 4 1 4 4 4 4 3 4 4 4 4 4 4 4 Jeff Davis ....... 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 J eflerson ________ 4 4 4 4 1 4 4 4 4 4 4 4 4 4 4 4 4 Jenkins ......... 3 3 3 3 1 3 3 3 3 3 3 3 3 3 3 3 3 Warren __________ 3 3 3 3 1 3 3 3 3 3 3 3 3 3 3 3 3 Subtotal _ _ 30 30 30 30 18 30 28 30 30 28 30 30 30 30 30 30 30 30 30 Low-death-rate area Cherokee........ 4 4 4 4 3 4 4 4 ........ 4 4 4 4 4 4 4 4 4 6 Fannin..._ _ ._ . 3 3 3 3 3 3 3 3 ........ 3 3 3 ........ 3 3 3 3 3 .......... Forsyth... 3 -3 3 3 3 3 3 3 1 3 3 3 3 3 3 3 3 3 __________ Gilmer... . 4 4 4 4 4 4 4 4 ________ 4 4 4 ________ 4 4 4 4 4 6 Hall ....... 4 4 4 4 2 4 4 4 ________ 4 4 4 4 4 4 4 4 4 6 Murray. . . 3 3 3 3 3 3 3 3 1 3 3 3 1 3 3 3 3 3 6 Pickens. _ _ 3 3 3 3 3 3 3 3 1 3 3 3 2 3 3 3 3 3 .......... Towns. . _ 3 3 3 3 3 3 1 3 ........ 3 3 3 l 3 3 3 3 3 6 Union ........... 3 3 3 3 3 3 2 3 1 3 3 3 ________ 3 3 3 3 3 .......... Subtotal . _ 30 30 30 30 27 30 27 30 4 30 30 30 15 30 30 30 30 30 30 Total ..... 60 60 60 60 45 60 55 60 34 58 60 60 45 60 60 60 60 60 60 l Tree samples were divided into stems and leaves, and these were analyzed separately; therefore the total number of tree samples is twice that shown. 2 Includes 44 samples of blackjack oak, 13 samples of post oak, and 3 samples of black oak. 3 Both cabbage and collards are included; they are botanically similar. GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA in northern Georgia than in southern Georgia—how- ever, a complete suite of samples could not be obtained in either area. Sweetgum was absent from only five sites. Of the six kinds of vegetables that were studied, only corn, green beans, and tomatoes could be found in all gardens that were sampled, although cabbage (or col- lards) was absent from only two of the 60 gardens. Regional differences in preference for vegetables were most obvious with blackeyed peas. This vegetable was found in all 30 gardens that we sampled in the south- ern area, but in only four gardens in the northern area. Similarly, lima beans were found in all southern gar- dens, but in only half of the northern gardens. Of the vegetables that were observed but not sampled because of their immaturity at the time of the field study and their infrequent occurrence, Irish (white) potatoes were rarely seen in the southern gardens, and sweet potatoes (yams) were uncommon in the north. Other vegetables that were observed, but which occurred erratically, in- cluded squash, peppers, beets, peanuts, okra, cucumbers, eggplant, and melons. Corn and tomatoes were found in almost all gardens that were examined and were by far the most common vegetables throughout the State. SAMPLING MEDIA PLANTS The kinds of plants that were sampled, and their scientific names (Fernald, 1950), are listed below: Blackgum, N yssa sylvatica Marsh. Blackjack oak, Quercus marilandica Muench. Black oak, Quercus velutina Lam. Post oak, Quercus stcllata Wang. Persimmon, Diospyros m‘rgim’ana L. Red maple, Accr ru brum L. Sassafras, Sassafras albidum (Nutt) Nees Sumac, Rims glabra L. Sweet gum, Liquidambar styracifiua L. Black cherry, Prunus scrotina Ehrh. Blackeyed peas, Vimza sincnsis Endl. (samples included sev- eral cultivated varieties) Cabbage, Brassica olcracea var. capitata L. Collards, Brassica olcracca var. accphala DC. Corn, Zea mays L. (samples included both field corn and sweet corn) Green beans, Phascolus vulgam's L. (samples included both green-pod and yellow-pod varieties) Lima beans, Phascolus limcnsis Macfad. (samples included both the bush form and the climbing form) Tomato, Lycopcrsicum csculcntum Mill. (all samples were of the common red-fruited form) The tree samples were stems (terminal parts of branches) about 12 inches long. The leaves were re- moved from the stems, and the leaves and the stems were analyzed separately. Both types of tree samples were analyzed, because it is known that some chemical elements are concentrated in stems, whereas others are 09 concentrated in leaves. The samples were air dried, pulverized in a Wiley mill, weighed, and burned to ash in an electric oven in which the heat was increased 50° C per hour to a temperature of 550° C and held at this temperature for 14 hours. The ash was then weighed to determine the ash yield of the dry plant sample and was used in all subsequent analytical procedures. The vegetable samples were prepared in the same manner as for table use; that is, the blackeyed peas were shelled and a few tender pods were left with the seeds, the cabbage and collards leaves were washed and cut, the corn grains were cut from the cobs, the pods of green beans were washed and broken into pieces, the lima beans were removed from the pods, and the toma- toes were washed and sliced. The prepared vegetables were packed in polyethylene bags as soon as possible after preparation and shipped by air to the Denver laboratories of the US. Geological Survey where they were oven dried to constant weight. Most vegetables, because of their content of sugar or colloids, or both, are difficult to air dry, in contrast to stem and leaf samples which air dry readily. The dried vegetable samples were pulverized and burned to ash in the same manner as the tree samples. SOILS In this report, soil-horizon terminology follows the traditional usage of the A and B horizons to designate divisions of the solum and of the C horizon to designate soil parent material (Baldwin and others, 1938). The uncultivated soils, at the tree-sample sites, were sampled from holes made by a clamshell digger to the depth necessary to include the C horizon of the soil. This depth ranged from less than 1 foot to 3 feet. The samples of the A and B horizons were taken from the sides of the holes by using a trowel, and the C horizon sample was taken from the bottom of the hole with the digger. Zonation ordinarily was well defined; some lithosols of the mountainous provinces and the regosols of the coastal plain, however, had indefinite horizons or no horizons and are termed azonal soils. Soils at two sites in the low-rate area were azonal lithosols, and soils at five sample sites in the high-rate area and two in the low-rate area were regosols. The lithosols were sampled by dig- ging to bedrock or to a zone so stony that further digging was impractical, and the depth of the holes in the regosols was arbitrarily set at 2 feet. These two types of soil were sampled just below the surface, midway down the hole, and at the bottom of the hole; and the samples were, for convenience, desigated A, B, and C horizons, respectively. 010 In order to obtain some measure of the homogeneity of soils at a site, two additional soil-sample holes were dug and the three soil horizons were sampled at each of five tree-sample sites in each of the different death rate areas. These additional sample holes were randomly located within a 30-foot radius of the initial soil-sample hole. The 60 additional samples, when added to the 30 samples of the initial sample holes at the sites,.provided 90 samples for the study of soil homogeneity. The re- sults of this study are given in table 4. The garden soils were sampled by taking the required amount from soil that was 1—2 inches below the surface. These soils were well pulverized when sampled and had been thoroughly mixed by cultivation practices that had extended over many years, therefore only the “plow zone” was sampled. No attempt was made to record past or present fertilizing practices; questioning the owners from place to place about fertilizing soon revealed that such information was either unknown, uncertain, or of doubtful accuracy in regard to kinds of fertilizers used, rates and frequency of application, and the years when the alleged applications were made. Widespread use of insecticides, principally lead arsenate and paris green (arsenic trioxide and copper acetate), is known to have been prevalent in past dec- ades, therefore heavy metal residues were expected to be found in garden soils. Analyses of these soils, however, did not reveal unusual concentrations of these metals. The soil samples were air dried, then sifted through a 200-mesh sieve and the —200 fraction used for chemical analysis. Some samples that had a large clay content were so hard when dry that it was necessary to grind them to powder in a ceramic mill before sieving. PROCEDURES USED IN CHEMICAL ANALYSIS The content of elements in plants that is reported was determined by analysis of the plant ash. The ash yield of the dry plant sample also is reported, therefore the element content of the dry plant can be computed. Contents of zinc and phosphorus were determined by colorimetric methods that were described by Ward, Lakin, Canney, and others (1963). Analyses for amounts of calcium were made by the EDTA titration method, and analyses for potassium by flame pho- tometry. Quantities of the other elements were deter- mined by semiquantitative spectrographic analysis (Myers, and others, 1961) after the ash of plant sam- ples had been diluted with an equal weight of matrix composed of sodium silica (10 percent Na). These val- ues were reported in geometric brackets having the boundaries 1.2, 0.83, 0.56, 0.38, 0.26, 0.18, 0.12, and so forth, percent; the brackets are identified by their respective geometric midpoints, such as 1.0, 0.7, 0.5, 0.3, STATISTICAL STUDIES IN FIELD GEOCHEIWISTRY 0.2, 0.15. Thus, a reported value of 0.3 percent, for example, identifies the bracket from 0.26 to 0.38 percent as the analyst’s best estimate of the concentration. The precision of a reported value is approximately plus or minus one bracket at the 68-percent level of confidence and plus or minus two brackets at the 95-percent level. Soil samples were analyzed by the same methods that were used for plants, except the pulverized soil was not burned to ash (most samples had a very low organic content) and a matrix was not added to the sample for spectrographic analysis. The chemical composition of soil samples was reported as parts per million or per- cent of the air-drv weight of the soil. The limits of detection of the analytical methods that were used follow (table 3). TABLE 3.—Analytical limits of detection [Values given are the lower limits, in percent, for the semiquantitative spectrographic method that was used, unless otherwise noted Element and chemical Detection Element and chemical Detection symbol limit sym ol limit Aluminum (A1) ____________ 0.001 Neodymium (Nd)____._._. 0.007 Barium (Ba). __. 0002 Nickel (Ni) ________ __ 0003 Boron (13).... 3002 Niobium (Nb) ..... u C N Calcium (Ca) 1 .001 Phosphorus (P). 2 .002 Cerium (Ce)..._ .015 Potassium (K).. . Chromium (Cr) ______ .0001 Scandium (Sc) _____ .0005 Cobalt (Co) .......... .0003 Sodium (Na) _______ .05 Copper (Cu) _________ .0001 Strontium (Sr) ..... .0005 Gallium (Ga) ........ .0005 Titanium (Ti) ..... .0002 Iron (Fe) ............. .001 Vanadium (V) _____ .0007 Lanthanum (La) ..... .003 Ytterbium (Yb)... .0001 Lead _________ . .001 Yttrium (Y) ..... ____ .001 Magnesium (Mg) ........... .005 Zinc (Zn) _________________ 2 .0025 Manganese (Mn) ........... .0001 Zirconium (Zr) ............ . 001 l Analyzed by ED’I‘A titration method. 2 Analyzed by colorimetric method. 3 Analyzed by flame photometry. The values given in table 3 are approximate lower limits of determination. Some combinations of elements in a sample, however, affect these limits. Concentrations somewhat lower than these values may be detected in unusually favorable materials, whereas these limits of determination may not be attained in unfavorable mate- rials. Upper limits in measuring the amounts of mag— nesium were exceeded by a few plant samples; these analyses were reported as >10 percent. METHODS OF STATISTICAL ANALYSIS The objective of the statistical analysis of the data has been to examine differences between plant and soil geochemistry in the high- and low-death-rate areas. The statistical significance of the differences between the two estimated means for each element concentration in each sample media was determined by analysis of vari— ance methods. Analysis of variance was also used in the estimation of total sampling and analytical error associated with an individual site at which the A, B, and C horizons of uncultivated soils were collected. Cor- GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA relation coefficients were computed to examine the co- variation among the compositions of the various sam- pling media among the sampling sites in both the high- and low-death—rate areas. Frequency distributions of most of the original chemical data, in units of weight percent, display high positive skewnesses and the variances of the distribu- tions tend to be proportional to their respective means. Log transformation of the original data proved effective in correcting both these situations to a satisfactory degree. The frequency distributions of the logs of the original data tend to be normal; the distributions of the original data tend to be lognormal. Log transformation also overcomes some effects of the geometric brackets in which spectrographic analyses are reported (Miesch, 1967b, p. B5). The geometric means were estimated by use of the antilog of the arithmetic mean of the logs. This is the appropriate measure of central tendency for a log- normal distribution and provides a “typical” or “characteristic” value for the element in the various sample media. A measure of variation is given by the geometric deviation, which is the antilog of the stand- ard deviation of the logs. About two-thirds of the area under the lognormal distribution curve lies in the range GM + GD to GM >< GD, where GM is the geometric mean and GD is the geometric deviation. The confi- dence interval about the geometric mean, the geo- metric error (GE), can be estimated from GE=10(‘ logGD/W) where t is Student’s t and N is the number of values on which the.geometric mean is based. Where t is se- lected at the 0.05 level of probability, the population geometric mean has the expected range GM ~:— GE to GM X GE, with 95-percent confidence. Many of the elements studied were not detected in part of the samples from each media. The elements are presumably present in the media, but in concen- trations below the sensitivities of the analytical methods used. Means and variances of the log concentrations, in these cases, were estimated by use of Cohen’s (1961, p. 535) technique for type I censored distributions. The estimators are: fi=aZ—>\(§;—xo) and &2=82+>\(5—xo)27 where ,2 and 62 are estimates of the population mean log and variance of the logs, respectively, :7; and 82 are the mean and variance of the logs of the concentra- tion values above the analytical sensitivities, (to is the log of the sensitivity, and A is taken from table 2 of Cll Cohen (1961, p. 538). The estimated geometric mean is GM=103 and the geometric deviation is GD=109. Estimated geometric means and geometric deviations are given in tables 5—7. Where the element was not detected in some of the samples analyzed, as indicated on the tables, the estimates were derived by Cohen’s technique. The statistical significance of differences between mean concentrations of constituents in various media in the high- and low-death-rate areas was estimated by analysis of variance methods; results are given in tables 5—7. Differences between means estimated from highly censored data were not tested; the required statistical tests are not available. Where the censored part of the distribution was small, the unknown values were as- signed to a semiquantitative spectrographic class im- mediately below the limit of sensitivity of the analytical method. The principal justification for replacing the unknown values is that the final answers are nearly independent of any reasonable values that may be used. The validity of analysis of variance results (tables 5—7) rests on three principal assumptions: (1) the sampling and analysis have been unbiased, or biases which may be present tend to be constant among sampling sites (Miesch, 1967a, p. A14), (2) the frequency distribu- tions of the data are drawn from corresponding log- normal population distributions, and (3) the geometric deviations of the population frequency distributions are equal. The correctness of the first of these assump- tions can only be inferred from the nature of the sam- pling techniques and methods of analysis, already described. The other two assumptions are only approxi- mately correct, as inferred from tests for normality and variance homogeneity; but the observed departures of the respective properties from the ideal conditions are not sufl‘icient to seriously invalidate the results. It is doubtful that any two areas, even areas in prox- imity, contain exactly the same concentrations of any particular element or other constituent in any particu- lar media. The analysis of variance, therefore, is an at- tempt to demonstrate the existence of differences we can be quite sure are present without the analysis, and fail- ure of the tests to demonstrate a difference merely indi- cates that the test lacks sufficient power (the power could be increased by taking more samples). The purpose of the analysis of variance, therefore, is not to demon- strate the existence of differences in composition between the two areas, but to find those differences which appear C12 to be large when compared with the variation within each of the two areas. The computational procedures used in the analysis of variance followed those described by Anderson and Bancroft (1952, p. 327—330) for designs containing un- qual numbers in subclasses. Estimates of experimental error in the data pertain- ing to uncultivated soils were obtained from replicate sampling at 10 sites, five in the high-death-rate area and five in the low-death-rate area. Three sample holes, at random locations as much as 60 feet apart, were dug at each of the 10 sites, and samples were taken from each of the A, B, and 0 soil horizons. A total of 90 samples were used. The statistical model employed was Xijk2fl+ai+6ij+7flk where X,,,, is the log concentration of an element in the kth sample (13kS3) from the jth site (13355) in the ith area (lsis 2), from either the A, B, or C horizon; a is the grand mean log concentration of the element in the A, B, or C horizon at all sites, a, is the deviation of the high- or low-death—rate area mean from the grand mean, 13,-, is the deviation of the site mean from the area mean, and 7”,, is the deviation of the log analytical value for a single hole from the mean for the site. The validity of the model depends on the absence of variable bias in sampling and analysis (Miesch, 1967a, p. A14). Estimates of the variance components for a, B, and 7 were derived according to Source Degrees of Mean square is freedom estimate of Between areas __________ 1 «,2 + 3(152 + 15m.2 Between sites . . _ _ _ - _ _ _ 8 «172+ 317,32 Between holes __________ 20 (1,2 Estimates of the individual components, (73,, 0,23, and 02, are given in table 4. In general, the variance be- tween the high— and low—death—rate areas (0,2,) is sufliciently large in comparison with variance com- ponents between sites and between samples that no difficulties are foreseen in identifying compositional differences between areas. With 30 sites per area and one sample per site, the within-area variance of the mean log for aluminum in the A horizon, for example 1s 0.021 +9119 1 30 — 0.0012. This value is a measure of the total error of the mean log aluminum content for an area and is satisfactorily low as compared with the variance between areas, 0.246 (table 4). STATISTICAL STUDIES IN FIELD GEOCHEMISTRY TABLE 4.—Components of logarithmic variance between samples, sites, and areas where uncultivated soils were sampled in the high- and low-death-rate areas [Numbers in italic indicate that the variance is significant at the 95-percent level of confidence] Component of logarithmic variance Between samples (cry?) Between sample sites Between areas (“2) 2 Element (we ) Soil horizons A B C A B C A B C 0. 031 0. 021 0. 038 0. 095 0. £46 0. 340 0. 216' . 014 . 017 025 045 125 106 . 091 062 073 078 118 593 701 619 022 077 063 059 060 233 190 047 022 033 047 067 081 006 O Q a: W O . . H .5 N h. m 0 O . . . 0 . 042 . 018 . 099 . 024 . 022 . 009 . 026 The variance among sample sites within the high— and low-death-rate areas is generally similar to the variance within sites (table 4). This appears to be true for all chemical elements in all three soil horizons. Conse- quently, if further studies were undertaken with an objective that depended on distinguishing among the sites within either area, it would be necessary to collect considerably more than one sample at each site. For example, in order to distinguish between two sites with respect to aluminum in the A soil horizon (table 4), it would be necessary to collect a sufficient number of samples, n, within each site so that the observed F ratio, estimated by (0.016)+n(0.021) (0.016) would exceed the critical F at a stated level of prob- ability for 1 and 2(n—1) degrees of freedom. In this case, for the 0.05 level of probability, n must equal or exceed 4; at least four samples from the A horizon at each sample site Would be required. Comparison of mean element concentrations (tables 5—7) for various plant and soil media in the high- and low-death-rate areas serves to identify covariation be- tween plant and soil compositions on a broad scale. For example, the generally lower concentrations of many elements in both soils and trees of the high-death—rate area (tables 6, 7), compared with those of the low-death- rate area, points to a broad-scale covariation between tree and soil chemistry. Such covariation, however, may or may not be present on a smaller scale—within either of the two areas. Covariation between soil and plant compositions within both the high- and low-death-rate areas was examined by means of product—moment cor— GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA relation coefficients (tables 8—11). Correlation coefli— cients were also used to examine covariation among soil horizons (table 12), and between the stems and leaves of trees (table 13) within each of the two areas. The correlation coefficients are based on logarithms of the element concentrations. In cases where the element was not detected in either the plant or the soil (or the stem or the leaf) at a sampling site, that site was not represented in the data used. Therefore, the bivariate frequency distributions on which the correlation coeffi— cients are based are censored at either or both the left- most or the lowermost extremes. The degree of censoring is indicated by the number of sites in which the element was not detected in either or both sampling media (tables 8—13). In cases where the element was detected in both sampling media at all sampling sites (that is, N =0 on tables 8—13), the bivariate frequency distributions of the logarithmic concentrations are uncensored and approx- imately normal. The statistical significance of the. corre- lation coefficients, in these cases, may be determined from widely available tables of critical values of r. How- ever, where the element was not detected in either or both sampling media. at one or more sites (for example, N 750 on tables 8~13), the statistical significance of the correlation coefficient is indeterminable, and the coeffi- cient must be regarded only as a relative index of covariation. RESULTS OF ANALYSES COMPOSITIONS OF VEGETABLES AND GARDEN SOILS The contents of certain elements in samples of vege- tables and garden soils from the two death-rate areas, expressed as geometric means (in percent), and the variations in contents expressed by geometric deviations are given in table 5. The ash content of vegetables also is provided to permit converting the amounts of ele- ments in vegetable ashes into amounts in the oven-dry vegetables. The weight of samples before dehydration (wet weight) was not recorded; therefore, the element content of “fresh” vegetables cannot be computed from these data. - COMPOSITIONS OF TREES AND UNCULTIVATED SOILS The contents of certain elements in samples of tree stems and leaves from the two death—rate areas, ex- pressed as geometric means (in percent), and the vari- ations in contents expressed by geometric deviations are given in table 6. The ash contents of stems and leaves are also provided to permit converting the amounts of elements in the ashes into amounts in the oven-dry stems and leaves. 368—640 0—70 3 C13 Data in table 6 show that leaves, when burned, yield about twice as much ash as do stems. This difference in yield is believed not to result from contamination of leaves by dust, but to be caused by the difference in element content of primary and secondary growth products. Leaves have only primary growth (originat- ing in primary meristem); stems have, in addition, secondary growth (originating in the cambium). Secondary growth produces an increase in “dry weight” as a result of the synthesis of organic compounds (lig- nin, among others) without a corresponding increase in amounts of the elements that constitute ash. Therefore, if equal weights of dried stems and leaves are burned the organic compounds are oxidized, and a greater weight of ash remains from the burned leaves. An examination of the element content in the ash of tree stems and leaves reveals the following tendencies in element concentration: Concentration greater in the stems—barium, cal- cium, copper (for several species), lead, and strontium. Concentration greater in the leaves—aluminum, iron, magnesium, potassium. titanium, ytterbium, and zirconium. Apparently equal or approximately equal concen- trations in stems and leaves—boron, chromium, copper (for most species), manganese, nickel (variable with species), and phosphorus. The element contents of uncultivated soils from the two death-rate areas are given in table 7. The contents (geometric means) of elements are printed in italic if the differences in content between the two areas are significant. If the differences in the content of an ele— ment are significant in one soil horizon, they are also significant in the other two horizons, except for ytter- bium in which only the A horizon has this difference. In garden soils (table 5) and in uncultivated soils (table 7) the same elements exist but in concentrations that are significantly different in the two death—rate areas. This fact suggests that cultivation practices over a period of years have not greatly altered the contents of many elements in garden soils. CORRESPONDENCE BETWEEN THE COMPOSITIONS OF PLANTS AND SOILS The correspondence between amounts of chemical elements in plants and those in soils in the areas studied can be examined, with the data available, on two scales. Correlation on a broad statewide scale may be examined by comparing the average plant and soil compositions for the high- and low-death-rate areas (tables 5—7 ). Correlation on a more detailed scale, between plant and 014: STATISTICAL STUDIES IN FIELD GEOCHEMISTRY TABLE 5.—Ash contents of vegetables and element contents of vegetable ashes and garden [GM , geometric mean (in percent); GD geometric deviation; N, number of samples in which the element was not detected; _ ___, insufficient data; geometric means are in John B. McHugh, Harriet Neiman, A. L. Sutton, Jr. Blackeyed peas Element, or ash High-rate area Lowrate area Kinds. localities, and numbers of samples Cabbage Corn High-rate area Low-rate area High-rate area Low-rate area 29 samples 4 samples 28 samples 30 samples 29 samples 30 samples GM GD N GM GD N GM GD N GM GD N GM GD N GM GD N Ash _____________________________________ 5.9 1.26 0 5.2 1.09 0 19.0 1.41 0 21.0 1.16 0 4.4 1.47 0 3.7 .55 0 Al _______________________________________ .059 2.03 0 .052 1.64 0 .11 1.90 0 .39 2.45 0 .066 2.03 0 .068 1.99 0 B _______________________________________ .022 1.54 0 .014 1.22 0 .0083 1.72 l .0053 2.29 5 .0056 1.78 3 .0037 2.25 8 Ba ______________________________________ .013 2.06 0 .0077 1.20 0 .032 2.95 0 .045 1.67 0 .0054 2.37 0 .0047 2.12 0 Ca ______________________________________ 6.6 1.31 0 4.7 1.29 016 1.38 0 20 1.19 0 1.8 1.44 0 1.7 1.51 o 9 4 8 9 30 8 4 28 29 30 7 3 20 0 0 0 0 0 4 30 0 0 4 30 0 0 29 30 14 0 0 .................. 4 29 __________________ 4 29 _ 30 .013 1.40 0 .062 2.34 0 .088 1.89 0 .0052 175 0 .0050 1.95 0 .0018 2.96 0 .0085 237 0 .025 282 0 .0020 500 3 .0021 3.79 1 ..28 __________________ 4 .0014 1 23 25 .................. 25 .................. 29 .................. 30 _____________ 4 25 .0000421 1 2 26 29 30 ............. 4 26 28 29 30 .065 1.09 0 .034 2.57 0 .021 2.39 0 .084 1.44 0 .085 1.38 0 4 .0000463 5.4 22 .00035 6.3 21 __________________ 28 __________________ 30 soil compositions Within the high- and low—death-rate areas, has been examined by means of correlation coef- ficients (tables 8—11). The concentration of some elements in soil (in an available form) is commonly assumed to determine the amount that the plant will absorb, unless the element is present in extremely large, or toxic, quantities or as toxic compounds. The major and minor essential ele— ments obviously must be absorbed in quantities sufficient for metabolism of the plant, regardless of their relative GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA 015 soils from two areas in Georgia that have difl'erent cardiovascular mortality rates italic where they are significantly different for the high- and low-death—rate areas at the 95-percent level of confidence. Analysts: John C. Hamilton, Thelma F. Harms, Barbara Tobin, and James H. Turner] Kinds, localities, and numbers of samples—Continued Green beans Lima beans Tomatoes Garden soils High-rate area Low-rate area High-rate area Low-rate area High-rate area Low-rate area High-rate area Low-rate area 30 samples 30 samples 30 samples 15 samples 30 samples 30 samples 30 samples 30 samples GM GD N GIII GD N GM GD N GM GD N GM GD N GM GD N GM GD N GM GD N 0.90 1.88 0 0 .0025 1.91 15 13 .0068 1.42 o o .080 2.16 o 0 .0036 3.37 25 25 .00013 2.68 26 2 .0016 2.00 0 . 0 .00099 1.85 0 . o .55 2.59 0 . o .................. 22 . o .041 2.30 0 . 0 .0018 2.86 19 0033 2.14 12 .027 1.84 o .24 1.95 o .0099 214 o .041 1.84 o .................. 30 30 .0013 1. 34 5 .0016 1. 52 5 .00018 4.02 19 .0018 1.70 0 .017 2. 1o 0 .045 1. 73 0 .00026 2.77 26 .0026 2.31 2 .00028 1.65 22 .00090 1.35 0 .00036 2.35 25 .0033 2.21 1 .19 1.36 0 .92 1-41 0 .0020 1.92 0 .0065 1.38 0 .0022 1.99 1 .0025 1.44 0 .00027 1.76 0 .00028 1.45 0 .................. 21 .0064 1.82 o 031 1 50 0 .018 1.48 0 abundances in the soil, or the plant could not live. It is by plants. Controlling factors in the absorption of the generally known, however, that many plants may absorb nonnutritive elements are poorly understood. A discus- larger quantities of these elements than are required for sion of the complex chemical and physiological proc- metabolic processes, if large amounts are available in esses involved in the absorption of elements by plants the soil—the so-called luxury consumption of elements is beyond the scope of this report. (See Sutclifl'e, 1962.) 016 STATISTICAL STUDIES IN FIELD GEOCI-IEMISTRY TABLE 6.—Ash contents of trees and element contents of tree ashes from GM, geometric mean (in percent); GD, geometric deviation; N, number of samples in which the element was not detected; .___, insufficient data; geometric means are in John B. McHugh, Harriet N eiman, A. L. Sutton, Jr., Kind, localities, and number 01 samples Blackgum Oak Persimmon Red maple Element, or ash, __ and kind of . sample High-rate area Low-rate area 11igh~rate area Low-rate area High-rate area Low-rate area High-rate area Low-rate area 30 samples 30 samples 30 samples 30 samples 30 samples 30 samples 30 samples 30 samples GM GD N GM GD N GM GD N GM GD N GM GD N GM GD N GM GD N GM GD N Al, stems ......... 0.52 1.69 0 0.63 1.41 0 0 0.33 1.98 0 0.17 2. 07 0 0.24 1.93 0 0.14 1.67 0 0.17 1.63 0 leaves... .91 2.12 0 1.1 1.58 0 0 .73 1.76 0 .24 2.50 0 .32 2.04 0 .19 2.44 0 .37 1. 92 0 Ash, stems. . 1.26 0 2.8 1.21 0 0 2.5 1.30 0 3.1 1.23 0 3.0 1.18 0 2.9 1.22 0 2.7 1.24 0 1.25 0 6.9 1.18 0 0 3.3 1.25 0 6.0 127 0 7.3 1.27 0 4.9 1.24 0 4.8 1.20 0 1.57 0 .036 1.40 0 0 .026 1. 51 0 .032 1 53 0 .022 1.71 0 .037 1.36 0 .035 1.34 0 1.75 0 .041 1.50 0 0 .044 1.54 0 .045 1 66 0 .036 1.80 0 .042 1.73 0 .036 1. 67 0 2.18 0 .98 1.67 0 0 .42 1.87 0 .25 2 29 0 .19 2.40 0 .35 2.82 0 .36 2.17 0 2. 54 0 .41 1. 73 0 0 . 19 1.97 0 .097 2 09 0 . 071 2. 57 0 . 14 2. 73 0 .13 2. 02 0 1.14 0 26 1.14 0 027 1. 22 023 1 24 0 20 1. 28 0 24 1.19 025 1. 20 0 1. 21 016 1.26 0 016 1.35 015 1 55 0 16 1.30 0 17 1. 21 016 1.26 0 3. 54 2 .019 6.14 3 28 .00067 2.38 14 _______________ 24 .00022 3.63 _____ 29 ............. 30 5.83 3 .040 7.67 3. 28 .00050 2.59 17 ............... 27 ..................... 30 _____________ 30 2.37 0 .00079 1.70 0 1 .00067 1.82 1 .00040 1.81 2 .00051 2.57 .15 3 .00046 1.77 1 2.13 1 .0010 2.00 0 0 .0011 1.81 0 .00034 1.79 3 .00050 2.02 .67 2 .00060 1.80 0 1. 88 0 .027 1. 80 0 0 .017 1.41 0 .023 2. 00 0 .025 1. 59 . 27 0 . 015 1. 51 0 1. 90 0 .010 1. 37 0 0 .017 1. 34 0 .0070 1. 52 0 .0064 1. 49 86 0 . 015 1. 45 0 1. 69 0 .19 2. 33 0 0 .19 1. 91 0 .10 1. 46 0 .16 2. 34 .47 0 .14 1. 59 0 1. 84 0 .26 2. 47 0 0 .38 1.89 0 .13 1. 56 0 .19 2.18 .96 0 .28 1. 89 0 1. 30 011 1.31 0 012 1.38 016 1.31 019 1. 25 .51 012 1.44 0 1. 23 019 1. 25 0 018 1.17 0 21 1. 24 0 23 1.19 . 26 015 1.31 0 _____ 29.....__ 27..____.. 28 .0022 4.41 17 00010 18.02 26 .00042 6.46 ._ 30___._.._ 30 ..... 30____.__. 29... 27 .0018 3.86 19 00024 11. 25 .0022 3.03 18 30.___.___ ..... 30 1.45 0 5.9 1.33 0 3.4 1.60 0 3.3 1.47 0 63 1.43 0 4.4 1.37 0 1.50 0 2.3 1.50 0 ..... ' 8.6 1.27 0 7.0 1.27 0 5.1 1.53 0 73 1.67 0 5.6 1.43 0 1.56 0 4.9 1.45 0 2.69 0 .95 1.86 0 .83 1.92 0 1.0 1.89 0 72 1.99 0 .62 1.86 0 1.80 0 .68 1.92 0 2.53 0 .96 2.11 0 1.4 1.93 0 1.3 1.76 0 89 2.35 0 .80 1.97 0 2.63 0 .67 2.37 0 _____ 29 30...... 30.......... 30...____.. 27 28 29...... 29 ..... 29 30..__.___ 30.... 30 29 30 29...... 30 2.24 1 .0060 2.01 1 .0011 1.72 0 0018 1.79 0 .0027 1.88 O .0043 2.20 0 .00075 2.26 5 .00086 2.22 3 3.31 1 .012 3.20 1 .0017 1.84 1 0034 2.29 0 .0018 2.25 0 .0019 1.87 0 .00040 3.15 12 .00050 3.13 10 1.37 0 1.2 1.21 0 .90 1.70 0 1 1.47 0 1.6 1.60 0 2.0 1.51 0 1.4 1.51 0 1.6 1.44 0 1.33 01.6 1.28 0 2.2 1.21 0 22 1.27 01.4 1.28 0 1.6 1.36 01.8 1.30 0 2.1 1.26 0 2. 88 1 .048 2. 34 0 .0096 2. 40 2 020 2. 90 0 . 011 2. 90 1 . 015 3. 85 1 .012 2. 59 1 . 023 3. 13 0 2. 01 1 .0096 2. 56 0 . 0070 2. 81 2 016’ 2. 40 0 0024 4. 13 8 .0068 2. 59 1 .0041 2. 92 4 . 010 2. 21 0 l. 93 0 . 5‘5 2.14 0 .15 1. 60 0 24 1. 72 0 .11 l. 91 0 . 11 1. 96 0 . 15 1. 87 0 . 18 1. 91 0 1. 72 0 .19 1. 04 0 . 082 1. 71 0 ('96 1. 66 0 . 054 2. 03 0 . 062 l. 98 0 . 068 1. 97 0 . 077 2. 04 0 1. 81 0 014 1. 81 0 . 00.91 1. 63 0 . 013 2. 13 0 .0096 2. 09 0 . 010 2. 18 f‘ . 0086 1. 81 0 . 0093 1. 85 0 2. 22 0 024 2 23 0 .024 1.62 0 .040 1.93 0 .013 2.27 0 .015 1.99 C .014 2.33 0 .021 1.84 0 2.24 23 00070 1 98 24 6 2.35 26 ............... 27 00046 2. 35 26 .00060 2.07 25 .00046 2.35 26 29 ..... 28 00058 280 23...... 27 .00061 2.99 22.............._ 29 .00040 2.65 26 . . 28. 28 7 35 24 ........ . 28 .00021 5.61 24 .00039 4.87 21 00010 19.05 23 .000074 8.55 26 ..... . 29 29 . . 9.71 25 .00010 6 55 26 .00038 7.06 20 .0014 2.89 11 .00014 18.93 22 .00054 3.38 20 ....... . 27 Yb, stems. . .0000059 10. 37 26 _____________ 26 ............. 29 .0000088. 7. 30 26 .0000084 10. 36 25 .............. 29 ............. 28 leaves... 0000090 9 71 25 ............. 25 ............. 27 .000067 3.37 18 000012 14 43 23 ............... 27 ............. 28 Zn, stems 091 1 65 0 .097 1 47 0 .034 1 53 0 .063 1.74 0 079 1 46 0 11 1 61 0 048 1 40 leaves... 026 1 39 0 .027 1 37 0 .044 1 33 0 .066 1.32 0 017 1 58 0 020 1 32 0 046 1 41 Zr, stems. . 00021 7.22 23 .00010 6 55 26 .00013 6 70 25 .000050 11.97 26 00024 4.61 24 .............. 29 ............. leaves .......... .00097 4.11 15 .00049 7 08 19 .00086 3 97 16 .00075 6.66 17 00027 4.91 23 .............. 28 .00065 4 19 GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA 017 two areas in Georgia that have difierent cardiovascular mortality rates italic where they are significantly different for the high— and low-death-rate areas at the 95-percent level of confidence. Analysts: John C. Hamilton, Thelma F. Harms, Barbara Tobin, and James H. Turner] Kind, localities, and number of samples—Continued Sassafras Smooth sumac Sweetgum Black cherry High-rate area Low-rate area High»rate area Low-rate area High-rate area Low-rate area High-rate area Low-rate area 17 samples 27 samples 30 samples 30 samples 28 samples 27 samples 30 samples 30 samples GM GD N GM G D N GM GD N GM GD N GM GD N GM GD N GM GD N GM GD N 0.39 2. 42 0 0.37 2. 67 0 0. 19 1. 73 0 0.23 1.90 0 0.20 1. 48 0 0.37 1. 77 0 0.15 2. 64 0 0. 24 2. 26 0 . 43 2. 26 9 .63 2.04 0 .26 1. 86 0 .61 2.12 0 .62 1. 54 0 .85 1. 77 0 .23 1. 90 0 .29 1. 94 0 1. 9 1. 31 0 1. 9 1. 26 0 3. 6 1. 22 0 3.8 1. 24 0 4. 8 1. 32 0 4.1 1. 37 0 2.0 1. 71 0 2. 3 1. 30 0 4. 7 1. 21 0 6. 7 1.13 0 4. 6 1.20 0 4. 8 1.32 0 5. 6 1.15 0 5. 6 1. 21 0 5. 4 1. 22 0 6. 5 1. 21 0 .032 1. 70 0 .025 1. 64 0 .030 1. 48 0 .023 1. 67 0 .026 1. 38 0 . 019 1. 39 0 . 036 1. 75 0 .034 1. 48 0 .030 1. 51 0 .016 1. 46 0 .032 1.54 0 .025 1.84 0 .040 1. 51 0 .028 1. 48 0 .029 1. 43 0 .023 1. 27 0 . 30 2. 26 0 .26 2. 37 0 .45 1.97 0 .32 2. 60 0 .25 1. 99 0 . 28 2.03 0 .36 2. 93 0 .30 3.12 0 .080 1. 99 O ' .071 1. 76 0 . 19 1. 84 0 .13 2. 42 0 .14 1. 99 0 . 13 2. 39 0 .24 2. 59 0 .21 2.80 0 21 1. 35 0 19 1.37 0 25 1. 27 0 24 1. 27 0 26' 1.17 0 23 1. 31 0 26 1. 17 0 25 1. 23 O 19 1. 25 0 16 1. 27 0 19 1. 21 0 18 1. 22 0 19 1.16 0 16 1. 23 0 23 1.16 0 21 1. 20 0 ________________ 17______...,_____. 26................ 29 .00014 4.02 25 .00020 3.72 22 .00025 1.92 24 .000055 10.11 25 .00037 2.39 21 17 ________________ 27 ________________ 30 _________________ 30 ................. 26 .................. 27 __________________ 30 ............ _ _ _ 30 1 .0010 2. 47 0 . 00052 1. 67 0 . 00040 2.07 3 .00046 1. 60 0 .00069 1. 73 0 . 00050 1. 69 1 . 00061 .99 2 0 .0013 1. 90 0 . 00050 1. 56 0 . 00072 1. 93 1 .00056 1. 62 0 . 00091 2. 19 0 . 00037 1. 83 2 . 00051 1. 92 2 0 .017 1. 34 0 .016 1. 56 0 .016 1.49 0 .013 2. 20 0 .015 l. 46 0 .017 1. 60 0 .018 1. 52 0 0 .012 1. 34 0 .0086 1. 47 0 .0084 1. 44 0 . 013 1.92 0 . 013 1. 42 0 .0077 1. 50 0 .0069 1. 39 0 0 .25 3. 21 0 .12 1. 66 0 .16 1. 92 0 . 081 1. 48 0 .11 2. 00 0 .15 1. 81 0 .19 2. 52 0 0 .35 2. 24 0 .15 1. 91 0 .29 2. 25 0 . 15 2. 06 0 . 19 1. 89 0 .14 1. 75 0 .21 1. 63 0 0 20 1. 33 0 14 1. 53 0 10 1. 41 0 6‘. 1 1. 89 0 11 2.10 0 12 1. 45 0 14 1. 51 0 0 23 1.22 0 10 1. 29 0 21 1.25 0 10 1.31 0 1.6 1.28 0 14 1.37 0 17 1.27 0 25 ________________ 28 . 0012 5. 37 21 .00018 12. 00 24 . 0022 3.04 16 ................. 27 .0014 5. 14 20 25 . 00020 13. 98 25 .0024 4. 59 17 ................. 25 . 0022 3. 23 16 00015 12. 79 26 .0017 4. 73 19 61 0 2.9 1. 53 0 2.3 1.41 0 2.6 1.46 0 3.1 1.63 0 4 8 1.73 0 4.0 1.49 0 46 0 5.2 1. 55 0 3.0 1.49 0 7.0 1.29 0 5.8 1.42 0 6 8 1. 73 0 6.7 1.26 0 14 0 .28 2.14 0 .20 2. 26 0 .41 1. 99 0 . 68 1. 74 0 57 2. 46 0 .56 2. 27 0 89 0 .14 1. 62 0 .097 1.85 0 .96 1.90 0 1. 3 1. 67 0 57 2. 44 0 .37 2.07 0 07 21 ________________ 29 ................. 29 ................. 28 27 _________________ 29 17 ________________ 26 ................ 28 ................. 29 ................. 28 27 _________________ 29 . . 0 . 0014 1. 72 0 .00074 2. 36 5 . 00079 3. 23 7 0012 1. 93 . 1 .0030 2. 34 0 00062 2. 94 8 . 3. 02 . . 0 . 0018 1.80 0 . 00042 3. 15 11 . 3. 06 9 0029 2. 07 0 .0063 2. 11 0 0014 1. 65 0 . 2. 24 . . 0 2. 5 1. 36 0 1. 7 1. 67 0 1. 41 0 .71 1. 7 0 1. 2 1. 66 0 2 1. 60 0 2. 3 1. 48 0 . . 0 2.2 1.28 0 2.2 1.35 0 1.23 0 1 4 l. 42 0 1.9 1.33 0 2 0 1.48 0 1.0 1.36 0 . . 0 . 014 4. 01 2 . 011 3. 91 3 3. 00 1 012 2. 86 1 . 011 3. 31 1 021 2. 71 0 . 027 4. 64 l . . 0 . 0095 2. 28 0 . 0067 2. 29 1 2. 25 0 0045 2. 47 3 . 0091 2. 07 0 0033 4. 65 7 0058 2. 40 1 . 2. 0 .17 1. 85 0 . 22 1. 71 0 . 22 0 12 1. 87 0 . 15 2. 03 0 10 1. 68 0 . 24 2. 29 0 . 2. 0 . 003 1. 91 0 . 11 1. 59 0 . 28 0 062 1. 80 0 . 075 1. 98 0 12 1. 74 0 . 16 2. 11 0 . 2. . 0 . 020 3. 64 0 . 015 2. 27 0 . 00 0 0077 1. 84 0 . 0060 2. 40 0 013 1. 82 0 . 011 2. 37 0 . 2. 17 0 .028 2. 42 0 . 020 2. 01 0 .47 0 012 1. 94 0 . 014 2.09 0 017 1 92 0 014 1. 95 0 . 1. 83 7 . 00074 3. 13 18 ................ 27 8 . ...... 27 00030 3 45 26 00067 1. 68 26 . 00070 2. 05 12 . 00062 2.83 20 ................ 27 25 ................. 28 30 16 27 . 000090 9. 71 25 20 _ 25 . 00020 0. 20 21 26 0 26 ________________ 28 29 22 25 000015 8. 42 24 ................. 13 0 074 1. 43 0 078 1 64 0 0 036 1. 33 0 037 1 44 0 20 00018 8 77 23 _________________ 28 27 21 00062 5 00 18 . 00013 14 62 23 25 C18 STATISTICAL STUDIES IN FIELD GEOCHEMIISTRY TABLE 7.—Element contents of uncultivated soils from two areas in Georgia that have difi‘erent cardiovascular mortality rates [GM, geometric mean (in percent); GD, geometric deviation; N, number of samples in which the element was not detected; _ __ _, insuflicient data; geometric means are in italic where they are significantly diflerent for the high- and low-death-rate areas at the 95- McHugh, Harriet Neiman, A. L. Sutton, J r., Barbara Tobin, and James H. Turner percent level of confidence. Analysts: John 0. Hamilton, Thelma F. Harms, John B. Kinds, localities, and numbers of samples A horizon B horizon C horizon Element High-rate area Low-rate area High-rate area Low-rate area High-rate area Low-rate area 30 samples 30 samples 30 samples 30 samples 30 samples 30 samples GM GD N G'M GD N GM GD N GM GD N GM GD N GM GD N 1.1 2.09 0 4.4 1.97 0 1.1 1.86 0 5.7 2.00 0 1.7 2.61 0 >6'.8 >1.7 0 . 0018 2. 55 20 .0022 3. 08 17 . 0020 1. 85 20 . 0025 2. 87 16 .0018 2. 03 21 . 0020 2. 94 18 . 010 1. 73 0 . 029 1. 85 0 . 0086‘ 1. 68 0 .027 2. 00 0 .0094 1. 83 0 . 029 1. 77 0 .12 1 80 7 .31 1.44 1 .10 1.51 9 .33 1.37 0 .12 1.81 7 .35 1.28 0 ..................... 24 . 0079 1. 58 25 . 0050 2. 19 26 .0058 2. 43 24 . 0076 1. 76 24 . 0058 1. 81 27 . 00015 3. 08 24 .00098 1. 74 2 . 00010 3. 19 26 .0010 1.81 l . 00019 2. 60 23 . 0011 1. 64 0 . 0011 1. 99 0 .0041 1. 79 0 .0011 1. 81 0 .0044 l. 63 0 . 0016 2. 08 0 . 0047 1. 58 0 . 00087 1. 77 0 . 0026‘ 1. 72 0 . 00088 1. 77 0 . 0029 1. 80 0 . 0011 2. 04 0 . 0033 1. 88 0 . 47 2. 27 0 2.1 1. 68 0 . 48 2.04 0 2. 6 1. 74 0 . 65 2. 62 0 3. 0 1. 59 0 . 0015 3. 57 20 . 0017 2. 32 1 . 00019 2. 31 20 . 0021 1. 96 0 . 00037 2. 21 11 . 0029 1. 85 0 .073 2.85 0 1.2 1.98 0 .079 2.84 0 1.1 1.99 0 .089 2.83 0 1.2 1.97 0 .0026 2. 40 14 .0034 1. 73 10 . 0024 2. 08 16 . 0037 1. 86 8 .0030 2. 03 13 . 0032 1. 64 10 . 026 2. 48 0 . 25 2. 19 0 .025 2. 17 0 .29 2. 16 0 .041 3. 05 0 .37 2. 09 0 .013 1. 86 0 . 032 1. 64 0 0085 2. 03 0 . 028 1. 95 0 .0061 l. 87 0 . 022 1. 76 0 . 0029 9. 61 24 . 21 2. 82 1 .................... 27 . 20 2. 86 1 .................... 27 . 20 2. 35 0 .0012 1.33 4 .0016 1.43 3 .0013 1.34 3 .0018 1.43 2 0013 1 51 6 .0019 1.40 1 . 00092 4. 85 26 ____________________ 28 .................... 28 . 0028 2. 33 24 .................... 27 ____________________ 29 . 000070 8.53 22 .0014 2. 10 1 . 0015 4. 25 20 .0017 1. 95 1 . 3. 3O 12 . 0021 1. 65 0 .0065 1. 86 0 .020 2. 09 0 . 0038 1.90 3 .014 2. 08 0 .0035 2. 28 4 . 014 2. 14 0 . 00028 4. 24 23 .0021 l. 83 3 .................... 27 . 0017 1. 68 5 .00026 4. 12 24 .0017 1. 58 4 . 00021 2. 64 22 . 00086 1. 47 1 . 00026 2.06 21 . 00099 1. 44 0 . 00050 1. 67 10 . 0011 1. 43 0 . 00057 2. 73 19 . 0030 2. 22 1 . 00065 1. 61 21 .0027 1. 98 1 . 00090 1. 84 13 .0029 1. 82 0 .17 1.94 0 .29 1.50 0 .18 1.64 0 .36 1.55 0 .21 1.65 0 .36 1.47 0 . 0015 2. 14 4 . 0057 1. 58 0 .0016 1. 90 4 . 0066 1. 40 0 .0025 2. 43 4 . 0077 1. 36 0 .0021 2. 04 1 . 0026 1. 83 0 .0020 1. 77 0 .0026 2. 00 0 . 0025 2. 30 1 . 0021 1. 57 0 . 00025 1. 95 0 . 00035 2. 01 0 00025 1. 68 0 . 00034 1. 97 0 .0031 2. 31 1 . 00028 1. 62 0 _______________________ 23 .0040 1.64 1 27 .0041 1.61 2 27 .0043 1.59 0 .026 1. 60 0 .020 1. 39 0 029 1. 62 0 .021 1. 51 0 .025 1. 60 0 .016 1. 60' 0 The diagrams in figures 3—6 provide graphical sum« maries of the correlation coefficients in tables 8—11. All the diagrams are similarly constructed. The diagram in figure 3, for example, summarizes the correlations between element contents in each of six vegetables and the soils of the gardens in which the vegetables grew. The garden soil is represented by the point at the center of the circular diagram. Each radius extending from this point represents a garden vegetable, as indicated. The distance of each point on the radius from the center of the circle is inversely proportional to the degree of positive correlation between the element content of the particular vegetable and the garden soil, among the sampling sites. Where the point lies near the center of the circle, the concentration of the element in the indi- cated vegetable shows a high degree of positive correla- tion with the concentration of the element in the soil. Where the point lies near the inner circle, the concen- tration of the element in the vegetable is estimated to be nearly independent of the amount of the element in the soil. Where the point lies near the outer circumfer- ence of the circle, the concentration of the element in the vegetable displays a negative correlation with the concentration of the element in the soil; that is, vege— tables containing higher concentrations of the element tend to occur in gardens where the concentration of the element in the soil is relatively low. Figures 4, 5, and 6 contain similar diagrams sum- marizing the correlations of elements in the stems and leaves of trees with the amounts of the elements in the C, B, and A soil horizons, respectively. They also sum- marize the correlations of element concentrations among the three soil horizons. I GEOCHENIICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA 019 TABLE 8.—Correlations between the amounts of elements in vegetables and garden soils in the high- and low-death-rate areas of Georgia [r, product-moment correlation coefficient between logarithms of concentrations; N, number of pairs ommitted because concentration of element in one or both kinds of samples (vegetable or soil) was beyond the range of measurement by the analytical methods that were used; ...., insufficient data] Blackeyed peas Cabbage Com Green beans Lima beans Tomatoes Element High- Low- High- Low- High- Low- High- Low- High- Low- High- Low— death-rate death-rate death-rate death-rate death-rate death-rate death-rate death-rate death-rate death-rate death-rate death-rate area, 29 area, 4 area, 28 area, 30 area, 29 area, 30 area, 30 area, 30 area, 30 area, 15 area, 30 area, 30 pairs pairs pairs pairs pairs pairs pairs pairs pairs pairs pairs pairs r N r N r N r N r N r N r N r N r N r N r N r N 0.38 0 —0. 68 1 0. 45 0 0. 19 4 0. 13 0 —0. 16 4 0. 26 0 0. 11 4 0. 13 0 0. 29 1 -0. 18 0 —0. 10 4 .05 14 ......... 2 .36 13 .18 15 —.01 16 .17 18 -.37 15 .12 13 — 24 15 -. 46 9 .05 6 —.36 18 .14 0 —.83 0 .53 0 -. 19 0 20 0 —.44 0 . 14 0 .13 0 13 0 —.l9 0 -.30 10 —.58 0 .06 0 —.75 0 .06 0 .25 0 .12 0 .00 0 .00 0 —.03 0 —11 0 —.29 0 .01 0 .34 0 .......... 28 ._....._. 4 ......... 28 ._...__.. 29 .._.__._. 29 ......... 30 ......... 29 .50 15 .00 28 —.94 10 ._._.____ 30 ...._..._ 29 -.08 17 ......... 3 .10 8 .11 4 .30 20 .25 20 -. 10 ll .14 10 —.34 26 .16 9 -.26 11 .42 25 —.05 0 .00 0 —.11 0 .50 0 —.10 0 .08 0 —.39 0 .49 0 —.31 0 —.65 0 .09 0 .65 0 —.09 0 -.04 0 .17 0 .22 0 —.08 0 .19 0 —.09 0 .05 0 ——.03 0 .06 0 --.26 0 —. 12 0 —. 39 0 .76 0 — 06 0 —.07 0 —. 41 0 .16 0 — 28 0 —.02 0 —.03 0 .29 0 .01 0 .24 0 .23 0 .77 0 40 0 .03 0 .28 0 .42 0 — 11 0 .21 0 -.08 0 —.22 0 -.02 0 —.01 0 —.09 0 —-.04 0 .17 0 .20 0 -—.08 0 —.10 0 —.27 0 .09 0 .08 0 .44 0 —.29 0 .01 0 .11 18 .65 l _________ 27 .50 25 —. 92 26 .45 14 .44 20 .07 2 —. 10 19 .15 0 ......... 28 ......... 30 .20 0 —.05 0 .ll 0 —.06 0 .07 0 -. 16 0 . 11 0 —. 02 0 22 0 -. 53 0 .13 0 .07 0 —. 72 24 . 89 1 —. 77 23 —. 08 l —. 41 24 —. 34 1 25 25 . 02 1 —. 01 25 . 18 l —. 33 26 . 01 1 ' —. l3 0 —. 53 0 —. l7 0 -. 09 0 . 26 3 . 01 l 07 0 01 0 —. 49 4 . 06 l —. 41 1 —. 10 6 V ________________________ 28 _________ 4 ......... 25 —. 37 25 ......... 29 ......... 30 ......... 29 ......... 29 ......... 30 _________ 14 —. 50 27 ......... 30 VEGETABLES AND GARDEN SOILS Comparison of the compositions of garden soilsfin the high- and low-death-rate areas indicates that those in the high death-rate area tend to contain significantly lower concentrations of nearly all elements studied (table 5). This difl'erence, which reflects the difference in the character of the bedrock geology in the two areas and the greater maturity of the soil in the high-death- rate area, is not apparent in the compositions of most of the vegetables, with the possible exceptions of cab- bage and green beans. Cabbages from the high-death- rate area, like the garden soils, tend to contain lower concentrations of aluminum, chromium, iron, nickel, and titanum; other elements, except for boron and molybdenum, appear to be about equally concentrated in cabbages from the two areas in spite of their gener- ally higher concentrations in soils of the low-death-rate area. The concentrations of several elements in green beans similarly appear to reflect the compositional dif- ferences in the soils between the two areas, but those in the other vegetables, in general, do not. The correlations between the element contents of vegetables and of garden soils within each of the two areas, as indicated by the correlation coefficients (fig. 3 and table 8), also fail to indicate any general relation- ship between vegetable and soil compositions. About half of the correlation coefficients are positive and half are negative, and there seems to be little or no con- sistency of correlation within elements, vegetables, or areas. The indication is that the greater number of the coeflicients are spurious, resulting largely from chance, and are not reproducible. It is reasonable to conclude, therefore, that the com- positions of garden vegetables within both the high- and the low-death-rate areas are generally independent of the compositions of the soils in which they are grown. Some correspondence between the composition of cab- bages and green beans and the compositions of the soils does appear to be present, on a ‘broader scale, between the high- and low-death-rate areas; these two vege- tables tend to exhibit some of the same compositional differences as do the soils from the two areas. C20 HIGH-DEATH—RATE AREA Bluckeyedl peas Green beans 0 ’e O e. \ STATISTICAL STUDIES IN FIELD GEOCHEMISTRY LOW-DEATH-RATE AR EA ALUMINUM BARIUM BORON CALCIUM HIGH-DEATH-RATE AREA LOW—DEATH-RATE AREA § § § COBALT COPPER FIGURE 3.—Correlations between the amounts of elements in garden soils and the amounts in vegetables from the high and low cardiovascular death-rate areas of Georgia. Distance of the point on the radius from the center of the circle is inversely proportional to the correlation of the amount of the element in the vegetable and the amount in the garden soil. GEOCHEMICAL ENVIRONMENTS, HIGH-DEATH-RATE AREA '15 0.5 -1 -0.5 0.5 MAGNESIUM MANGANESE NICKEL PHOSPHORUS LOW-DEATH-RATE AREA FIGURE 3.—Continued Wfi®® CARDIOVASCULAR MORTALITY RATES, HIGH-DEATH-RATE AREA -1 40,5 0 0.5 -1 41.5 0 0.5 -1 45 o 15 TITANIUM -1 —o.5 o 0.5 C21 LOW-DEATH-RATE AREA thkeyedlpaas GEORGIA 3’ ‘3: f k g. 2. Q. 9:» POTASSIUM Gm" W“ -l -0.5 0 0.5 STRONTIU M VANADIUM 022 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY HIGH-DEATH-RATE AREA LOW-DEATH-RATE AREA A hori_zlon HIGH-DEATH-RATE AREA LOW- DEATH-RATE AR EA .1 ‘1 -05 '(15 E a 3 5 a’ “a w E -k a.“ s -\\‘ I‘ 9 4, Maple Yd ALUMINUM -1 gfié COBALT -l %fi .1 ,1 % % BARIUM COPPER -1 -] ‘l ‘05 415 '05 0 BORON GALLIUM .1 -l -05 '05 0 CALCIUM -1 -l -l -05 0 CHROMIUM LANTHANUM FIGURE 4.—Correlations between the amounts of elements in the C horizon of uncultivated soils (center of circles) and the amounts in the A and B soil horizons and in the stems and leaves of trees. Triangles represent soil correlations; dots represent stems, squares represent leaves, of trees. Distance of the symbols on the radii from the center of the circles is inversely proportional to the correlation of the amounts in the samples. GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA C23 HIGH-DEATH-RATE AREA LEAD MAGNESIUM -1 --05 o MANGANESE NICKEL —1 was 4) NIOBIUM LOW-DEATH-RATE AREA FIGURE 4.——Continued HIGH-DEATH-RATE AREA -1 415 0 LOW- DEATH—RATE AREA A horiztlm I" ’6‘, PHOSPHORUS MBP'e -1 --a5 0 POTASSIUM -1 --n5 0 as SCANDIUM SODIUM STRONTIUM 024 STATISTICAL STUDIES IN FIELD GEOCI—IEMISTRY HIGH-DEATH-RATE AREA LOW-DEATH-RATE AREA HIGH-DEATH-RATE AREA LOW-DEATH-RATE AREA Ah ' ‘d organ 0 -1 -1 -1 c.“ 4.5 415 415 f a o; a U’ E 39 § ’0 6'9 " Maple V TITANIUM YTTRIUM -1 -1 —1 -| —us -us 415 «a VANADIUM ZINC 'l 'l -l .45 -a.5 «15 fl 0 YTTERBIUM ZIRCONIUM FIGURE 4.—Continued GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA TREES AND UNCULTIVATED SOILS Comparisons of the element content of the three hori- zons of uncultivated soils (table 7) and of eight species of trees (table 6) from the two death-rate areas indicate that concentrations of most elements in both soils and trees tend to be higher in the low-death-rate area. Similar trends were shown earlier in this report for garden soils but the trends were not apparent, for the most part, in the garden vegetables. The ability to dis- criminate between the two areas on the basis of tree analyses as a whole contrasts with discriminations based on vegetable analyses, and the better discrimination using trees may be caused by their growth habit which also contrasts with that of vegetables. All vegetables that were sampled are annuals that grow only one season and are then harvested before they are mature. Thus they have only a short period of time available for ab- sorbing elements from the soil, whereas trees may absorb elements for periods of many years. Vegetables have shallow ‘root systems, in contrast to trees which have deep and widespread roots. Trees, therefore, may be more effective than vegetables in reflecting the chemical nature of the soil on which they grow. Of the eight species of trees, oak appears to be the most sensitive to the variation in element content of soils from the two death-rate areas. This relationship of oak to uncultivated soils resembles that of cabbage to garden soils. Sweetgum trees follow oak in sensitivity to soil C25 composition, and sassafras and smooth sumac seem to be the least sensitive species (table 6) . Stems and leaves appear to vary independently in containing amounts of elements that are significantly diflerent in the two areas. Only some elements (for ex- ample, nickel and lead) in some trees (for example, blackgum and oak) show significant differences in con- tents in stems and leaves simultaneously. Caution should be used in making generalizations of tree-soil relationships based on data presented in this report. The concentrations of the elements that were studied in Georgia soils are all believed to be within the “normal” range. The fact is well known that some trees, if growing on soils that contain highly anomalous amounts of certain elements (for example, soils over— lying mineral deposits), contain anomalous amounts of the elements in which the soil is enriched. Correlations of the amounts of elements in tree stems and in leaves and the three horizons of uncultivated soils are given in tables 9—11 and are illustrated in figures 4—6. These correlations fail to indicate any general relation- ships between tree and soil compositions. As with the correlations of vegetables and garden soils, about half of the correlation coeflicients given in these tables are posi— tive and half are negative, and little or no consistency is apparent within elements, trees, or death—rate areas. These results indicate that the greater number of coeffi— cients result largely from chance and are not reproducible. 026 STATISTICAL STUDIES IN FIELD GEOCHEMSTRY TABLE 9,—C'orrelat7'ons between the amounts of elements in trees and in the [r, product-moment correlation coefficient between logarithms of concentrations; N, number of pairs omitted because concentration of element in Blackgum Oak Persimmon Red maple High-death— Low-death- High—death- Low-death- High-death- Low-death- High-death- Low—death- rate area, 30 rate area, 30 rate area, 30 rate area, 30 rate area, 30 rate area, 30 rate area, 30 rate area, 30 Element, and plant part analyzed pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of stems and stems and stems and stems and stems and stems and stems and stems and leaves leaves leaves leaves leaves leaves leaves leaves r N r N r N r N r N r N r N r N A], stems. —0 31 O —0. l7 9 —0. 24 0 —0. 11 9 —0 23 1 —0. 26 9 —0. 13 0 9 leaves. — 28 0 —. 19 9 -—. l9 0 —. 22 9 - 11 0 —. 24 9 —. 18 0 9 . 36 21 . 41 18 —. 13 21 —. 35 18 — 09 22 —. 02 18 —. 30 21 18 . 25 21 . 10 18 —. 25 21 . 16 18 . 18 21 . 24 18 . 15 21 18 .20 0 —. 13 0 .10 0. —.54 0 1 -.33 0 .23 0 0 .16 0 —.33 0 .19 0 —.33 0 25 0 —.34 0 .34 0 0 —. 36 7 . 24 0 . 05 7 —. Z} 0 03 7 —. 14 0 —. 08 7 0 —. 12 7 .08 0 .02 7 —.31 0 — 10 7 —. l5 0 —. 12 7 0 . 02 23 . 22 3 .......... 30 —. ll 14 .......... 29 —. 07 23 .......... 29 30 . 65 24 . 30 3 .......... 30 . 21 17 .......... 28 .......... 30 .......... 29 30 . 16 0 -.01 0 .05 1 .11 1 . 01 3 —. 16 3 . 10 3 l .28 1 .08 0 .14 0 —.26 0 .27 3 —.15 1 —.13 2 0 —. 17 0 .31 0 —.11 0 .11 0 .19 1 .09 0 —.37 0 0 . 18 0 —. 08 0 —. 34 0 —.06 0 —. 13 0 . 18 0 —. 13 0 0 —.11 0 —.23 0 .02 0 —.02 0 —-.09 l —.27 0 .10 0 0 . 22 0 —. 38 0 . 13 0 —. 22 O .05 0 -—. 09 0 . l5 0 0 —. 36 0 —. 15 0 . 13 0 —. 10 0 —. 09 0 . 19 0 —. 24 0 0 . 16 0 —.34 0 .38 0 —. 5O 0 —. 10 0 —.07 0 —. 05 0 0 .02 1 .15 3 —.30 0 .20 0 .35 2 .01 0 —.09 0 0 .25 24 . 43 18 —. 04 6 . 16 3 —. 02 12 --. 11 2 —. 28 2 0 —. l5 0 —.06 0 —.25 0 .03 0 . 10 l —.05 0 -—. 14 0 0 —. 06 0 .03 0 —. 35 0 . 07 0 —. 07 0 —. 01 0 —. 13 0 0 -—. 19 13 .38 1 —. 43 12 . 30 0 . 14 13 —. 08 0 . 62 15 3 . 34 13 .45 l . 04 13 . 23 0 . 25 12 —. 26 0 . 57 20 0 .15 4 —.00 0 .12 0 .11 0 .17 4 .24 0 .02 4 0 .04 4 .26 0 .38 0 .13 0 .16 4 .19 0 .21 4 0 —.22 24 .11 4 .13 24 —.13 4 .06 24 .15 5 —. 57 4 00 24 19 4 —. 17 24 .12 4 —. 07 25 . 28 5 —. 18 4 21 13 04 0 . 58 13 —. 03 0 . 26 14 .03 0 .59 0 02 13 23 0 .47 13 . 10 0 . 15 13 . 07 0 . 59 0 —- 24 0 — 17 0 —. 03 0 —. 17 0 -. 34 1 -. 16 0 . 17 0 — 09 O - 15 0 .06 0 —. 16 0 —. 23 0 —. l7 0 —. l3 0 44 24 29 24 .......... 26 . 91 27 .......... 28 .......... 25 —. 00 29 __________ 28 08 23 —.26 27 .16 22 29 —.12 26..."... 28 -— 62 27 — 20 0 .......... 27 —- 11 0 —.01 27 —.31 0 .50 0 . ______ 27 — 30 0 .......... 27 — 34 0 —.93 27 . 18 10 . 26 0 TABLE 10.—Correlations between the amounts of elements in trees and in the [r, product-moment correlation coefficient between logarithms of concentrations; N, number of pairs omitted because concentration of element in Blackgum Oak Persimmon Red maple High-death- Low-death- High-death— Low-death- High-death- Low-death— High-death— Low-death- rate area, 30 rate area, 30 rate area, 30 rate area, 30 rate area, 30 rate area. 30 rate area, 30 rate area, 30 Element, and plant part analyzed pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of stems and stems and stems and stems and stems and stems and stems and stems and leaves leaves leaves leaves leaves leaves leaves leaves r N r N r N r N r N r N r N r N Al, stems .................................... -0. 12 o —0. 33 7 0 —o. 18 1 —0. 19 7 —0. 18 0 0.06 7 ............. —.17 0 —. 17 7 0 —. 19 0 -.09 7 .01 0 —. 27 7 ............. —. ll 20 .09 16 20 —. 19 20 31 16 —.61 20 .06 16 ............. —.41 20 --.17 16 20 -.11 20 .24 16 .04 20 .44 16 _____________ .32 0 .01 0 0 -. 59 1 —.58 0 .30 0 —. 12 0 ___________ .35 0 —.40 0 0 —. 52 0 —.56 0 .33 0 -.33 0 ___________ —.13 9 .08 0 9 .08 9 —.19 0 —.44 7 .03 0 ........... —.16 9 —.02 0 9 .02 9 —38 0 —29 9 .18 0 ........... .93 26 .09 4 30 —.16 30 36 23 ... 30 ..... 30 ___________ .68 26 4 30 .32 7 30 30 .......... 30 ..... 30 ___________ .21 0 0 1 . 13 1 3 01 3 .29 3 .09 l . 18 1 0 0 -. 02 0 3 03 l .04 2 . 14 0 —. 08 0 0 0 . 12 0 1 — 02 O 07 0 —. 12 0 .01 0 0 0 —.11 0 0 .14 0 —.03 0 -.20 0 ....... —. 04 0 0 0 . 11 0 1 — 03 0 —. 04 0 —. 13 0 .12 0 0 0 —. 05 0 0 —. 01 0 . 13 0 . 10 0 —.38 0 0 0 —.14 0 0 16 0 —.20 0 .14 0 ...... .18 0 0 . 0 —.53 0 0 —.14 0 —.12 0 —.03 0 ........ .08 7 3 —.33 0 .13 0 3 - 04 0 —.29 0 .11 0 .......... .16 24 18 —.19 6 .18 3 12 06 2 —.22 2 .28 0 .......... —.l5 0 0 —.25 0 —.l6 0 1 — 11 0 -.10 0 —.60 0 __________ —.08 0 0 —-.32 0 —.27 0 0 — 03 0 —-.l9 0 —.62 0 Ni. stems... ........ .51 21 2 .39 20 .12 1 21 — 07 1 .51 20 .30 4 leaves ........... . 64 21 2 50 20 . 17 1 . 20 — 25 1 . 47 22 . 52 11 P,stems__ ...... .27 3 0 28 3 —.06 0 . 3 17 0 .24 3 —.05 0 leaves. . . .......... .40 3 o .40 3 —. 11 o -. 22 3 14 0 . 38 3 .02 0 Pb.stems. ............ —.99 27 5 —.5o 27 —.02 5 —.99 27 .26 6 -.82 27 .20 5 leaves.. . .......... —. s7 27 5 —. 73 27 . 1o 5 —. 98 27 22 6 —. 87 27 21 5 Sr, Stems.. ........ .01 21 1 —. 21 21 —. 02 1 —. 27 21 05 1 —. 01 21 09 1 leaves... ........ 31 21 1 . 03 21 —. 01 1 -.1e 21 . 11 1 -.11 21 22 l T1,stems. __________ — 28 o o .03 o —.22 o —.1-1 1 —.01 0 .14 0 - 35 0 leaves. — 09 0 0 . l6 0 — 04 0 -—. 15 0 -.05 0 —. 06 0 04 0 V stems — 43 24 24 —.00 26 28 — 80 25 —.00 26 .......... 29 leaves.. .......... — 01 2s 23 —. 7s 27 29 —. 07 26 —. 01 28 .......... 28 Zn stems ________________ 27 2 __________ 27 27 —. 29 2 __________ 27 18 2 leaves _______________________________________________ 27 2 __________ 27 27 23 2 __________ 27 — 09 2 GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA C horizon of uncultivated soils in the high- and low-death-rate areas of Georgia C27 one or both kinds of samples (tree or soil) was beyond the range of measurement by the analytical methods that were used; . .. ., insufficient data] Sassafras Smooth sumac Sweetgum Black cherry High-death- Low-death- High—death- Low-death- High—death- Low—death- High—death- Low-death- rate area, 17 rate area, 27 rate area, 30 rate area, 30 rate area, 28 rate area, 27 rate area, 30 rate area, 30 Element, and plant part analyzed pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of stems and stems and stems and stems and stems and stems and stems and stems and leaves leaves leaves leaves leaves leaves leaves leaves r N r N r N r N r N r N r N r N 0 —0 32 8 0 —0.04 0 —0.08 9 —0. 23 0 —0. 18 9 0 — 16 8 0 —.27 9 0 -—.25 9 —.03 0 —.21 9 11 — 10 6 21 —-.32 18 20 —.29 16 .09 21 . 02 18 11 —.30 6 21 —. 13 18 20 —.23 16 .12 21 .27 18 0 —.32 0 0 —.25 0 0 —.26 0 .24 0 —.32 10 0 —. 0 0 —.27 0 0 —.21 0 .32 0 —.24 0 4 —. 0 7 —. 25 0 7 22 0 -. l9 7 . 02 0 4 —. 0 7 —.04 0 7 .30 0 —.10 7 —.23 0 17 .. 26 30 .20 25 ... 26 92 24 — 00 28 —. 17 21 17 . . 27 30 .......... 30 . _ _ 27 __________ 27 __________ 30 __________ 30 1 .24 0 0 20 3 . 0 07 0 .14 l —. 23 2 0 08 0 0 l5 1 . 29 0 . 09 0 . 13 2 —. 14 0 0 .11 0 0 —.08 0 —.06 0 — 35 0 —.24 0 .11 0 0 . 00 0 0 05 0 —. 01 0 — 15 0 —. 06 0 . 25 0 0 31 0 0 09 0 . 05 0 — 02 0 . 19 0 —. 34 0 0 06 0 0 — 11 0 .19 0 — 15 0 .27 0 —.18 0 0 06 0 0 — 11 0 —.26 0 —.01 0 .38 0 —.12 0 0 — 08 0 0 — 01 0 —.07 0 .20 0 .23 0 —.20 0 0 .00 0 —.23 0 — 04 0 —.21 0 08 0 .24 4 —-.08 0 3 —.l2 2 —.08 3 — 25 0 .04 6 .30 3 .23 9 —. 16 6 0 04 0 —.23 0 — 01 0 .05 0 — 26 0 -.10 0 —.05 0 0 —.01 0 —.52 0 16 0 .09 0 -.26 0 -.l7 0 —.08 0 7 12 0 —. 23 15 . 29 7 — 12 13 .33 0 . 26 17 . 43 9 7 .40 0 —.21 20 09 9 —.19 12 31 0 —. 19 12 .53 2 1 —.07 0 .75 4 — 10 0 .53 4 —. 14 0 .20 4 .21 0 l 23 0 15 4 —. 18 0 . 38 4 24 0 . l7 4 20 0 14 23 6 — 86 25 .23 5 —. 27 22 06 5 .26 24 14 5 14 20 4 — 14 24 .28 4 —. 64 22 — 20 4 —-. 19 25 34 5 8 —.26 0 34 13 .44 0 .69 12 14 0 .46 13 — 07 0 8 —. 32 0 .42 13 -. 02 0 .68 12 — 09 0 .67 13 —. 01 0 0 —.32 0 —.13 0 -.05 O .09 0 — 20 0 —.10 0 --.15 0 0 —.36 0 —.07 0 —.35 0 —.17 0 — 43 0 —.03 0 —.27 0 8 -. 09 18 . 50 27 .......... 28 . 00 25 __________ 27 —. 45 26 —. 58 20 13 . 47 20 __________ 27 . 25 23 .......... 27 __________ 25 .......... 28 __________ 36 16 -— 12 0 — 32 27 —. 34 0 90 25 — 19 0 —. 82 27 23 0 leaves _______________________________________________ 16 07 0 — 78 27 —. 43 0 __________ 25 — 02 0 —. 00 27 . 50 0 B horizon of uncultwated 801,18 m the htgh- and low-death-rate areas of Georgm one or both kinds of samples (tree or soil) was beyond the range of measurement by the analytical methods that were used; . __ _, insufficient data] Sassafras Smooth sumac Sweetgum Black cherry High-death- Low-death- High-death- Low-death— High-death- Low-death- High-death- Low-death- rate area, 17 rate area, 27 rate area, 30 rate area, 30 rate area, 28 rate area, 27 rate area, 30 rate area, 30 Element, and plant part analyzed pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of stems and stems and stems and stems and stems and stems and stems and stems and leaves leaves leaves leaves leaves leaves leaves leaves r N r N r N r N r N r N r N r N A1, stems .................................... —0 23 0 0 —0 09 7 0 —0 45 7 leaves. ............. - 18 0 0 —. 21 7 0 — 26 7 . ............. 10 20 10 13 20 40 16 ............. 10 20 — 37 13 20 36 16 Ba stems ............... 0 0 - 30 0 0 — 46 0 ............... 0 O — 41 0 0 — 47 0 _______________ 7 9 — 11 0 . 9 - 03 0 _______________ 7 9 03 0 9 — 00 0 ....................... 17 29 58 24 29 - 23 22 ....................... l7 __ 30 _._.______ 28 27 30 .......... 30 1 . 0 15 3 .08 0 15 0 1 —. 18 2 0 . 0 ()0 1 . 15 0 20 0 2 . 01 2 0 —.08 0 —.11 0 15 0 - 32 0 0 —. 19 0 0 45 0 -. 10 0 l8 0 —- 23 0 0 —. 10 0 0 10 0 -. 06 0 .18 0 - 16 0 0 —. 24 0 0 06 0 —. 08 0 33 0 - 51 0 0 —. 02 0 0 06 0 —. 19 0 —- 26 0 —. 00 0 0 —. 12 0 0 — 04 0 —. 04 0 — 08 0 . 23 0 0 —-. 20 0 0 —— 16 0 —. 13 0 -— 17 0 07 0 4 . 16 0 3 —-.09 3 -.31 0 — 14 6 .30 3 9 .37 6 0 —.26 0 —.09 0 01 0 —.30 0 0 .07 0 0 — 40 0 —.15 0 06 0 —.27 0 0 .07 0 9 25 22 25 8 60 19 . 35 1 22 . 42 10 9 —. 64 23 . 14 10 48 19 .28 1 20 .48 2 l 74 3 — 24 0 47 3 .02 0 3 . 24 0 l 47 3 —. 27 0 41 3 . 13 0 3 . 16 0 l7 — 50 27 . 20 6 —. 94 25 17 6 27 . 26 6 17 -— 85 27 . 40 5 — 91 25 — 01 5 28 . 47 6 12 . 31 21 . 24 1 — 22 19 19 1 21 . 07 1 12 . 16 21 . 12 1 — 0 21 . 11 l 0 —. 25 0 14 0 — 0 0 - . 02 0 0 —. 17 0 — 36 0 0 0 -. 00 0 9 — . 0 __________ 28 27 26 . 04 26 13 —. 0 —. 03 23 25 28 .......... 30 Zn, stems. _____________ 15 27 —. 52 2 2 _ 27 .03 2 leaves __________________ _ ____________________________ 15 27 —. 36 2 2 _ _ _ 27 .56 2 C28 HIGH-DEATH-RATE AREA ALUMINUM LOW— DEATH-RATE AREA STATISTICAL STUDIES IN FIELD GEOCHEMISTRY LOW-DEATH-RATE AREA HIGH-DEATH-RATE AREA as COBALT -1 COPPER BARIUM GALLIUM -1 BORON -1 IRON CALCIUM LANTHANUM CHROMIUM stems, squares represent leaves, of trees. Distance of the symbols on the radii from the center of the circles is inversely FIGURE 5.—C0rrela~tions between the amounts of elements in the B horizon of uncultivated soils (center of circles) and the lamounts in the A horizon of soils and in the stems and leaves of trees. Triangles represent soil correlations; dots represent proportional to the correlation of the amounts in the samples. GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA 029 HIGH-DEATH-RATE AREA LOW-DEATH-RATE AREA HIGH-DEATH-RATE AREA LOW-DEATH-RATE AREA A horizon .1 -1 -1 - -as 415 415 o o o 05 o 55 o 9%.: 3r 4' e°° 9‘ f 30 LEAD PHOSPHORUS ”W: W” -x -1 -1 -1 ~05 -05 '05 415 o o 0 o 05 0.5 05 05 MAGNESIUM POTASSIUM -] -1 -1 ’l -0.5 -n5 -(15 '05 0 o 0 0 0 5 05 0 5 0 5 MANGANESE SCANDIUM -1 -1 -1 -1 415 415 415 -(15 0 0 0 0 05 o 045 05 NICKEL SODIUM -1 -1 -1 415 415 415 0 0 0 o 5 0.5 0.5 NIOBIUM STRONT‘UM FIGURE 5.—Continued C30 HIGH-DEATH-RATE AREA A horizon 3"0 e B s .s 5 Q0 80-9: \e "'4: w TITANIUM -l 415 0 0,5 -l 415 0 0.5 YTTERBIUM VANADIUM STATISTICAL STUDIES IN FIELD GEOCHEMISTRY LOW- DEATH-RATE AREA -l 45 0 0.5 415 0‘5 415 0.5 HIGH-DEATH- RATE AREA -1 -fl§ 0 05 -1 -Q5 0 0.5 ZIRCONIUM FIGURE 5.——Continued LOW-DEATH-RATE AREA -1 -&5 05 GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA CORRESPONDENCE AMONG THE COMPOSITIONS 0F SOIL HORIZONS Correlations of the amounts of elements in samples of the three horizons of uncultivated soil are given in table 12 and illustrated in figures 4—6. The correlations of ele- ments between the A and the C horizons (fig. 4 and table 12) are all positive except for boron, gallium, and strontium in the high-death-rate area and lanthanum in the low-death-rate area. The correlations of the amounts in the B and C horizons (fig. 4 and table 12) are all positive except for gallium and strontium in the high- rate area. The correlations of the amounts in the A and 031 B horizons (fig. 5 and table 12) are all positive except for lanthanum in the low—death-rate area. These data indicate that the amounts of most ele- ments in the three soil horizons are strongly interde- pendent. The elements that have negative correlations are nonnutritive, except boron which is a micronutrient; their differences in amounts in soil samples from the two death—rate areas are not significant «at the 95-percent level of confidence (table 7). The results of this study also suggest that in search- ing for differences in amounts of most elements in the soils of the two death-rate areas, any one of the three soil horizons could be sampled alone. C32 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY HIGH-DEATH-RATE AREA LOW-DEATH-RATE AREA Blackgllu'n I 4) Swaetgum Persimmon o x"e Sassafras ALUM INU M é BARIUM BORON CALCIUM —0.5 (15 CHROMIUM HIGH-DEATH-RATE AREA LOW- DEATH-RATE AREA COBALT COPPER IRON -1 -us (15 LEAD -1 415 as MAGNESIUM FIGURE 6.———Correlations between the amounts of elements in the A horizon of uncultivated soils (center of circles) and the amounts in stems and leaves of trees. Dots represent stems, squares represent leaves. Distance of the symbols on the radii from the center of the circles is inversely proportional to the correlation of the amounts in the samples. GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA 033 HIGH-DEATH-RATE AREA LOW- DEATH-RATE AREA HIGH-DEATH-RATE AREA LOW- DEATH-RATE AREA -I Blackglum -05 as E E a s g 5 09% “ MANGANESE smounum Sassafras -1 -1 ’- -x -0.5 ~05 -Cl5 o 5o (15 I15 I: I NICKEL TITANIUM -l -1 -n5 '05 II 05 I15 PHOSPHORUS VANADIUM -| _1 415 -as (15 0.5 POTASSIU M ZINC FIGURE 6.—Continued C34 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY TABLE 11.—C'orrelations between the amounts of elements in trees and in the A [r, product-moment correlation coefficient between logarithms of concentrations; N, number of pairs omitted because concentration of element in one Blackgum Oak Persimmon Red maple High-death- Low-death- High-death- Low-death- High-death- Low-death- High-death- Low-death- Eiement, and plant part analyzed rate area, 30 rate area, 30 rate area, 30 rate area, 30 rate area, 30 rate area, 30 rate area, 30 rate area, 30 pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of stems and stems and stems and stems and stems and stems and stems and stems and leaves leaves leaves leaves leaves leaves leaves leaves r N r N r N r N r N r N r N r N A1, stems .................................... —0. 42 l —0. 21 4 —0. 39 1 —0 29 4 —0. 04 2 —0. 25 4 —0. 21 1 —0. 01 4 —. 32 l —. 24 4 . 1 — 20 4 .06 1 —. 18 4 --. 15 1 —. 18 4 . —. 10 20 . 69 17 20 41 17 .12 21 . 36 17 —. 44 20 -. 06 17 . —-.38 20 .08 17 20 46 17 —. 25 20 .69 17 —. 11 20 .36 17 _ .06 0 .08 0 0 —.38 0 .17 l —.21 0 .20 0 .02 0 . .11 0 —. 14 0 0 —.21 0 .04 0 -.27 0 .24 0 ——.04 0 — 21 7 .16 1 7 — 01 1 —.26 7 .32 1 —.44 7 —. 16 1 _ —.11 7 11 1 7 —.08 1 —. 12 7 .25 1 —.46 7 .20 1 . -.23 24 .17 5 _._. 30 —.46 16 .......... 25 .21 23 .......... 29 .......... 30 . .54 25 .26 5 .... 29 05 19 .......... 29 .......... 30 .......... 30 __________ 30 _ —.03 0 . 18 0 1 .14 l 24 3 .09 3 .48 3 19 1 . . 09 1 . 24 0 0 -- 01 0 13 3 . 05 1 . 06 2 l9 0 . .03 0 .25 0 — 16 0 .11 0 .03 l .09 0 —.04 0 — 10 0 . —. 08 0 —. 13 0 — 40 0 — 14 0 —- 01 0 .18 0 —. 15 0 — 18 0 _ —21 0 -.10 0 —10 0 .06 0 —00 1 .03 0 —.08 0 —17 0 . 03 0 —. 19 0 — 02 0 —. 11 0 -— 01 0 —. 10 0 .28 0 — 08 0 . —.37 0 —.07 0 00 0 —.17 0 -.03 0 .26 0 -. 15 0 10 0 . 18 0 —.27 0 27 0 —.57 0 01 0 —. 09 0 -. 12 0 -— 01 0 . - 01 l 13 3 -— 50 0 . 23 0 08 3 -—. 05 0 —.44 0 12 0 _ — 51 24 71 18 — 39 6 . 24 3 -— 27 12 . 01 2 -. 41 2 29 0 . — 12 0 —.18 0 — 07 0 —.17 0 20 l —.07 0 -.01 0 -—.69 0 . — 17 0 03 0 — 05 0 — 18 0 — 20 0 .07 0 —.16 0 —.75 0 _ 51 22 - 10 2 — 24 22 . 17 l 18 22 —. 00 l . 23 22 . 24 4 . 60 22 —. 12 2 12 22 . 14 1 — 10 22 —. 21 1 . 06 22 . 66 11 . l7 0 .17 0 06 0 .20 0 .35 0 .19 0 .21 0 —.02 0 . 23 0 .31 0 21 .18 0 —.09 0 07 0 .11 0 .04 0 . — 57 23 . 46 3 — 38 23 . 38 3 —. 10 23 .30 4 .31 24 . 33 3 _ 39 23 . 55 3 10 23 . 43 3 -. 21 24 28 4 . 01 24 . 56 3 _ — 21 19 .05 1 —.52 19 .05 1 —.22 19 18 l —.23 19 .16 l . — 05 19 .29 1 —. 55 19 . 11 1 —. 08 19 16 1 —. 43 19 . 21 1 _ — 27 0 —. 17 0 .20 0 —.26 0 —.29 l —.07 0 .24 0 —.20 0 . — 17 0 —.20 0 .37 0 -.l3 0 —. 14 0 —.l7 0 —.ll 0 .03 0 . —.58 25 .01 24 —.00 26 —. 91 27 —.00 27 - 25 25 —.00 26 .......... 29 . - 01 28 —. 03 23 -. 78 27 .55 22 .......... 29 —. (77 26 __________ 28 __________ 28 . — 46 23 —.24 l .21 23 —.28 1 —.73 23 —.26 1 70 23 —.05 1 ........................... — 09 23 06 1 .01 23 —.25 l .26 23 —.07 l 64 23 -.17 1 TABLE 12.—Correlations _between the amounts of elements in hori- zons of nncultwated smls m the high- and low-death-rate areas of Georgia [r, product-moment correlation coefficient between logarithms of concentrations; N, number of pairs omitted because concentration of element in the sample from one or both soil horizons was beyond the range of measurement by the analytical methods that were used; ...., no data available] High-death-rate area, 30 pairs Low-death-rate area, 30 pairs AandB AandC BandC AandB AandC BandC horizons horizons horizons horizons horizons horizons r N r N r N r N r N r N Element 0. 82 8 0. 73 10 0. 73 11 . 55 18 . 70 19 . 60 18 . 70 0 . 81 0 . 70 0 . 39 1 . 19 l . 51 0 . 84 3 . 79 2 . 86 1 . 86 0 . 80 0 . 79 0 . 81 0 . 73 O . 78 0 . 82 0 . 60 0 . 53 0 . 77 l . 53 l . 71 0 . 96 0 . 95 0 . 97 0 —. 06 13 —. 15 13 . 03 13 . 91 0 . 67 0 . 64 0 . 78 0 . 62 0 . 78 0 . 78 1 . 74 l . 85 1 . 55 3 . 78 3 . 79 3 . 87 l . 83 1 . 85 1 . 79 0 . 82 0 . 70 0 . 38 5 . 30 4 . 71 5 . 74 1 . 79 1 . 78 0 . 86 2 . 83 1 . 89 1 . 79 0 . 75 0 . 67 0 . 77 0 . 50 0 . 45 0 . 39 0 . 60 0 . 36 0 . 76 0 . 49 0 . 43 0 . 71 2 . 52 1 . 47 2 . 34 0 . 34 0 . 54 .0 CORRESPONDENCE BETWEEN THE COMPOSITIONS 0F STEMS AND LEAVES OF TREES The correlations that indicate the correspondence of stem and leaf analyses are given in table 13 and illus- trated in figure 7. The correlations are all positive ex- cept for ash, lead, and zinc in cherry, lead in sassafras, and magnesium in sweetgum, all in samples from the high—death-rate area. The correlations are most highly positive for aluminum, barium, calcium, manganese, and strontium. These data indicate that the amounts of most elements in stems and leaves are strongly interdependent, and that in searching for differences in element content of trees from the two death-rate areas, either stems alone or leaves alone, could be used. SUMMARY 1. Of the 159 counties in Georgia, the nine counties of northern Georgia that were chosen for geochemical studies because of their low rate of death due to cardio- vascular diseases also are among the 20 counties having the lowest rate for all causes of death, as well as being GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA C35 horizon of uncultivated soils in the high- and low-death-rate areas of Georgia or both kinds of samples (tree or soil) was beyond the range of measurement by the analytical methods that were used;_._., insufficient data] Sassafras Smooth sumac Sweetg'um Black cherry High-death- Low-death- High-death- Low-death- High-death- Low-death— High-death- Low-death- Element, and plant part analyzed rate area, 17 rate area, 27 rate area, 30 rate area, 30 rate area, 28 rate area 27 rate area. 30 rate area, 30 pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of pairs each of stems and stems and stems and stems and stems and stems and stems and stems and leaves leaves leaves leaves leaves leaves leaves leaves 7 N r N r N r N r N r N r N r N A1, stems. - 0 —0. 47 3 —0. 18 1 —0. l5 4 1 4 —0. l6 1 —0. 30 4 leaves__ 0 —. 15 3 --.02 1 —. ll 4 1 4 —. 13 1 —.37 4 12 . 21 15 —. 32 20 . 20 17 18 14 —. 55 20 . 22 17 12 . 12 15 —. 46 20 . 42 17 18 14 —. 37 20 . 36 17 0 —.23 0 .38 0 —. ll 0 0 0 —.02 0 —. 15 0 0 —. 25 0 . 38 O —. 16 0 0 0 .07 0 —. 13 0 4 15 1 . O2 7 —. 08 1 7 1 - 18 7 10 l 4 12 1 . 21 7 . 04 l 7 l — 25 7 . 11 1 17 __________ 26 __________ 29 37 25 25 25 —. 01 27 — 25 21 17 .......... 27 .......... 30 .......... 30 27 27 .......... 30 .......... 30 —. 22 l —. 15 0 .02 0 28 3 0 . 0 —.09 l — 20 2 —. 19 0 . 05 0 . 33 0 06 1 0 . 0 —. 02 2 07 2 —. 60 0 —. 22 0 05 0 —. 19 0 0 . 0 —. 23 0 — 10 0 —. 02 0 —. 13 0 — l4 0 —. l3 0 0 . 0 —. 10 0 . 12 0 —.2l 0 —.45 0 02 0 —-. 19 0 O . 0 --. 15 0 —.23 0 —. 10 0 —-.22 0 02 0 —.32 0 0 . 0 —.11 0 -. 17 0 —. 24 0 .06 0 04 0 —.18 0 — 32 0 .06 0 .53 0 —.06 0 26 0 —.12 0 —.04 0 —.04 0 -— 18 0 .23 0 .35 0 —.18 0 -—.68 0 —.15 0 —.27 0 —.15 0 — 18 0 .04 0 — 05 4 21 0 —.25 3 —.06 2 —-.44 3 -.27 0 — 36 6 .17 3 —.25 9 39 6 08 0 -.23 0 -.26 0 —.22 0 14 0 —.24 0 02 0 14 0 .18 0 —.23 0 -. 20 0 —.27 0 25 ______ —.l7 0 22 0 13 0 . 20 12 —. 04 l 68 23 11 8 36 20 . 23 l — 16 23 . 53 10 — 15 12 .33 1 74 25 07 10 31 20 .13 l 09 22 .54 2 08 0 .13 0 38 0 —- 15 0 .25 0 .07 O .06 0 .17 0 28 0 .16 0 16 0 -.18 0 .26 0 .23 0 .20 0 .23 0 — 99 13 . 30 5 —. 44 23 . 39 4 . 18 21 .37 4 — 05 23 . 24 4 11 13 . 34 3 —. 09 23 45 3 -. 45 21 30 3 — 42 24 36 4 . — 76 11 —. 17 1 .15 19 14 1 —. 46 17 27 1 -— 19 19 14 1 — 50 11 —.28 1 .09 19 .13 l —.43 17 13 1 — 26 19 24 1 _ — 48 0 —.43 0 —.00 0 —.13 0 .12 0 — 35 0 — 11 0 - 11 0 -29 0 —.44 0 .03 0 —36 0 —.13 0 - 55 0 —03 0 ~23 0 V stems.. — 11 7 —.39 18 .11 27 -.01 28 —.00 25 .......... 27 — 57 26 16 26 — 73 12 —. 17 18 -. 50 27 — 07 23 __________ 27 .......... 25 .......... 28 __________ 30 - 43 13 . 05 1 . 27 23 — 46 1 37 22 01 1 — 01 23 20 1 leaves. . — 91 13 22 1 .26 23 — 26 1 37 22 25 l — 03 23 55 1 among those with the very lowest rate for the cardio- vascular diseases generally, and coronary heart disease specifically. The counties chosen for study in central and south-central Georgia are all among the 20 counties having the highest rate for all causes of death, and have mortality rates for the cardiovascular diseases roughly twice as high as those for the low-rate counties that were studied. In this one State, contrasts in mortality rates for cardiovascular diseases are found which are almost as great as can be found between any two coun— ties in the United States. These contrasts are not due to random error, and almost certainly were not produced by the methods of data collection and classification. 2. The counties that have high cardiovascular death rates are located where Paleudult, Quartzipsamment, and Ochraquult soils predominate. These soils are de- rived from unconsolidated or slightly consolidated sands, sandy clays, and clays of the Atlantic Coastal Plain region. The counties with low death rates due to the same cause are located within the area of Haplu- dult soils that were derived from the weatherng of rocks in the Appalachian Plateaus, Valley and Ridge, Blue Ridge, and Piedmont provinces. 3. The frequency of occurrence of the plants that were sampled differed in the two death-rate areas. The vege- tables found most frequently in both areas were corn, green beans, tomatoes, and cabbage, and the most com- mon trees were blackgum, persimmon, red maple, smooth sumac, and black cherry. 4. Estimates of experimental error in sampling soils indicated that in general the variation between the high- and low-death-rate areas is sufficiently large, in comparison with the variation between sites or between samples within sites, that compositional differences be- tween soils from the two areas appear to be identifiable. 5. The concentrations of the elements that were studied in both garden soils and uncultivated soils tend to be significantly different in the two death-rate areas and are generally higher in the low-death-rate area. 6. Because of the similarities in element content of the soil horizons at a sample site, the differences in con- centrations of almost all elements in uncultivated soils from the two areas could have been determined by sampling any one of the three soil horizons. 7. The cultivation of garden soils apparently has not greatly altered their content of the elements that were C36 HIGH-DEATH-RATE AREA STATISTICAL STUDIES IN FIELD GEOCHEMISTRY LOW- DEATH-RATE AREA Blackgrm -l s E a : Sassafras ASH -l -1 -0.5 415 >0 0 05 05 ALUMINUM -1 415 0 0.5 BARIUM ~l -l 415 -(l5 -0 0 05 (15 BORON ‘ —1 ~06 0 Q5 CALCIUM HIGH-DEATH-RATE AREA -1 -0.5 -0 0.5 CHROMIUM COBALT COPPER IRON LEAD LOW- DEATH-RATE AREA -1 415 -0 PQS -1 >115 -l :5 FIGURE 7.-—Corre1ations between the amounts of elements in stems and in leaves of trees. The center of the circles represents stems, and the dots represent leaves. portional to the correlation of the am Distances of the dots on the radii from the center of the circles is inversely pro- ounts in samples of stems and leaves. GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA HIGH-DEATH-RATE AREA -1 415 -0 I15 MAGNESIUM MANGANESE NICKEL PHOSPHORUS POTASSIUM LOW-DEATH-RATE AREA -l 4.5 r0 HIGH-DEATH-RATE AREA —1 (l5 FIGURE 7.—Continued Swoamum STRONTIUM TITANIUM VANADIUM ZINC C37 LOW-DEATH-RATE AREA Blackg‘am Persimmon 4% ‘0 Sassatvas C38 STATISTICAL STUDIES IN FIELD GEOCHEMSTRY TABLE 13.—Correlations between the amounts of elements in stems [r, product-moment correlation coefficient between logarithms of concentrations; N, number of pairs omitted because concentration of element Blackgum Oak Persimmon Red maple Element or ash High—death- Low-death- High-death- Low-death- High—death— Low-death- High-death- Low-death- rate area, rate area, rate area, rate area, rate area, rate area, rate area, rate area, 30 pairs 30 pairs 30 pairs 30 pairs 30 pairs 30 pairs 30 pairs 30 pairs r N r N r N r N r N r N r N r N 0. 65 0 0. 65 0 0. 64 0 0. 61 0 0. 85 1 0. 80 0 0. 72 0 0. 66 0 .49 0 .73 0 .49 0 .70 0 .41 0 .12 0 .53 0 .41 0 .61 0 .54 0 .42 0 .35 0 .41 l .52 0 .40 0 .33 0 .91 0 .26 0 .84 0 .71 0 .79 l .74 0 .88 0 .77 0 .67 0 .76 0 .64 0 .77 0 .40 0 .78 0 .86 0 .65 0 . 93 3 . 94 3 .......... 29 . 46 18 . 95 27 __________ 30 __________ 30 .......... 30 .53 l .72 0 .54 1 .68 1 .45 5 .48 3 .30 4 .64 1 .32 0 .54 0 .54 0 .56 0 .28 1 .31 0 .50 0 .73 0 .54 0 .92 0 .44 0 .69 0 .62 1 .64 0 .56 0 .73 0 .44 0 .50 0 .32 0 .49 0 .72 0 .60 0 .63 0 .50 0 . 67 .-4 .02 18 .36 6 .19 3 .43 13 .46 2 .39 2 .33 0 .85 0 .72 0 .81 0 .51 0 .73 l .80 0 .89 0 .91 0 .77 1 .87 l .64 1 .51 0 .39 1 .75 0 .82 13 .82 11 .62 :0 .31 0 .55 0 .67 0 .18 0 .54 0 .49 0 .45 0 .30 ‘ l .62 0 .35 3 .73 0 .56 9 .54 1 .66 5 .36 0 .65 0 .83 0 .41 0 .69 0 .93 l .89 0 .87 0 .87 0 .58 0 .83 0 .50 0 .54 0 .77 1 .58 0 .57 0 .40 0 ............ 28 .57 26 ....._.___ 28 .......... 28 _._______. 30 __..___.__ 28 .......... 29 .......... 29 .61 0 .27 0 .41 0 .33 0 .20 0 .31 0 .80 0 .42 0 studied, if judged by the concentrations of these ele- CONCLUSIONS ments 1n uncultivated soils. 8. The difference 111 element composition of soils from the two areas is not apparent in analyses of most vege- tables from the two areas. Only cabbage and green beans tend to reflect, on a broad scale, the difference in amounts of certain elements in soils firom the two areas. 9. Correlations between the elemeint content of gar- den soils and of vegetables fail to indicate a general rela- tionship between the composition of vegetables and the soils on which they grew. 10. Within each area, the results fail to demonstrate any consistent correlations between the element content of trees and the composition of any soil horizons at the sites where the trees grew. 11. The concentrations of elements 1n stems and 111 leaves of trees are strongly related. Where higher con- centrations occur in the stems, higher concentrations are also found in the leaves. 12. Tendencies in the concentrations of certain ele- ments indicate that stems concentrate barium, calcium, copper, lead, and strontium; leaves iconcentrate alumi- num, iron, magnesium, potassium, titanium, ytterbium, and zirconium; and that stems and leaves are approxi- mately equal in their concentrations of boron, chromium, manganese, nickel, and phosphorus. 13. In the low-death—rate area of northern Georgia the concentration of most elements hi1 tree samples tends to be higher than in the high-death-rate areas of Georgia. Therefore, trees appear to be more sensitive than vegetables to the variation in element content of soils from the two areas. 1. The greater abundance of many chemical elements in soils from the low-death-rate area (northern Geor- gia), compared with their abundance in soils from the high-death-ra-te areas (south-central Georgia), is be- lieved to reflect the differences in the character of the bedrock which provided the materials for soil forma- tion. The sediments in the high—rate area probably 'had been extensively weathered and leached during their original deposition. Soils formed on these sediments, consequently, are relatively deficient in many elemental constituents. The present soils in the low-death-rate area, 011 the other hand, are being continously supplied with elements from the weathering of fresh igneous and metamorphic rocks. 9. The analyses of tree samples and some vegetable samples tend to reveal the geochemical differences in soils from the two areas only when examined 011 a broad scale. 3. The difference between the two areas in the levels of various chemical elements present in vegetables ap- pears to be so slight as to be unlikely to contribute ap- preciably to the observed differences in death rates in middle-aged humans. These data show that garden veg- etables do not clearly reflect the chemical compositions of the soils on which the vegetables grew. The chemicals in soils, however, may enter the human food chain in water, in food plants that were not sampled in this study, and in meat and milk from animals that have fed on cultivated forage plants and native vegetation. 4. If geochemical differences between the two areas do, in fact, have a causal relationship to death from GEOCHEMICAL ENVIRONMENTS, CARDIOVASCULAR MORTALITY RATES, GEORGIA and leaves of trees in the high- and low-death-rate areas of Georgia C39 in the sample (stems or leaves) was beyond the range of measurement by the analytical methods that were used; _ . . _, insufficient data] Sassafras Smooth sumac Sweetgum Black cherry High-death- Low—death- High—death- Low-death- High—death- Low-death- High-death- Low-death- Element or ash rate area, rate area, rate area, rate area, rate area, rate area, rate area, rate area, 17 pairs 27 pairs 30 pairs 30 pairs 28 pairs 27 pairs 30 pairs 30 pairs r N r N r N r N r N r N r N r N 0 0. 70 0 0. 69 0 0. 69 0 0. 68 0 O. 32 0 0. 63 0 0. 73 0 0 .19 0 .38 0 .51 0 .60 0 .56 0 —.17 0 .43 0 0 .35 0 .42 0 .40 0 .59 0 .33 0 .78 0 .25 0 0 .82 0 .73 0 .89 0 .83 0 .66 0 .89 0 .94 0 0 .75 0 .72 0 .56 0 .75 0 .87 0 .69 0 .69 0 17 .......... 27 .......... 30 __________ 30 .......... 26 __________ 27 .......... 30 .......... 30 1 .62 0 .67 0 .62 4 .25 0 .33 0 .44 2 .40 3 0 .46 0 .53 0 .53 0 .91 0 .70 0 .33 0 .61 0 0 .73 0 .83 0 .73 0 .64 0 .49 0 .57 0 .19 0 0 .79 O .72 0 .42 0 .83 0 .79 0 .58 0 .61 0 3 .59 2 . 51 3 . 72 0 —. 21 6 .56 3 . 60 10 .39 6 0 .88 0 .65 0 .82 0 .64 0 .74 0 .91 0 .81 0 0 .57 0 .61 13 .68 11 .64 1 .89 0 .43 8 .54 9 0 .64 O .45 0 .52 0 .84 0 .57 0 .76 0 .66 0 0 . 67 2 .69 3 .59 1 . 34 4 . 56 1 —. 07 7 . 58 2 0 .88 0 .69 O .89 0 .84 0 .52 0 .74 0 .93 0 0 .72 0 .85 0 .42 0 .54 0 .78 0 .61 0 .62 0 13 . 11 22 .......... 28 .......... 28 .......... 28 __________ 27 .......... 29 .......... 30 0 . 5O 0 44 0 61 0 . 65 0 . 57 0 —. 39 0 . 52 0 cardiovascular diseases, the cause would appear to be a deficiency, rather than an excess, of the elements that were studied. LITERATURE CITED Anderson, R. L., and Bancroft, T. A., 1952, Statistical theory in research: New York, McGraw-Hill Book Co., 399 p. Baldwin, Mark, Kellogg, C. E., and Thorp, James, 1938, Soil classification, in Soils and men: U.S. Dept. Agriculture Yearbook 1938, p. 979—1001. Carter, R. L., and Giddens, Joel, 1954, Soils of Georgia—Their formation, classification and management: Georgia Univ. College of Agriculture, Expt. Sta. Bull. 2, 68 1). Cohen, A. 0., Jr., 1961, Tables for maximum likelihood esti- mates—Singly truncated and singly censored samples: Technometrics, v. 3, no. 4, p. 535—541. Cramer, H. W., 1967, Environmental science in early Georgia: Georgia Acad. Sci. Bu11., v. 25, no. 4, p. 215-221. Enterline, P. E., Rikli, A. E., Sauer, H. I., and Hyman, Merton, 1960, Death rates for coronary heart disease in metropoli- tan and other areas: Public Health Repts., v. 75, no. 8, p. 759—766. Fernald, M. L., 1950, Gray’s manual of botany [8th ed.]: New York, American Book Co., 1632 p. Georgia Division of Mines, Mining and Geology, 1939, Geologic map of Georgia: Georgia Div. Mines, Mining and Geology, 1 sheet. Laurence, R. A., 1968, Ore deposits of the Southern Appala- chians, p. 155—168, in Ridge, J. D., ed., Ore deposits of the United States, 1933—1967—The Graton~Sa1es volume: New York, Am. Inst. Mining, Metallurgical, and Petroleum En- gineers, 991 p. Lilienfeld, A. M., Pedersen, Einar, and Dowd, J. E., 1967, Can- cer epidemiology—Methods of study: Baltimore, Johns Hopkins Press, 165 p. Linder, F. E., and Grove, R. D., 1943, Vital statistics rates in the United States 1900—1940: U.S. Dept. Health, Education, and Welfare, 1051 p. Miesch, A. T. 1967a, Theory of error in geochemical data: U.S. Geol. Survey Prof. Paper 574—A, 17 p. 1967b, Methods of computation for estimating geochemi- cal abundance: U.S. Geol. Survey Prof. Paper 574—B, 15 p. Myers, A. T., Havens, R. G., and Dunton, P. J., 1961, A spectro- chemical method for the semiquantitative analysis of rocks, minerals and ores: U.S. Geol. Survey Bull. 1084—1, p. 207— 229. Sauer, H. I., 1962, Epidemiology of cardiovascular mortality—— Geographic and ethnic: Am. Jour. Public Health, v. 52, no. 1, p. 94—105. Sauer, H. 1., and Enterline, P. E., 1959, Are geographic varia- tions in death rates for the cardiovascular diseases real?: Chronic Diseases J our., v. 10, no. 6, p. 513—524. Sauer, H. I., Payne, G. H., Council, C. R., and Terrell, J. C., 1966, Cardiovascular disease mortality patterns in Georgia and North Carolina: Public Health Rpts., v. 81, no. 5, p. 455— 465. Shacklette, H. T., and Sauer, H. I., 1964, The association of cardiovascular mortality rates in Georgia with the abund- ance and distribution of certain elements in rocks, soils, and plants: U.S. Geol. Survey open-file report, 21 p. Soil Survey Staff, 1960, Soil classification—A comprehensive system, 7th approximation: U.S. Dept. Agriculture, 265 p. 1967, Supplement to soil classification system (7th ap- proximation) : U.S. Dept. Agriculture, 207 p. Sutcliife, J. F., 1962, Mineral salts absorption in plants: New York, Pergamon Press, 194 p. U.S. Geological Survey, 1968, Landforms of the United States: U.S. Geol. Survey pamph, 15 p. [U.S.] Soil Conservation Service, 1969, Distribution of principal kinds of soils—Orders, Suborders, and Great Groups, in National Atlas of the United States of America: U.S. Geol. Survey, sheet 86, 2 1). Ward, F. N., Lakin, H. W., Canney, F. C., and others, 1963, Analytical methods used in geochemical exploration by the U.S. Geological Survey: U.S. Geol. Survey Bull. 1152, 100 1). World Health Organization, 1957, Manual of the international statistical classification of diseases, injuries and causes of death, 7th revision: Geneva, World Health Organization, v. 1, 393 p. U.S. GOVERNMENT PRINTING OFFICE: 1969 0—363-640 , 7 DAY >E7b 74 WDElemental Composition of Surfieial Materials in the Conterminous United States I TGEOLOGICAL SURVEY/PROFESSIONAL PAPER 574—D , I .r _,._,—- \J Elemental Composition of Surficial Materials in the Conterminous United States By HANSFORD T. SHACKLETTE, J. C. HAMILTON, JOSEPHINE G. BOERNGEN, and JESSIE M. BOWLES STATISTICAL STUDIES IN FIELD GEOCHEMISTRY GEOLOGICAL SURVEY PROFESSIONAL PAPER 574—D An account of the amounts of certain chemical elements in samples of soils and other regoliths UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON : 1971 UNITED STATES DEPARTMENT OF THE INTERIOR ROGERS C. B. MORTON, Secretary GEOLOGICAL SURVEY William T. Pecora, Director Library of Congress catalog-card No. 71—610288 For sale by the Superintendent of Documents, US. Government Printing Office Washington, DC. 20402 - Price 70 cents (paper cover) CONTENTS Page Page Abstract _________________________________________ D1 Collection and analysis of geochemical data—Con. Introduction _____________________________________ 1 sampling media ------------------------------ D4 Acknowledgments ____ _____________________________ 1 Chemical analys15 procedures __________________ 4 . Methods of statistical analysis _________________ 5 Revxew of the literature ___________________________ 2 Results ___________________________________________ 6 Collection and analysis Of geochemical data ————————— 2 Discussion and conclusions ________________________ 6 Sampling plan _______________________________ 2 References cited _ 7 ILLUSTRATIONS Page Page FIGURE 1. Map showing location of sample sites FIGURES 2—31. Maps showing element content of sur- in the conterminous United States ficial materials in the conter- where elements not commonly de- minous United States—Continued tected in surficial deposits were 14. Lead ______________________ D36 found, and the amounts of the ele- 15. Magnesium __________________ 38 ments present ———————————————————— D10 16. Manganese _________________ 40 2—31. Maps showing element content of 17- Molybdenum ——————————————— 42 surficial materials in the contermi— 18. Neodymium _________________ 44 nous United States: 19_ Nickel _______________________ 46 2. Aluminum / _________________ 12 20. Niobium ___________________ 48 3. Barium _____________________ 14 21. Phosphorus __________________ 50 4. Beryllium __________________ 16 22. Potassium _________________ 52 5. Boron ____________________ m- 18 23. Scandium __________________ 54 6. Calcium ___________________ 20 24. Sodiumiv ____________________ 56 7. Cerium ____________________ 22 25. Strontium __________________ 58 8. Chromium _________________ 24 26. Titanium ___________________ 60 9. Cobalt _____________________ 26 27. Vanadium __________________ 62 10. Copper ____________________ 28 28. Ytterbium _________________ 64 11. Gallium _____________________ 30 29. Yttrium ___________________ 66 12. Iron _________________________ 32 30. Zinc ________________________ 68 13. Lanthanum ________________ 34 31. Zirconium __________________ 70 TABLES Page TABLE 1. Average contents, and range in contents, reported for elements in soils and other surficial materials ______ D3 2. Analytical limits of detection ______________________________________________________________________ 5 3. Geometric mean compositions, and geometric deviations, of samples of soils and other surficial materials in the conterminous United States _______________________________ ___ _____ ___ 7 III STATISTICAL STUDIES IN FIELD GEOCHEMISTRY ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS IN THE CONTERMINOUS UNITED STATES By HANSFORD T. SHACKLETTE, J. C. HAMILTON, JOSEPHINE G. BOERNGEN, and JESSIE M. BOWLES ABSTRACT Samples of soils or other regoliths, taken at a depth of approximately 8 inches from locations about 50 miles apart throughout the conterminous United States, were analyzed for their content of elements. In this manner, 863 sample sites were chosen, and the results of the sample analyses for 35 elements were plotted on maps. The arithmetic and geometric mean, the geometric deviation, and a histogram showing fre- quencies of analytical values are given for 30 elements. Surficial materials of the western half of the United States generally contain more calcium, magnesium, strontium, potassium, sodium, aluminum, and barium, but contain less titanium and zirconium than do those of the eastern half. Surficial materials in the Atlantic Coastal Plain tend to have much lower concentrations of most metals than are common in those of other regions, whereas these materials in the Basin and Range province, in parts of the Rocky Moun- tains, and in Maine and adjacent States generally have high metal concentrations. Some smaller patterns of element abun- dance can be noted, but the degree of confidence in the validity of these patterns decreases as the patterns become less extensive. INTRODUCTION The abundances of certain chemical elements in soils and other surficial materials are determined not only by the element content of the bedrock or other deposits from which the materials originated, but also by the effects of climatic and biological factors that have acted on the materials for various periods of time. The diversity of these factors in a large area is expected to result in a corresponding diversity in the element contents of the surficial materials. At the beginning of this study, few data were available on the abundance of the elements in surfi- cial materials of the United States as a whole. Most of the early reports discussed only the elements that were of economic importance to mining or agricul- ture in a metallogenic area or State; and the data, for the most part, cannot be evaluated with reference to average, or “normal,” amounts in undisturbed materials because they were based on samples of de- posits expected to have anomalous amounts of cer- tain elements, or were based only on samples from cultivated fields. We began a sampling program in 1961 that was designed to give estimates of the range of element abundance in surficial materials that were unaltered or very little altered from their natural condition, and in plants that grew on these deposits, throughout the conterminous United States. Because of the great amount of travel necessary to complete this program, geologists and others of the U.S. Geological Survey were asked to assist by collecting samples when traveling to and from project areas and to contribute appropriate data that they might have collected for other purposes. The response to this request, together with the samples and data that we collected, resulted in obtaining samples of surficial materials and plants from 863 sites. The locations of these sites are shown on the maps of element dis- tributions in this report. The elemental compositions of only theisurficial materials are given in this report; the data on’analy- ses of the plant samples are held in files of the U.S. Geological Survey. ACKNOWLEDGMENTS This study was made possible by the cooperation of many persons in the U.S. Geological Survey. We thank Messrs. D. F. Davidson, A. T. Miesch, and A. T. Myers for their interest in, and continued support of, this study. The sampling plan was sug- gested by Mrs. Helen L. Cannon, who also contrib- uted analytical data from her project areas and many samples from her travel routes. We thank also Messrs. E. V. Post and W. R. Griflitts for the large number of samples that they collected for this study. Others who collected samples, and to Whom we express gratitude, follow: F. A. Branson, R. A. Cadigan, F. C. Canney, F. W. Cater, Jr., Todd Church, J. J. Connor, Dwight Crowder, J. A. Erd- man, G. B. Gott, T. P. Hill, E. K. J enne, J. R. Keith, D1 D2 Frank Kleinhampl, A. T. Miesch, R. F. Miller, R. C. Pearson, M. H. Staatz, T. A. Steven, M. H. Strobell, V. E. Swanson, R. R. Tidball, and J. D. Vine. We acknowledge the analytical support provided by the following U.S. Geological Survey chemists: Lowell Artis, Philip Aruscavage, S. D. Botts, L. A. Bradley, J. W. Budinsky, Alice Caemmerer, E. Y. Campbell, G. W. Chloe, Don Cole, E. F. Cooley, N. M. Conklin, W. B. Crandell, Maurice DeValliere, P. L. D. Elmore, E. J. Fennelly, J. L. Finley, Johnnie Gardner, J. L. Glenn, T. F. Harms, R. G. Havens, R. H. Heidel, M. B. Hinkle, Claude Huffman, Jr., L. B. Jenkins, B. W. Lanthorn, L. M. Lee, K. W. Leong, J. B. McHugh, J. D. Mensik, V. M. Merritt, Wayne Mountjoy, H. M. Nakagawa, H. G. Neiman, Uteana Oda, C. S. Papp, G. D. Shipley, Hezekiah Smith, A. L. Sutton, Jr., J. A. Thomas, Barbara Tobin, J. E. Troxel, J. H. Turner, and G. H. Van- Sickle. We were assisted in computer processing of data by the following persons of the U.S. Geological Sur— vey: W. A. Buehrer, Allen Popiel, M. R. Roberts, W. C. Schomburg, G. I. Selner, R. C. Terrazas, and George Van Trump, Jr. REVIEW OF THE LITERATURE The literature on the chemical analysis of soils and other surficial materials in the United States is extensive and deals largely with specific agricultural problems of regional interest. Many of the papers were written by soil scientists and chemists asso- ciated with State agricultural experiment stations and colleges of agriculture, and most reports con- sidered only elements that were known to be nutri- tive or toxic to plants or animals. Chemists with the U.S. Department of Agriculture prepared most early reports of element abundance in soils for large areas of the United States. (See Robinson, 1914; Robinson and others, 1917.) The 1938 yearbook of agriculture was devoted to reports on soils of the United States; in this book McMurtrey and Robinson (1938) discussed the importance and abundance of trace elements in soils. Amounts of the major elements in soil samples from a few soil pro- files distributed throughout the United States were compiled by the soil scientist C. F. Marbut (1935) to illustrate characteristics of soil units. The use of soil analysis in geochemical prospecting began in this country in the 1940’s, and many reports were published on the element amounts in soils from areas where mineral deposits were known or sus- pected to occur. Most of these reports included only a few elements in soils from small areas. This early STATISTICAL STUDIES IN FIELD GEOCHEMISTRY geochemical work was discussed by Webb (1953) and by Hawkes (1957). In succeeding years, as soil analyses became an accepted method of prospecting and as analytical methods were improved, many elements in soils were analyzed; still, the areas studied were commonly small. An estimate of the amounts of elements in aver- age, or “normal,” soils is useful in appraising the amounts of elements in a soil sample as related to agricultural, mineral prospecting, or health and disease investigations. Swaine (1955) gave an ex- tensive bibliography of trace-element reports on soils of the world, and he also summarized reports of the average amounts of elements as given by several in- vestigators. The most comprehensive list of average amounts of rare and dispersed elements in soils is that of Vinogradov (1959), who reported the analyti- cal results of extensive studies of soils in the Soviet Union, as well as analyses of soils from other coun- tries. He did not state the basis upon which he estab- lished the average values; however, these values are presumably the arithmetic means of element amounts in samples from throughout the world. Hawkes and Webb (1962) gave average amounts of certain ele- ments in soils as useful guides in mineral explora- tion. In their discussions of the principles of geo- chemistry, Goldschmidt (1954) and Rankama and Sahama (1955) reported the amounts of various elements present in soils and in other surficial materials. A recent report on the chemical characteristics of soils was edited by Bear (1964). In this book, the chapter on chemical composition of soils by Jackson (1964) and the chapter on trace elements in soils by Mitchell (1964) give the ranges in values or the average amounts of some soil elements. The average amounts of elements in soils and other surficial materials of the United States. as deter- mined in the present study, are given in table 1, with the average values or ranges in values that were reported by Vinogradov (1959), Hawkes and Webb (1962), Jackson (1964), and Mitchell (1964). The averages from the present study in table 1 are the arithmetic means. Although the averages were com- puted by the methods described by Miesch (1967), the values obtained are directly comparable with the arithmetic means derived by common computational procedures. COLLECTION AND ANALYSIS OF GEOCHEMICAL DATA SAMPLING PLAN The sampling plan was designed with the empha- sis on practicality—in keeping with the expenditures ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES D3 TABLE 1.—Aoe’rage contents, and range in contents, reported for elements in soils and other surfic'ial materials [Data are in parts per million; each average represents arithmetic mean; ________ , no data available] Hawkes and Webb (1962) Present report chemical prospecting) (elements useful in geo- Vinogradov Jackson (1964) Mitchell (1964) (1969) (presumably, Element avfelrgges Average Range Average Range gggg‘ig: “Typical” average, Razgfesigttéggfient or range in values surface soils Al _______________ 66,000 700—>100,000 ________________________ 71,300 10,000-60,000 _____________ B ________________ 34 <2o_300 10 _____________ 10 30 _____________ Ba _______________ 554 15—5,000 500 100—3,000 500 ____________ 400—3,000 Be _______________ 1 <1—7 6 ____________ 6 ____________ <5—5 Ca -A _____________ 24,000 <150—320,000 ________________________ 13,700 7,000 _____________ Ce _______________ 86 <150—300 ________________________ 50 _________________________ Co ________________ 10 <3—70 8 1—40 8 ____________ <2—80 Cr _______________ 53 1—1,500 200 5—1,000 200 ____________ 5—3,000 Cu _______________ 25 <1-300 20 2—100 20 ____________ <10—100 Fe _________________ 25,000 100->100,000 ________ 14,000—40,000 38,000 7,000—42,000 _____________ Ga _______________ 19 <5—70 ________________________ 30 ____________ 15—70 K ________________ 23,000 50~70,000 ________________________ 13,600 400—28,000 _____________ La _______________ 41 <30—200 40 ____________ 40 ____________ <30—200 Mg _______________ 9,200 50—100,000 ________________________ 6,300 <6,000 _____________ Mo ________________________ <3—7 2 0.2—5 2 1—10 <1—5 Mn _______________ _ 560 <1—7,000 850 ZOO—3,000 850 ___________ ZOO—5,000 Na _______________ 12,000 <500—100,000 ________________________ 6,300 _____________________________ Nb _______________ 13 (10-100 ____________________________________________________________________ Nd _______________ 45 (70—300 ___________________________________________________________________ Ni ________________ 20 (5—700 40 5—500 40 ____________ 10—800 P ________________ 420 20—6,000 ________________________ 800 500 _____________ Pb _________________ 20 <10—700 10 2—200 10 ____________ <20—80 Sc ________________ 10 <5—50 _________________________ 7 ____________ <3—15 Sr _______________ 240 <5—3,000 _________________________ 300 ____________ 60—700 Ti _______________ 3,000 300—15,000 4,600 1,000—10,000 4,600 1,200—6,000 ______________ V ________________ 76 <7—500 100 20—500 100 _____________ 20—250 Y ________________ 29 <10—200 ________________________ 50 ____________ 25—100 Yb _______________ 4 <1—50 ____________________________________________________________________ Zn _________________ 54 <25—2,000 50 10—300 50 ______________________________ Zr _________________ 240 <10—2,000 ________________________ 300 _____________ 200—>1,000 of time and funds available—and its variance from an ideal plan has been recognized from the begin- ning. .Because the collection of most samples was, by necessity, incidental to other duties of the samplers, the instructions for sampling were simplified as much as possible, so that sampling methods would be consistent within the wide range in kinds of sites to be sampled. The samples, other than those from certain project areas, were collected by US. Geo- logical Survey personnel along their routes of travel to areas of other types of field studies. The locations of the routes that were sampled de- pended on both the network of roads that existed and the destinations of the samplers. Sampling intensity D4 was kept at a minimum by selecting only one sample site every 50 miles along the routes. The specific sample sites were selected, insofar as possible, to represent surficial materials that were very little altered from their natural condition and that sup- ported native or cultivated plants suitable for sam- pling. In practice, this site selection necessitated sampling away from roadcuts and fills. In some areas, only cultivated fields were available for sam- pling. Contamination of the sample sites by vehicular emissions was seemingly insignificant, even though many sites were within 100 yards or less of the roads. However, we had no adequate way of meas- uring any contamination that may have occurred. (See Cannon and Bowles, 1962.) Many of the sam- pled routes had only light vehicular traffic, and some were new interstate highways. Routes through con- gested areas generally were not sampled; therefore, no gross contamination of the samples was expected. The study areas that were sampled were as fol- lows: Wisconsin and parts of contiguous States, southeastern Missouri, Georgia, and Kentucky, sam- pled by Shacklette; Kentucky, sampled by J. J. Con- nor and R. R. Tidball; Nevada, New Mexico, and Maryland, sampled by H. L. Cannon; and various lo- cations in Arizona, Colorado, Montana, New Mexico, Utah, and Wyoming, sampled by F. A. Branson and R. F. Miller. Sampling techniques used in these areas varied according to the primary objectives of the studies being conducted, but, generally, these techniques were closely similar to the methods used in sampling along the roads. In general, the sampling within study areas was more intensive than that along the travel routes. To make the sampling intensity of the two sampling programs more nearly equal, only the samples from selected sites in the study areas were used for this report. The selected sites were approximately 50 miles apart. Where two or more samples were col- lected from one site, they were assigned numbers, and one of these samples was randomly chosen for evaluation in this study. SAMPLING MEDIA The material sampled at most sites could be termed “soil” because it was a mixture of comminuted rock and organic matter, it supported ordinary land plants, and it doubtless contained a rich microbiota. Some of the sampled deposits, however, were not soils as defined above, but were other kinds of rego- liths. These regoliths included desert sands, sand STATISTICAL STUDIES IN FIELD GEOCHEMISTRY dunes, some loess deposits, and beach and alluvial deposits that contained little or no visible organic matter. In some places the distinctions between soils and other regoliths are vague because the mate- rial of the deposits is transitional between the two. Samples were collected from a few deposits consist- ing of mostly organic materials that would ordinarily be classified as peats, rather than soils. To unify sampling techniques, the samplers were asked to collect the samples at a depth of approxi- mately 8 inches below the surface of the deposits. This depth was chosen as our estimate of a depth below the plow zone that would include parts of the zone of illuviation in most well-developed zonal soils, and as a convenient depth for sampling other sur- ficial materials. Where the thickness of the material was less than 8 inches, as in shallow soils over bed- rock or in lithosols over large rock fragments, sam- ples were taken of the material that lay just above the rock deposits. About half a pint of this material was collected, put in a paper envelope, and shipped to the US. Geological Survey laboratories in Denver. CHEMICAL ANALYSIS PROCEDURES The soil samples were oven dried in the laboratory and then sifted through a 200-mesh sieve. Finally, the sifted, minus-200-mesh fraction of the sample was used for analysis. Some of the samples having a high clay content formed a hard mass when dried and had to be pulverized in a ceramic mill before being sieved. Contents of zinc and phosphorus were determined by the colorimetric methods described by Ward, Lakin, Canney, and others (1963). Analyses for amounts of calcium were made by the EDTA titra- tion method, and analyses for potassium by flame photometry. Quantities of the other elements were determined by semiquantitative spectrographic analysis (Myers and others, 1961). The values obtained by spec- trography were reported in geometric brackets hav- ing the boundaries 1.2, 0.83, 0.56, 0.38, 0.26, 0.18, 0.12, and so forth, percent; the brackets are identi- fied by their respective geometric midpoints, 1.0, 0.7, 0.5, 0.3, 0.2, 0.15, and so forth. Thus, a reported value of 0.3 percent, for example, identifies the bracket from 0.26 to 0.38 percent as the analyst’s best estimate of the concentration present. The pre- cision of a reported value is approximately plus or minus one bracket at the 68-percent level of confi- dence, and plus or minus two brackets at the 95- percent level. ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES The limits of detection of the analytical methods that were used are given in table 2. The values given in table 2 are the approximate lower limits of determination. Some combinations of elements in a sample, however, affect these limits. For example, concentrations somewhat lower than these values may be detected in unusually favorable materials, whereas these limits of detection may not be attained in unfavorable materials. METHODS OF STATISTICAL ANALYSIS The values obtained by analysis for each element in each sample, and the location of the corresponding sample site, expressed as degrees and minutes of latitude and longitude, were entered on automatic- data-processing cards. 'The analytical values were transformed to logarithms through a computer pro- gram which also determined the minimum and maxi— mum values, as well as the basic statistics. In addi- tion, the program reported all occurrences of missing data, indicated the concentrations that were beyond the limits of detection of the analytical methods, and printed both a histogram of the analytical values and an accompanying table of frequencies and cumulative frequencies for each class designation on the histo- gram. By examining the table of frequencies, we were able to divide the range of reported values for most elements into five classes, so that approximately 20 percent of the values fell into each class. (See figs. 2—31.) However, the limited ranges in values for some elements prohibited use of more than two or three classes to represent the total distribution. Class-interval code numbers for each element were then assigned to each analytical value: values in TABLE 2.—Analytical limits of detection [Data reported in parts per million. Analyses made by semiquantitative spectrographic method, except as indicated] Lower limit of Lower limit of Element detection Element detection (ppm) (ppm) Aluminum (A1) _____ 10 Molybdenum (Mo) ___ 3 Barium (Ba) ________ 2 Neodymium (Nd) __1 70 Beryllium (Be) ______ 1 Nickel (Ni) _________ 5 Boron (B) __________ 20 Niobium (Nb) ______ 10 Calcium (Ca) _______ 1150 Phosphorus (P) _____ ’2 Cerium (Ce) ________ 150 Potassium (K) ______ a50 Chromium (Cr) _____ 1 Scandium (Sc) ______ 5 Cobalt (Co) ________ 3 Sodium (Na) ________ 500 Copper (Cu) ________ 1 Strontium (Sr) _______ 5 Gallium (Ga) _______ 5 Titanium (Ti) ______ 2 Iron (Fe) ____________ 10 Vanadium (V) ______ 7 Lanthanum (La) ___ 30 Ytterbium (Yb) _____ 1 Lead (Pb) __________ 10 Yttrium (Y) ________ 10 Magnesium (Mg) -_-_ 50 Zinc (Zn) __________ ’25 Manganese (Mn) ____ 1 Zirconium (Zr) ______ 10 1 Analyzed by EDTA titration method. 3 Analyzed by colorimetric method. 3 Analyzed by flame photometry. D5 the lowest class were assigned the code number 1; values in the highest class were assigned the code number 5. These class-code numbers, together with the lati- tude and longitude of the site where the sample was collected, were then directed from the computer to the automatic plotter, which located each sample site on base maps of the United States by printing the class code for each element at each sample site. These class-code numbers on the base maps were then re-' placed with symbols, as shown in figures 2-31. The geometric means and geometric deviations given in figures 2—31 and in table 3 are antilogs of the arithmetic means and standard deviations, re- spectively, of the logarithms of the analytical values. Where some of the concentrations for an element were determined to be less than the sensitivity of the analytical method (table 2), the mean and standard deviations of the logarithms were estimated by means of a censored-distribution technique devised by Cohen (1959). The geometric mean is a measure of central tendency of the frequency distribution and, as such, is an estimate of the typical or most common concentration for the element. The range from the geometric mean multiplied by the geometric deviation to the geometric mean divided by the geo- metric deviation generally includes about two-thirds of the analytical values. Approximately 95 percent of the values occur in the range from the geometric mean multiplied by the square of the geometric devi- ation to the geometric mean divided by the square of the geometric deviation. For example, the geometric mean aluminum content of surficial materials of the United States is 4.5 percent and the geometric devia- tion is 2.41 (fig. 2; table 3). It is estimated that approximately two-thirds of the samples have alumi- num contents in the range 4.5—:-2.41 (1.87 percent) to 4.5x2.41 (10.84 percent), and about 95 percent have aluminum contents in the range 4.5+(2.41)2 (0.78 percent) to 4.5x (2.41)2 (26.1 percent). Further discussions of the treatment of censored frequency distributions of geochemical data and of the use of geometric means and geometric deviations were given by Shacklette, Sauer, and Miesch (1970). The arithmetic means of the analytical data given in table 1 were derived from the estimated geometric means and geometric deviations by using a technique described by Miesch (1967), which is based on methods devised by Cohen (1959) and Sichel (1952). The arithmetic means in table 1, unlike the geometric means shown in figures 2—31 and in table 3, are esti- mates of geochemical abundance (Miesch, 1967) and are directly comparable to arithmetic means com- D6 monly employed in presentations of geochemical averages in the literature. Arithmetic means are always larger than corresponding geometric means (Miesch, 1967, p. B1) and are estimates of the fractional part of a single specimen that consists of the element of concern rather than of the typical concentration of the element in a suite of samples. RESULTS The results of this study are presented on maps of the conterminous United States which give, for each element, the locations of the sample sites, and the concentration of the element at the site. Several elements were detected in only a few samples. The lower detection limits, in parts per million, of the semiquantitative spectrographic method for these elements are as follows: Antimony, 150; bismuth, 10; lithium, 50; silver, 0.5; and tin, 10. The localities of these samples and the contents of these elements in the samples are given in figure 1. The concentrations of 30 elements were sufficiently large to be detected in samples from many sites. The results of analyses of these samples are given in figures 2—31; the concentrations of elements are indicated by symbols, as classified in the accompany- ing frequency histograms. Although we intended to analyze these 30 elements in all samples, through error, certain elements were omitted in the analyses of a few samples. Some elements were looked for in all samples but were not found. These elements, analyzed by the semiquantitative spectrographic method, and their lower detection limits, in parts per million, are as follows: Arsenic, 1,000; cadmium, 20; germanium, 10; gold, 20; hafnium, 100; indium, 10; platinum, 30; palladium, 1 ; rhenium, 30; tantalum, 200; tellurium, 2,000; thallium, 50; thorium, 200; and uranium, 500. If lanthanum or cerium was found in a sample, the following elements, with their stated lower detection limits, were looked for in the same sample but were not found: Dysprosium, 50; erbium, 50; gadolinium, 50; holmium, 20; lutetium, 30; terbium, 300; and thulium, 20. DISCUSSION AND CONCLUSIONS The data presented in this report may reveal evidence of regional variations in abundances of elements; single values or small clusters of values on the maps may have little significance when con- sidered alone. Apparent differences in values shown between certain sample routes, such as those across the Great Plains and the Central States, suggest the possibility of systematic errors in sampling or in laboratory analysis. Some gross patterns, neverthe- STATISTICAL STUDIES IN FIELD GEOCHEMISTRY less, are evident in compositional variation of rego- liths, as shown in figures 2—31. The lower abundances of some elements (notably ‘ calcium, magnesium, strontium, potassium, sodium, aluminum, and barium) and the greater abundances of titanium and zirconium in regoliths of the Eastern United States indicate a regional pattern of the largest scale. The abundances of these elements differ markedly on either side of a line extending from western Minnesota southward through east- central Texas. This line is generally about the 97th meridian and corresponds to the boundary proposed by Marbut (1935, p. 14), which divides soils of the United States into two major groups—the pedalfers which lie to the east, and the pedocals to the west. Marbut (1928) attributed the major differences in chemical and physical qualities of these two major groups to the effects of climate. A line approximating the 97th meridian also separates the Orders, Sub- orders, and Great Groups of moist to wet soils in the Eastern United States from the same categories of dry soils which lie to the west, as mapped by the [US] Soil Conservation Service (1969). Geometric mean compositions of regoliths west and east of the 97th meridian are given in table 3. Superimposed upon this large—scale compositional variation pattern are several features of intermedi- ate scale. Most notable of these are (1) the low concentrations of many elements in soils of the Atlantic Coastal Plain, and (2) the general excep- tion to this low concentration in the part of the Coastal Plain crossed by the Mississippi River. Several small-scale features of compositional vari- ation have been noted; among them, the low con— centrations of many elements in regoliths of lower Michigan, northern Nebraska, and the part of Texas adjacent to the southeast corner of New Mexico are prominent. In contrast, large amounts of some elements tend to occur in regoliths of Maine, parts of the Rocky Mountains, and the Basin and Range province. The concentrations of certain elements, especially boron, cerium, neodymium, niobium, ytterbium, and yttrium, do not exhibit well-defined patterns of dis- tribution; and the regional concentrations of some other elements cannot be evaluated, because they are not present in detectable amounts in most of the samples. The degree of confidence in regional pat- terns of element abundance is expected to be in direct proportion to the number of samples included in the region. As the patterns become smaller, the probability increases that the observed character- ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES D7 TABLE 3.—Geometm’c mean compositions, and geometric deviations, of samples of soils and other surficial materials in the conterminous United States [Geometric means reported in parts per million. Too few molybdenum values were available to make a statistical evaluation] The conterminous United States Eastern United States (east of 97th meridian) Western United States (west of 97th meridian) N~492 N = 863 N z 371 Element . . Geometric Geometric Geometric Geometric Geometric Geometric mean deviation mean deviation mean deviation Al ,,,,,,,,,,,,,,,,,,,,,, 45,000 2.41 54,000 2.02 33,000 2.70 B ___________________ 26 2.05 22 2.09 32 1.92 Ba ______________________ 430 2.06 560 1.80 300 2.19 Be ___________________ 0.6 2.49 0.6 2.47 0.6 2.53 Ca ,,,,,,,,,,,,,,,,,,,,, 8,800 3.92 18,000 2.93 3,200 2.87 Ce _____________________ 75 1.67 74 1.64 78 1.70 Co ____________________ 7 2.21 8 2.01 7 2.55 Cr ___________________ 37 2.32 38 2.16 36 2.52 Cu ,,,,,,,,,,,,,,,,,,,, 18 2.28 21 2.00 14 2.54 Fe ____________________ 18,000 2.30 20,000 1.90 15,000 2.76 Ga ____________________ 14 2.11 18 1.71 10 2.53 K ,,,,,,,,,,,,,,,,,,,,, 12,000 2.71 17,000 1.60 7,400 3.56 La ____________________ 1.85 35 1.81 33 1.90 Mg ___________________ 4,700 3.19 7,800 2.21 2,300 3.39 Mn ___________________ 340 2.70 389 1.94 285 3.65 Na ______________________ 4,000 4.11 10,200 1.98\ 2,600 4.11 Nb 1111111111111111111 1 12 1.66 11 1.74 13 1.54 Nd ___________________ 39 1.72 36 1.81 44 1.61 Ni 11111111111111111111 14 2.26 16 2.03 13 2.60 P _____________________ 250 2.74 320 2.33 180 3.03 Pb ,,,,,, . ,,,,,,,,,,,,,,, 16 1.96 18 1.93 14 1.96 Sc ....................... 8 1.79 9 1.74 7 1.85 Sr ________________________ 120 3.39 210 2.12 51 3.56 Ti ,,,,,,,,,,,,,,,,,,,,,,, 2,500 1.87 2,100 1.82 3,000 1.84 V ____________________ 56 2.16 66 1.91 46 2.41 Y ,,,,,,,,,,,,,,,,,,,,,,,, 24 1.77 25 1.66 23 1.93 Yb ____________________ 3 1.81 3 1.67 3 2,03 Zn ,,,,,,,,,,,,,,,,,,,, 44 1.86 51 1.78 36 1.89 Zr ___________________ 200 1.90 170 ’ 1.78 250 1.95 istics which form the patterns are the result of chance. Some small- and intermediate-scale features of element-abundance patterns are known to reflect geological characteristics of the areas that the soils overlie. A few soil samples with high phosphorous content, for example, are associated with phosphate deposits in Florida, and a single sample with high copper content from the Upper Peninsula of Michi- gan is known to be of soil that occurs over a copper deposit. Samples from most of the regoliths overlying basic volcanic rocks of Washington and Oregon con- tained higher than average concentrations of iron and of a few other elements. These data do not provide consistent evidence of north-south trends in elemental compositions that might be expected to relate to differences in tempera- ture regimes under which the surficial materials developed. There is, moreover, no evidence of signifi- cant differences in element abundances between glaciated and nonglaciated areas (the general area of continental glaciation includes the northern tier of States from Montana to Maine and south in places to about lat 40° N.). The world averages of abundance for some ele— ments in soils, as given by Vinogradov (-1959) and by others (table 1), do not correspond to the averages of abudance for those elements in soils of the United States, according to the data presented in this report. The world averages are too low for the amounts of boron, calcium, cerium, lead, magnesium, potassium, and sodium in United States soils, and too high for beryllium, chromium, gallium, manganese, nickel, phosphorus, titanium, vanadium, and yttrium. This report presents, for the first time, averages of the abundance of niobium, neodymium, and yttrium in soils. ' REFERENCES CITED Bear, F. E., ed., 1964, Chemistry of the soil [2d ed.]: New York, Reinhold Publishing Corp., 515 p. Cannon, H. L., and Bowles, J. M., 1962, Contamination of D8 vegetation by tetraethyl lead: Science, v. 137, no. 3532, p. 765466. Cohen, A. 0., Jr., 1959, Simplified estimators for the normal distribution when samples are singly censored or trun- cated: Technometrics, v. 1, no. 3, p. 217—237. Goldschmidt, V. M., 1954, Geochemistry: Oxford, Clarendon Press, 730 p. Hawkes, H. E., 1957, Principles of geochemical prospecting: U.S. Geol. Survey Bull. 1000—F, p. 225—355. Hawkes, H. E., and Webb, J. S., 1962, Geochemistry in mineral exploration: New York, N.Y., and Evanston, 111., Harper & Row, Publishers, 415 p. Jackson, M. L., 1964, Chemical composition of soils, in Bear, F. E., ed., Chemistry of the soil [2d ed.]: New York, Reinhold Publishing Corp., p. 71—141. McMurtrey, J. E., Jr., and Robinson, W. 0., 1938, Neglected soil constituents that affect plant and animal develop- ment, p. 807—829, in Soils and men—Yearbook of Agri- culture 1938: Washington, U.S. Govt. Printing Office, 1232 p. Marbut, C. F., 1928, Classification, nomenclature, and map- ping of soils in the United States—The American point of view: Soil Sci., v. 25, p. 61—70. 1935, Soils of the United States, pt. 3 of Atlas of American agriculture: Washington, U.S. Govt. Printing Office, 98 p. Miesch, A. T., 1967, Methods of computation for estimating geochemical abundance: U.S. Geol. Survey Prof. Paper 574—B, 15 p. Mitchell, R. L., 1964, Trace elements in soils, in Bear, F. E., ed., Chemistry of the soil [2d ed.]: New York, Reinhold Publishing Corp., p. 320—368. Myers, A. T., Havens, R. G., and Dunton, P. J., 1961, A spec- trochemical method for the semiquantitative analysis of STATISTICAL STUDIES IN FIELD GEOCHEMISTRY rocks, minerals, and ores: U.S. Geol. Survey Bull. 1084— I, p. 207—229. Rankama, K., and Sahama, T. G., 1955, Geochemistry: Chi- cago, Chicago Univ. Press, 912 p. Robinson, W. 0., 1914, The inorganic composition of some important American soils: U.S. Dept. Agriculture Bull. 122, 27 p. Robinson, W. 0., Steinkoenig, L. A., and Fry, W. H., 1917, Variation in the chemical composition of soils: U.S. Dept. Agriculture Bull. 551, 16 p. Shacklette, H. T., Sauer, H. I., and Miesch, A. T., 1970, Geo- chemical environments and cardiovascular mortality rates in Georgia: U.S. Geol. Survey Prof. Paper 574—0, 39 p. Sichel, H. S., 1952, New methods in the statistical evaluation of mine sampling data: Inst. Mining and Metallurgy Trans., v. 61, p. 261—288. Swaine, D. J ., 1955, The trace-element content of soils; Eng- land, Commonwealth Agr. Bur., Commonwealth Bur. Soil Sci. Tech. Commun. 48, 157 p. [U.S.] Soil Conservation Service, 1969, Distribution of prin- cipal kinds of soils—Orders, Suborders, and Great Groups: U.S. Geol. Survey National Atlas, Sheet 86. Vinogradov, A. P., 1959, The geochemistry of rare and dis- persed chemical elements in soils [2d ed., revised and enlarged]: New York, Consultants Bur. Enterprises, 209 p. Ward, F. N., Lakin, H. W., Canney, F. C., and others, 1963, Analytical methods used in geochemical exploration by the U.S. Geological Survey: U.S. Geol. Survey Bull. 1152, 100 p. Webb, J. S., 1953, A review of American progress in geo- chemical prospecting and recommendations for future British work in this field: Inst. Mining and Metallurgy Trans., v. 62, pt. 7, p. 321—348. FIGURES 1—31 D10 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 6 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108“ 106° 104° 102° 100° 98° 96° 48 \7 I / I I I I I I I I I I I I I I \\\ \~\ 7 r‘~~\--\ 46°\ /, (I ~““\... ‘5 H 1 ~‘_—~[”"““‘”‘~--—~~————-___.. \ ‘ ._.._, G T 0 N I, 1“ F I ~~~~~ ‘9, M 0 N I \ 1 ~- T ., , x 44\ SW15) \ [I ANA (NORTII DAKOTA‘ CA I I , 1 I \ G 0 \t t L Q r—..._______________J 42°\ / b.~.r©“\~~‘.-_ \ i 11) A I O \--J l I H O I ' 1 l /\ / 7:L ISOIITHQDAKOTAI \ I \42 AgI3 I » . _ I 40°.N “ ‘ )o I W I (I M I N G I‘“~ 1) g... 7 ‘\\/ --—~fFI——-\V I I )> “NI” I S11(15) ’ ' / olA\5‘~ \ I N E B R-X‘ * K5 380\ NYE l g() 7 ‘~ ~-_~L“'~‘1 DKA I V A l ' D A / 1: I 0 SnII5I ‘- T A H ’ ° 5 \ m I I C o “(20) —--——————__—_. I I oAgI2),Sn(15) I. 36° '7 / Z .SnI15) ' \ \ I C 0 L 0 R ‘1 I '15» U‘ I I ’\ D O ' K A \1 9 A 0 \2I\ l 7 . '~° “-~-2\ I \ r1 T“——\_f‘_ 1 34°\ \ ‘ Z ‘ ~-—-———-————.........— \\ ARIZONA ’, U3 ,l ““‘--i ‘0 2’ 74/ .Ag(3)ll i llOELAHOMA 32“\ ,J’SHUO). ' N E W _ O I M E I I . X I C 0 , ~~~a .3115) I 2 \ I I kJMA~r¢L \\ Q I I U) 30K \ A N G i .Li(70) .Li(70) \\~\-Lf“‘° ~--.__d ' T E x A s \ 28°\ \1 \\ /h~\ \\r \‘ 26°\ \ \\ 1". 24°\ k. 22°\ l / I 'I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° FIGURE 1.—Location of sample sites in the conterminous United States where elements not commonly ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 96° 94° 92“ 90" 88° 86° 84° 82° 80° 78“ 76” 74° 72° 70° 68° 66° 64° I I l I I I I I \ \ I \ \ \ \ \ \ /48° /"’~\ /46" ,44“ ,42" I I NLAN“ “"“—" ~ /40° l 5 v \ _— " \ESYX)V \ 10 W A f T “381‘“ ‘1. (I \ “ §\ “A“ / 38° RN“ * 2k / j | \ 0 3H.) I \IAH4SM410 —"'_" “ILLINOIS‘ ,A ’L/‘y’ ,N“ --—\A \ 'Sn(15)\”‘ NDIAI‘ i! {figs'l‘ ’ 7x 3 < f vnw G‘N ‘I-Af \ \l \ “‘w‘vs 36° I 4‘; f f / / IMISSOURII , U. \ I‘A ( : \\‘ ”‘QMMCKY fvaR’L-ffi/‘Z, .1 \ ~J Sb(500) fii’jpfl—fl/ D ___..... ’j “'W—fl'fl’ I» ROL N\ v /34 I._—-————, -———“ x GA Er , f SQ 3E ’ N \ L”? '1‘ EN 1 “IN E “ { -'-""""\.’-\ V .— \ I KANSA‘I T‘"’T‘ ‘ ETRQ AR L r—v" q\ 50 ‘ ,32° ero) f; I \ \gAROIIXNA I 1 I \ \ fl \Lw ’ I I \ GEORG‘A\~ ., I ________3 MISSIb-IALABAMA\ \/ , a [— g SIPPI I s «V .~ ,30 I ’4 I ( 9 ' N : ‘I 6—69 P " 3“. MILL—C —- ~——-——-» \ o A, LOUISIANA K .L y »— x /?8 b "' I <3 k“ L1(.50) \\ // — .43 \» , f ‘. ,26" " o 4: ‘5 x a o v I 24° 0 500 MILES ‘ I; / L I I l I I ‘3 “9"" ,22" I . I I I I I I \ \ I I \ 96° 94° 92° 90° 88° 86° 84° 82° 80“ 78° 76° 74° detected in surficial deposits were found, and the amounts of the elements,in parts per million. present. D11 D12 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° I I I I I I I I I I I I I [\\““~\ I (1 ‘~\§‘\ . ~-‘-~ ’"—:_.. 0/. 1 7 --———--—..__ _. I as S I .k,’ a. I 99" o 9 o e \ ~\I , O 0. o I 0 OH \ I 0 o I o o . e / (’1 o o e o [I e 9 O O .I\o ’ -° 0 G. 3 L. ° ~~\ .I‘ ”‘”‘—“T."9 . o 0 v f£ “~C-__‘W=’ 0 \ 0 a ’ 0 o O °I I I O I o o O G 0% $0 O 0’0 o o 9 I I o . i :1er a , ° 9 9 ° ° 1"“ ‘w\2 ., 9 . GI‘“————~——_. ° , a I . . “m, 6690 00, 0o f 9 0 0 . o q o O G o \5 o lo 0 60 0.0\ 7 .9“_‘ oi . 0 e 5 I o .5. '5‘“? 0 ‘° f 000 900 09. . 9 I 9 o . o . K— c, o 09 I 90. <3 L__* \ I 0° 0 0 0| 0 0 o o 0—; o_‘ / ' '0 I ' ::- . . I o e o o e 0 I Lo 0 f o ' e o I \~0 o ‘ a- o 0 o o o o O I o e 01‘ W+‘\.‘ O . or o , o ‘1 _.~‘_ °.I_ 0 o o o 0‘“ ——-——_._.._____._.__ e 0 I ° . . 00 J . 0 0 e o ““‘w-v O . I o 0 ' o I o l o o O 9 O o I e o 9 O o l 0 G 0 | . .0 o o a 0 j a 0 l O 9 o 0 l O I o O o O G . , o o o . J o 0 e 0 5w 0 0 My ’ I 0 '9 0 UMAN/ga‘. 0 0! e I 0 \ o a 7 o o \\ ° 0 ‘ 0 0 0 o o G o o \\ G 000' ‘3. o o o o 0 o 0 0 300 ‘\~-L‘] ‘LXOF 194—40 o \o o e 0 0° 0 o o \ O O G 0 28°\ \ o o 200 I o O 0 5 \ A» 0 o O E \ f \ O I) 0 E \\’ \\ O 0 G 26 ‘ 100 \ \K‘ 1 24‘; ° 1 m \L do‘c’oooo—émmmrxga ~\ ALUMINUM, IN PERCENT Geometric mean, 4.5 D Geometric deviation, 2.41 22 \ Numberofsamplesandanalyses, 791 I I I I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96" ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 96" 94° 96° 94° 90" i 92" 88° 90° 86“ 84° 82° 80° 78° 76° 74° 72° i \ \ \ \ \ A\ \ 500 MILES J 88 ° 86 " 84° 82 ° 80 ° FIGURE 2,—A1uminum content of surficial materials. 70" \ \ 78" \ 76° 64° \ 74° / 48° I 38“ ,36" / 34° / 24° D13 I314 , 128° 48°\ 7 FREQUENCY 22“.\ 126” 124° STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 122° 120° 118° 116° 114° 112° / / l I I I I I I moooo ~Nmm~ o c GeomeIric mean. 430 Geometric dewahon. 2‘06 110° 108° 106° 104“ 102° 100° 98" 96° I I I I I I Number of samples and analyses, 863 I I I I I I I I I I I I 118" 116° 114° 112” 110° 108° 106” 104° 102° 100° 98“ 96" ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76" 74° 72° 70° 68° 66° STATES 64° I I I I I I I I \ \ \ \ \ \ \ 500 MILES 5 —0 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76" FIGURE 3.—Barium content of surficial materials. \ \ / 46° 44° 42° 40° 38° 36° 34° 32° 30° 28° 26" 24° 22° D15 D16 44" 42° 40° 38° 36° 34° 32° 30° 28° 26° 24° 22° STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° I / I I I I I I I I I I I I I I I \\ \\‘ o ‘7-‘ 0 I I‘“\~- 0 o I ’ \‘~\ 0 I ”‘\—_ n o / N -- —__ C o X ‘1 “—--—-— —— I, ox Y 0 I 00 e o o 0 \ .~--.\~ \HO 0 9 O ‘4 I O 2 a I) 0 o\o r» \ I O O 0 g 0 I I (I o e 9 0 g// e \‘ o o 6 09 I e O O 0:0 0 L o o o t—~_____ 0 ° 'quJ-~‘\‘— o ”:— 0 0 9 0 e ” ~e"‘\—9.I 0 \I II 0 e o I, 0 o o C’I ° I / O o o 0!) e 000 o 0 IO 6 I\“\ ' 0 o o I o I) O o I -‘ o 9 ‘ O 0 I O O o Y...— ‘U\Q O O G OI—“——-—.._.__ $6 00 I 0 ‘i O O o I " ‘§ 0 090° 00’ 00 f o o o o q o o O O 0 lo 0 - 0 0C, 9 O o o 0 OO 060 7\@‘ ~ 5 ' o O 7L5‘1 . ‘0 o r 000 [000 000 e on g o 0 . .L ' o 0 ol 0 00 I 00. e L________________\_ . o I o o. 0 0 ol 0 o o o o o o o e I . o O r 9 0 .0 o I \ ' o o 0 O 01 0 ° ° 0 I I . O O I \ o o O or) O\ 0 LD\. 9 O O l ‘ 2 ' o o o e o / o 001‘ ~T+~ .0 00 of) O o o e \ N] I .0 :‘h—..___ OJ. 0 O O (‘0 0 o 0“ —~—~—-———-—~—— o 9 I 0 0 0 o O o o ‘ 00 0O 0 o r) ‘--——... o k I 0 o I0 o o 0 , O I) O o o o O I . o I o 9 o I o 0 I J o o ’ o o o 0 3 ° 9 , ”Do 0 OI o e o I I G O O I O 9 J‘ 9 . O G O —- ‘0 ° 0 46 o o 0 d O \ T O O O . 0 3‘1. x I O o 0 ana/d» \ o 0' e 0 I \\ O O (7 O O G 0 o o o SYMBOL AND PERCENIAGE \ o 0' 9 o o o i0 g 60 2 (O) o e 9 OF 10m SAMPLES \‘0 0 ..~ L‘cfi‘u o \‘i‘w’ " ‘0 “-0-40 0 \o O o 0 00 o o O \ o o o o \ . o O , I o 0 O o g \ A... o o w \ g \\r( \\\O O O O O x ‘\ “2M“ ‘ V \ BERYLLIUM, IN PARTS PER MILLION \K Geometric mean. 06 ppm ~\ Geometric deviation. 2.49 Number oI samples and analyses. 847 l l I I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES D17 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76“ 74° 72° 70° 68° 66° 64° 48° I I l l l \ \ \ \ \ \ \ \ \ \ \ \/ ’46” / 44° , 42° , 40° ,38“ , 34° , 32° / 30” ,28' o 500 MILES " / I , ,22‘ 96" 94° 92° 90° 88° 86° 84° 82° 80" 78° 76° 74° FIGURE 4.—Beryllium content of surficial materials. D18 46° 44° 42° 40° 38° 36° 34° 32" 30° 28° 26" 24° 22° STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° ~ I / / I I I I I I I I I I I I I I \\ ~o\‘~ o , ’I‘ ~\-‘\~ \ o 0 I I \.~ ‘ o / “‘— 0 .. '--—_ .l o 1 7 *-———-——..$___ 0\ I d I 05 e 0 0 ° \ .~_~,\“J \L‘o o o (I 0 e \ \ o \ II o o o g I 0 I90 O J I; G 0 ° 9 o o o 0 / \‘ 0 o O 0 ‘I ° . 0 01° 0 I o O I L 0 e o rl‘“———____a_,) e o '4 r{~‘\“—s- 0 0° \ \ o. a“. 5-2-3.. e . I 0 o 6 <3 I 0 e o I I / O o 9 o O}; . Ca. 6 0 I6 0 I [\‘\\ I . I O O I H‘ o G 0 . I O O o ‘ ‘~ I o o g—w \ ‘j\i O O 0 OI—_ 'h———.__ ’\ e I O 0 O I V ( o o q o H . 00.9 '9’ o. ‘f a 9 o e o C, 0 K1 0 lo 0 \- O 0 i O 0 o O I 00. 0 7\-O§ ‘- .L 9 ‘ \ 0 9 9 o 3 ““9 9 ‘6 7 oo . .0 00 <3 0 o 9 e 9 e L G e . O o 90 I 00 (3 L \ e 0 I I o ““—--——-——— \O 0 I . o O 0 9 GI e 0 O o o o o O O O I O O O r 0 0 00 a I \ \ 0 o o o o o, . o o 0 e I O \ 9 o I e 9 o 0 I . 9 O o 0' .0 \ O/lw-\O‘Q?~‘ f9 0 e G O P G O G g 0 G O 0 ~. 9 o T+~\Q‘e 9 o \ I J I e. e h_-\~__ OI. 6 0 0 o ————..... \ o o e . O O O 0 ~_———-— 0 I o 6 ° 10 oo 0 0 o '“fi‘fi . O o 0 G O O 0 ' I O O O \ I G I e l O o ' o l’ o e e I G o 0 j o 3 ° I o \ o 30 ° e: 9 9 o o . o I O I o g (3 I 0 0 0 o e _ ‘9 O o O o o o' *\ “T O 0 O J . 0 W1. \ G I 0 O O G fi‘vthr‘flx \\ G 0 0| 0 O l g SYMBOL AND PERCENTAGE \ o 0’ 0 o 0 9 \ OF TOTAL SAMPLES \\ G ' O o :9 O 0 g 6. e 0 o o o e __ -‘_ O o o 0 30° OIC‘IIGl'l \~O\°‘-I-\I ‘kfo o U~~€hdo a o O o 9 90 G e O \\\O O 0 o o \ 200 1 O o > O o E’ \ A... O 9 o e .5 \ 8 \ , \\C\) o O O O \ 100 5' \ °\ 1 0 oocoooooo \ \ <3~~M~2222 \N BORON, m PARTS PER MILLION "K. Geometric mean. 26 Geometric deviation, 2.05 Number of samples and analyses, 863 / / I l I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76° 74° 72° 70° 68° 66° 64° I I l l \ l \ \ \ \ \ \ \ 0 500 MILES 4 96° 94° 92° 90° 88° 86° 84° 82° 80° FIGURE 5.—Boron content of surficial materials. \ \ 78° \ \ 76" \ \/ 74° 46° 42° 40° 38" 36° 34° 32° 30° 28° 26° 24° 22° D19 D20 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 0 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° 48 \ / I I I I I I I I I I I I \.\“ o I P\‘\~‘\ 46 \ . I I ‘~\- I ‘\~.. 0 / \-— -_‘ e 9 1 7 --—-——.._._ _.__._ I, 0\0 Y 00 I... g o e \ a u~_~.\~\\1 ‘50 o 0 3 e .\0 44 \ I o a o 09 o > " 0 o o I 0 o o / (r\ . . o o I a , . 0 0'. O I e ‘ . I O. \ L o ‘ 0 r ““——-———_TQ__J 42° } 0 '«pr ‘\~~ ° ° \ \ 0 ° 0 w“\-9J o I I, o o o I, o o g I 9 o 6 o o I e I / o G . A ° '8 o o l “I o o o I o . L . 9 I 40°\ \“\,.: o I o o . __ 0 Y‘— ‘6\£ . 0 . \*—~—_._ I o ‘I 0 0 . I “NV-\é.‘ 000' 00’ f 0 g 0 0 . Q o e o o lo 0 ° \- 9 0* 9 g g X D O .C .3 O 7\—D‘ 0 z, 38 \ a '\‘-‘_J_-‘ ‘ o o o ‘1: 5 oo . 7,.9 090 . o I e . 0 9 e 0 . o, o O. I, 090 o L__~——~ \ o e —-‘_“‘ I 6 0 e, g 0 0 O 3 Q a o 0 .° r ' “.00 o I e 36 \ O / o 0 o o 0 e I I o I ' o 9 e o g o I ‘4 9 9 . .I 0/ . °}‘~\+ 0 96 o of o o 0 e o G{) <3 . ‘l G ‘\_-‘ O o i I " o ' “\-_ °_I_ 34 \ ° 0 e O. O O O G —“"'—_———-—--——-— I o e 9 o 9 ° -- o . e 0 0 e e ‘——. . k 0 0 '0 O o G O I O L 0 O O O , . I o o o ' o o . 0 I 0 o I o o a. 0 e 0 j e I 32 \ ,fi 0 o e ' o o I o I . ° Q 0 . (1L 9 0 o 0 ° \. 0 o J o . o . \ T 0 ' 0 o o ‘3" \ I o . 6\V¥\p\/J\. \ O! 0 Q \ G O Q 0 30°\ \\ 70 O 99 . O 0 O O 9 Q 0 \ o 3 GI ° . o . 0 ° 0 0 0 \-\0\_L‘J~_\Lofi_‘q§_, . ‘ a . e e 0 e e \ . ' o SVMBOL AND PERCENTAGE OF TOTAL SAMPLES \ 0 . 0 O 28 \ 8‘ \ o . o I 0 0 . \ A- o > \ 0r \ O . . o O O E ~f \\ 26°\ 3% \ ‘\ 1 24°\ QENmWNODOO \L CALCIUM, IN PERCENT HAN", "K. Geometric mean. 088 Geometric deviation, 3192 0 Number of samples and analyses, 855 22 \ I I I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° ELEMENTAL 96° 94° 96° y—O 94° 92° 90” 92° COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 88° 86° 84° 82“ 80° 78° 76° 74° 72° 70° 68" ‘ T \ \ \ \ \ \ \ \ \ 500 MILES " 4. 90° 88° 86° 84° 82° 80° 78° 76" FIGURE 6.—Ca1cium content of surficial materials. 66° Y \ 74° ,44" , 42° , 4On / 38° ,36" /34" ,32‘ /30° ,28" ,. 24° D21 D22 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 8° 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° 4 I / / / I I I I I I I I I I I I I \\ \\\ o ‘7‘- O I I‘“\‘ \ 46° 0 00 I I ‘~\~- 0 o / ’ \“\—~ -_ o, o 1 ‘I --———-——._ .__... I 0:) I C? . .~——..,\\ \1 \HO 0 000 o 0 o o o O \\ 44” o \ I o o o O O o \ I O 0 o o I) O J C\ O I O O I o o 0 l/ o I o O o o I o o O \o o o I L o - ° 0 r&—~————--oo ~I 42° I; O Lira!" ~\"‘~e_ % O \ O 0 IO ° 0 0“““NJ o I o O! I 9 O o l o / O o o G 0/) q) o o, o I /\“\J O O I o (1') O O ' ~ 0 o 40.. \~‘ 3 o I e O O ’—- O y_ ‘\~ 0 o -~~~—.___ GO I ‘l o 0 o OI d\V \(fi 0 o o o 0 oO / o\ \D o OCI o o o 0 K5 38" I e 000 07 um...___‘_ _ 0 ‘10 o 0" \ o o o I) o 990 O 9 T a O o o O k- 9 I 90 0 L \ e o I o I Q -~-—-—-—— __ ._ __ _ __ O\OOO IQ 0:6'99'0990000 36° 0 o / 0 O I0 0 o I e e o e O O I O C o \O o I O O I oo \ 0 Lo; Q 0 OI o/ T~~ O o 0 OP 0 o o o o o o F \ Ox] 00 ~E‘f‘~~—..Q‘~o o I 0““\~ OJ. 34° \ e O o o o o ‘ —---_.~_._._.._._.____ 0 O I o o O 0 e 0 oo 0 o o o ‘*-—~——q \ ° 0 '0 O o .' O J o o o 0 C e ’2 e o l O O O o O 9 I O o oo 0 32° [>0 00 OI r) O :1 o ' o I o l g e o I e 9 O (L 0 o o ~~~<>° ° e I0 o 0' a~ o e I O I O A \. \x 0 Q, 9 0' 6‘3. "x” 30° \\ O O c, O o 0 e e o o SVMBOLANDPERCENIAGE \ O I O O 0 0 f0 2 00 ° 3 0 ° 0 or IOIAL SAMPLES \NO 00 .. 2 L~O\ _ o O N no \‘14 O 0 174—4 0 on H \ o o e o o 0 an” \O 0 000 o O O D 200-. x 28 V \ 5 , I o z \ h~ o g »\ \ O 0 gm \ ,f \\ 26° 1 \ x \ o x ”522% \ 24° CERIUM, IN PARIS PER MILLION \v‘ Geometric mean, 75 —..\ Geometric deviation, 1.67 Number of samples and analyses, 753 22" I I l I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° 96° I 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 94° 92° 90" 88° 86° 84° 82" 80° 78° 76" 74° 72° 70" 68° 66° 64° l I \ l \ \ \ \ \ \ \ \ \ \ \ /48° 46° / 40" , 28° .3- n , / 24° 0 500 MILES 4 /22° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76“ 74° FIGURE 7.—Cerium content of surficial materials. D23 D24 48° 46" 44° 42° 40° 38° 36° 34° 32° 30° 28° 26° 24° 22° STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° / / / I I I I I I I I I I I I I I \‘ ~.\\\ - 7 "r~-\-- I "‘ ~ 0 . e 0 o / I \. ‘\‘_\-- 9 e ) T“‘*-———__ __ I, ex. 3' I 0‘” g o e o \ “N-.. o o a {\“J ‘7 O .6 (5 9 0‘0 e > , o o G o e .I o o . / 6‘ o 0 o '0 I 0 0 O 0 .IO 0 ° / O ‘ o I Q“ \ 0 I; L. o ~§ G or ~~——_.____Fo__J \_§ 9 O G ’0 O 0 41Mf£ flo‘_‘.~a G \‘ I O o . II 0 0 O .I I o I O o Q 0I' ° ”15’ O O I“ I “ e F \N'\ o 9 I 0 o A e I “ ° ° ' ° ° ° I r" O 7. )TQ‘I o o e o 0‘ WV _ $1 0 e G 0000 Col 00? . O G 6 CI 0 O O 0 Go IO 0 00 o\‘ \4 0 0CI 0 O 0 0 I 0 ‘\__ L s o a“ o Ia o f 0007000 Cog 00 0-1.0 o o o 0 0L 0 o e e, o 90 I 90 o __ \ 0 0 o O “ -- ~—- —— _ __ I 0 or 9 0' 0 G 0 o 0 <3 0 ° 6 o / ' 6 ° I ° ° '0 e \ O o 0 o 0 o e o 0 I 9 \ o I G ’0 a 0 O I a e 0 9‘ 9 o o o O \ 0/L0\‘_9T‘ f9 . G G 0'? 0 e . 0 O \ O‘v e 'o ‘T+—\Q_O o 0 ° 0 o I ’90 ' ““—- 0_I_____ o ________ ° K o 2’ e 0 I o o o e o o 0 00 0 J30 0. GI g G 0 j G I e I 0 g I (3 O : O O O 4‘ e O O 0 0 __ \o 9 10 e o e o ..\ T 0 0 0 o O \EOVL \ I o O G d‘vaANflx \ O .1 O 9 \\ 0 o 7 o o 0 \ O a 00 ° 0 O 0 o o o 0 \ O 0 ol 1 ° 0 O o o o o o \- O -L F-\I\O‘1-—-o~.a__, 0 ~ ‘J \ o \O 0 e e a. o O G \ o o 0 o \ o 0 o I o O o o \ "- o e > \ ff \\0 O O . ‘z’ * \ E» \ E x ‘\ 1 .n \ ~\ Geometric mean, 37 Geometric deviation, 2.32 Number of samples and analyses, 863 I / l l I I I I I I I I 118° 116° 114° 112° 110° 108‘ 106° 104” 102° 100° 98" 96° D25 ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 66° 64° . 68" \ 84° 82° \ \ 86° ‘ 96" o 6 4 / / 44° , 42° , 40° / 38° ’36° / 34° /32° / 30° , 28° ’26” ,. 24° 500 MILES ,22" 74° 76n 780 80° 82° 84“ 86° 88° 90° 92° 94° 96° FIGURE 8.—Chromium content of surficial materials. D26 46" 44" 42° 40° 38° 36° 34° 32° 30° 28° 26° 24° 22" 128° / / 118° STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° / / I I I I I I I I I I I I I I \\ \\N o ‘7‘ . I [\“\“ o o . [I “\.... ~ . . / ‘“~ -—.~-_ _~ . .1 —~~-....~-___ ..__ 01 I \ I .\H (ed, 0 G g o ‘ u~—~.\~\\! o 0 o I; 0 \ o I 9 e o e 0 ‘3 o > (I, o o 0 ° 0 I o e o ' / \‘ o 9 09 'I 9 e 0 oI\0 . 0/ O L 0 G O Q.______ o J . 3 e 'v f£‘\ ‘~-9. QI— ° \ o o I. o 0 5 I “~~—-..OZI Q I I 0 o o I ° 0 o I Q . O I I / o 0 [3 o .09 G O 0 o . I I x“ t o ' \~‘\ 0 0 I e o 5 \“ a C I O O G g.” ‘ \ 0 e f‘“'~-—-.._...__,__ o '7. 3 ‘~/ . O 0 G 0 6‘v ( .0. I 0 0 I KM, e o 0 .006 0’ 09% G O 0 s q 0 O O 9 lo 0 N.‘ o as! 0 o O 09.0 ‘4‘ 0 3 I o 0 “~ 'L"-- e I. o 0 <3 *9 e O 0 GI 0 00 II no. 0 L__ _ \ o ' I . e (I: o 0 OI . o o . . I, . o 0 0 9 0 o a. e l 9 9 o O 0 e 0 I e 0 0 o \ O I o e I ° 0 o ' o o.e\ 9 Lo‘- 3 o 9 I’ 0/ W‘.‘ o 9 G 0 g o O 0 O o .. 0 ° 0 TI‘NQ 0 0 ° 0 \ l _ I Go . h—_- ‘H \ . O O 0 0 G O fi —-— —— —__. __ ._ .. __ 0 e I o o g o o O 0 9° 0. 9 o 0 0 ——~.~..___, 0 k . . I 0 O ' o I e L 0 0 0 ° 0 e . I o o 9 0 o G . 00 g j 0 I )0 O OI e e o o I O 1 0 l 0 e O (L G o 0 G G -‘ \o ' a J. ’ o o . c' 3‘ Z 0 o \ . l c 0 o O 9 d‘kl‘LAN/dl\ \ o 0: e a g o \\ o e o o \ 9 . .0 0 C G e o o o 0 GI 1 ° 0 o o O o O 0 \‘0 O H‘ ‘km—Ug~4 O ‘ o x o o c o o o SYMBOL AND PERCENTAGE OF TOTAL SAMPLES \ ° 0 o 0 o 0 e z 2 22 :3 \ O 200 oI 0 10m 0 I \ 0 o I I | I I O o o I I 1, c. O O o 5 \ or \o O o z I ‘\ \ o O 0 I J 2100 \ E I \ u- | I \\ | I ' \ m a '1 V _#~mm.\ \ COBALT. IN PARIS PER MILLION \N Geometric mean,7 a. Geometric deviaIIon, 221 Number of samples and analyses, 855 I l I I I I I I I I I 116° 114° 112° 110° 108° 106” 104° 102° 100° 98" 96" 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 94° 92° 90° 88° 86° 84° 82° 80° 78“ 76° 74° 72° 70° 68° 66° 64° 96° 94° I 92° 1 \ \ \ \ \ \ \ \ \ \ *Y 500 MILES f |— 90° 88° 86° 84“ 82" 80° '780 76° FIGURE 9.—Cobalt content of surficial materials. \/ 46° 44° 42° 40° 38” 36° 34° 32° 30° 28° 26" 22° D27 D28 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° 48 \ I / / I I I I I I I I I I I I I I \‘ ‘8‘“ 7 I\“\~ 0 \~ 46 \ 0 g 0 I (I ‘\___. ~ 0 0 / ‘ ~\_- __~‘ 9 . ) 7 --——-..—_.. __ II “a I 0 I 0. u~...~..\ J \H. 00 be 0 o 0 O \ 44x 0 “ I 0 °. ° 0‘0 o > ’ o e 0 e 0 . I o 0 / C} 0 9 0 e 0 I e e O o 0‘0 0 °’ 0 L o a... o G - o r —'-"-"—“—"'J 42° L, “I; “~~~..e_ ° 0 \ o o l0 0 e . 5 I 0 “—!J o \.I I o o g I o . o 0| o 0 e I I / o o 0 IG a 000 e e o I I\‘\ I g I O . ' \N, o I 9 (3 0 ‘ G o 0 .. 40x \7‘ I o 0 e _ . g— ‘\S_‘/ O 6 0 O ~""“"‘~-~—-—.. I o " I 6‘“ \( 0° 9 q o N’ 0 38° 0 l ‘3 go... ‘7~—._9‘ “(I o O O *3 \ I O 0 7.1.0“? 9 ‘0 o f 00 g [0.0 90. ° 0 o 0 ° L o 0 o 0 ol 0 90 'I 000 o I-——.__._ _____ \ O O I 0 9 . ° ° 9' ° ° ° 0 o o 0 36° 0 O I o O 0 r e o '0 e 0 o \ \\ 0 I o 'l O o a. o o I o o 0 0° 0\ o Lbi‘g f a o 0 CI 0/ 0 eT“ . 0 OP 0 o G 0 O 0 G G \ o ‘1 O T+~\e_‘o . a i o I 90 o F‘-\__ o 34 .\ ° 0 Co a 0 ° 0 o ‘‘‘ ‘— ** —‘ "" — o I o o o 00 (10 “Ge 9 o a -—_.._., o \ 0 0 e ,o o o 0 I o L 0 o o o l 0 ° 1’ ° ' I O O 0 o 0 ° ' O o O Q. 0 G 0 j 0 I 32 .\ J30 c o 0 e , 0 I 0 ' 0 O o (1‘9 e o 0 0 .‘__()0 ° . l e 9 o 0 o d ;\ o o 9 \ .I a so a I 0 WV» 0 o 0 I o a \\ ' .I g 0 3O \ \ 9 ” o ‘30 0 O 9 o o e o a \ 0 o .1 0 ° 0 O o 0 9 e o \§\G~—L ~-\Loo\o—-G-.o-_, 0 \\o o o e 0 0" o o e \ O o O 9 28°\ \ o 0 o 1 a 0 \ A- o 5 o \ O 0 g \ [if \\\ G G O 0 26°\ g \ \\\ Q 1 24o\ VH-wasszaasgggg \ COPPER, IN PARIS PER MILLION LN \ Geometric mean, 18 Geometric deviation, 228 Number of sampies and analyses, 855 22°\ / l I I I I i I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° 96" I 96" ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 94° 94° 92“ 1 90° i 92" 88° 86° 84° 82° 80° 78° 75° 74° 72° 70° 68° 1 l \ \ \ \ \ \ \ \ \ soo MILES f 90° 88° 86° 84° 82° 80" 78° 76“ FIGURE 10.—Copper content of surficial materials. 66" \ 64" \ / \ 74° 48° 46° 44° 42° 40° 38° 36° 34° 32° 30“ 28° 26" 24° 22" D29 D30 44" 40° 38° 36° 32° 30° 28° 26° 2?.” STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128° 126" 124° 122° 120° 118° 116° 114° 112° 110° - 108° 106° 104° 102° 100° 98" 96° ~. I / / / / / I I I I I I I I I I I \‘ ‘.\\‘ o \[NI‘“\ I I ‘~\-__ \ ' °° I I “‘“~-_- 0 o / k\*~ -_‘_ e . ) -_—.~-__ __ I 9‘ ‘5 I G I d! o o 9 o 0 \ ~ 0 .~_-,\\\1 \Ho 3 o é 0 GM \ . I o o o .0 o t, I o o 0 ° 0 I o I; . o / (/1 e o o e e I o 9 (3 e ’10 9 / O a f 1‘ 0 O o i“‘---—~.......__O__3 ° I . wa~~\w 0. 0e \ \ o O I. o . I 0....“5‘ G I I O . Q I O G O I I o / ° 0 o 9/9 o 0.. 9 o la 0 [\\‘ , ' O I 0 o I \u‘ 0 0 I 9 Q . I \ \~n._ e O. I O G O gh— 7~NI a . . J~~-~-~~~—.\V .. I o 0 O I \(fi .0 .... 0.] .0 11 e e G o o 9 O 0 G o ‘5 o o .0 l0 9 .0. o\‘ \4‘ .1 o 0 9 x \ I “--..L.____~_ 1; o o - o C o f 099 70.0 '0. o O '. O Q o o 0L ° . o o, e 09 ,1 0°. 0 I— —— __ __ __ _ 5 . o I ' o 0 0I o ' 0 To“:- \9 0 '0 .90 0 0 . . I .0 o o . g o. . I \ \\. ° . I 0 ’ ’ ° . ° ° I .0 o\. . LO; Q i e . o o ' I o ‘ \“ 0 . . r o g o e o \\ O \1/ o G 00 T+~é\9~.~e' O 01 O . 0 N \ I0 0 o I 09 9 -7-“ 2 ‘~——-—-————-——- . o 0 \I o . 9°. J° “0° 0 .GI'Gv:*-——a 0 O O 0 O O o \ o I o G ’I 9 l g o o O O o [I o g I 0 e 0 : ° 1 \ . It“ . Go 0 G 'I O O 3 0 ’ o I 0 I G O I 0 0 J\ 9 O o O O ____ ‘0 ° . 0 ° 9 a o 0‘ 0 , '2’ O o o O 0 W \ I o O 0 6MA~,,.~ \ O . °' 0 G I e swam AND PERCENTAGE \\ . o G \ 0? 10m SAMPLES \\ . 9 o :0 G G O O 00 a o o o o OI - o 0 O O 300 ‘~'\‘_L\f“-1\00 ‘o‘U--&~.J o O ‘ o o 0 \' O a g o o o \ o O O O \ 200 \ . O O 5 \l 6 O O z \ A- o o o ‘5 \ 8 \ ff ix\0 0 9 G O 5 ~ \ \ 100 \ \\\ 1 0 mmomoooo \ N v ~_Nmm~ GALLIUM, IN PARTS PER MILLION L~ \ Geometnc mean, 14 Geometric deviation, 2.11 Number of samples and analyses, 860 l / I I I I I I I I I I 118° 116” 114C 11.2” 110° 108° 106° 104° 102° 100° 98° 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES D31 96" 94° 92° 90° 88° 86° 84° 82" 80° 78° 76° 74° 72° 70° 68° 66° 64° 48° I I 1 $ X \ \ \ \ \ \ \ \ \ \ \ \/ 46° , 42° /30° O O 00 €300 0 o 500 MILES 7, 96° 94° 92" 90° 88° 86" 84a 82" 80° 78° 76" 74° FIGURE 11,—Ga11ium content of surficial materials. D32 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 48° 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 196° N I I I I I I I I I I I . [\2-\\ 46°\ I \~~\‘ ~\“ . ‘- “"~_ \ 7 3 9 0° ‘50 . 00° ‘IO 0 e o o . \\ 44°\ I, . 0 o . o o C 6 . o I G G I, \ o ‘ o 0 o o I 9 o .|\o L 0 ° e Fin—«-“TLJ o I “\_ 0 O 42 \ 4rf ‘Nfi‘- 0 \ e e \__1 0 l e . I o . I I e o 0 6F o $9 0 0 I0 . O I 0 £ 0 . I 0 40°\ 0 ’ l o o 0 J— . Y— ‘3\g o 0 ‘~———____ o ‘I O O I “V‘Kg 9 '0 f 0 e 0 O 6 q 0 O O O /‘3 o \- . ° # o o 0 KS 0 O 90. 0 \fi‘ 0 .3 38 \ I o "‘~‘° 1'5“.» o ‘6 f 000 [000 “9. o 9 I 9 G o o 9L o o 00 I °00 G L__ x o —_.____-. . ’ 00 r 0 9000 0 0 0: 0 9 0 0 o o e e o 36 \ o O o, 0 o 0 o I 0 0 o o I i o 0 ° 9 o g ‘9\ 9 o e e o e o 00 ‘WT‘f—‘g 00 . o I' 0 o o o o I 9. o ‘F‘~\-_ 01 34 \ O o o 0 o ’ “~‘--—-—-—-—--—-- 0 o O 9 O 3 O 00 0 O O 0 *“—-fi 0 0 0 1° 0 O 0 1 G l o O G O 0 (3 o I o e e I o o ‘ rs o G 0 32° cl 9 9 . ' . \ o l . G o I o e (L o o O o 0 ’ o 0 e o 0 «~\ 9 9 e G “a" I A \ 0] 0 e G O 5“} “r, a o O 30 \ 0 19 o '0 O O o 0 a 0 o o \ 0 o 0’ p- :\ O o o o O O 0 ‘\G~_L\l \koo 0‘04-.. 0 ‘ O o o o \. o O o o o o O a 2 2m: 2 x 0 ° 28 \ o o \ . o I e O O \ A- 0 e o o 5 \ ff \\0 O 0 0 0 3 ~ \ 26°\ § \ E \ ‘\ 1 24°\ 0 \k man. In PERCENT 7 "‘\ Geometric mean, 1.8 Geometric deviafion, 2.30 Number of samples and analyses, 861 22°\ / l I l I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, C‘ONTERMINOUS UNITED STATES 96" 94° 92° 90° 88° 86° 84° 82° 80° 78° 76° 74° 72° 70° 68° 66° 64° _ I'__ 96" r—O 94" I l 92" 1 \ \ \ \ \ \ \ \ \ \ 500 MILES " 4’ 90° 88° 86° 84” 82° 80° 78° 76“ FIGURE 12.—Iron content of surficial materials. \ \ /48° / 46° ,44° /42" , 40° , 34° / 24° D33 D34 48° 46° 44° 40° 36° 34° 32° 30° 28° 26° 24° 22° STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° 7 I I I I I I I I I I I I I I I I \s \O\‘\ O 7 ‘Il\“\- I ‘\~~ e O O I l \“\—-_ 0 \~- ~._ 0/. 1 -—~—_..____ __ I O\ I I a 00 I 0° 0 e 0 5 \ .~_-_,\\ \HO 0 0 o O ‘4 I o 00 e d O\0 o \ ’ e G 6 ° I l C e o O o 2/ \1 ° 9 ° 0 ° I O o 0 0 01° 0 O o / o O ‘I 1'0 ~ ° oF°‘“"-~-r:—9 ~\_“ ° D I0 . ' 0 “Ukr‘f 4~T\‘§-I e \I l O o o I 0 e o I I o / o 9 . 9/) e c§9 e 9’0 O [\K‘ I . . I O o I \N‘ 0 G I o O O I \ \ g o I . O o If" ‘b\g 0 e I———~_._____ o . 7 e 4 . o ° °. “av—M o ' . o l o q o H e 90. 9 o f o o o o o 0 O .0 O O G O 9* e 0 o 9 009 0 7‘3‘ § 0 \ I 8 o ‘TLUMa o H I so 0 0 O I O O \ e [00 0, o e o . L x O . 0 0’ O 9’ I] .0. G L~~ -_ \ g\. 0 o I, .0 Core .°:.o°° 10.00000 . . o o . o . \ o o I o 0' e e o e o '1 \ . O O O 9. °\ . Lo; f . e °I ‘0 WIT 00W 0 ° "C‘+~\O o g 0 \ «I I .0 ;—.‘__fi 0 . 0 o 0 O 0 .~; 0 “-~——~~.-___________ o 0 Q I . . 00 J. 0.0 . e 9 ;-m_q . ° 9 0 ° 9 e o \ ° ' 0 I 9 l O o e G o o O o o I ll 0 e I O 0 I . e . e O J g , 2 e c a 0 o o o l O I O I . O O O o G O O I 0 ___ \0 ° . 4 o o o O o ;\ e o T\ 9 I O O O . fixvt"'\,r"‘ \ O 0' o I I e \ . e G SVMBOL AND PERCENTAGE \ 7 . . o o o . OF TOTAL SAMPLES \\ o I e .o o O e 00 06 e o o o 228 2 9 ,._ e 300 Oleo 0 I \3‘11‘: \ng‘j‘g—Jl-do 0 I \ 0 O O 0 e 0 | \ . 000 e 0 | \\ O I o I O s 200 l 1 o 0 O 9 z \ I‘- . o O “5 | \ f \o e e g i \ \‘r \‘ O O | 100 I \ I ‘\ 1 \ “one“. ‘\ v~m~223 LANTMANUM, IN PARTS PER MILLION \‘~\ Geometric mean. 34 Geometric deviatlon, 1.85 Numberofsamplesandanalyses,837 I I l I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 96° 94° 92° 90° 88° 86° 84" 82° 80° 78° 76° 74° 72° 70° 68° 66° 64° 48° I I I I I I I \ \ I \ I \ \ \ \ \/ . , 24° 0 500 MILES I" I I I I I I I I I \ \ \ 96° 94° 92° 90° 88° 86° 84° 82° 80“ 78° 76° 74° FIGURE 13.—Lanthanum cdntent of surficial materials. D35 D36 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY a 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° 48 \ I / / / I I I I I I I I I I I I I I“~\-‘ 46°\ I \~~\_~ \fi‘ \__ _-~__ \ 7 I a \He 0 o 1005 e e o o e \ 44°\ I O 9 0 a & 0‘0 I O o 9 G e I o (1‘ 9 0 e o 9 I, o e o a 0:9 0 1 o 0 ““~--———_ @— D L. ~~\ el— 1': I 42\ 45hr ‘-~a.__\ 34 G \ o o I, 0 ° o o i o o of) o “a. 9 519 9 ° I l 0 e g o o p 400\ 7.:E .9 I e e 9 g I"--——_ at... \~ 0 o ”'“--— , o ‘l e 6 ' , ‘°\\-I"\\(fl o o o .0 f 0 0 0 q o o a o /o o \‘ 0 .+ o G 0 8° 00 0.0 \J‘\‘_ L 0 5 3 \ 0 T 5“!) 0 ‘0 f 009 7..e 0.. e e I 9 O , 9 ok- 0 0 <3 00 ’I .0. ° L____ \ . . ol . (“‘“T‘ . o .0 r 0 ’0'. I. I 0 o o o 6" o 3 \ e e o e, . . .0 . . I, f O o ' . o, ‘0 . ‘ .1 o o o o o o 0:" “'G-+~\g~o° . ol 0 . o o o I°o o °~-—~~— e1 34 \ e 09 9 e 0 . 9 ~—-————_____.__ 0 9 Jo °e e . o ___5__‘ 0 e , o O 9 9 g I e l G O G o I O 0 ° ° I 0 o 0 e g I o G 0 GI 0 ° 3 . . e ' 32 x 0 l 0 O l O o 1‘ ' O . . o 9 o 0 ° . J o o o O 0' -\ o I e o O J o \B’Ld A“ ~ V 6l . O O l N; ’1’” a g 7 e 9 30 \ o o 00 o o 0 o . o o o . “ cl 9 O 0 <3 0 3 o 9 0 \ F- "' ~ I ‘\1.Lq “if?“ ““QH'O 9 e 0 O o o o I o o o 8 I \\e O o o 2 °\ I 0 O | i 1 o O O > \ 1“. 0 o 0 2’ I \ { “\a e G 9 0 ° 3 | r \ 6 3 | ‘ 2 °\ E | . \ I \ I \ g 1 24°\ \k N\ Geometric mean. 16 o Geometric deviatcon. 1.96 22 \. Numberofsamplesandanalyses.863 I I I I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 92° FIGURE 14.—Lead content of surficial materials. 96” 94° 90° 88° 86° 84° 82° 80° 78°._ 76° 74° 72° 70° 68° 66" 64° I i l l l \ \ \ \ \ \ \ \ \ \ \ \/ / ._._J‘ , ’ I I I L—_———_— / I s 5 / 0L 1—‘20 0 Ce , -__J / O / ‘5' / \LG. 3 / O 0 I 9 / ’ o 500M|LES " .f ’ I I I I '. I l l l l 1 \ \ \ \ \ \ 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76° 74° 46° 44° 42° 40° 38° 36° 34° 32° 30° 28° 26° 24° 22“ D37 D38 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 6 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° 48 \ I / / I I I I I I I I I I I I I I \\ \G\“ . 7 ~I\~‘\~ 46" I I \‘~\ \ 0 o Q ~-\ 0 o / l ‘T “— __ e a 1 7 —~—~————~ —— I, 0‘0 I 00 I 0. o 9 O 0 \ .~—~4\“\3 kg,“ 9 0 L. Q \ 44°\ . I o e O 0 o o b l e 0 0 9 I I. 0 <3 0 ' / (I\ 0 o . O o I 9 . o o I o o I I 0\ 0 I a 2 I L Q C _‘“~'“-—-—-—__Q__J o . k I “\‘ <3 G. 42 L 0 . . o o . ”VAL- ~~.-__~_ . \I II 0 o o I 9 o 9 O 8' 0 ° 6 I I o o 0 0° 0 o o I'\‘ / O . I. 0 I. o I ‘\ ' . I 40° “‘\:~ 0 e 0 II 00 5 0 O b 14"" \ —-_ 71k; 0 . , I M-_____.\. O O 0 I v--'\. ( (3 I O . o I 0 Q 0 K.) o 00.0 o e, f o g o e O o C O \S G. l. . ...I3 ‘ \43 Q .I o O O 380 7 ‘ \~- 1.. 5 \ I G ‘7;- "9“.1 O ‘0 ° I W .9 WL e I 90 o L \ .0 . oI . .0 I e G 0 I ”~—-——————._.._____. 0 I e 0 G a 6 o 36 \' o o / ° 0 .GI ° “0 ’0 o 6I a o 0 . O o . O 9 I O\\ 0 . I 0 £0 a 60 O O I '0 .‘o 0 Lo; I 0 0 ° I 0 \ ‘o-~- o o I° 0 O 0 ‘3 0 0 o «I o G 0 fir-I-‘sp‘ e o o 0 g \ l I .0 G‘.‘ _\‘ 01 34° Io <3 . G o ‘ ~-_______ 0 I . ”a o J. O o o —"—“— . 00 I GO 9 e G —‘ ‘“' O . e I0 0 6 3 o l o (1 o o o K 0 I o o 0 o I o I c 0 0 e 0 o I o ’ a 0 o e :I 9 I 32" f . 0 . .’ 0 <3 0 I ° 0 I I Q 0 (3 O J‘ O O O G O ‘5 \. 0 9 J 9 o e . .A\ T 0 ' o 9 ° \ I O o O 0 bWA~/\. ' \\ ‘ o .' 0 ° 30° \ <3 70 o d ° 0 0 \ 0 .90 . o 9 G e o o e \ g G at o G a 9 o O 0 \‘\L.L..IM\1:UP‘°‘~‘-~ 0 \ o \e o G 0 <3 0 o 0 SYMBOL AND PERCENTAGE OF TOTAL SAMPLES \ 0 g e 0 28° 2 z ‘2: E’. \ 200 I o IOIOI o | I ° ° 0 I I l I | o e | | \ A- o o I l | \ f A“ 0 0 g G O 53 I | | r \ ° 5 I | ~ \ 260 §100 | \ E | \ | \ | I \ 0 1 24° 2 o \‘k MAGNESIUM. IN PERCENT A ~\ Geometric mean. 0.47 Geometric deVIatIon, 3.19 Number of samples and analyses. 850 22° / / I I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96" ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76“ 74° 72° 70° 68° 66“ 64" 40 I l l I \ \ \ \ \ \ \ \ \ \ \ \ \/8 46° /38° v- . /240 0 500 MILES g 96° 94° 92° 90° 88" 86° 84° 82° 80° 78“ 76" 74° FIGURE Iii—Magnesium content of surficial materials. D39 D40 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 0 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104“ 102° 100° 98” 96° 48 ~ I / / I I I I I I I I I I I I I I \~ ~\\ 0 ~ __ C 7 F\‘\‘ \ 46°\ 9° . . I (I “\_‘\~‘ . 0/ ) “ ~—~..._____ . ____. __ II ox 7 y 0 o 1 9° o . o o \ 44° under-«\J K; o 9 $0 0 0‘ \ o I o o 00 9 . \ I G e O . 0 I k I C e . O o . / \l o 9 e '0 ‘1 ' e o .'\o o , o . i L o O O l”‘~———__.__O__} o . 3 1 1‘ ‘~\‘ 0 9. 42 \ O O . O O 9 “W ‘4§“—Zl . \‘ II o g . II o . o O! l 0 O / ' e o I, o $0 0 .IO 0 /\\\ I . 0 l O Q I \N‘ o o I o o o 1 I: ~‘ 0 . I 9 ~— 40 \ 7‘ o o I»~ . ‘a\i 0 “‘-—-—— e I . O 0 ° °: “0—H o . o q o \5 g 00. 0 :01 .o f o o e e o o o . 9 0 \ . .4 o O a o .0. e 7\g_\ . o s 38 \ I '0 o “713*.” . s. O f 609 Iago 0.. ol 90 e 90; 6 . 6 el o ,0 I, 00 9 L——___ \ 6 G . 0 0 l o Q —_—~_' I . a 0. 9 O O o o o o a o o l 0 O r 9 ° 0 e l 36 \ . . O 0 .I 9 . C O i \ I O O . .. O I o. e\ o IL°;~S ‘ e u 9 o; o -‘ o o 0 0 e o x 6 ° 0? T+~\Q 0. o 0 e 0 0 ° 0 \ I l I G. G ‘I‘~~___.~_ OJ- 34 ,\ 0 0 o 00 . O . 0 ~‘_"—-——"—— ° I o 0 g 6 ‘30 0 ° _.__ O 9 Co 0 0 e ---. o \ . . ,0 ° ’ o I o L e o o 9 0 o e o I o 0 o . 0 l/ O 00 l 0 e 0 I 32° ,9 0 0° 0 G! o O 3 o I o I l O o l e 9 FL 0 o 0 __~ ‘0 O O J. 0 O O . O, G .n\ O O O . 9 T\\ G I 0 O 0 (I! 0 Lg‘v-LAfi/w\ . O! o o o u \\ . e 7 0 O 30 \ \ o o '0 G o g 0 . , 0 . \ 9 cl 2 e o o o 0 o 0 0 \~°\1.L\F“K"W‘D—4-a o ‘0 9 e O o \0 O 0 e o O 0 G o SYMBOL AND PERCENTAGE OF TOIAL SAMPLES \ 0 28° 9 Dec 1 \ . e 200 8 2.12;. 2’ I o I I : 1‘ 0 “:3 a a . O \ a : \ [ \‘sf 0 0 ° O 2 26° 3 | E I \ .1 | \ | \ I \ 1:2: I I o m I I \ 2“ r~~W~222382§§§§§§§§§§§§ L MANGANESE. IN PARTS PER MILLION " fl N m m N "x Geometric mean. 340 Geometric deVIatIon. 2.70 22 a Number of samples and analysesI 861 I l I I I l I l l I l 1 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES FIGURE 16.—Manganese content of surficial materials. 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 75° 74° 72° 70° 68° 66° 64° I I I I I I \ \ ‘I \ \ \ \ ‘I \ W \ /‘ / __._J ‘ , I I I I L___.__'___~ ,. I 1 ., \ / a L— —-—-\ M ‘ I ___.__J / O , /—-'\.H\L / . ’ o O J o C O o 500 MILES " 4" / l I l l l i l l \ l \ \ \ \ \ \ 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76° 74° 48" 46° 44° 40° 38° 36° 34" 32° 30° 28° 26° 24° 22° D41 D42 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 48° 46° 44° 42‘ 40° 38° 36° 34° 32° 30° 28° 26° 24° 22° 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° ‘~ I / I / I I I I I I I I I I I I I \\ ‘\ 0 ‘~ 0 7 r‘“\~~ \ o o I I \‘-—-.- O O o I . “N. 0/0 ) “““'--——-———...__ __ I ox I? I o 00 I 000 o 0 o O \ .~-—._.\\\1 \HO o b « o \ \ o \ II o o o 80 O 0 0 I C o o o o I O . ‘ o O o 0 o I O 0 0 I0 I ’3 O \ O 0 / O I Q. I L o 0 ”~——____ 0 _J o 'r— ? o ‘ r[‘““~‘_o or O \ . . ,o 0 J21“ —~~._<3_, O \' I O O O I, O o O OI o O o I /~\ I o O O o I) . 9°C a OIO o . I ‘\~I\ o O I O o q; 0 I ~~ O 0 I o 0 0 "' \ 7mg 0 o O I~——-»—.—--._o O. I o O O I Vmé’ o 00. 0 DO 0’ ‘6 f O o O O 0 q o O O O 0o I0 0 000 o\‘ ‘9‘ O OI O O o 0 X5 C \‘__._ \ I (I O LUfim O I. ' f 0°07ooor4o o 0' 00 000k 0 o o 0, 0 00 I 00. O I—-—~ _ ___ \ O O I o o of? O 0 CI 0 O 0 o o——o 0—- o o I 00 o 00F 0 2 00 o I \ \ . ° 0 I O a O 0 0 I . \ O I o O o ‘ O o I oo o\ o Lo\\9 f0 0 I O 0 r, OT“~W‘+ 00 no 0 of O O o O O O O O V. 4 ~\“, \\ fol O I 00 9~____~.~~ OI. 0 ‘~———..__..______ \ o O I o oo 0 00 C: 0 o o o o 00 u o o ----—~~-—..1 o I\ O , o O o o ,y o J) o o 3 o I O r) I o r) o I o o l o o o o ’ 00 ‘0 O o j 0 O l \ PO 0' O O O O I O I o O o O [\o 0 JO 0 o o o‘ O ( \ O O O .__‘ o o 'h o o o T\ O I O O 0 CI 0 gd‘végAN-‘rfl" \ o 0’ c O I \ ° C, . o \ O . (I ”x ‘\ SYMBOL AND PERCENTAGE \ O 0 ’ ’ I“ 0 0 O O 0 OF TOTAL SAMPLES \\ o 0 0., .._ I? “fl 0 E O O O o O C 3 as \~J_§1 ' \O 35 “3—40 O 7670 O \0 o o o o O o o o 200 O 0 F) O x \ \ o o o 5 I o o o E \ 1‘- 0 o O 3 \ 3100 \ ,‘f \\o O O O o E ~ \ \\ ° \ MOLYBDENUM, IN PARTS PER MILLION t \ \ Geometric mean and geometric ‘1‘ deviation not computed be- -. cause of insufficient data ‘ Number of samples and analyses, 844 -\ l I l I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76° 74° 72° 70° 68° 66° STATES 64° 96° 94° 1 1 92° 1 I \ \ \ \ \ \ \ 500 MILES 4 l i \ \ \ \ 90° 88° 86° 84° 82° 80" FIGURE 17.—Molybdenum content of surficial materials. \ 78° \ \ 76° \ \ /' 74° 48° 440 42° 40° 38° 36° 34° 32° 30° 28° 26° 24° 22° I)43 D44 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 0 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° 48 x I / / I I I I I I I I I I I I I I \‘ ‘\\‘ o 7 [\\‘\‘~\ 46o\ o o o /I II ‘~\~_\“ o I “ T~~—--—-——— ~— II ‘o I O c O \ o ..\_....,\\\‘I \__ J O I o o \ 44 \ o I o 00 o > 1 o 0 o O o I I) O / (’1 O O O O I C 0 o 0" o I e o (L I 0 “—————.T9_ 42° /> L‘ r “\—~ I— 0 J \ 0 G O 0 9 «‘5‘ Ne-‘u‘K‘J O \‘ II o e I o O I o 9 O I I / O o o l o {JO 0 o O O ‘ \‘ I /\ \J O o I 0 ® 0 I 40°\ \7 I - g"— ~b\2 O O O’— ~--‘__-—O‘O 06 I 0 TI 0 I V‘KKg 6 o I I q 0 o 0699 0 o e o /O o I 0 <1 0 O 0 K3 0 O O oocr C ‘7\_“_ e o 5 38 \ I ““‘ri—~_‘ 0 so 9 W o [000 060 V a 9 ' 9 o o o 9 0 . o OI O O 'I 900 9 L _______ \_ 0 O I O 0 ° 9 9 ol 9 O o g 0 o o 36 ° / O f O C‘ °° 0 o o O o O 9 \ 0 \\e o I I o o o o o I o 00 \ L e f e o 0' I \~Q “\ o O . P o o o o o o o \ ‘1 O O ‘II‘\Q_0 o i 0 34°\ I0 0 o n o 9‘: -~-————-—--——-——- ' . 9 e 0 (2) J O o O --‘-“ o k ,0 0 0 O I L O o o o o O I o o 0 ll . O I m a 0 I 32° 6 0 O 0 fl 0 0 j “ V I \ I O t I 0 o O 'I e 9 o 4‘ G 9 9 .‘£0 0 a . J G o o o ;\ 1. \ O p O 0 O O I G u‘v-LA~/,.a\. \\ O 9 I Q o I . I 0 30°\ \\ G I9 o 90 e 0 e o SYMBOL AND PERCENTAGE \ o OI o O I r, o q 9 OF TOIAL SAMPLES \‘n\o 4""‘xk‘0‘3‘0‘4 __, ,) 9 ‘ :2 p, “ \O o 455,%° ' \° 0 C C F (O o o O " ’3 V \ o I 28°\ ‘1 o o o o E \ O’r""\€ e 0 O u.) n g \‘r \\ O 0 26°\ 5 \ \\ {3'0ch \ :N2282 1 24°.\ NEODVMIUM, IN PARTS PER MILLION \L Geometric mean. 39 ~\ Geometnc deviation. 1.72 Number of samples and analyses. 616 22°\ I l I I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 96“ 94° I 1 96° r—O 94° 92° I 90° l 92° 88° 86° 84° 82° 80° 78° 76° 74° 72° 70° 68" l \ \ \ \ \ \ \ \ \ \ 500 MILES “ 4' 90° 88° 86° 84° 82° 80" 78° 76° FIGURE 18.—Neodymium content of surficial materials. 66° \ 64° \ 74° /48° , 44° / 42° /38n /36° , 34° / 30° ,28" ,26" D45 D46 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 0 128" 126" 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° 48 \ I / / I I I I I I I I I I I I I I \\ 1‘?“ a 7 I‘~~\- 46 \ ’ O I ll ~-\._~ .- 0 e / ”K—h _‘ . .1 7 -——~~——-—--~ .1. I, 9‘0 $- 00 I 0’ o a e .~-~,\\ 4 \Ho . ’30 o G \\ 44°\ 0 ‘ I - o ‘ 9 0 o \ I o o C o ' I I. ) (I ‘ o 9 o . / “ a G e 9 a ’1 ° 0 9 g .'\G 0 c . I, L 0 0 0 r‘——~—_.__..,___I_J 42° . f o '«wff“\“% ‘ 0' \ \ 0 ° I. 9 o “‘N~—:I o 1 o / e o 0 ./° . '9' s eh O /\“ I O . I O . I \~. g o I o a h ' a a 40°\ ~~7__ G I G 0 ° 0 9"“ . gm“ 3\‘ ~--——~___ O ‘ I O ‘l 0 . 6 O ’ e‘v-m ( ‘ 6 _ ( 69] e 9 Q o M 0 use 6 Q 9 e G 0 o O O u [9 ° 0 4 O o 0 G 9,. G ~7\_" 3 O O ”S 38 \ I ’ ‘~-(;-—.L..6h._._‘O 0 ‘. o f 0- 7,0; )9. o g: e G . . .k— r; 0 . 0’ e '0 ’I 999 0 1—1.... __ x 9 o I . a. 0: 0 e d 0 9 9 o e o 9 a o o r e e e 0 I 36°)\ 3 , O l u 0 0 o 0 . ' \ U 0 I 0 I . o 9 a I o 9 e I e o o o 0. K O/Lf¢\6‘:? ‘ f O . 0 O O; ’ O O 0 0 e c \ a ‘1 v 9 Win-.mgh‘e . . e I I. o F~—~__ oi 34 \ \ i. 9 0 ea 9 0 . O “~~-—————w____ O (a \I . . ' 3 . Ge 0 0 e “*h—q o o 0 e 9 o o o I a o o 9 \ ° 0 o o e I o a L 0 0 ° 6 <3 ’1 o o. o l o 0- I 32°\ ,6 r2 '0 6 <3 0 ° 0 ' o I o I O I o 9 4‘9 . 0 0 6 ~ \g 0 0 ° 0 a d O -\ “T O 9 o . . . \\ 01' O o 0 O ‘é‘wo'AN/d‘ 9 I \ e 9 O o BOON. \\ 9 G o '0 o O O 0 o o o o o o SYMBOI AND PERCENTAGE OF 10m SAMPLES \\ g “361 ,__ L3» _“ O o 0 O 0 9 e m «—« EnN m ‘\ N 0 “‘4 (3 1 N N- N ~ \.I o 3001 o I o IGIOI o | “ o c O 0 e o. e c. o I I I | 0 G O 2 a I I | \\ o 8 \ | I | O O 200 I I 1 o O 0 G a I I \\ - \ o o o a l I \ If \\0 G O 0 0 § : ~ \\ 26°\ E 100 I \ I \ l \ | 1 24° ' \ \ 0 L ~~~~m~222232 ~¥ NICKEL IN PARTS PER MILLION GeomeIrIc mean. 14 D Geometrlc dewation‘ 2.26 22 \ Numberofsamplesandanalyses, 862 l l I I I I I I I I I I 118° 116° 1.14” 1.12” 110° 108° 106° 104° 102° 100° 98a 96° D47 ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES ‘ /.48° 54° 78° 76° \ \ 88° 86° 80° \ l \ 92° 3 94° I 96° 0 6 4 / ,, 44° ,42° /.40“ ,,38“ ,—36° ,a34° ,.32° / 30“ ,.28° ,. 26° ,, 24° 500 MILES 0L ,.22° 74° 76“ 78° 80" 82° 84° 86° 88° 90° 92° 94" 96° FIGURE 19.—Nicke1 content of surficial materials. D48 46° 44° 42° 40° 36° 34° 32° 30° 28° 26° 24° 22° STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128" 126” 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° 7 / / I I I I I I I I I I I I I I \~ \\‘ o ~7~~ o I [\“\—- I \~~ 00 I I O O I Gd} 0 0 (3 G I “~-~~’\\‘\I \‘w0 o b 0 0‘ o . I O G G . \ I o 0 G o I l (I o o 0 Q . / \‘ Q G o O Q I O O 0 8‘0 0 O / 0 L o I Q. \ o 0 _“-——-—3—°—-;I O 0 bro—~51.“ “~e gr 0 \ e O l0 . I ~.—“c3" o I I 0 o c I 0 o O I Q . O O I I / . e o I; o 00 0 o o <3 /\\‘ I 0 I 0 o I \N‘ o I e | e o 9 ‘~7 I O o 0 o t— ‘w\2 o 0 0 ‘~—————___ 09 I 0 ~/ 9 0 I 6“” \x‘, e o o o o 9 I O 0060 “I-.o-‘ 0 I o 0 O 5 I -“'-L — 0 § 0 o o 0 O .6 7° 0 \ f O o oo 00 o 9 0 0 e o 9 L x G e o e, 0 00 I, 0.. G I___ \ G O o 0 I . 0 O: 9 0 0| 0 0 o . . 6 o . O I O 9 0 Q 6 I \ . 0 9 <3 0 0 o I o I o I 0 . . e \ . 9 O o I 90 0\ 0 \e o . 0 9' \ 0/ O ‘;W~~ o . o o I5 o 0 9 0 a e a e o o . o G I e o o <3 K e O I L 0 0 ° 0 , . o I o e . e o I o ’ e 9 <3 9 j o I j? G O G o O 9 O l 0 I G C ' 9 0 O (L 9 0 0 o G __ ‘0 ° . 0 o 0 «~\ T o o e 0 e \ l o . O o O fi‘V'LAK/N ‘~ 9 OI O I G SYMBOL AND PERCENTAGE \\ ° <1 0 . 0 or TOTAL SAMPLES \ 0 o 0 g 0 O o o o o o o \ e 0 e I g 0 o o o o o o “\iiqh‘L‘W‘U‘G-d 0 ‘ O o e o o o o \ o o a \ o o O o \ O 0 o z 1 O O 0 z \ h- . g e g x f \ o 0 Q 0 g \‘r \‘ O \ ‘\ 1 ocmoooog \ V H NIOBIUM. IN PARIS PER MILLION \‘-..\ Geometnc mean. 12 GeameInc deviatlon, 1.66 Number of samples and analyses, 813 l l I I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 96° 94° 92° 90° 88“ 86° 84° 82° 80° 78° 76° 74° 72° 70° 68° 66° 64° . l l i l X I \ \ \ \ \ \ \ \ \ \ \/48 , 44° ‘ __ ‘ ,24° 0 500 MILES , 96° 94° 92° 90° 88° 86° 84“ 82° 80° 78° 76° 74" FIGURE 20.—Niobium content of surficial materials. D49 D50 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 9 128" 126“ 124° 12‘2” 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° \ I / / I I I I I I I I I I I I I I \‘ “\~ 9 ~7~~ 6 I [\“\“\ 46°\ 9 O I (I ‘~\‘_ 6 o / fi§'~‘~— -__ .I 0 § 7 -—__-___$_~_ ex I o o ‘9.» no I 9 s g e 0 \ ~ 0 u .~_-..-\~‘\5 \HC‘ 0 e I; o 0‘ 44 \ e I 0 0 so 9 e \ , 0 ~ 0 - ~ I I I , ; g, 0 . 0 o i / . “ o o '9 [I ' 9 o 0". / . I L g 0 o .“--—~.______0_J 42" . / o lawf£~‘\‘~... " 99 \ \ ° ' Ie ° ' I o—““&‘ 9 I I 9 O (o I . o O 9' I o / G o 0 0I6 . Q,“ o GI0 o \‘ ' o I o I \J 0 I e 0 b 0 | m°\ I“: ° ' I o o 0 ‘* ‘?\‘.‘l 9 0 0 Ji—h““——a~e o. I 9 0 e I V \(fl 0 I G O o o 00.0 o o. 0 o o (9 a Q 3 V \S L 7 38° ,. I . "e. -I\‘s‘\- “i e I O ‘5 \ . I 00 3 e 0 .6 o —I’ ’ " f 0. G. or, 9 o o e ’L G g . .I U 99 I, .00 O ——-——__._...._._.____ \. 0 ' I . o. 0.: 0 ° d 0 ° ° 9 o 0 0 36° 0 I .0 . o . .9 o I \ \ e “ I e 'I o . e G o I \ e o . I o. o\ o .L e ' o ‘I \_ c e 0:), a '7 ‘T+“.Q_Q CI, 0 ° 9 0 Go 0 o \ I I I .0 <3- “-—_ CI. 34 \ \ 0 0 ——-——__..____...____ Q I 0 e ..O 0 U G 0 .\ o . 90 0 9 0 o 0 "‘“—-—-n . k 0 0 IO 0 0 ca 0 1 _, L 0 3 a o 6 o o e ' o 9 O 0 ll . O I O O | O Q G! 9 0 q I 32: 2 . . a o 0 e . I e O I o O ' a ' o (L O 0 O 0 O -_. ‘0 e lo a o o ..\ T o o . 0 ° 0‘ \ 1 . 0 » 6MA~fl¥ \ . ol 0 G a D \\ O 7 I H p 30 \ \ e 9 r: ’0 9 0 O a o a a 3 \ Q 0 of 3 9 3 3 Q 0 g I) 0 \~\1.L‘j—-- Lop—73 ’4‘ O K x \o o O 9 4 60 0 o O SVMb‘UI AND PERCENMGI m mm SAMPIIV \ o 0 3 r. 28\ 200 3 35; N \ “a I 3 ° \ A» 0 x) \ . o a L: ff \\: O G G I 260\ g I00 \ L. \\\‘ “MM 31% ‘ ‘ M°\ §§§§§§§§§§s§g233~2NMmN \~ PHOSPHORUS. IN PERCENT "\ Geometric mean. 0.025 Geometric dewatlcn‘ 2.74 22° \_ Numberof samples and analyses 856 I / I I I I I I I I I I 118° 116° 114° 1125 110° 108° 106° 104" '102° 100° 98° 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES, 96° 94° 92° 90° 88° 86° 84° 82" 80° 78° 76° 74° 72° 70° 68° 66" 64° 48° l I l K \ 1 \ \ Y \ \ \ \ \ \ \ \/ 44° , 32° 0 500 MILES 4' 96" 94° 92" 90° 88° 86° 84” 82° 80° 78° 76° 74° FIGURE 21.——Phosphorus content of surficiai materials. D51 D52 48" 46° 44° 42° 40° 38° 36° 34° 32° 30° 28° 26° 24° 22° STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96" I I I I I I I I I I I I I I I I I - \\ 1\‘ o ‘7‘- ‘3 I [\“\~2 I \~~ 96 I \ ° 0+: T“ ——————— r- I o 99 e0 I o g e 0 \ u~-~_~\\\1 \‘fl. 0 o 9 O \ O I O O O o 0 O 9 \ I O Q 0 O F ’ C o e 0 O / \‘ . e 0 9 0 II 0 0 0 a? o , g i L 0 0 0 Fl~~—-—-—-____O_J I ‘~\‘ 0 . G O 0 . waf —‘g__“:’ 00 \l [I e O 6 I, 9 o o 'I l 0 ° 0 o G a 00 e 0' o l\“ t 9 ° 0 IO 0 o I \N‘ O Q 9 O I 6 Q 9 I “7_ I o o 0 0 Y— ‘§\g . a e —~——~_.___ 00 I 0 ‘l o o 0 I ’V \(‘I o 000 o 00! .0 i‘ 0 o 0 o q o e O o \S 0 o e .0 l 6 0°00 ‘ ‘3‘ O I 0 ° 0 5 I ‘ 0 L3 "‘0 9 ‘0 0 0097996 90. o 0’ 0° 9,0L ° . o a, e 0. II .'. . L——__ \ o O I o o - o: o 0 3| . o o o o 9 9 o . o r o O . ' I \ . o o 9 o o I . \ o . I o 1 ¢ 0 0 o . o I ' o o 0 a 0e G\ . g 0 0 IL 0/ ‘ W‘~ . o o g a o o o x O ° 0 ‘G‘+~\1 u o 0 ° 0 0 \ 1 so 0“-“ .1 \ O 0 9 I O . 60 0“ —._______ O O —“—— e o \g o o 0 0 J. 99 9 0 o ‘F._____' o O 9 O '0 O 0 0 e I 0 O O K G I o o 9 . , o O I o o o g , I o ’ 0b e 0 e j e 9 I f ' I o e 0 I o o J‘ ‘ O c ’ 0 9 0 o o .‘ ‘0 ° . Jo ° 0 o 0 o' e - T 9 ° 0 o O 0“ \ l e o 9 (SWAN/v" \ O 0‘ Q 0 I o \\ o . 7 o o O \ g 0. 9 O o o e e o o SYMBOL AND PERCENTAGE or TOTAL SAMPLES \\ . O o! .— E9 G t o o G 0 o o 0 _ qu 0 ~\ “ \G‘td-d o N _~m _. o 0 010m 0 \ o 300 | I \. o o 90 G o o o 0 I \\ O G | I ° ° 200 I 1 O 0 'O o I ‘ "' e o O a ‘ 0 E I \~/( \\‘ G 0 O O o | E | \ 100 I \ | \ I \ I 1 I \L 0 N\ Geometric mean. 1.2 Geometric deviatmn. 2.71 Number of samples and analyses. 861 I I I I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100“ 98" 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES D53 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76° 74° 72° 70° 68° 66° 54° 48° I I l I I I I I \ \ \ \ \ \ \ \ \A 46" / 30° I O O 0 O 00 9,0 . C\ L." i ..‘ “o” —\3———-—-"""L , " O I o I ,28 I I I '~ I o OI» OoO\O 03”}!90 I O I O - n ‘ / 24° 0 500 MILES I. 96° 94° 92° 90° 88" 86° 84“ 82° 80° 78° 76" 74" FIGURE 22.—-Potassium content of surficial materials. D54 48" 46° 44° 42° 38° 36° 34" 32° 30° 28° 26° 24° STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96" I / I / / 7 I I I I I I I I I I I \\ ~0\‘~ O 7 7“\\~ I \~~ 0 e . . I I \‘~‘“-—~—- 0/. l ‘ “*-——-—-——_.. ...._. I 0» Y I “I I I 00 0 o 0 0 \ h~~~¢\__ \110 O 9 a ‘\I\ [I o 00 o 3 I) O 0‘ o o O o e l e o . O / (1‘ O o G 0 O [I G 9 O 9 0‘0 9 G O / o 9_ o I L! O - o or ‘“—"““""T’—';I ~\_ 0 . a O 0 “WAI- ‘~.Q,__~~..~zj 00 \‘ I, 0 o e I 0 e 0 0I I / G 0 .0 0/3 0 Q3! (3 9’0 0 /\\‘\ I O 0 ' O O : . ‘1‘\3‘ 9 O O I, O G O 0 . ‘— ‘-a 2 o #‘“‘——————— 9 $- 0 7 ~: I e o o o . aw a On: ’3 ‘30! a G q 0 H e 00 G O 0 o 0 e o o 0 KS /0 o 9 § 0 o o o I 00 .’o 7‘..9_ ~‘ J- o 5 o 0 _~ 0 \a \ 9 f '00 I000 009 0 GO 9—? 9 <3 0 o 0L x O o O 0, O 0o I 000 O L“~~ \ o\000 I o 00 0:00910900900 0 l 9 o 0 ° 0 0 I <3 0 o o o, e o o 9 e I I 9 O I 6 \36 L9 0 O O G 9' OR\ 0/ \o‘sfir~ O 9 a 6 o I3 9 o 0 9 9 0 O T‘O‘+~\g 0‘ Q 9 g G \ O I 90.0-— o 9 ° 9 I o L o o O \ 9 o 9 e I o 0 o 0 O l I O I G I o 1 o 0 o 3 j 9 e l 'pe a 0' e ‘3 o I O o G I e G o (L 0 0 o o __ \9 ° . 4 e ' e o e d o -\ ’ T O O o o ‘ 0 wt. \ I o g 0 dwa‘x/n ‘\ G o 0I G e I SVMBOL AND PERCENTAGE \ o 7 9 . o o o 0 0 or TOTAL SAMPLES \ 0 e O G 0 0 9 C 0 «Imp—‘V \\9009; .— k. 0 00 0 o o o N NNN _~ -.‘ _ ‘ 300 o mom 0 | \~J_\1 \OO 6 13—4 _.o o | | I I l \o o o 0 00 0 o 0 | l | I I \ a O O o I I | | I \ | " I | o 0 I I I go o > g I \ 1‘- 0 0 Q O m \ O 9 g I \ ,‘f \\ o 0 0 E I ~ \ 100 l \ l k\ I I \I | 1 “gums-22:3 \\ SCANDIUM. IN PARIS PER MILLION —..\‘ Geometric mean 8 Geomelnc deviatIon 1.79 NumberoI samples and analyses 848 I I I I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76“ 74° 72° 70° 68° 66° 64" I I I I I I I I \ I I \ \ \ \ \ \ / / ._._.nn‘ " N "‘ kn / o o G o J, r k / ‘ I x I I L......_---—-—---"'." . / a \I“ O a O O ? $0 / 0 K— _____'_ \ -;~-°? o 0 e 0 K / i o I o G 9' 9 -__..J / I —o— o o 0" e 0 I o 0 Ir ° / fly." I \Lo 0 <3 0 l ..__. I o I—‘fl— L I / 0 ¢ I a o o I0 O 00 O I) O O o ' 0 o o \ O 9 I I 0 0 O 1 e00 I “ado—4° __._.— L.<3u——-s ‘ ( A Q....——- I“ \ I ( r \ z 0 ° 0 L If 0 0 m . * o 0 o a e - x: r I, 0 ° 9 0 '. k : 'fi 0 ‘ O I " . O O \\ / . I. o o \ v V O / , ‘ 0 Jo (j c 500 MILES ” f / I I l l .0.” I I I I I I \ \ \ \ \ 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76° 74° FIGURE 23.—Scandium content of surficial materials. 44° 42° 40° 38° 36° 34° 32° 30° 28° 26° D55 D56 48° 46° 44° 42° 40° 38° 36° 34° 32° 30° 28° 26° 24° 22° STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° I / ’ I I I I I I I I I I I I I I I \‘ \\‘ o , ‘7‘- . I I‘“\T‘\ . 0 O I I, ~~\“\._ o o / ”N‘— -_‘ . .1 7 ————-——— ~— [I O\ I I a so I 0" o e 0 u~—-——\~ \1 M10 0 O ‘ O 0 0 \\ 0 o ‘ I o o ,7 .0 ° 9 O \ I (3 . ~ 0 G I l (I - O o O 0 ° / \‘ o 9 o ' 0 I 9 9 o 9" 0 . ’ . 0 Q. 0 (f L ° ~1 ° .r "‘““"‘r°~-I O .4 r \‘~\.g, 0 9 \ O 9 0 9 3"“ “\——I o ’ o O o 9! l I O G 0 I 0 G I I . 0 ¢ 0/9 0 Q90 0 OI<3 o I\\\ O a I l 0 O I “‘\~°‘ 0 ° 0 1 oo 0 ° 0 OI: '— ‘u\g . 0 ““——~——.—._ 9 3|— 0 7 o ‘I o e . I 9‘0“ ( ° 9 o I o q 0 \fi 0 099° 0 '0 f G a o o 0 o X o o o lo a 9°.o\‘7\_9.. 0‘! o o 0 x I O \“\L ‘ § 0 e U -o ‘3 0 0 f 000 7OQQ “a. 9 0! o G o o ok— ° 00 o 0, o 0. I 0’0 0 I_______________\. o. I ... core 0.:6 00° 0: o 0 <3 co 0 o 0 \ . . I l (3 O O o, . . I O I o \. I e 0 .0 o I um . L,\- o f ° . ”I, o "‘ “ o o . o o o \\ 0 \/I 9 G? W+‘\E__°. 0 '01 O G 0 I9 0 o I a :‘fi“ 0 —~......_.________ ° 1 o e 0 e ‘3 o . so 0 w o --———. O k G 6 o I O 9 G (3 I 0 l O 0 O O . I, o 0 I o o o ' o o g ’ ° 0 o o o o ffi O .9 GI O 0 6 j 0 ' o I O o . ' 0 G o CIR O O O G O -‘ ‘6 ° . J0 o e e (I 3‘ T O 0 O o I \3’1. \ I g a g c g‘v-LMx/fin \ o 'I e o 1 0 \‘~ 0 o a, o o \ o .0 0 O G 0 0 O o 0 \ o 0 GI ° 0 o o ( 0 c o 0 ‘~\!. "‘\k° 1‘n—3—4 o ‘ ‘0 O O O 0 \0 O a 0 0 (~ 7 \ o o e o \ o o 1 <3 0 a \ A- 0 o o o \ E \ ,f \{I o o o 9 § \ L \ ‘\ 1 \ NK Geometric mean. 0.40 Geometric deviation. 4.11 Number of sampIes and analyses. 788 I I I I I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° -r__- 95° | ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 94° I 92° ! 90° I 88" 86° 84° 82° 80° 78° 76° 74° ‘72" 70" V 68" 1 \ \ \ \ \ \ \ \ \ \ 500 MILES " If 90° 88° 86° 84° 82° 80° 78° 76° FIGURE 24.—Sodium content of surficial materials. 66° \ 64° 74" 46° 440 42° 40° 38° 36° 34° 32° 30° 28° 26° 24° 22" D57 D58 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY a 128° 126“ 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96" 48 ~ / / I I I I I I I I I I I I I I I 46°\ 44°\ 42°\ 40°\ 38° 36°\ 34°\ 32°\ 30° 28°\ 26°\ , Q 2 ‘“ 100 2 o w n: .. 24°\ Geometric mean, 120 Geometric deviation. 3139 Number 01 samples and analyses. 862 22 ° \ / / l l l I I l I l I 1 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96" ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 703’ FIGURE 25.——Strontium content of surficial materials. 96" 94° 92° 90° 88° 86° 84° 82° 80° 78° 76" 74° 72° 68° 66° 64° 1 I l I \ \ \ \ \ \ \ \ \ \ \ \ \ / / ._.._._I ‘ , l /‘ I I L.._.._.,._-_ | I‘ i - s /‘ o k— __ ———-x / -_~_l / 0 / /““\L / G D V ‘ n O / l i O O J o ‘ o O o 500 MILES ‘ 1' / I_ J I I °-' I i l l \ l \ \ \ X \ \ 96° 94° 92° 90° 88° 86° 84n 82° 80° 78° 76° 74° 46° 44" 42° 40° 38" 36" 34° 32° 30° 28° 26° 22" D59 D60 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° 48 \ / I / I I I I I I I I I I I I I I \‘ \\\ O \ o 7 I‘“\~2 a I \~- 46 \ g o .. I l \“ ~‘. . / \ \_- __‘__ 'I ° 3 ‘1 “"1”“ I . I 3° 9 o o _~_~,\‘ \H“ o o (5 e o o \\ n o o 44 \ . [I o a a '0 I o ) (I o 0 o o 0 0L 0 . / 1 o 0 o “e '1 ° 0 o oI\o o O (I 0 L 9 O G I‘l“—_h_”‘T.—J 0 ~.. 42°\ . o . e o 0 4rd; \—‘~e__ _§J 0° \‘ o I, ° ° 9 ' I. ° ° 0 6' I / e . o o of, o ‘90 0 0’0 0 I \ o ]\ \\J I I 9 . ’ ~\e 0 o 9 o e O 40o\ ~‘ 2 I e o J-_ o g..— 7 0\2.7I 0 0° 0 “---——9,V_\ 09 I O O l \(l, e e G coo: 6° 0’ 90 (i o o o o . q o O o O C \ a IO 0 00 o o\‘ \3 ’4 o O 0 38° I o ““‘“~~—.L s N O o 0 B‘qo (3 ‘0 ° 000 000 '0. o ‘ o o 9 o L \ o 0 I o. o L G 0% 00 O 0, o o I o . -——_ __ __ .. __ _ \ o 0 o 361 0' ° / o o o . 0 e \\ e I o I O 6 . o o 3 '1 o o ck G ' ° ’1 O/LO‘\O‘£‘O“~ f0 9 0 O 0" ° 6 o 0 ° ° \ 0‘, O O ‘0‘+;\1‘0 0 ° o o F“‘ 34‘; \ I. 0 O ' .0 'o e I.“ O —.___-.____-__ 0 G 1 0 C o 0 O a 6 S o 0 9 o —.____‘ a . G o , ° 0 \ o .0 0 g I e o o a 9 o 2’ ° 0 I O a G ‘0 O ’ I e O o .0 3 o 32"\ [co (’0 a e e 9 . ' G a I o I g o o I o ‘ 0 g 0 o o ___ \n . o o 6 «q T '3 ° . o 6 O \\ G: e 0 Z 0 9 ' O $‘ngN/fin \ 9 a 30°\ SYMBOL AND PERCENTAGE \\ 0 70 . C, 0 0 O o o o o e . or Tom SAMPLES \ o 6, 3V 0 O o c, o o o o 5.22 a \ “\11 ~~m~m-~ o O \ ( O 0 0 O o I \° 0° 0 o 28 \ i \ . 9 o > I 1 O o o o ‘2’ | ‘\ A” o 0 g I ff \\0 O 9 6 0 ‘9‘: ~ \ 26o_‘ .1 \ x ‘\ 1 24°\ \L TITANIUM, IN PERCENT N‘- Geometnc mean, 025 Geometric deviatlon, 1.87 22 ° _\ Number of samples and analyses. 861 l l I I I I I I I I 1 l 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES D61 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76° 74° 72° 70° 68° 66° 64° 48° l I l I 1 \ \ \ \ \ \ \ \ \ \ \ \/ 46° /44° , 40° /38“ ,32" ”28° 0 500 MILES l 96" 94" 92° 90° 88° 86° 84" 82° 80° 78° 76“ 74" FIGURE 26.—-Titanium content of surficial materials. D62 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 0 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° 48 \I I I I I I I I I I I I I I I I - I 46°\ \ l 0 o o. I 0 a e o v~-~~_‘\1 \‘o . e (50 G e 1‘ 44°\ 0 I e e o e O O 0 j ’ o o 9 ° 6 I o o o )/ (A‘ . e a so 'I e O 0 ° .l\g O / O o o l-.. _ - L o -1 ~ or ~—-~--r°—I 42°\ 9 Drag \‘“~6-- 0 ° \ ° . I. ° ° I 0 T“?! ° , 9 e o I 0 e o I ..“ O . 0 O 0 0 0 ~v~ 0 C‘ O. 0 O ‘0 0 . 0 40°\ 38° 36°\ 34° 32°\ 3001 28“\ FREQUENCY 26°\ 24° VANADIUM. IN PARTS PER MILLION Geometric mean. 56 Geometnc deviatlon. 216 Number of samples and analyses. 863 / I I l I I I I I I I I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98“ 96" 22°\ ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76° 74° 72° 70° 68° 66" 64° 48° l I i l l I \ \ \ \ \ \ \ \ \ \ \/ ,44" ,42° , 40° /38° / 36° /34° /32° ,.28° ,26" N k , 24° 0 500 MILES I, /22° 96° 94° 92" 9C“ 88° 86° 84° 82° 80° 78° 76° 74° FIGURE 27.—Vanadium content of surfici‘al materials. D63 D64 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° 48K I I / I I I I I I I I I I I I I I \‘ \\‘ O ‘ e \I 7‘“\ o I I ‘~\~- 46 \ a o I I \_~\_ 0 e / ‘~‘- -_~ 9 e 1 7 ‘_-‘-—" —" II “a I 09 0 ”2....“ J \19 000 l o e e o I , ~2 0 é o O\ o 44 \ o I e e O o e o > I o o 9 o I o ‘9 0 O / C‘ e G o 0 e [I 9 O O O I\0 o , o g L 0 ° «3 rm--________9_4 42°\ . G O G DUMFI ‘ “—9.__ ‘09.! O \ o I e e e O I e e G " 0' l e l / e e 0;, e G.“ ' OIe e I [\\\ I O . I G 9 I \~‘ 9 0 I o e a 40°\ ‘~ : O I o o 0 I__~ o I?" 7 WTQ‘I o 00 e 0' _-‘-_e\v_.. ( 0 I 09 g 6 GI . 0 q 0 \° 0 00 e o O 0 o o 0 o o x 90 lo ‘3 o o\‘ ‘3 e 6I G 0 0 a e e 7 ~_~__ 5 38 \ I ‘ \J.__ ‘. C e O "‘ 9 e f °°e logo 000 o ‘3' 0 0 o o 'L X o e a 0, 0 r0 ; °°, ° L __ _ _____ ‘. g \g e o I o 0 o 0 CI 0 O 0 . . o o o o o I 0 0 00 r e 0 ° 9 I 36" o e 0 a G \ \ O o I o I e o o 0 o I 09 2A 0 o g0 f 0 ° 0 o I O o ‘£T~- ° 0 o I' 0 o o o o e o o a, ° 0 77+~\1_o o , \ I F-— __ 01 34c \ O O . -—-——-———____._ \ G O o 0 O ___— e ‘ o 0 e e e 0 o G o "——-1 e . ° 0 I O 0 0 ° I e e o 0 O O G I O o 9 ° 2’ ' ° I 0 ° 0 0 ° ° I O 90 o 9 o J ° I 32°\ ,1“ 9 9' o o 0 0 I o I <3 I o g 4‘ o . o . o . o a o .. -~ ‘0 . J 0 o o 0 I \ T G o 0 e j 0 W. \ ' 0 o SVVVMF" \ 0 01 0 g 1 a SYMBOL AND PERCENTAGE \ 0 e 0 30°\ or 1mm SAMPLES ‘\ 9 7. . 00 0 O o o c o o o \\ 9 0 9" F— k‘d‘ e O O O G O 9 O ‘L‘J \‘3 o 9 \o o O 0 o o o 0 \ o o o 0 28°\ ‘ o o > 1 G C o o u \ "- o O E \ o g \ ff \\ G 0 0 0 E ~ \ 26°\ \ \\\ 1 24°\ m. \L 3~~~~m~22223 ~ VTTERBIUM. m PARTS PER MILLION “ Geometric mean, 3.0 Geometric deviation, 1.81 22°\ Numberoisamplesandanalyses. 795 l l I I I I I I I I l I 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 96° 94° 92° 90° 88° 86° 84° 82° 80° 78° 76° 74° 72° 70° 68° 66° 64° 48° 1 l i I i T \ \ \ \ \ \ \ \ \ \ \/ / 46° o 500 MILES " , /22° 96° 94° ‘ 92° 90° 88° 86° 84° 82° 80° 78° 76° 74° FIGURE 28.——Ytterbium content of surficial materials. D65 D66 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 0 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° 48 \ I I / I I I I I I I I ‘ I I I I I I \" \\‘ o ‘ -- a 7 F\‘\~- . 46° I I \~-.\ 00 l I "“-~__- 0 -_ “‘—_ 0/ o 1 7 —————$__... Ox I 0 I 90 00 a g o 0 \ ..~-.-...\\\I \Ho 0 b o 0‘ 44° 9 I o g 30 O o \ ’ o o e o I a I (I 0 o I e 0 l \ 0 o G / 0 ‘ o 6 g o I . o o\ 3 / L o o e 9——._____ , J 42° 0 o LIUMI- “~K"~‘e, oer o \ o o I. o ‘-~_G.J e l O O 0 I O O 0 I I o I o o ' 9I” ° 0'. a °I° o I \ I . I O . I \N Q 0 I 0 . 40“ ~‘ ’ ° I e o o t“.- 7“:\2~I G 6 o J‘“““‘*--—~—-——,° o I . e o I Vflg O o O O I . 0 q 0 o 0 O0 00 e G e G G G I O O O 0 KS 0 o l 0 .030 ‘ \4.‘~_ a o O 0 x 38 I O TLC—‘2. . I. \ 0 f 0097000 0,0 9 o I o O . . 0L. \ a” 0,000 I’ .00 o L ““ - -- - ~~ —\- ~ o o e 0 I .0 g 0 OI o 0 O o o 0 0 - / ° ° ° I ° ‘ '° ° ' 36° 9 e g . o 0 \\o . I o I a a e e o I o 0 o 00 9‘ 9 o f o o 0' \ o/Le\o‘3T~— . ° 9 OF ° 9 0 ° ° 0 o co .1 ' o 771%.. 1.0 o o 0 \ i J G I D. G 3“——-‘__ O 34 0 0 o a 000 ° 0 ° 0 0 ‘----—~--———-————- I o . ‘0 1 Go . o G o "‘-—--u . O 0 e G O I G O o 0 \ o ’ o o O 0 ’ .0 l 9 O o O o ’1 . one I j o ' 0 I 32° ,5 0 °0 0! o 9 ° 0 o , . a I o 9 I O o g 1‘ O o o o 0 fi‘T'W e 0 ° 0 e 0 ° d :\~v1. \ O ' O O 0 O 4 . .‘Wfiy‘r’\ \ 0 9' o \ 0 7 o 0 30° \\ e o '0 9 0 o o o o o o \ e 0 a! 3. 9 o e o 9 o 0 0 SYMBOL AND PERCENTAGE \ ._~ ii "-Nk m— Uq_ _, 0 OF TOTAL SAMPLES \0 o o o o o 2 :2: 2 ° 0 0 9 o 000 o \ O O G 0 a W363 \\ O 28 200 I I j | ‘ o o o I ' I I e o \ h... o . o > | I \ ° E | | \ {I \\0 O O b e 260 3100 | I ‘ ‘\ E | \ | \ | I \ oéomcoooée I 24° v__Nm..-.Ngfg \ VTTRIUM, IN PARIS PER MILLION \\ Geometric mean, 24 N‘ Geometnc deviation, 1.7 Number of Samples and analyses. 863 22° 1 I I I l I I I I I I I 118° 116° 114° I12° 110° 108° 106° 104° 102° 100° = 98° 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 96" 94° 92° 90° 88° 86° 84° 82° 80° 78° 76° 74° 72° 70° 68° 66° 64° 48“ l | l i l \ \ \ \ \ \ \ \ \ \ \ \/ 46° /. 40° ,. 32° 500 MILES o l 96" 94° 92° 90° 88° 86° 84° 82° 80° 78° 76° 74° FIGURE 29.—Yttrium content of surficial materials. D67 D68 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 0 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° 48 \ I / I I I I I I I I I I I I I I I \‘ ‘g\“ o 7°~I\“\ a I I ‘~\‘ 46 .\ o 0 e I I ‘\_‘ O o / ) “\-_ __g o . ‘-————__ .__- I, O\O 7 Y o I 9" o o 0 o O \ a u~-~.,\\\1 \L‘o . o I: e O\ o 44 \ e I o o o O \J l 0 9 O o . o I 0 0 ° / C} c 0 o ‘ 0 I 0 . 0 ° .I\o 9 , o z I 1‘ 0 ° ' FL‘~———-—_TLJ 42° } O Li‘m!‘£“‘“~& a . \ \ o O I° o “\-J o I I o g . I o . o ‘I ° I / 0 . . °/. . so . el. . o . - I I‘\\\\ I I 0 O | ~.. 0 0 . I o o o 40°\ ~‘ 9 I o e 3 J_ t“ s O Q —“"-‘————_—_ 7. 1\“/ O O 0 g §V_\( . I ° ' M o 0.0. 00’ 0 o o 9 O 0 e o o o o 0 KS 0 l0 0 \~ 9 0 § 0 O O o .0. O 7\_’§ 9 5 38K I 2 "“~~L—.. I. . 0 O O G (3 I) . 'fi'. 0 f o o o. .0 <3 G e o o L 0 a o “I 0 90 I .90 0 L—‘——- \ ° ° ' 0 ° 0| o 0 0 —o_-o—-3——' a a I] o 0 °° r ° ‘.'° o I ' 36 .\ e . 0 e .I o . o o I \ . I o . o . o I o o I~°~o w ~ 'I. o ‘ ‘§ 0 o o . '? W+~ g o. . O O 0 ° \ ‘1 I 00 O_h~" 01 34°\ . G o 00 3 O 0-.“ O -—‘_——‘——————— 0 o I o c o o (In ‘0 o o ;;____I o o o 9 I 0 ’ o o o I 9 o O 0 ‘K I 9 o O . , . 0 l 0 O 0 I [I 0 I o / 0 o G o J 0 ° I 32" f 0 ° 0' o ’i o \ 9 I l e O I . 0 I G J ' 0 ,3 I o s e ”‘f. 0 0 ° I 0 o G G d 3 ”six 0 \\ OI : a O 4 d‘vgn‘qu \ o . 0 30¢ \ 0 79 , 0 0 O \ SYMBOL AND PERCENTAGE OF TOTAL SAMPLES \ o o O (I 0 a 9 c O 2::; “N” \ 0 ° '1 3 o 9 c: 3 ’ ° 9 O 300 oIoIem o I \~\1_L\I““~K”O"’a 'imfi-.. O G I I \o o 0 o .0 0 o O I \ o 3 O a 28°\ I \ g 0 200 | o z; I I a ” G 0 E I \ "~ 0 0 2 \ f “g a o E I \\r \\ o 0 ’ 26°\ 100 I \ I “ I \ ‘l 0. \ 24°\ 33533223222~N L V zmc. IN PERCENT " Geometric mean 0.0044 Geometric deVIaIIon. 1.86 22° \ Number of samples and analyses. 863 I I l I I I I I I I I l 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° 96" ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 94° 92° 88° 86° 84° 82° 80° 78° 76" 74" 72° 70° 68° X \ \ \ \ \ \ \ \ \ 500 MILES " 4 J l \ \ \ \ \ \ 90° 88° 86° 84° 82° 80" 78° 76" FIGURE 30.—Zinc content of surficial materials. 66" 64" \ \ 74° /48° 46° /44° D69 - D70 48" 46° 44° 38° 36° 34° 32° 30° 28° 26° 24° 22° STATISTICAL STUDIES IN FIELD GEOCHEMISTRY 128° 126° 124° 122° 120° 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98" 96° ~ I / / I I I I I I I I I I I I I I \\ ~\‘ 0 ‘7“ I“~\~ I, I ~\~‘\ \ 0 O o 0 / I ~- -- ~_ 0 o ) “"'~-———---.. __ I, ox Y ° 00 o I o g e o ‘ \. __. o o I ~ o-Nx‘J \‘7 9 Ge 6 9 0‘0 \ e \ 1 o o 0 <3 G I I0 ) C o 0 O 0 / \‘ o o g 00 I 9 o o O Glo o O ’ o L o l ‘1— e \ (3 O l O J.“ ‘~ OI— ~~—_'U—6"J \ e o e e “If“ a “en‘x: 0 \, II 6 o (3 I, G g 0 CI 0 I / 0 e o 0 ”I3 9 0,0 0 °Io o I \ o /\ ‘\J\ 0 o I o G (‘0 O o I \ \“ e O I 0 O 0 {~— ‘7;\2 0 ,I“"~-———.—___ ° 90 7 o o O o O , ‘uv—x \(5 o o 0 009000 0’ 09% O G 0 G 9 q 0 O 0 o 0 K5 6 0 IO 0 0000 ‘ \J‘ 0* O 9 O ‘5 I O \‘—\L \ O 5—— . O \G o f 00.7000 0.0 9 O-I , e 0 0 Ok- 0 o O o o I so 0 x o I o l 9 —— — .. __ __ _._ __ 90 II a 00/90 06:.°°°:’°'oooo .0 o e ‘3 \ O G O G . G 0 ° \\ 0’ 0 I O I o O 0 09 O I m . Les: 0 I . ° ° °I3 o “ 2- o o o o o e \ o / o 0?- W1»- 00 o o o o o \ \j I g. :‘h—___ OJ. \ O 0 o . o 9‘- “.‘——————.—..__.__ 0 l o 00 o o . 0 3 . o 0 00 9 O O —‘——' o \ ' ,o 0 ° 0 I o L o o o 0 0 l’ 0 0 l 9 . 9 lO 6 I O G o .6 j e \ ”a 00 Q 0 Go I 0 I o o Ix 0 O '9 O o o 9 Q 9 1‘: ’ . 0 ~_ 0 Q T e a O I 0 o 9 j o 0 A ‘ \ e 1“} “r” x G O! g 0 I \\ 9 o C, e o \ \ o a O 0 e o g 9 , . \\ o Gaol F‘ ‘ e O G 0 0 e V ’ ° ‘~\..J_ ‘Ldfia‘I—fi—a o . o O a O u 0 \ . 06 u o \ o 9 \ \ o o o I o O o > \ he. . . O \ o E \\ ff \\ O 0 G 0 o ‘ ‘ 2 g \ x ‘\ 1 \ ~\ Geometnc mean. 200 \ Geometric devnahon‘ 1.90 Number of samples and analyses. 863 I l I I I I I I I I I 1 118° 116° 114° 112° 110° 108° 106° 104° 102° 100° 98° 96° ELEMENTAL COMPOSITION OF SURFICIAL MATERIALS, CONTERMINOUS UNITED STATES 96° 94° 92° 90° 88° 86° 84° 82° 80" 78" 76" 74° 72" 70° 68° 66° 64° 4 a I I l l i 1 \ \ \ \ \ \ . \ \ \ \ x / 8 ‘46‘ 44" ,42° V- ‘ / 240 0 500 MILES g 96° 94" 92" 90° 88° 86° 84" 82° 80° 78° 76G 74‘ FIGURE 31,—Zirconium content of surficial materials. 9 11.5. GOVERNMENT PRINTING OFFICE: 1971 O - 410-960 D71 Gila-y ‘5': A , 7&3? o, '5 5- '97" 35w ' 3? M” [75’ 74 57?”? ‘ Airborné Chémical Elements in Spanish Moss GEOLOGICAL SURVEY PROFESSIONAL PAPER 574—E Airborne Chemical Elements in Spanish Moss By HANSFORD T. SHACKLETTE and JON J. CONNOR STATISTICAL STUDIES IN FIELD GEOCHEMISTRY GEOLOGICAL SURVEY PROFESSIONAL PAPER 574—E A study of local and regional variation in airborne materials as indicated by chemical analysis of an epiphyte UNITED STATES GOVERNMENT PRINTING OFFICE, “IASHINGTON : 1973 UNITED STATES DEPARTMENT OF THE INTERIOR ROGERS C. B. MORTON, Secretary GEOLOGICAL SURVEY V. E. McKelvey, Director Library of Congress catalog-card No. 73600178 \ For sale by the Superintendent of Documents, US. Government Printing Office Washington DC 20402 — Price $1.25 (paper cover) Stock Number 2401-02456 CONTENTS Page Page Abstract ............................................................................... E1 Methods of study — Continued Introduction ....................... 1 Data presentation E6 Responses of plants to airborne chemical elements ........... 2 Elemental composition of Spanish moss ............................... 9 Morphology and ecology of Spanish moss ............................ 3 Discussion of results 42 Methods of study... ........... .. 4 Summary and conclusions 44 Sampling plan ....... .. .. ...... 4 References cited 45 Analytical methods ....... 6 ILLUSTRATIONS Page FIGURE 1. Drawings of growth habit and morphological features of Spanish moss (Tillamdsia usneoides L.) .............. E5 2. Photograph showing growth of Spanish moss on the tops of pond cypress (Taxodium ascendens Brongn.) and on the more shaded lower branches ............................................................................................................. 6 3. Map showing Spanish moss sample localities ...................... 7 4. Map showing localities of Spanish moss samples containing elements not commonly detected and the con- centrations of the elements ............................................................................................................. 10 5—35. Maps showing ash yield and element content of Spanish moss samples: 5. Ash ............................................................................................................. 11 6. Aluminum ..................................... ........ 12 7. Arsenic ....................................................... 13 8. Barium .............. ..... 14 9. Boron ....................................................................................................................... 15 10. Cadmium. ..................... .. 16 11. Calcium ................................................................ 17 12. Chromium.. . _ .................... 18 13. Cobalt .............................................................. ‘ ..... 1 9 14. Copper .................................................................................. 20 15. Gallium.... ........................ 21 16. Iron ............................................................ 22 17. Lanthanum ......................................... ..... 23 18. Lead ............ ........ 24 19. Lithium ...................................................... 25 20. Magnesium ................................................. 26 21. Manganese... ..................... 27 22. Molybdenum ........ 28 23. Nickel ................................................ 29 24. Phosphorus... ..................... 30 25. Potassium... ................. 31 26. Scandium.... 32 27. Silver ..................................................................................................................................................................... 33 28. Sodium ......................................................................................................... 34 29. Strontium... .. 35 30. Titanium ................................................................................................................................................................ 36 31. Vanadium .......................................................................................... 37 32. Ytterbium... .. _ _. . 38 33. Yttrium ............................................................................................................ 39 34. Zinc ............................................................................................................. 40 35. Zirconium .............................................................................................................................................................. 41 III IV TABLE CONTENTS TABLES Page 1. Spanish moss sample localities and sampling dates ........................................................ E7 2. Analytical limits of detection .......................................................................................................................................... 9 3. Chemical composition of Spanish moss samples and samples of soil-rooted plants from the conterminous United States _. ...................................................................................................... 43 44 4. Elements found in concentrations greater than average in Spanish moss samples from four kinds of sites STATISTICAL STUDIES IN FIELD GEOCHEMISTRY AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS By HANSFORD T. SHACKLETTE and JON J. CONNOR ABSTRACT Spanish moss (Tillandsia usneoides L.), collected from its geographic range in Southern United States, was analyzed for 38 chemical elements in 123 samples. Although Spanish moss is an epiphyte and must obtain all of its element load from the atmosphere, most elements in samples of this plant occur in concentrations similar to those of ordinary soil-rooted plants. Analyses of Spanish moss samples collected at rural, residential, highway, and industrial locations reflected signifi- cant differences in concentrations of metals. Samples from industrial and highway locations are characterized as con- taining greater-than-average amounts of arsenic, cadmium, chromium, cobalt, copper, lead, nickel, and vanadium. The high levels of lead found in some samples from highway loca- tions are especially noteworthy. Many samples from sites near the seashore contained greater—than-average amounts of so- dium that is thought to have been derived from ocean spray. Samples from rural locations commonly contain low concen- trations of the metals usually associated with industrial or urban activity but may contain large amounts of the elements that are ordinary constituents of soil dust. Four of six sam- ples containing detectable amounts of tin were collected within 50 miles of the only tin smelter in the United States; this result suggests that elemental analyses of Spanish moss samples can provide an economical and rapid method of esti- mating the kind and relative degree of local atmospheric metal pollution. INTRODUCTION The importance of airborne materials in affecting the quality of the environment for organisms has prompted many investigators to search for materials and methods most suitable for measuring the con- centrations and distribution of these materials. Com- monly, the methods consist of various procedures for collecting the airborne materials by filtration and then chemically analyzing the materials, or for continuous instrumental monitoring of gaseous elements or com- pounds in the air. Recently, however, the degree and extent of airborne contamination have been evalu- ated by means of observation or chemical analysis of plants that either respond in some visible way to the airborne materials or have the ability to col- lect airborne materials by their physical structures, physiological functions, or both. The use of suitable plants for measuring con- centrations of airborne materials provides the ad- vantages of (1) an integration of the periodic fluctuations in amounts of these materials that occur over relatively long periods of time, and (2) econ- omy in sampling. A disadvantage of using plants, as opposed to filtration or direct monitoring, for this purpose is that analyses of the plants can give esti- mates only of the relative amounts of airborne ma- terials at different locations, rather than the amounts in a known volume of air. This study reports the use of Spanish moss (Til- landsz'a usneoides L.), an epiphyte (commonly called an “air plant”) that is common in the Atlantic and Gulf Coastal Plains, for evaluating local and regional variation in airborne materials. Specimens were collected from industrial areas and near major highways, where man—related contaminants were ex- pected to be abundant, as well as from rural areas, where more natural atmospheric burdens were be- lieved to occur. We thank Messrs. Todd Church, J. A. Erdman, J. R. Keith, R. R. Tidball, and J. D. Vine, of the US. Geological Survey, and Messrs. D. B. Ames, W. G. Haag, and Richard Harter, and Mrs. M. A. Erdman for assistance in collecting samples. County Agricultural Extension Agents in each of the States also submitted samples; their cooperation is grate- fully acknowledged. Mrs. J. G. Boerngen assisted in computer processing of data, and Mrs. J. M. Bowles prepared the drawings from living plants. All statis- tical analyses were performed on the US. Geological Survey’s IBM 360/65 computer and were based on computer programs in the Survey’s STATPAC sys- tem (Sower and others, 1971). Analyses were made by the following Geological Survey chemists: Leon A. Bradley, Thelma F. Harms, Harriet G. Neiman, Clara S. E. Papp, and James H. Turner. E1 E2 RESPONSES 0F PLANTS TO AIRBORNE CHEMICAL ELEMENTS The effects of excessive concentrations of certain airborne materials on many types of organisms may be either deleterious or beneficial. The responses of plants to airborne toxic substances have been re- ported many times as evidence of aerial pollution, and the effects of this pollution were reviewed by the European Congress on the Influence of Air Pol- lution on Plants and Animals (1969). Some plants, particularly lichens and mosses, are very sensitive to sulfur dioxide (and perhaps other gaseous compounds) in the air. More than a century ago Nylander (1866, p. 365) reported [translated], “lichens provide, through their behavior, a measure of the healthfulness of air, and are (one may say) very sensitive ‘hygiometers.’ ” The sensitivity of both mosses and lichens to air pollution was dis- cussed by LeBlanc (1969). The extent and physio- logical condition of lichen populations were used to map the degree of pollution in urban and industrial areas by Sloover and LeBlanc (1968). Many plant species are sensitive to concentrations of airborne fluorine. Symptoms of fluorine poison- ing in plants were discussed by Brewer (1966, p. 180—196), and the “normal” fluorine contents of many species were given by Garber (1968, p. 42—48) for use in evaluating the extent of fluorine pollution from industrial emissions. Although additions of extraneous materials to the air generally are considered to be undesirable, in- creased atmospheric concentrations of certain ele— ments may be beneficial to plants. Ingham (1950) discussed the importance of the mineral content of air and rain to agriculture. Egnér (1965) reviewed the importance to soil fertility of sulfur compounds that are supplied by atmospheric precipitation, and Riehm (1965) described the role of this source in supplying the biologically essential elements calcium, chlorine, boron, potassium, magnesium, nitrogen, and sodium to soils in Europe. The absorption of atmospheric ammonia by soils in New Jersey was measured by .Malo and Purvis (1964). They found that 8.2 pounds of ammoniacal nitrogen per acre was deposited from precipitation in a year, and they believed that nitrogen from this source was a factor in the high yields of maize that was grown without the addition of nitrogen fertiliz- ers from 1958 to 1960. The direct absorption of atmospheric ammonia by plant leaves was demon— strated by Hutchinson, Millington, and Peters (1972), who stated, We believe that our data have broad implications in regard to both plant nutrition and air pollution and water pollution STATISTICAL STUDIES IN FIELD GEOCHEMISTRY control. Calculations [based on their data] indicate that the annual NH3 absorption by plant canopies could be about 20 kg per hectare. This rate of NH3 supply is large enough to contribute significantly to the nitrogen budget of a grow- ing plant community and could exert a prodigious influence on the long-term behavior of an ecosystem. Shacklette and Cuthbert (1967, p. 42) calculated that the total iodine content of certain soil-rooted plants could be obtained from the exchanges of at- mospheric gases that accompany the process of photosynthesis. The total amount of atmospheric iodine added to the soil each year in England was estimated to slightly exceed the amount of iodine removed from the soil in a crop of kale (Chilean Iodine Educ. Bur., 1956). The “normal” concentration of carbon dioxide in air is generally considered to be about 300 ppm (parts per million), and amounts much greater than this may be considered undesirable for animals. Yet experiments have shown that above-normal concen- trations of this gas in the atmosphere increase the growth rate of plants. Holley (1965) found that in- creasing the carbon dioxide content of the air in greenhouses to as much as 1,000 ppm was economi- cally feasible, and at present the use of carbon dioxide generators is a common practice in the pro- duction of some greenhouse crops. Many other reports have pointed out the impor- tance to land plants of nutrient elements that are obtained directly or indirectly from the atmosphere. However, reports apparently lack conclusive experi- ments on the relative amounts of certain chemical elements absorbed by land plants from the soil solu- tion, as opposed to amounts of these elements ab- sorbed directly from the atmosphere (Shacklette and Cuthbert, 1967, p. 41). The problem of element source does not arise in discussing Spanish moss because this plant obtains all nutrients from atmo- spheric gases, precipitation, and airborne particulate matter. Most kinds of airborne materials cause no appar- ent damage to plants that are subjected to them. However, the plants may respond by accumulating large amounts of certain of these materials either within the tissues, in deposits on the surfaces of the plants, or both. Ordinary terrestrial plants may in- corporate airborne materials into their tissues di- rectly through leaves or indirectly by root absorption of fallout material that accumulates in‘soils. W'ohl- bier (1968, p. 142) noted [translated], “Plants take fluorine from the soil by means of their roots, and from the air through the stomates of the leaves.” Garber, Guderian, and Stratmann (1968, p. 41) con- cluded [translated], The variation in fluorine enrichment in plants from experi— AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS mental soil that has not been polluted is insignificant in comparison with the possible enrichment through fluorine- containing air pollution. Therefore, analyses of the fluorine content of plants can be of consequence as an important ad- junct in diagnosing the effects of fluorine in areas of airborne industrial emissions. The mechanisms of element absorption by, or fixa- tion on, leaves and stems of plants are not fully understood; doubtless, the mechanisms vary greatly, depending on elements and plant species. Generally, the amounts of airborne substances absorbed and incorporated in the plant tissue cannot be distin- guished from the amounts held only on the surfaces of leaves and stems. Washing the samples with water or other solvents may remove some of the surficial deposits that are not firmly bound to the plant sur- faces. Goodman and Roberts (1971, p. 298) reported, Earlier experiments clearly showed that it was not possible to wash any significant amounts of metals either from moss samples or grass material when it was obtained outside the moss desert [an area so heavily polluted from atmospheric fallout that mosses could not grow in it]. Grass samples taken inside the desert, however, bore significant quantities of wash- able metals. * as much as 45% of analysed metals could be removed by washing. MacIntire, Hardin, and Hester (1952, p. 1368) found analyses of Spanish moss useful in measuring relative degrees of atmospheric fluorine at different locations, but they did not determine whether the increases in fluorine content in the experimental plants were due to metabolic functions of the plants or to chemical or physical fixation on the plant. In a study of air pollution as measured by analysis of the moss Hypnum cumessiforme, Goodman and Rob- erts (1971, p. 291) stated, Further information is, however, required about the relative bonding energies of the metals on the exchange surfaces of Hypnum and the modifying influences of rainfall pH. It is also important to know the relative contribution to uptake made by dry deposition (sedimentation and diffusion of vari- ous particle sizes) and wash-out processes by rainfall. Practical applications of the pronounced ability of mosses and lichens to accumulate airborne contami- nants have been made recently in Europe. By analyzing specimens of the mosses Hylocomz'um splendens, Pleurozimn schrebem’, and Hypnmn cu- pressiforme that were preserved in the Botanical Museum in Lund, Sweden, and that had been col- lected at intervals from 1860 to 1968 from the same locations, Riihling and Tyler (1968) were able to record the effect of human activity on the concentra- tion of atmospheric lead. They wrote (1968, p. 321), “From values of c. 20 ppm [dry weight basis] in the years 1860—1875 the concentration of lead was more than doubled between 1875 and 1900. During the first half of the 20th century no measurable E3 was a new strong increase to a present average of c. 80—90 ppm.” Riihling and Tyler (1968, 1969, 1971) and Ruhling (1969, 1971) conducted studies of re- gional distribution of other airborne heavy metals by analyzing samples of mosses and other plants, and they found this method effective in evaluating aerial pollution. Jaakkola, Takahashi, and Miettinen (1971) ana- lyzed specimens of a lichen (Cladom‘a alpestm's) to measure the airborne cadmium that was released by a recently constructed zinc refinery in Finland. Goodman and Roberts (1971) used samples of the moss H ypnum cupressiforme to measure the airborne heavy metals in some industrial areas in Wales. They concluded (1971, p. 291), “Our methods are much more rapid, inexpensive, and probably more mean- ingful than spot-sampling of air by filtration, for which prohibitive resources are needed for a few months’ operation.” MORPHOLOGY AND ECOLOGY OF SPANISH MOSS Certain morphological features of plants enhance their ability to accumulate airborne materials; there- fore, as the physical structures among different species vary, so do responses to airborne materials. Osburn (1963) found moss and lichen mats to be effective filters of radioactive fallout in snow melt water in the Rocky Mountains. In a study of radio- nuclide fallout from the atmosphere, Watson, Han- son, Davis, and Rickard (1966) stated, “Direct foliar interception by plants is regarded as one of the prin- cipal sources of contamination from fallout materi— als. * * * Foliar surface area and plant density are therefore important factors requiring consideration in evaluating fallout accumulation.” In the same study the writers pointed out (p. 1176) that greater accumulations of atmospheric fallout occur in plants (1) with persistent above-ground parts (perennial plants), (2) capable of obtaining water and nutri- ents from the atmosphere, and (3) having great absorptive characteristics. Mosses, lichens, Spanish moss, and certain other plants have all these attri- butes; moreover, Spanish moss has the advantage, for use in evaluating the kind and amount of air- borne materials, of having no roots or rootlike or- gans and hence of having no direct contact with the soil. Results of chemical element analyses of plant sam- ples are expressed, on the basis of weight, as the proportion of the element of concern in relation to the total sample. Therefore, plants with a great foliar surface in proportion to their weight tend to changes were observed, but after about 1950 there 1 accumulate greater concentrations of elements from E4 the air than do other plants because their large sur- face area favors increased absorption and surface deposition of the airborne materials. Thin, small, numerous, and finely divided stems and leaves (or stemlike and leaflike organs, as occur in lichens) are morphological expressions of great surface area relative to the weight of the plant. Spanish moss is classed as an epiphyte because it most commonly grows on trees. It is not parasitic because it gets all essential elements and water from the air and produces its own food by photosynthesis. It is in the pineapple family (Bromeliaceae), and most species in this family are epiphytes with highly specialized morphological adaptations that assist in obtaining water and other materials from the air. The morphology of this plant was outlined by Garth (1964), who also gave the major references in the literature. The plant body consists of a sinuous pendant stem that bears curved filiform leaves at nodes on the stem (fig. 1A). The entire plant body, except the flowers and seed capsules, is closely invested with overlapping translucent scales (fig. 1H ), each scale being attached to the stem or leaf at a central point (fig. 1G). These abundant scales greatly increase the surface of the plant that is exposed to the air. Aso (1909), on the basis of experiments he conducted on lithium nitrate absorption, concluded that salts could pass through the scales into the plant. The ability of the scales to entrap airborne particulate matter is obvious. The scales also assist in water absorption by the body of the plant and contribute to the ability of the plant to withstand drying even when placed in a strong dessicant (Billings, 1904). Small yellow flowers (fig. 13) borne terminally on short branches that arise from leaf axils (fig. 1A) appear usually in March and April and probably are self-pollinated. The seed capsules (fig. 1A and C) begin to form in June but do not mature until the following December and January (Garth, 1964). The capsules then open (dehisce) , and each releases 10 to 20 seeds (fig. 1D). Each seed bears a plume of long epidermal hairs (fig. 1E) that add buoyancy to the seed and enable it to become airborne. If the seed falls on a suitable substrate, such as the rough bark of a tree or a mass of Spanish moss plants, the small barbs at the joints of the hairs (fig. 1F) assist the seed in adhering to the substrate (Billings, 1904, p. 107). After germination of the seed, rootlike or— gans (rhizoids) attach the plant to the substrate (fig. 11) but do not function as absorbing organs (Garth, 1964). These anchoring organs soon degen- erate, and the plant is supported only by its densely intertwined stems and leaves. STATISTICAL STUDIES IN FIELD GEOCHEMISTRY The factors responsible for the occurrence of Spanish moss at particular locations, and for its re- gional distribution, are imperfectly understood, al- though the ecology of the plant has been investigated extensively. In many large areas Spanish moss is extremely abundant and grows on almost any kind of support, whereas in others it occurs at discontinu— ous locations or is found only closely associated with ponds and streams. It grows on both living and dead trees and in full sun exposure or in shade (fig. 2). Masses of the plants are torn from the supporting trees by winds, and if dropped at suitable locations, they continue growth and establish new colonies. Garth (1964) suggested that the distribution of Spanish moss in the United States is related to major storm paths which arise in Mexico and move later- ally over the coastal plains. The northern limit of the distribution of Spanish moss (fig. 3) is controlled in some manner by sea- sonal temperatures, although the plant will endure short periods of freezing weather and snowfall. Transplant experiments have proved that it will sur- vive and reproduce at favorable locations as much as 75 miles north of its natural range (Garth, 1964). Its western limit, in east-central Texas, probably is controlled by the frequency of rainfall. Garth (1964) demonstrated that the plant could survive only about 3 to 4 months if deprived of rain, even if natural high humidity was maintained around the plant. METHODS OF STUDY SAMPLING PLAN The collection of samples for this study was begun in 1965 and was continued through 1970 as opportu- nities arose while we were engaged in other field studies. The general plan was to obtain samples from as many areas as possible at sites spaced about 50 miles apart throughout the range of Spanish moss. In order to obtain samples from some areas that we could not visit, County Agricultural Extension Agents in selected counties were requested to collect and submit samples, and 22 agents responded. In addition, single samples were contributed by several other persons. Samples were obtained from a wide variety of habitats, including roadsides, industrial and urban areas, remote locations in forests and swamps, ocean beaches, and agricultural areas. Special efforts were made to procure some specimens from sites at the western and northern limits of the plant’s range. The samples were pulled from trees or other supports, and Visible extraneous matter was removed from the samples before they were placed in cardboard boxes and shipped to the US. Geological Survey labora- tories in Denver. AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E5 FIGURE 1. —— Growth habit and morphological features of Spanish moss (Tillandsia usneoides L.). A, mature plant bearing a seed capsule; B, flower; C, immature seed capsule; D, dehiscence of the mature capsule, which releases the seed; E, air- borne seed, with plume of epidermal hairs; F, distal portion of a single epidermal hair; G, a peltate epidermal scale; H, portion of stem, showing imbricated translucent scales; 1, young seedling attached to a branch by means of rhizoids. E6 STATISTICAL STUDIES FIGURE 2.——Growth of Spanish moss on the tops of pond cypress (Taxodium ascendens Brongn.) and on the more shaded lower branches. Okefenokee Swamp, Charlton County, Ga. Photographed July 23, 1963. The sampling sites are plotted in figure 3 and described in table 1. All samples were analyzed in the winter of 1970—71. Although the gain or loss of elements during storage of the samples could not be precisely evaluated, we noted no correlation between storage time and concentrations of the more volatile elements. ANALYTICAL METHODS In order to minimize the effects of analytical drift, the samples were arranged in a randomized order before being submitted to the laboratories, and they were analyzed in the same sequence. The unwashed samples were ovendried and then pulverized in a blender. A wet-digestion method was used to prepare the samples for determining their arsenic, mercury, and selenium concentrations. For determining the concentrations of other elements in the samples, por— tions of the pulverized plants were transferred to IN FIELD GEOCHEMISTRY ceramic crucibles, weighed, and burned to ash in an electric muffle in which the heat was increased 50°C per hour to a temperature of 550°C and held at this temperature for about 24 hours. The ash was then weighed to determine the ash yield of the dry plant sample. A colorimetric method was used to analyze the ash for phosphorus content, and an atomic ab- sorption method was used for determining cadmium, calcium, lithium, potassium, sodium, and zinc. Con- centrations of the remaining elements in ash were determined by a semiquantitative emission spectro- graphic method (Myers and others, 1961). The values obtained by spectrography were re- ported in geometric brackets having boundaries, in percent, of 1.2, 0.83, 0.56, 0.38, 0.26, 0.18_, 0.12, and so forth; the brackets are identified by their respec- tive geometric midpoints, 1.0, 0.7, 0.5, 0.3, 0.2, 0.15, and so forth. Thus, a reported value of 0.3 percent, for example, identifies the bracket from 0.26 to 0.38 percent as the analyst’s best estimate of the concen- tration present. The precision of a reported value is approximately plus or minus one bracket at the 68-percent level of confidence and plus or minus two brackets at the 95-percent level. The approximate limits of detection of 'the ana- lytical methods that were used are given in table 2. Some combinations of elements in a sample, however, affect these limits. For example, concentrations somewhat lower than these values may be detected in unusually favorable materials, whereas these lim— its of detection may not be attained in unfavorable materials. DATA PRESENTATION The concentration of each element in each sample (or its ash) and the location of the corresponding sample site expressed as degrees and minutes of latitude and longitude were entered on automatic- data-processing cards. The analytical values were transformed to logarithms through a computer pro— gram which also determined the minimum and maxi— mum values and the basic statistics. In addition, the program reported all occurrences of missing data, indicated the concentrations that were beyond the limits of detection of the analytical methods, and printed both a histogram of the analytical values and an accompanying table of frequencies and cumula- tive frequencies for each class designation on the histogram. The table of frequencies could be used to divide the range of reported values into five classes so that as far as possible about 20 percent of the values fell into each class. For some elements the limited ranges in values prohibited use of more than two or three classes to represent the total distribution. Class-in- AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E7 ’ 210 / FORTH 21/fi52'2" fl CAROLINA l—_ “ —-| F——_—-——:’ SOUTH CAROLINA 52:3. 1‘ l \ 1 ’ ’ ’ “x/ag 13 I _. ’ 214A 40 I ' i ARKANSAS ""“T—‘flwY " [2152' / 4 .' L l ' GEORGIA\ / 62. i “ 1.. I I ALABAMA\ /.\\ 55 l “WW“ \L} __ ISSISSIPPII \222 37 41333 ‘~ :-. 217 N I . :23; 233. I 248 Lajfi. S o o ApprOXImate western and northern, Il— 29292 282\ '246 /*fil$. . o 220 J 218 _ i limit of Spanish Moss distributiond * ’ . ' (199 20036.33 . 219 k ‘-- — - .. ' 289. '1290 2865 3250 632231'1 0 .32'34 f 35 ‘\ TEXAS I, \ 9- 2'288. ' oyfm‘ P 7} ._ _\ 53 .223 ‘ 6287. 24 —22 ”—22; , \ 4/ 274 2 3 9 27-2260 ' x \ ,—--— 7§. 273 239' 41:23 .24 r .2. 7 o \ l / 277' '7 271’ 237' ’ o W’ ’ 96'23 ‘ 7o . " \ A 7219- 238' 226960 - ' 72 "4 '2 it, ' 2 232 23° ‘ 54 '23? 67 \ f _\ a k f . :‘l \J \ 5/ 959.2672‘9851’,‘ 270 3" FLOR‘DA 28 203.232 \ \ ‘49 56 2 . , LOUISIANA 2;; v 31299 .74 ' 2: 5; - ‘1 251 1/‘ 206 ‘S '268 \ \{25 y 263 51 . 3° 1 207 o \ O ' 81 \ ‘ 254 7 4 fix‘ \ O 100 200 300 400 500 MILES . " x | I l L L l I ' r 1 I l l I I "90-” 0 100 200 300 400 500 KILOMETERS "' ' FIGURE 3. —— Spanish moss sample localities (described in table 1). Seashore sites are plotted offshore. TABLE 1. — Spanish moss sample localities and sampling dates Locality Laboratory No. Sampling Locality (fig- 3) (D414—) date State County or parish Area 29 ....................... 3‘76 ...................... "June 1965 l nuisinnn Natchitoches 2 miles E. of Clarence. 30' 330 Alahnmn Monroe U.S. Route 84, at Perdue Hill on Alabama River. 31 237 ,,,,, do do Henry State Route 10, 2 miles E. of Shorterville. 32 299 .. dn Georgia lrwin State Route 32, 10 miles E. of Ocilla. dn Tefi Davi U.S. Route 23, 5 miles NW. of Hazelhurst. d0 Bacon U.S. Route 1, 4 miles S. of Alma. ‘10 Glynn St Simons Island, near Fort Fredrica. do Dodge State Route 117, at Rhine. dn RIM-Hey S's-tn Route 87, 12 miles N. of Cochran. dn Emanuel State Route 192, 3.5 miles W. of Stillmore. dn Burke U.S. Route 25, 10 miles N. of Waynesboro. 40 409 do North garnllina Erunswick State Route 211, 3 miles W. of Supply. 41 203 do South are ina .................... orry ..................................... U.S. Route 17, at Little River. 42 1'13 dn d0 FlorPW‘P State Route 378, 2 miles W. of Lake City. 43 "129 d0 NOI‘th Cfil‘filiml ODSIOW State Route 17, 1 mile S. of Verona. 50 ........................ 411 ...................... September 1965.... Tex-as Brazoria K miles IEW. 1:): Angleton, Farm Road 521, at Oyster ree . 51 201 do Florida . Sarasota 0n Wilkinson Road. . 52 “R dn MiSSiS_SiDD' LRWI‘GYIC" A miles S. of Monticello, near Pearl River. 53 912 d0 Georgia Baker Mixed pine-hardwood forest. 54 414 Sn £10121??? 1_ 811‘)": t Bacityleirka Cross City. 55 Q46 n Du am "In al‘ 85 on ............................. 15120 S an . 56 109 d0 TPX‘“ ViCtOT‘R Bank of Guadalupe River, near Victoria. 58 368 ......... dn Florida Palm Beach .......................... Junction of Pike and Belvedere Road, West Palm Beach. Texas Karnes... ...Bank of San Antonio River. Florida St. John Tn St. Augustine. do Highlands In Sebring. South Carolina Georgetown. ..In Georgetown. Alabama Covington... ...7 miles SW. of Opp. Florida Volusia In DeLand. do Orange At 2350 E. Michigan Ave., Orlando. Louisiana Terrebonne ........................... In Gibson. October 1965 ..... ..Florida Wakulla '2 .5 miles W. of Crawfordville. ..Louisiana .............................. East Baton Rouge .............. R. 1 E. .8 S. "September 1965 do Florida D9 de Paradise Key, Everglades National Park. do North Carolina ................... Tyrrell ................................... Near Albemarlc Sound shore, 0.5 miles N. of Columbia. .......... do........... .........................Indian River........................At Vero Beach. October 1965.. Rroward On Prospect Road, Fort Lauderdale. ...March 1966.... Colorado ................................ At rest stop, State Route 71, 8 miles NW. of Columbus. On Fort Walton Beach. U.S. Route 280, 8 miles E. of Cusetta. State Route 49, 3 miles S. of Andersonville. Hillsboro ................................ StaTtea Route 60, in Limona, about 5 miles E. of amp ...State Route 33, at Eva. ...Seminole.. U. S. Route 17— 92, at margin of Lake Jessup. Alachula ...U.S. Route 441, 1 mile S. of Micanopy. Levy ......... ..U.S. Route 41, 2 miles S. of Morriston. ....Polk ........... E8 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY TABLE 1. —— Spanish moss sample localities and sampling dates — Continued Locality Laboratory No. Sampling Locality (fig- 3) (D414—l date State County or parish Area 206 810 do do Manatee U.S. Route 41, undisturbed lot at N. edge of Bradenton. 207 875 do do Charlotte U.S. Route 41, 5 miles W. of Murdock. 208 “‘79 ...do do Highlands U.S. Route 27, 5 miles S. of Sebring. 209 '24"! . ...do do Polk State Route 60, at edge of Bartow. 210 ........................ 370 ...................... January 1970.... ...North Carolina... .Pasquotank... ....U.S. Route 158, 10 miles E. of Sunbury. 211 21R do do Dare U.S. Route 158, 3 miles N. of Kitty Hawk. 2‘? "‘40 1‘" do Washington State Route 64, 8 miles E. of Rober. 21? 200 do do Robeson Interstate Highway 95, at Lumber River, 3 miles S. of Lumberton. 214 817 do South Carolina Dillon U.S. Route 301, 1 mile N. of Latta. 215 851 do do Marion State Route 501, at Little Pee Dee River. 216 953 do do Charleston U.S. Route 17, at W. city limits of Charleston. 217 266 do do Reauford Hunting Island State Park. 218 287 do do do Folly Field, 200 yards from the beach, Hilton Head Island. 219 ‘193 do Georgia (‘hatbam Savannah Beach, near the lighthouse, Tybee Island. 220 ‘213 do do Emanuel U.S. Route 80, 2 miles E. of Adrian. 221 264 .......... do do Iiaurens U.S. Route 80, at Ford Branch, 3 miles W. of Dublin. 222 242 do do Houston Interstate Highway 75, 2 miles S. of Perry. 222 898 do do Tiowndes Interstate Highway ‘75, 2 miles N. of Valdosta. 224 219 do Florida Hamilton Interstate Highway 75, at Jasper-Madison exit. 225 225 do _ .. d Suwanee U.S. Route 90, at Wellborn. 226 262 do Iefl‘erson U.S. Route 90, 1 mile W. of Monticello. 227 407 do do Gaflsr‘nn U.S. Route 90, 5 miles W. of Gretna. 228 878 do do Holmes State Route 79, 3 miles N. of Bonifay. 229 285 do Alabama Geneva County road, 5 miles NE. of Hartford. 230 '{15 do Florida Bay U.S. Route 98, W. side of Panama City. 231 298 do do Walton U.S. Route 98, bridge at Philips Inlet, 2 miles W. of Sunnyside. 222 805 do Alabama Ral-lwin 0n the beach at Josephine. 233 409 do do Mobile Interstate Highway 10, 5 miles W. of Mobile. 234 271 (In Mi issippi Hanonok In Bay St. Louis. 235 296 do Louisiana St. Tammany ....................... Junction of U.S. Routes 90 and 190, 4 miles W. of Perlington. do Tanginabon U.S. Route 190, at Robert. do St. Landry At Krotz Springs. do Avoyelles State Route 1, at Maureville. do Pointe Coupee ..................... State Route 1, 2 miles S. of Batchelor. do Ascension U.S. Route 61, at Sorrento. do St. Tammany ....................... State Route 25, at Folsom. MI i ippi Walthall State Route 5, at Mississippi-Louisiana Stai line. do Marion U.S. Route 98, at Columbia. ...do Forest U.S. Route 11, on bank of Leaf River, at Hattieshurg. do Perry State Route 15, on bank of Leaf River, at Beaumont. 246 '19? do do Greene State Route 63, on bank of Chickasawhay River, at Leakesville. 247 'K45 do Alabama Washington“ .State Route 56, 7 miles W. of Chatom. 24R #67 do do ..... Clarke... .U.S. Route 84, 1 mile W. of Whatley. 249 250 do do ..... Conecuh ..U.S. Route 84, 2 miles E. of Belleville. 250 406 .......... do do ...Lowndes.. .Interstate Highway 65, 3 miles S. of Letohatchee exit. 251 244 do ....... do ..Elmore U.S. Route 231, at the Tallapoosa River. 254 ........................ 400 ...................... April 1970 Florida Collier Coximba area, Marco Island. 259 823 May 1970 Texas San Patricio Rank of Nueces River, near Sandia. 261 294 d do Live Oak State Route 59, at the Nueces River. 263. 286 .......... do Arnnsa: State Route 35, 1 mile N. of Aransas Pass. 265 295 do Calhoun State Route 35, near the Guadelupe River. 266 29* do Maiflgorda State Route 35, 2 miles E. of Blessing. 267 416 do Fort Bend In Sugar Land. 268 231 do .......... do Harri Interstate Highway 10, 5 miles E. of junction with Interstate Highway 610. do Chambers Interstate Highway 10, at E. side of Trinity River, Wallisville. do Iefi'ersn" Interstate Highway 10, at W. city limits of Beaumont. do Orange Interstate Highway 10, at W. side of Orange. Louisiana Calrasien U.S. Route 171, 4 miles N. of Lake Charles. Vernon State Route 8, at the Sabine River bridge. Iasper U.S. Route 190. 1 mile E. of Neches River. Tyler U.S. Route 190, 1 mile E. of Polk County line. Polk U.S. Route 190, 1 mile E. of San Jacinto. Walker State Route 30, 10 miles W. of Huntsville. . Grimes State Route 90, 1 mile S. of Navasota. ...do Washington U.S. Route 290, 5 miles W. of Brenham. Florida Pins-"as At Crystal Beach. Mi i inpi Yazoo U.S. Route 49W, 6 miles W. of Yazoo. do Hinds Interstate Highway 55, at Pearl River, 2 miles S. of Jackson. 286 227 do do Jeffer on State Route 28, 18 miles W. of Union Church. 287 401 do do Adams U.S. Route 61, 4 miles W. of Natchez. 288 288 do Louisiana Concordia U.S. Route 84, 2 miles W. of Wildsvilie. 289 269 do do ....... La Salle ................................. U.S. Route 165, 5 miles NE. of Tullos. 290 415 do do Caldwell U.S. Route 165. at Columbia. 291 989 do do Franklin State Route 4, 1 mile NE. of Winnsboro. 292 854 do do Richland State Route 17, 4 miles N. of Delhi. 29'! 982 do Arman-um Ashley State Route 8, 1 mile W. of Parkdale, on Cutofi' Creek. 294 252 .......... do do Lafayette U.S. Route 82, 2 miles E. of Red River. AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS terval code numbers were assigned to each analytical value; these numbers were entered on base maps by an automatic plotter at the proper geographical loca- tion of the corresponding sample and then replaced with symbols as shOWn in figures 5 to 35. The geometric means and geometric deviations given in figures 5 to 35 are antilogs of the arithmetic means and standard deviations, respectively, of the logarithms of the analytical values. Where some of the concentrations for an element were determined to be less than the sensitivity of the analytical method (table 2), the means and standard deviations of the logarithms were estimated by means of a censored-distribution technique devised by Cohin (1961). Means estimated by the use of this tec - nique may be lower than the limits of detection f r certain elements, as is illustrated by the mean ar— senic content of 0.79 ppm in Spanish moss (fig. 7), whereas the limit of detection for arsenic is 1 pp (table 2). Further discussions of the treatment f censored frequency distributions of geochemical data and of the use of geometric means and geometric deviations were given by Miesch (1967) and Shac lette, Sauer, and Miesch (1970). All data analysis in this study is based on log:- rithms of reported concentrations because minor el - ments commonly tend to exhibit positively skewed frequency distributions. The geometric mean is t ‘e best measure of central tendency in log normal y distributed data and, as such, is an estimate of the typical or most common concentration for the el - ment. The range from the geometric mean multipli d by the geometric deviation to the geometric mean divided by the geometric deviation generally includes about two-thirds of the analytical values. About 95 percent of the values occur in the range from the geometric mean multiplied by the square of the geometric deviation to the geometric mean divided by the square of the geometric deviation. For ex— ample, the geometric mean cadmium content of ash of Spanish moss is 7.9 ppm, and the geometric deviation is 1.65 (fig. 10). Thus, probably about two-thirds of a large group of randomly selected samples will have cadmium contents in the range 7.9+1.65:4.8 ppm to 7.9><1.65—_—13.0 ppm, and about 95 percent will have cadmium contents in the range 7.9+1.652:2.9 ppm to 7.9)(1.65'-’:21.5 ppm. ELEMENTAL COMPOSITION OF SPANISH MOSS Chemical analyses of Spanish moss were performed first by Wherry and Buchanan (1926) and later by Wherry and Capen (1928); the analyses indi- E9 TABLE 2. — Analytical limits of detection [Analyses made by semiquantitative spectrographic method, except as indi- cated. Dry plant material was used for arsenic, mercury, and selenium analyses; plant ash was used for analyses of all other elements. Data re- ported in parts per million] Lower limit of detection Lower limit of detection 20 Mercury ............ ‘0.5 Molybdenum.... Element Element Aluminu Arsenic.. ..... 11 Barium.. 3 Neodymium ..... Beryllium... 2 Nickel ..... 10 Boron ..... 50 Niobium 20 Cadmium 2.5 Phosphor 12 Calcium 2150 Potassium. 250 Cerium 300 Scandium 10 Chromiu 2 Selenium 11 Cobalt _____ 7 Silver ...... 1 Copper... 2 Sodium. 2100 Gallium.. 10 Strontiu 10 German: 15 Tin ..... 20 Iron .................. 20 Titanium ..... 5 Lanthanum 70 Vanadium.. 15 Lead ............ 20 Ytterbium... 2 Lithium ...... 24 Yttrium.. 20 Magnesium 50 Zinc ......... 225 Manganese. 2 Zirconiu 20 1Analyzed by colorimetric method. 2Analyzed by atomic absorption method. cated that the plant obtains a wide range of chemi- cal elements from the atmosphere. MacIntire, Hardin, and Hester (1952) transplanted specimens of Spanish moss that contained about 27 ppm fluo- rine in ash when growing in Florida to sites in Tennessee at different distances from factories that emitted fluorine from their stacks. After 3 months the fluorine content of the transplanted specimens ranged from 100 ppm to as much as 2,418 ppm on a dry-weight basis, the amount in the specimen being inversely proportional to the distance of the sample from the factory. Shacklette and Cuthbert (1967, p. 43) reported that the iodine content of five Spanish moss specimens averaged 5 ppm in dry matter and ranged from 4 to 7 ppm. Martinez, Nathany, and Dharmarajan (1971) reported lead analyses of eight Spanish moss sam- ples from Baton Rouge, La., from sites near heavily traveled highways and from sites more distant from heavy traffic. They found that the lead concentra- tion was greatest in the samples from sites near highways, with a maximum of 0.085 percent (850 ppm) in the dry samples, whereas samples from sites more distant from highways contained as little as 0.0051 percent (51 ppm). The percentages of ash obtained by burning the samples were not given; if the ash yield of these samples was similar to the mean ash content of samples analyzed in the present study (4.5 percent of dry weight), their maximum lead concentration would convert to about 18,700 ppm in ash, and their minimum lead concentrations would convert to about 1,122 ppm in ash. These values are well within the range (70 to 50,000 ppm lead in ash) given in figure 17 of the present report. E10 Benzing and Renfrow (1971) analyzed samples of Tillandsz'a circinata Schlecht., a small epiphyte closely related to Spanish moss. They reported con- centrations of the nutritive elements calcium, mag- nesium, nitrogen, phosphorus, potassium, and sodium in samples of prefruiting, fruiting, and post- fruiting plants from six sites in southern Florida. Although considerable variation was found in the concentrations of these elements among sites and growth stages, plants in the prefruiting stage gen- erally contained more potassium, magnesium, and phosphorus than plants in other growth stages. Trends in growth-stage concentrations of the other elements were not pronounced, although the post- fruiting stage at one sampling site contained the largest amounts of magnesium, calcium, and sodium. Shacklette (1972) reported that the cadmium concentrations in the Spanish moss samples dis- cussed in the present report ranged from 2.2 to 27 ppm in ash. The concentrations were thought to be related to the degree and kind of aerial pollution at the locations where the samples grew. The ash yield and the concentrations of selected elements in the samples of Spanish moss collected STATISTICAL STUDIES IN FIELD GEOCHEMISTRY in this study are presented in figures 5 to 35. Sample localities indicated by symbols in these figures are referred to by locality numbers in figure 3 and are described in table 1. Concentrations of all elements but arsenic, mer- cury, and selenium (figs. 4—35) are given as parts per million in ash. The percentages of ash obtained by burning the dry samples are given in figure 5. The parts per million of an element in ash can be converted to approximate parts per million in the dry material by means of the following equation: Element (ppm) in dry plant : element (ppm) in ash ash content (percent) 100 The elements beryllium, cerium, germanium, mer- cury, neodymium, niobium, selenium, and tin were not commonly detected in the samples. The localities where samples containing these elements were col- lected and the concentrations of the elements in the samrles are shown in figure 4. Some elements were looked for in all samples but were not found. These elements, analyzed by the semiquantitative emission spectrographic method, and their lower detection limits, in parts per million, -/". / ,JNORTH CAROLINA ) Be 2 T‘““" l'““"" ’ / yNb 15 \ V L ’ ____— .' ‘ ! \ r? J.{' "‘~ I I } r..-—T—--"‘\' —\ SOUTH .' L I ARKANSAS ' I \ \gAROLINA l “ 1. I Ii" I \ GEORGIA ~ “"““/“‘"\L 'ISSISSIPPI ALABAMA .~ , se 1 N ,' ‘. I \ Nd 150. .' K:‘~-~-a u\ Ce 300% I‘ Be 2' (I ' ' Nb 20 Nd 150 Ce 200 . Ce 200 TEXAS I A ___.— \\ Nb 20 L”- . 36 3362 ‘ \ NdlSO ‘ , LOUISIANA ‘\ > > r '7 ‘ l Sn 20 Hg 05 . . Sn 20 - .Hg 0.5 Hg 0.5 ‘\ f"“\ 5“ 20. 1. I'Hg 0.5 "8 0-5 \\ Hg 0.7 \\r \\ Sn 30.. 1 _ ‘ Se 1 Sn 20 Ge 15 ‘ \ 0 ~ \ ‘ FLORIDA \ v I 1 \\ o 100 200 300 400 500 MILES Hg 0.5 ‘x.\ . i VII l I In I 1] | -. 0 100 200 300 400 500 KILOMETERS 3.1?" FIGURE 4.—-Localities of Spanish moss samples containing elements not commonly detected and the concentrations of the elements. Mercury (Hg) and selenium (Se) values are expressed as parts per million in dry plant material; beryllium (Be), cerium (Ce), germanium (Ge), neodymium (Nd), niobium (Nb), and tin (Sn) values are ex- pressed as parts per million in ash. AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS are as follows: Gold, 40; hafnium, 200; indium, 20; platinum, 60; palladium, 2; rhenium, 60; tanta- lum, 400; tellurium, 4,000; thallium, 100; thorium, 400; and uranium, 1,000. If lanthanum or cerium was found in a sample, the following elements, with —‘__ ‘__, E11 their stated lower detection limits, were specifically looked for in the same sample: Dysprosium, 100; erbium, 100; gadolinium, 100; holmium, 40; lutetium, 60; terbium, 600; and thulium, 40. Though looked for, none of these elements was found. 7.1— l | N. st. W a: \\ \ \\ [h—\ \\r ‘\ \ o \\ O \ O 1 x , \‘~ o 1oo 200 300 400 500 MILES ‘ x I l l I ° l ' l l ' 0 100 200 300 400 500 KILOMETERS 90y" Symbol, and percentage of total samples represented by symbol 25 o 10 |°ml o z o E Z: 2 1 :‘3 | l 1 20 — | I | l 5 Hr ' z | L” D :5 I.” C: u. l 7 fd:i . . . 7 a 3: I l I | ". ". 9. N <- . I U | z "5 I C! 20* I I.” 1 IL 10— FIGURE 6. —-— Aluminum content of Spanish moss. olllllllllllll 8888888888888 cacao—u—Nmfiédd /\ ALUMINUM, IN PERCENTAGE IN ASH Geometric mean, 4.08 Geometric deviation, 2.21 Number of samples and analyses, 123 AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E13 r‘“——q /’I ‘Iv-‘LUJ" I If [\i I I J? \ } OI I I A | I tJVLAN/EH‘ }——__ Symbol, and percentage of total samples represented by symbol 70 3'3' 9 0&0: o w— 0 10|0 200 300 400 500 MILES I I I | | I | ' I I I I I 0 100 200 300 400 500 KILOMETERS I I I I I 50— I I 40— FREQUENCY 30- 1° _ FIGURE 7. — Arsenic content of Spanish moss. I I I O. “I q .— u— N <1.0— ARSENIC, IN PARTS PER MILLION IN DRY MATERIAL Geometric mean, 0.79 Geometric deviation, 1.55 Number of samples and analyses, 116 E14 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY Symbol, and percentage of total samples represented by symbol 40 a I :eI z; IaI : o | o I O :o'l o I | l | I I 0 100 200 300 400 500 MILES —“ | | I I—II‘I I I I1 I 1I I | | o 100 200 300 400 500 KILOMETERS 30— | I I | I | I I I >- I I U | z I '5‘ 20— I I # ‘3 | I I.“ z | IL I | I I 10— I FIGURE 8. — Barium content of Spanish moss. 0 l | | I I J l | I I I Q :3 O a O a O O O O Q "° " 53 “.2 8 8 S E 8 S 8 .— — N BARIUM, IN PARTS PER MILLION IN ASH Geometric mean, 564 Geometric deviation, 2.32 Number of samples and analyses, 123 AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E15 Symbol, and percentage of total samples represented by symbol 70 I I I I WI .':| 3| fil Ln 0 I o | O I o I o I I I I (I) 1(IJO zoo 3oo 4pc 500 MILES I 60-- I f I l “I I I I I I | o 100 200 300 400 500 KILOMETERS | | | | 50- I I I I I I I I 5 40— I I 2 I | Lu 2 I | c’ I E I u. 30— I 20* FIGURE 9. — Boron content of Spanish moss. 10— 0 l I 1 I I ~ 5 a § § BORUN, IN PARTS PER MILLION IN ASH Geometric mean, 150 Geometric deviation, 1.34 Number of samples and analyses, 123 E16 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY I‘“—-- ————— I I I I I : L“ I W I I k.‘—‘\—-tl ‘\ \ \ \\ /h-,\ \\r \‘ \ O \\ O \I O I x \‘ o 1?0 200 300 400 500 MILES ‘- I I l | I | x I I I I o' 0 100 200 300 400 500 KILOMETERS - ... "5"" Symbol, and percentage of total samples represented by symbol 25 I | I I I I I I on I Gm 928: .ZII .17 | | | 2“" | I l | | I | l I I I Z 15— I |_ I E I I 8 I | E: | u. H]— 5— 0 I l I I I I | I I I I I I l I I I I I I L catwaqwrfimcatquzsthqozquqw. @NNNMMQfim‘DNwm:‘:::2:g CADMIUM, IN PARTS PER MILLION IN ASH Geometric mean, 5.9 Geometric deviation, 1.65 Number of samples and analyses, 122 FIGURE 10. — Cadmium content of Spanish moss. AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E17 Symbol, and percentage of total samples represented by symbol 2 | NI 3| 2| :5 O f0 | g l0 l . o 100 200 300 400 500 MILES —— erT l ‘ IL II J I : l o 100 200 300 400 500 KILOMETERS 30— l _l >. o 2 g 20 0 Lu :1: u. 10 - . . FIGURE 11. — Calclum content of Spamsh moss. o—ll 1 1 1 I llll Nuvmw—mcc —.—Nn CALCIUM, IN PERCENTAGE IN ASH Geometric mean, 10.07 Geometric deviation, 1.56 Number of samples and analyses, 123 E18 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY -’/”I. _— .0 )0 FR“, r—--~--< / -Iv ’ I ' \/"\ I I I - __,I-J/ . ~ I L I l T. \ \ 0. I ,\ l gJ | \ \‘ O I _ I x ' \ G I “MAN/“- \Lfi > I \ \x I N ' '9_____!_2 I \ €000 .f' I :0 I 0 5° ' 00 o" r~--~~ . .9; I . o e \ \ 0 o , o I 09. I o \ \ 0:0 .009 ___O_J—Aw__Q___L \\ 0 G }.—— z r—va . O 00 \\ 1 (30 G O I 09 .99" G ’ ‘, - G 1 \ f“"\ ” . o ' ° / G 3 \ \J \ 0 . 01, c c 0° \\ \‘ 0O o’)/ I ‘.°Q ‘ \\ O (,4, \3, O 1\ I - O“ x k \ . I \ 99". Symbol, and percentage of total samples represented by symbol 50 I o IOIOI039 o 5 113|30—“ ‘3 | I | (l) 1(|)o 200 300 400 500 MILES I I fill I II I 1I l 40- I I o 100 200 300 400 500 KILOMETERS I I | _ I I I >_ 30- | ‘-’ I Z L“ I D c! I LLI a: | ”' I 20— I I 10— . FIGURE 12. — Chromium content of Spanlsh moss. U—III I I I I I I I I I I I IE r\ D In D D D ° C O D D O D D O V— F- N M LO N O LO C D C D Q Q CHROMIUM, IN PARTS PER MILLION IN ASH Geometric mean, 57 Geometric deviation, 1.73 Number of samples and analyses, 123 AIRBORNE CHEMICAL ELEMENTS IN ////0 r’" so 0 F“—-~ r—“—"’—g /f / I ’ _____. - I ll \ I? ’J—J” V\/~5 I .___——— ' - I! I I 'r———T— T *\ 00¢ ‘ Law I {J I \ \\ O "-'v I \ [I tgwAN/ghnifi "L l \ 6\ -f N . Io___9.§ I x 00000 0 I [— 5». ' 59° 0 O K““ O l . 3 O o ~~ . .0; o o a v0 \ \ o O . o I o I, \ ago a ,0 .0 00_O_9_—\ L- , \\ °\‘o_":’7 0 ~ 0 77—6 00 L O ’ . :‘ 2 "Q ( \ 1 00°90 1- 9.?ng O 5/ O 0 o \ A- - Q' I - O \ - _. G \ If \\\ O . 0‘ Q . 00 \\ \ .9 0 -40 ‘~ 000 0\ \\ (3 ¢ 1/ ’0‘ O \ ° 7’" I - 1 ~ 0 \ v- I o ‘ \\_‘ \\ oal’; \ ya”. Symbol, and percentage of total samples represented by symbol 4“ ImI “I I §|5|3I§I o: I | {I I! | 0 100 200 300 400 500 MILES I I | | Ifill l I II I 1I I I | —- I 0 100 200 300 400 500 KILOMETERS I P 30— I I | I >. U z 3 20— CI __ I.” I LL 10— FIGURE 13. — Cobalt content of Spanish moss. 0 I I I I l I l I 5 e a 52 2 a a 2 COBALT, IN PARTS PER MILLION IN ASH Geometric mean, 9.8 Geometric deviation, 1.83 Number of samples and analyses, 123 SPANISH MOSS E20 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY Symbol, and percentage of total samples represented by symbol 40 I o .olfi‘ol o 14 I“ Nil 16 | I o 100 200 300 400 500 MILES | l |[ 1.] I I 1I 1‘ I 4 I I o 100 200 300 400 500 KILOMETERS l 30— i 1 __A l I | >. | U | z “3‘ | o 20— _I LIJ C: H. 10— FIGURE 14. — Copper content of Spanish moss. ollllllllllllI—l‘ D a o c: c a a D a o c a c N M m N 6 ID a D O C o O O '- " N "’ “’ " ‘3 1‘2 3 COPPER, IN PARTS PER MILLION IN ASH Geometric mean,l48 Geometric deviation, 2.01 Number of samples and analyses, 123 AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E21 Symbol, and percentage of total samples represented by symbol :I .‘1‘I NI 3| 3 ’ I I I 0 1c|)o 200 300 400 500 MILES I | l I I I | | I . I I I I I I TI o 100 200 300 400 500 KILOMETERS I 30— I I l I I >' I U z I g zo~ a __I DJ E: Ll. 10— FIGURE 15. — Gallium content of Spanish moss. 0 l I l l | I l {7 " E 3‘3 S S S GALLIUM, IN PARTS PER MILLION IN ASH Geometric mean, 10 Geometric deviation, 1.71 Number of samples and analyses, 114 E22 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY Symbol, and percentage of total samples represented by symbol 50——W was ._ ._ E o lololol o I I l 0 100 200 300 400 500 MILES % I I I II [I] l 1 Il I ll l l 40_ l 1 — 0 100 200 300 400 500 KILOMETERS | | | I l | | l | l >_ 30~ | | g l I m | | g l | Lu | '3: l “- 20» l 10 E FIGURE 16. — Iron content of Spanish moss. 0 I l i l | l | l l l E 1’ 8 8 S E 8 8 8 8 8 o' c o d :5 d .—‘ -—' N m ‘9' IRON, IN PERCENTAGE IN ASH Geometric mean, 1.58 Geometric deviation, 1.87 Number of samples and analyses, 123 AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E23 Symbol, and percentage of total samples represented by symbol 90 0‘” 025} '4 80 — 1 I o 100 200 300 400 500 MILES I IL' II I I ‘I I I‘ I l 70 - I o 100 200 300 400 500 KILOMETERS | | 60 — I >' I g 50 [ Lu 3 I c: I-u I CC 40' ‘ L I | 30 — | FIGURE 17. — Lathanum content of Spanish moss. LANTHANUM, IN PARTS PER MILLION IN ASH Geometric mean, 14 Geometric deviation, 3.80 Number of samples and analyses, 123 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY E24 r--——~ ----- , ' I l I I l .' L“ I " I k‘\_“—-II \ \ \ \ \ m. \\\r/ \\‘ \ o \\ g \ o 1\ \\~\ . I_L “IO 2(|)o 3oo 4cl)o soc MILES I ' I I I | I 9.4 O 100 200 300 400 500 KILOMETERS Symbol, and percentage of total samples represented by symbol 20 | | l | 013 I 022' 92 | | | | | 10* FREQUENCY 5% 37 300 500 ~ 700 — 1000 — 1500 e 2000 — 3000 — 5000 — LEAD, IN PARTS PER MILLION IN ASH Geometric mean, 4,170 Geometric deviation, 3.16 Number of samples and analyses, 123 FIGURE 18. — Lead content of Spanish moss. AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS _—/”/”" ,r” 00/30 ,- F"“*, r—--~--7 / / —? ’ _,.———‘v . I I \ ’Cl” I \ \\ V3: ' 'O___9_; l \ 0.000 ,-' I [— L 0 I so . O O 4* I:~—- 0* I ° 9 '° ~~ I 00;! o o o O , ‘ \ o o g 0 I 009 I o \ Geo o 00 I0 ___QJ——\ 3,- -‘ \\ e\___.C9_Q—I G O ‘i—‘Q 0 L \ 000 I0 0 . . a tn ‘ ‘0 ‘ I 0 I 00 o gum» - ‘ o o \\ fn_\ 0° . t O __‘ ’ O S \ 0 o ‘ o ' 00 I \\ 0 _ -/ u‘O'. O. \ o v“ I: I 1 \ I, o k \I \ 9“". Symbol, and percentage of total samples represented by symbol 5“ o IoioI .— 9 '1540 27| 9 | I cl) lcl>0 200 300 400 500 MILES I I III I I I I J I I | I i o 100 200 300 400 500 KILOMETERS 4o- : ' | I ‘ I | _ | >_ 30— | u I Z w l 3 | c L“ | K ‘L I 20— I 10— _ FIGURE 19. — Lithium content of Spanish moss. 0 J_ l J l l I l I l f‘ O In C D C O c u— -— N" in IN a .— LITHIUM, IN PARTS PER MILLION IN ASH Geometric mean, 22.4 Geometric deviation, 1.53 Number of samples and analyses, 122 E26 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY h- Symbol, and percentage of total samples represented by symbol 50 : lemme o IG|0:0IO I I — I o 100 200 300 400 500 MILES I I I I I I I | r "I I I I 40— | | o 100 200 300 400 500 KILOMETERS | % | I | I 30* I >' | U z | LU 3 I c: I“ | I LL | 20— I 10_ FIGURE 20. — Magnesium content of Spanish moss. 0 l l l l | I I l I m. N. =4 m. =2 <=. 6. a. D. D c 1— r— N {’0 L0 N 2 MAGNESIUM, IN PERCENTAGE IN ASH Geometric mean,4.5 Geometric deviation, 1.73 Number of samples and analyses, I23 FREQUENCY 30 20 10 AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E27 r———__ Symbol, and percentage of total samples represented by symbol I 33 i§|8lglg o |0|olo|o i | ll 1 l '__| l l l I l _ I —. | l _ l lllllllllll 88888888888 NmmNOchDQq F—Nnml‘c MANGANESE, IN PARTS PER MILLION IN ASH Geometric mean, 2,300 Geometric deviation, 2.60 Number of samples and analyses, 123 0 1(|)o 200 300 400 500 MILES I I I | l I | ' I T I l o 100 200 300 400 500 KILOMETERS FIGURE 21. —— Manganese content of Spanish moss. E28 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY Symbol, and percentage of total samples represented by symbol 100 El 2‘ N ole} o | | _| | o 100 200 300 400 500 MILES I l l I | I J | l v | r 1 1 I | o 100 200 300 400 500 KILOMETERS 90? I | | g l >- | g | 3 20~ b o I.“ I IL 10— __ FIGURE 22. — Molybdenum content of Spanish moss. 0 I I l I I l h h D in a V .— .- N MOLYBDENUM, IN PARTS PER MILLION IN ASH Geometric mean, 3.7 Geometric deviation, L80 Number of samples and analyses, 123 AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E29 Symbol, and percentage of total samples represented by symbol 40 a ‘ Ialsl :21 N o :omwl 0 ~ I l | o 100 200 300 400 500 MILES l I I I‘ I I I JT | | 30b | o 100 200 300 400 500 KILOMETERS | | | I >' I U z | ‘5 20— o ‘ _ Lu a: _ u. 10» “ FIGURE 23. —— Nickel content of Spanish moss. I olllllllllll e 2 "2 a a s 2 8 S 8 .— — N NICKEL, IN PARTS PER MILLION IN ASH Geometric mean, 39 Geometric deviation, 1.74 Number of samples and analyses, 123 E30 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY I’" 00 ° /~ 7““? r—-“""_'—:I ,/ O I I \ ‘7‘ #""'\/« l g, \0 I I ' t1 ""T'J C 00 ' I r”'T— \ \ O I L I ' I \ O I A {J \ \ I | X I \ G , - ““V"/‘“‘"\L‘ > I \ \ 5 N I I0 9.; I \ 00090 .3‘ l—“—'— I0 I 500 90 " - 9* l O O I O 7 k ~——‘_4 I of e 9 0 I, ‘ \ 9 000,00 0 I0009 0,9 ’4 I \ 9L_39—-000 -' 0 5—1-1?“ ‘ 9, 0 9I 9‘ 24.». - 0 I. 1 000 I 00 <3ch 0 9/ o I “x 00 ‘ 0' G I 3 2 \ \\ ff \\ O 00‘ Q ‘ 0_ I 00 \\ ‘ ‘ O ’ O r \\ 00 O‘/ I ‘V g 1 c K‘ 0 ¥ ‘5 . / 1 ‘ .- o \ n . - I \\_~_ \I u. ’ x 90". Symbol, and percentage of total samples represented by symbol 55 o loleI o I o 20 |28| 41‘ 11 | 2 | | 50— I Ll l | | | | I I o 100 200 300 400 500 MILES I l I I I i | | I | IT I I I I 40 | I o 100 200 300 400 500 KILOMETERS _ I | I I I I > I I o I I Z '— 3 30— I c: | L” n: | “L I I I I zu~ I I I I I I 10— I FIGURE 24. — Phosphorus content of Spanish moss. 0 I I I I I I I I II'T D D D c C D D D O D O D O D D D D D O D on a: <2! 30— i u E l g T m a: u. 20_ 10— __ FIGURE 25. — Potassium content of Spanish moss. 0 I I I l | I I I I manqqqchq POTASSIUM, IN PERCENTAGE IN ASH Geometric mean, 9.41 Geometric deviation, 1-74 Number of samples and analyses, 123 E32 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY Symbol, and percentage of total samples represented by symbol 40 I I SII '5 I 0) 0| 0 | o I I I o 100 200 300 400 500 MILES I I I I I I J _| | WI I I I I _I 0 100 200 300 400 500 KILOMETERS 30— >- U E —I 3 b c: 20 LU fl: IL 10— FIGURE 26. — Scandium content of Spanish moss. (I l l l I | l in m h D ID 9 V v— -- N SCANDIUM, IN PARTS PER MILLION IN ASH Geometric mean, 5.85 Geometric deviation, 1.89 Number of samples and analyses, 100 AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E33 k— _ \ —‘__, FREQUENCY ?i 110 100 10 Symbol, and percentage of total samples represented by symbol N F F O . gl OI I I I | I | I I l I | l I | l I I I I I | | l I | I | I I I I | I =2 =2 m. a. 9. =2 =2 0 =2 e. .— .— .— N n in N e In D V -— — N SILVER, IN PARTS PER MILLION IN ASH Geometric mean, 0.12 Geometric deviation, 5.42 Number of samples and analyses, 123 0 100 200 300 400 500 MILES L I I | I I I I ' I I I I 0 100 200 300 400 500 KILOMETERS FIGURE 27. — Silver content of Spanish moss. E34 Symbol, and percentage of total samples represented by symbol 2“ I I I I s: I SI :2 I 2| : O I 0 | O | O I 0 I I I I I I > I U 2 LLI D C! L“ I LI. 1 l I l I I C D D D D D Q =2 a. c c, 5. =2 9. N 4‘") IO N a L!) D F- — N SODIUM, IN PERCENTAGE IN ASH Geometric mean, 2.2 Geometric deviation, 4.05 Number of samples and analyses, 123 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY O 100 200 300 400 500 M|LES I I I I l I I | I I I I J O 100 200 300 400 500 KILOMETERS FIGURE 28. — Sodium content of Spanish moss. FREQUENCY AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E35 r——__ ‘__§__ Symbol, and percentage of total samples represented by symbol 40 ji c: I 3' 8] SI '- o Io|o|oI o . — I I l 0 100 200 300 400 500 MILES I L l I l I I I 30— I I ' I I I I I | 0 100 200 300 400 500 KILOMETERS I I 20— 10— FIGURE 29. — Strontium content of Spanish moss. I) I l I I I I I I I II I D O O D D O D O c D e O O In D a D D e D D D D O '— ‘— N on LB N O LO Q C D .— — N m in l\ STRONTIUM, IN PARTS PER MILLION IN ASH Geometric mean, 704 Geometric deviation, 1.85 Number of samples and analyses, 123 E36 FREQUENCY STATISTICAL STUDIES IN FIELD GEOCHEMISTRY TITANIUM, IN PERCENTAGE IN ASH Geometric mean, 0.23 Geometric deviation, 1.96 Number of samples and analyses, 123 ,,/ I _—/ o. \‘ ,~ , I—————~l r——-————;, ,/ . / I ‘7 ,.———v\ K ’ / \. i I \‘ (1 __,_.r—’[!’ C O. . I I 'r”"T‘ \ \ . ' I L I I \ o ' -\ {X ' \ \ I ' I ‘ °\ ‘ J‘WAK/hH‘ > I \ .4 N I 1 x I,‘ I I9 0 I \ 03.00 K“ _._——— L Q | 5Q . . ['5- I v I . 0 I °, 0° |\ 0.. 0° \ 0 .I . o O ' 0:0 0 .0 IO °°__'.Q.——\____O_._L I' oL— . O O \ 00¢, 90 CI 3;.hx . - o I‘ ’0 ) 00.. I I O l/ O O O h_’\ C - ‘ . I O o I‘ \f/ \‘ g 004 c (-. 00“ S O . \ \ o ' < 00 ( \ 0 f?/ “.K. O \\\\ O V ‘1 o 1\ O “I I 0 \ I a?" Symbol, and percentage of total samples represented by symbol 4° I a, ImImI a o | o I°|°I o i I | I 0 100 200 300 400 500 MILES | |_ i [II I I II I 4 30_ | 0 100 200 300 400 500 KILOMETERS | | | | I l 20— i | l l I | 10— FIGURE 30. — Titanium content of Spanish moss. I I I I I I I I I I I o In D D O D C a D O c O S S 8 8 '5 E 2 8 8 8 E a" c‘ :5 :5 c' c' o' :5 e' o' :5 AIRBORNE CHEMICAL ELEMENTS IN SPANISH Moss E37 Symbol, and percentage of total samples represented by symbol 5° 5: I§I a! :l a o |olol I o I I ' ' | I I o 100 200 300 400 500 MILES f g L I I l I | 40_ | I I f I I I I _ I I l o 100 200 300 400 500 KILOMETERS | | _I I l 30 — | > I U 2 l I.“ 3 | (:1 W | K H- I 20 — I 10 — FIGURE 31. — Vanadium content of Spanish moss. [I I I I l I I O a C a D D " S 3 8 8 S VANADIUM, IN PARTS PER MILLION IN ASH Geometric mean, 91 Geometric deviation, 1.73 Number of samples and analyses, 123 E38 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY I---——~ r—""—"_:I ,/ O r/ ,—’v K I I \ 1 J—’( \/ \O ' I f, _-——-r’ x I I _——— - 00 I 'r T \ O .' L“ | {I . I \ o I 1. I I I \ O\\ o “WM/”“xL > I \ = N I ‘ “ “34 I 'O #9.; I \ 00.00 '0 "" r0 | 50. 0 '¢ I Oé I o G ( C 3 k“-—~—a I 0'}, 0 o o o 0 "o \ \ 0 O 9 I O I I \ 0:0 ' oo '9 “flu—*wuL—I \ \ 0L_9.Q. 0 e . o o \ (3 ,0 0 e I EL y 1:“ . O 1 00°0 *_ 0000 ‘1'?" o T, o o \‘ fh_h\ (C; O ‘ O a o 1 ‘II 0 O ‘\\ \~f \\ GO . O \\ 1 ° 0 \\ O(3 0‘ / L .000 J \\ e '1”) ‘3 o I I/ . \ o“ ‘ . INN \ 0/ x 9". Symbol, and percentage of total samples represented by symbol 4“ oIoIe ml 0 8| fifl °'| ‘° I I I_ I o 100 200 300 400 500 MILES I I 'II 1I II I 1I I Q _~ I o 100 200 300 400 500 KILOMETERS 30- | I | | > I U 2 I "5 20~ l ° I LIJ °‘ I LL I I | 10- FIGURE 32. — Ytterbium content of Spanish moss. (I I I I I I I I I If N N M In 1‘ 5 Ln a c F .— N co V YTTERBIUM, IN PARTS PER MILLION IN ASH Geometric mean, 2.4 Geometric deviation, 1.81 Number of samples and analyses, 123 AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E39 I r--——~ r— ————— ‘3» ,/ _ o / [I l \ 1 ’£{___(’ \/\50 I 'I : §"'T—~’T\— “\ 00‘ ( II In. I fr" | \ \‘ O \ | 1 ' \ O - I] “WAN/4‘"\L‘ > I \ \\ , N . I0____Q_§ . \ .0000 .0 [—— L O I 50 . O ‘ _‘ I 0‘ I 0 O O G a" k ——~... 1 0° f O ‘ O . 00 \ O 0 0'00 0 \0 0° 0.....A 0 _ x \ L £900 " o T’s—0’ ‘~ 0 ,9 00 '0 l - »_-.¢ ¢ 0 \ 00 O 3 O G O .fiwsa . ' O O . r"‘\ 09 ‘ ° W / O 8 \\ e O Q ( \\f \\ GO / . 06 \ \ G O .. 0.0 .\ \ 0 .GJ/ ‘V o l/ \ ° 1 x‘ ; K \ o"-‘- 1 . \x \x ,4 x ”'89". Symbol, and percentage of total samples represented by symbol 40 o 0 0310 o 24' 2f— 11' 6 J | [—— l o 100 200 300 400 500 MILES l [L 'll ll [1 T II I *1 g o 100 200 300 400 500 KILOMETERS 30— | | l l | >' I U z | LLI g 20— | ”-' I I *- l I l 10— FIGURE 33. — Yttrium content of Spanish moss. 0 I l I I I I E’. 8 8 S E 8 8 V YTTRIUM, IN PARTS PER MILLION IN ASH Geometric mean, 25 Geometric deviation, 1.78 Number of samples and analyses, 123 E40 FREQUENCY STATISTICAL STUDIES IN FIELD GEOCHEMISTRY f"/ .0 39 f/" I Fr" r—--~"7 ,/ . / J... v . g \ g 42/ we I __.’—--r’ ( I . . ,r—r \ n l’ L,\ I {I I \ \\ O ’ 1. l 1 I \ 0‘ - d\\rLAI\/_,.I»"\L > I \ \ - N I ‘ ‘ <"7O I 10 Qi | \ 00000 a I "’ 2" I . °. - G a T“ I 00 lg . l G C; O O 00 \ O O O I 00 O " 0' O \9 (19—4 # ‘ \ I 99 .09 —--' 0 w—"" \. 0L.-— 0 . 0 \ 00 ,O G o I - z, ,_ ‘ ¢ 9 ' ()0 5. O O O (Ff—x . O ‘ 1 O O 00 . 0 f. e O A- -0 ' o f \\ o O o a o O. \ r \ , ° - C O r O O\ \ 00 O‘/ I “o Go \\ O .l/ ‘3 O 1 I/ ‘ o \ \ ‘va ~ 0 K \ ./, ‘ “we" Symbol, and percentage of total samples represented by symbol 40 I 8 I RI 3; El 3 o {0:0‘0‘ o I I o 100 200 300 400 500 MILES ll—I }'1|1IIIIII|4‘ |_ 1 0 100 200 300 400 500 KILOMETERS 30— I I I I I I 20— A 10* FIGURE 34. — Zinc content of Spanish moss. 0 1 l I I I I 1 I o D D D D D O D D D LO 6 D O O D a O c: D '- N m m " 2 2‘3 8 8 S ZINC, IN PARTS PER MILLION IN ASH Geometric mean, 1,235 Geometric deviation, 1.83 Number of samples and analyses, 123 AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E41 Symbol, and percentage of total samples represented by symbol 0 100 200 300 400 500 MILES I_r I I I I I 30 o o I o Iol . I I I I I I I m Q I ‘3 I :3 N 0 100 200 300 400 500 KILOMETERS I I I I I I I I 20— I i I I I I FREQUENCY FIGURE 35. — Zirconium content of Spanish moss. I | I l I | E 8 8 8 8 8 8 V .— -— N In In N ZIRCONIUM, IN PARTS PER MILLION IN ASH Geometric mean, 110 Geometric deviation, 2.12 Number of samples and analyses, 123 E42 DISCUSSION OF RESULTS Although most species of flowering plants are believed to absorb only minor amounts of nutritive elements through their leaves (Fried and Broeshart, 1967, p. 119) and therefore must depend largely on absorption by roots, Spanish moss must obtain all nutritive elements by means of leaf and stem absorption. Some gases, including carbon dioxide, oxygen, sulfur dioxide (Berger, 1965, p. 232), chlo- rine (Berger, 1965, p. 248), and ammonia (Hutch- inson and others, 1972), that come in contact with ordinary plant leaves are absorbed directly and are used in metabolic processes. These gases are thought to be similarly utilized by Spanish moss, although such utilization has not been proved experimentally. Particulate matter and materials in solution that become lodged on Spanish moss plants are analogous in function to the soil in which ordinary rooted plants grow. Analyses of Spanish moss samples re- veal not only the chemical elements in the plant tissue but also those that are lodged on the plant surfaces, and the concentrations of elements in the two groups cannot be differentiated with certainty. Metabolic processes probably have only a minor effect on the concentrations of most elements dis- cussed in this report. This effect may be largely the immobilization of certain nutritive elements in tissues of the plant, so loss of these elements through leaching by rainfall is reduced. However, both es- sential and nonessential elements may be held by chemical bonding to organic molecules in the plant, as was suggested by Goodman and Roberts (1971, p. 289). Therefore, the action of Spanish moss in ac- cumulating airborne materials probably consists of more than simple atmospheric filtration. Dissemination of Spanish moss occurs most com- monly by wind transport of vegetative fragments of the plant rather than by seed dispersion. Most masses of Spanish moss that were collected as samples probably originated from such fragments, and for this and other reasons they cannot be traced back to their origins or ages as individual plants. A sample of Spanish moss, a perennial plant, is a composite of tissues of different ages, and there is no practical method of determining the age of the oldest tissue that is included. The increase in mass of a plant of this type tends to be exponential rather than linear; therefore, the samples are weighted to a substantial but unknown degree in favor of tissues of more recent growth. Although the effects of this loading cannot be measured readily, they must be considered in evaluating the role of this plant as an STATISTICAL STUDIES IN FIELD GEOCHEMISTRY For these reasons, analyses of Spanish moss samples probably reflect most strongly the atmospheric bur— dens of recent months or years, although some por- tion of a sample may be 10 or more years old. In order to determine if the morphological and physiological features which adapt Spanish moss to an epiphytic habitat result in accumulations of kinds and amounts of elements different from the ac— cumulations in ordinary soil-rooted (terrestrial) plants, comparisons can be made by using data that are now available. A study of the chemical composi- tion of soil samples from 912 sites throughout the conterminous United States (Shacklette, Hamilton, Boerngen, and Bowles, 1971; Shacklette, Boerngen, and Turner, 1971) did not report _analyses of the plants that were sampled concurrently with the soil samples. Summary analytical data for these plant samples and for Spanish moss samples are given in table 3. The terrestrial plants that were sampled included a wide variety of life forms—trees, shrubs, broad— leafed herbs, and grasses—as well as different plant parts. Element concentrations in these plant samples ranged widely; moreover, the suite of samples was heavily weighted in favor of woody plants (trees and shrubs). For these reasons, we believe that in comparing element concentrations in this heterogen- eous group (terrestrial plants) with those of an entirely homogeneous group (Spanish moss), geo— metric mean values in Spanish moss are better compared with the central two-thirds ranges of the terrestrial plant analyses than with the geometric means. The average percentages of ash obtained by burn- ing dry plants of the two groups are very similar. However, the typical concentrations of aluminum, cobalt, chromium, gallium, iron, sodium, lead, titan— ium, vanadium, and zirconium in ash of Spanish moss samples exceed the upper limit of the expected 67- percent range in ash of the soil-rooted plants. The extremely high concentrations of lead in Spanish moss undoubtedly reflect the fact that most samples were collected at sites near highways. Some samples of Spanish moss had a greater concentration of one or more of the elements chromium, copper, gallium, lead, vanadium, and zirconium than was found in any sample of soil-rooted plants. One sample of Spanish moss contained 15 ppm germanium in ash, an element that is very rarely reported to occur in plants. No element concentrations in Spanish moss were unusually low, although concentrations of the mac— ronutrient elements potassium and phosphorus are integrator of time increments of airborne materials. I near the lower end of the expected 67-percent range AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS E43 TABLE 3. —-— Chemical composition of Spanish moss samples and samples of soil-rooted plants from the conte'rminous United States [Geometric means and ranges of elements reported as parts per million in ash. GM, geometric mean; GD, geometric deviation; ratio, number of samples in which the element was detected to total number of samples; ...... , no data available] Spanish moss Soil-routed plants Element, Central and 8511 GM GD Range Ratio GM GD Range Ratio 67-percent range 5.42 <1—20 16 123 ............ <1—70 88 1,125 2.21 1,500—>100.000 120 123 6,500 3.52 <150—>100.000 1,109 1,117 ,000 1.55 1—2 54 116 .............................. 1.34 70—300 123 . 123 240 2.10 <30—3,000 1,135 1,150 110—500 2.32 50—2,000 123 123 390 3.75 2—70,000 1,151 1,151 loo—1,500 ...... 12 <2—100 8 1,153 ........................ <20—30 4 . 1,125 .56 18,000—320,000 120,000 2.69 1,600—430,000 988 988 45,000—320,000 1.65 .8—27 .............................. ...... 200—300 1,000—1,500 2 1,117 1.83 (7—70 98 123 94 5.83 (5—300 232 1,122 .16—5.5 1.73 7—1,500 123 123 9 6 2.85 <1—700 1,096 1,139 3.4—27 2.01 20—2,000 123 123 100 1.98 5—1.500 1,153 1,153 51—200 1.87 1,000—50,000 123 123 3,600 2.52 100-70,000 1,153 1,153 1,400-9,100 1.71 (7—50 95 114 20 9.83 <3—30 150 1,105 .02—2.0 ...... 115 1 123 ...... <.5—.7 8 116 1.74 6,000—200,000 123 123 130,000 1 82 8,800—450,000 1,006 1.006 71,000—240,000 3.80 (50—150 36 123 ............ <50—1,500 46 1,125 ...... 1.53 8—90 122 122 .............................. 1.73 5,000—100,000 1232123 30,000 2.05 ,.1,000—>100,000 1,120: 1,153 15,000—62,000 2.60 ZOO—10,000 1231123 1,100 4.38 30—100,000 1.153 :1,153 250—4,800 1.80 (7—20 29 2123 4.2 3.94 <2—500 459 : 1.124 1.1—17 4.05 1,100—240,000 1232123 4,600 4.00 400—360,000 277 :277 1,200—18.000 ...... 15—20 2 1123 130 111.120 1.46 1150 4:36 32 6.21 <150—1,500 11 :50 5.2-200 1.74 7—200 123:123 16 2.84 (5—500 1,028: 1,139 5.6—45 1.55 2,400—48,000 1231123 20,000 2.27 1,600—400,000 9912991 8,800—45,000 3.16 70—50,000 1232123 86 6.17 <10—15,000 980 : 1,145 14—530 1.89 (5—20 68 : 100 ............ (7—30 43 : 1,153 ...... ...... 11 42123 1.80 20—30 6:123 ............ <10—70 22 : 1,152 ...... 1.85 loo—7,000 123:123 880 3.78 <10—20,000 1,145 : 1,152 230—3,300 1.96 150—7,000 123:123 260 3.49 (7—15.000 1,122: 1,149 75—910 1.73 15—500 1231123 11 4.15 <7—300 694: 1,123 2.7—46 1.78 (20—150 942123 ............ (IO—700 161 :1,128 ...... 1.81 <2—30 91 :123 ............ <1—70 101 : 1,109 ...... 1.83 140—4,600 1231123 450 2.73 <25-5,800 642 Z 643 170—1235 2.12 <20—700 121:123 14 3.45 (20—500 470:1,152 4.1—48 Ash, percent?" 4.6 1.46 22—34 1231123 5 3 1.98 .43—65 1,152:1,152 2.7—11 1All analyses were the same in samples in which the element was detected. ”Analyses reported on dry weight basis. for soil-rooted plants and concentrations of mag— nesium and calcium are nearer the center of the range. Cerium, potassium, and niobium were the only elements that occurred at lower levels in a Spanish moss sample than in any sample of soil- rooted plants represented in table 3. The capability of Spanish moss to serve as a long- term integrator of local airborne element loads is suggested by data in figure 3. In most plant ash, tin generally occurs in concentrations below the limit of analytical detection, and tin was quantita- tively recorded in only six of the 123 Spanish moss samples. Four of the six samples were collected in the Houston-Galveston area in Texas. Because all 123 samples were analyzed in a sequence randomized with respect to geographic location, the probability of four samples in a localized area containing detect- able concentrations of tin by chance alone is quite small. We hypothesize, therefore, that the atmo— sphere in this area carries unusually high concen- trations of tin. This hypothesis seems confirmed by the fact that a tin smelter, the only one in the United States, is located at Texas City, Tex. (Lewis, 1971, p. 1066), across the bay from Galveston. In order to characterize possible differences in the local airborne element load, analyses of five samples from each of four kinds of areas, classified according to their principal economic uses, were examined, and the elements that occurred in concentrations greater than the geometric mean are shown in table 4. More complete descriptions of the sample localities are given in table 1. Table 4 shows that high concentrations of arsenic, cadmium, chromium, cobalt, copper, lead, nickel, and vanadium in Spanish moss samples occur in areas where high rates of vehicular and industrial emissions are expected. The influence of ocean spray on sodium concen- trations in the samples is indicated in figure 28. All samples (except one from Texas) that have greater—than-average sodium concentrations are from locations where airborne saltwater is expected E44 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY TABLE 4. —E'lemcnts found in concentrations greater than average in Spanish moss samples from four kinds of sites Locality Locality ‘ County or Elements (fig- 3) State parish Area Rural sites (agricultural and recreational uses) 32 Georgia Irwin . ._.Al, Ba, Fe, Mn, Sc, Li, and Zr. 35 do Glynn St. Simons Island Ii. Li. Mg, P, K, Na, and V. 59_ Tera Karnes San Antonio River.. Al, Cr, Ga, Fe, Sc, Ti, and Zr. 72 Florida Dade ..... Paradise Key ............ ....Ca, Mg, and Na. 294. Arkan as Lafayette Co and P. Sites near highways with heavy traffic 213 ............ North Carolina... Robeson ........... Interstate 95 ............ Al, As, Cd, Cr, Co, Cu. Ga, Fe, Pb, Li, Ni. K, Sc, Ti, V, Yb, Y, and Zr, ...Hamilton.. ..Interstate 75. ...Cd, Ca, Pb, Mg, Mn, and Zn. Mobile .............. Interstate 10 .................... Cd, Co. Cr, Cu, Ga, Fe, Pb. Li, Mg. Mn, M0, Ni, K, Sc. Na, Sr, Ti. and V. Lowndes ......................... Interstate 65 ............................... A], B, Ca, Cr, Co, Ga, Fe, Li, Mo, Ni, Sc, Sr, Ti, V, Yb, and Y. Harri Interstate 10 ............................... B'l. Cd, Ca, Cr, Co. Cu, Fe, Pb, Mg, Mn, Ni, Sr, and Zn. Urban sites (residential and business uses) 54.. Florida ...Dixie ...... .Cross City ..... ....Ca. .do... Palm Be West Palm Ca, Na, and Sr. "Louisiana... East Baton Rouge. ..Florida.... 1 ,.Baton Rouge. ..Bradenton.. ...... Columbia Ba, ll, Cd. Co, Fe, Mg. Mn, Ni, K. and Sr. Ca, Li, Mg, P, Na, and Sr. ...Ba, Cd, Co, Fe, Mn, K, Sr, and Ti. Urban sites (industrial and other uses) ...Limona industrial park Charleston ..... .. anama City. Beaumont As, B, Cd, Cu, Fe, Mo, Ni, P, Na. V, and Zn. Cd, Cr, Co, Cu, Mg, Ni. Na, and V. Al, Cd, Cr, Cu, Fe, Pb, Mg, Na, V, and Zn. Al, Ba, Cd, Ca, Cr, Co, Cu, Ga, Fe, La, Pb, Ni, Sc. Sr, and Ti. 287 ............ Mississippi ...................... Adams ............................. Near Natchez .............................. Ba, Cd, Cr, Co, Fe, Pb, Mn. and K. to be present at times, and no sample that contains less—than—average amounts of sodium is from such a location. Elements that are common constituents of soil dust, such as aluminum, calcium, magnesium, and iron, are found in samples from all areas. Multivariate statistical analysis methods are being used to provide further interpretations of element distribution patterns in these Spanish moss samples, and the results of this study are planned to be presented in a subsequent paper. SUMMARY AND CONCLUSIONS 1. Airborne chemical elements, which are ac- cumulated by many species of plants, may contribute to the nutrition of the plant or may produce toxic effects. These elements are carried as gases, solutes, or particulate matter and may be actively or pas- sively accumulated by the plant. 2. The concentration of airborne elements in the plant is thought to be determined by the concentra- tions present in air or airborne materials, the in- herent ability of the plant to absorb the elements, the ratio of plant surface to total mass, and the length of time that the plant is exposed to the air. 3. Perennial plants which have a high surface— to-mass ratio and which have no direct connection to the soil by means of a conductive system are effective integrators of airborne materials over time. Analyses of lichens, mosses, and Spanish moss have been used to estimate the degree of atmospheric contamination. 4. Elemental analyses of Spanish moss samples collected from rural, residential, highway, and in- dustrial locations reflect significant differences in concentrations of metals. Samples from industrial and highway locations can be characterized as con- taining greater-than-average amounts of arsenic, cadmium, chromium, cobalt, copper, lead, nickel, and vanadium. The high levels of lead found in some samples from highway locations are especially note- worthy. Of 123 samples analyzed, only six were found to contain tin; four of these grew in the Houston-Galveston area and are believed to reflect the presence, in the area, of a tin smelter, the only one in the United States. 5. Many samples from sites near the seashore contained greater-than-average amounts of sodium that is thought to have been derived from ocean spray. Samples from rural locations commonly con- tain low concentrations of the metals usually as- sociated with urban or industrial activities. 6. Results of this study indicate that elemental analysis of Spanish moss can be used as an econom- ical and rapid method of estimating the kind and relative concentration of airborne chemical elements among locations over a period of months or years. AIRBORNE CHEMICAL ELEMENTS IN SPANISH MOSS REFERENCES CITED Aso, K., 1909, Konnen Bromeliaceen durch die Schuppen der Bl'atter Salze aufnehmen? [Can plants in the Brome- liaceae family absorb salts through the leaf scales?]: Flora, v. 100, p. 447—450. Benzing, D. H., and Renfrow, A., 1971, The nutrient status of populations in south Florida, part 1 of The biology of the epiphytic bromeliad Tillmzdsia circiazata Schlect.: Am. Jour. Botany, v. 58, no. 9, p. 867—873. Berger, K. C., 1965, Introductory soils: New York, Macmil- lan Co., 371 p. Billings, F. H., 1904, A study of Tillandsia usneoides: Bot. Gaz., v. 38, p. 99—121. Brewer, R. F., 1966, Fluorine, in Chapman, H. D., ed., Diag- nostic criteria for plants and soils: California Univ., Div. Agr. Sci., p. 180-196. Chilean Iodine Educational Bureau, 1956, Geochemistry of iodine: London, The Shenval Press, 150 p. Cohen, A. 0., Jr., 1961, Tables for maximum likelihood esti- mates— Singly truncated and singly censored samples: Technometrics, v. 3, no. 4, p. 535—541. Egnér, H., 1965, Die Bedeutung der Schwefelverbindungen in der Luft fiir die Bodenfruchtbarkeit [The importance to soil fertility of sulfur compounds in the atmosphere]: Agrochimica, v. 9, no. 2, p. 133—143. European Congress on the Influence of Air Pollution 0n Plants and Animals, 1969, Air pollution: European Cong. on Influence of Air Pollution on Plants and Animals, 1st, Wageningen, The Netherlands, 1968, Proc., 415 p. Fried, Maurice, and Broeshart, Hans, 1967, The soil-plant system in relation to inorganic nutrition: New York and London, Academic Press, 358 p. Garber, K., 1968, Uber den Fluorgehalt der Pflanzen [The fluorine content of plants], in Fluor-Wirkungen—For- schungsergebnisse bei Pflanze und Tier [Fluorine effects —Results of research with plants and animals]: Wies- baden, Franz Steiner Verlag, Inv. Rept. 14, p. 42—48. Garber, K., Guderian, R., and Stratmann, H., 1968, Unter- suchungen iiber die Aufnahme von Fluor aus dem Boden durch Pflanzen [Investigations on the uptake of fluorine from the soil by plants], in Fluor-Wirkungen— For- schungsergebnisse bei Pflanze und Tier [Fluorine efl‘ects —-results of research with plants and animals]: Wies- baden, Franz Steiner Verlag, Inv. Rept. 14, p. 28—41. Garth, R. E., 1964, The ecology of Spanish moss (Tillandsia usncoides) —- Its growth and distribution: Ecology, v. 45, no. 3, p. 470—481. Goodman, G. T., and Roberts, T. M., 1971, Plants and soils as indicators of metals in the air: Nature, v. 231, no. 5301, p. 287—292. Holley, W. D., 1965, The CO: story, in Ball, Vic, ed., The Ball Red Book [11th ed.]: Chicago, Geo. J. Ball, Inc., p. 94—98. Hutchinson, G. L., Millington, R. J., and Peters, D. B., 1972, Atmospheric ammonia ——- absorption by plant leaves: Sci- ence, v. 175, no. 4023, p. 771—772. Ingham, G., 1950, The mineral content of air and rain and its importance to agriculture: Jour. Agr. Sci., v. 40, p. 55—61. Jaakkola, Timo, Takahashi, Hiroshi, and Miettinen, J. K., 1971, Cadmium content in sea water, bottom sediment, fish, lichen and elk in Finland: in Trace elements in rela- tion to cardiovascular disease: World Health Organiza- E45 tion Mtg. Rept., Geneva, Switzerland, Feb. 8—13, 1971, 16 p. Lewis, J. R., 1971, Tin: U.S. Bur. Mines Minerals Yearbook 1969, v. 1—2, p. 1065—1080. LeBlanc, Fabius, 1969, Epiphytes and air pollution, in Air pollution: European Cong. on Influence of Air Pollution on Plants and Animals, 1st, Wageningen, The Nether- lands, 1968, Proc., p. 211—221. MacIntire, W. H., Hardin, L. J., and Hester, Winnifred, 1952, Measurement of atmospheric fluorine — Analyses of rain waters and Spanish moss exposures: Indus. Eng. Chem- istry, v. 44, no. 6, p. 1365—1370. Malo, B. A., and Purvis, E. R., 1964, Soil absorption of atmo- spheric ammonia: Soil Sci., v. 97, p. 242-247. Martinez, J. D., Nathany, Madhusudan, and Dharmarajan, Venkatram, 1971, Spanish moss, a sensor for lead: Na- ture, v. 233, no. 5321, p. 564-665. Miesch, A. T., 1967, Methods of computation for estimating geochemical abundance: U.S. Geol. Survey Prof. Paper, 574—3, 15 p. Myers, A. T., Havens, R. G., and Dunton, P. J., 1961, A spectrochemical method for the semiquantitative analysis of rocks, minerals, and ores: U.S. Geol. Survey Bull. 1084—1, p. 207—229. Nylander, M. W., 1866, Les Lichens du Jardin du Luxem- bourg [The lichens of the Luxembourg Garden]: Soc. Bot. France, v. 13, p. 364-372. Osburn, W. 8., Jr., 1963, The dynamics of fallout distribution in a Colorado alpine tundra snow accumulation ecosys- tem, in Schultz, Vincent, and Klement, A. W., Jr., eds., Radioecology — National symposium on radioecology, lst, Colorado, 1961, Proc.: New York, Reinhold Publishing Corp., p. 51—71. Riehm, H., 1965, Pflanzenn'ahrstofl'e in den atmospharischen Niederschlagen unter besonderer Beriicksichtigung des Schwefels [Plant nutrients in atmospheric precipitation, with particular reference to sulfur], in L0 zolfo in agricoltura—Atti del V Simposio Internazionale di Agro- chimica, Palermo, 1964: Inst. di Chimica Agraria, Col- lana della rivista “Agrochimica,” 7, p. 453—472. Riihling, Ake, 1969, Tungmetallfororeningar inom Oskar- shamnsomradet [Heavy metal pollution within the limits of the Oskarshamn area]: Forskningsrapport, Eko- logiska tungmetallundersiikningar, Avd. for ekologisk botanik, Lunds Universitet, 11 p. fl, 1971, Tungmetallfiiroreningar inom Stor-Stock- holmsomradet [Heavy metal pollution within the limits of the Greater Stockholm area]: Rapport nr. 25 fran ekologiska tungmetallundersékningar, Avd. for ekologisk botanik, Lunds Universitet, 34 p. Ruhling, Ake, and Tyler, Germund, 1968, An ecological ap- proach to the lead problem: Botaniska Notiser, v. 121, p. 321—342. 1969, Ecology of heavy metals —a regional and his- torical study: Botaniska Notiser, v. 122, p. 248—259. 1971, Regional differences in the deposition of heavy metals over Scandinavia: Jour. Appl. Ecology, v. 8, p. 497-507. Shacklette, H. T., 1972, Cadmium in plants: U.S. Geol. Survey Bull. 1314—G, 28 p. Shacklette, H. T., and Cuthbert, M. E., 1967, Iodine content of plant groups as influenced by variation in rock and E46 soil type, in Cannon, H. L., and Davidson, D. F., eds., Relation of geology and trace elements to nutrition—A symposium, New York, 1963: Geol. Soc. America Spec. Paper 90, p. 31—46. Shacklette, H. T., Boerngen, J. G., and Turner, R. L., 1971, Mercury in the environment— Surficial materials of the conterminous United States: US. Geol. Survey Circ. 644, 5 p. Shacklette, H. T., Hamilton, J. C., Boerngen, J. G., and Bowles, J. M., 1971, Elemental composition of surficial materials in the conterminous United States: US. Geol. Survey Prof. Paper 574—D, 71 p. Shacklette, H. T., Sauer, H. I., and Miesch, A. T., 1970, Geo— chemical environments and cardiovascular mortality rates in Georgia: US. Geol. Survey Prof. Paper 574—0, 39 p. Sloover, Jacques De, and LeBlanc, Fabius, 1968, Mapping of atmospheric pollution on the basis of lichen sensitivity, in Misra, R., and Gopal, B., eds., Symposium on recent advances in tropical ecology, 1968, Proc.: Varanasi—5, India, Internat. Soc. Tropical Ecology, pt. 1, p. 42—56. STATISTICAL STUDIES IN FIELD GEOCHEMISTRY Sower, F. B., Eicher, R. N., and Selner, G. I., 1971, The STATPAC system: [U.S.] Geol. Survey Computer Contr. 11, 36 p. Watson, D. 0., Hanson, W. C., Davis, J. J., and Rickard, W. H., 1966, Radionuclides in terrestrial and freshwater biota, chap. 42 in Wilimovsky, N. J., and Wolfe, J. N., eds., Environment of the Cape Thompson region, Alaska: US. Atomic Energy Comm., p. 1165—1200; avail— able only from U.S. Dept. Commerce, Natl. Tech. Inf. Service, Springfield, Va. 22151, as Rept. PNE—481. Wherry, E. T.,. and Buchanan, Ruth, 1926, Composition of the ash of Spanish-moss: Ecology, v. 7, no. 3, p. 303—306. Wherry, E. T., and Capen, R. G., 1928, Mineral constituents of Spanish-moss and Ballmoss: Ecology, v. 9, no. 4, p. 501—504. W6hlbier, W., 1968, Zusammenfassung und Ausblick [Sum- mary and outlook], in Fluor-Wirkungen—Forschung- sergebnisse bei Pflanze und Tier [Fluorine effects—re- sults of research with plants and animals]: Wiesbaden, Franz Steiner Verlag, Inv. Rept. 14, p. 142—147. ‘5’ U.S. GOVERNMENT PRINTING OFFICE: 1973 07543—576/8 :33“: 7 DAV E’ 76’ (574+— Background Geochemistry of Some Rocks, Soils, Plants, and Vegetables in the Conterminous United States GEOLOGICAL SURVEY PROFESSIONAL PAPER 574—F Ir :‘z'w "I “3",“: r we“ V Lara- ! LiurCEfl'il’ I Eiiii‘4'{?I§Z="{ 0? CU 3’0“?“5ifi JUL 2 1 1975 U.S.S.Do TV VT 7 Background Geochemistry Of Some Rocks, SOils, Plants, and Vegetables in the Conterminous United States By JON J. CONNOR and HANSFORD T. SHACKLETTE With sections on FIELD STUDIES By RICHARD J. EBENS, JAMES A. ERDMAN, A. T. MIESCH, RONALD R. TIDBALL, and HARRY A. TOURTELOT STATISTICAL STUDIES IN FIELD GEOCHEMISTRY GEOLOGICAL SURVEY PROFESSIONAL PAPER 574-F Geochemical summaries for 147 landscape units sampled in 25 field studies UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON : 1975 UNITED STATES DEPARTMENT OF THE INTERIOR ROGERS C. B. MORTON, Secretary GEOLOGICAL SURVEY V. E. McKelvey, Director Library of Congress Cataloging in Publication Data Connor, Jon J. Background geochemistry of some rocks, soils, plants, and vegetables in the conterminous United States. (Statistical studies in field geochemistry) (Geological Survey Professional Paper 574—F) Includes bibliographical references. Supt. of Docs. no.: I 19.162574—F 1. Geochemistry—United States. I. Shacklette, Hansford T., joint author. II. Ebens, Richard J. 111. Title. IV. Series. V. Series: United States Geological Survey Professional Paper S74-F. QE515.C73 551.9 75—619124 For sale by the Superintendent of Documents, US. Government Printing Office Washington, DC. 20402 Stock Number 024—001-02684—8 CONTENTS Page Abstract ............................................................................................................................................... F 1 Introduction ........................................................................................................................................ 1 Methods of study. ..... 2 Objectives .......................... 2 Collection procedures ............. 2 Descriptions of field studies ........................................................................................................ 3 Methods of analysis ............................................................................................................................ 10 Geochemical summaries ................. 12 Organization and use of data... ..... 12 Concluding remarks ................ 16 References cited ................................................................................................................................... 17 ILLUSTRATIONS Page FIGURE 1. Map showing location of study areas where data were obtained .......................................................................................................... F4 2. Map showing vegetation-type areas in Missouri, and location of quadrangles where plants and soils were sampled for Study Nos. 17, 20, and 24 ............................................................................................ . . 5 3. Graphs of 1' as a function of number of analyses and the geometric deviation ................................................................................... 15 TABLES Page TABLE 1. Analytical methods used .......................................................................................................................................................................... F11 2. Elements commonly looked for, but rarely or never detected, by multi—element spectrographic analysis, and their approximate lower limits of determination ....................................................................................................... 12 3. Percentage of ash obtained by burning dry material of cultivated plants ............................................................................................. 13 4. Percentage of ash obtained by burning dry material of native plant species ......................................................................................... 14 5—53. Concentration of certain elements in rocks, unconsolidated geologic deposits, soils, and plants: 5. Aluminum ........................................................................................................................................ 21 6. Antimony (in soils only) ..... 25 7. Arsenic ...................................................................................................................... 26 8. .................................................................................... 28 9. Beryllium .................................................................................................................................................................................... 33 10. Bismuth (in soils only) ............................................................................................................................................................... 35 ll. Boron ........................................................................................................................... 35 12. Cadmium ..................................................................................................................... 40 13. Calcium ................ 43 14. Carbon (carbonate) .. ............................................. 47 15. Carbon (organic).. .................................................................. 49 16. Cerium ........................................................................................................................................................................................ 50 17. Chromium .................................................................................................................................................................................. 52 18. Cobalt .......................................................................................................................................................................................... 57 IV CONTENTS TABLE 5—53. Concentration of certain elements—Continued 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. Copper ................................................................................................................................................................................. Fluorine .. . Gallium .................................................................................................................. Gold (in rocks only) .............................................................................................. Iodine ................................................................................................................................................................................... Iron ...................................................................................................................................................................................... Lanthanum Lead ..................................................................................................................................................................................... Lithium ................................................................................................................................................................................ Magnesium Manganese .................................................................................... - ............................... Mercury ........................................................................................................................ Molybdenum. ............................................................................................................................................. Neodymium ............................................................................................................................................. Nickel ......... Niobium ........................................................................................................................... Phosphorus ..................................................................................................................... Potassium ................................. Rubidium (in plant ash only).... Scandium ................................. Selenium.... ........................................................ Silicon.. ........................................................ Silver ....... Sodium ............................................................................................... Strontium ........................................................................................... Thorium (in soils only). ................. Tin ........... Titanium.... Tungsten .............................................................................................................................................................................. Uranium (in soils only) ....................................................................................................................................................... Vanadium ...................... Ytterbium... Yttrium ...... Zinc ............ Zirconium ............................................................................................................................................................................ Page F6] 66 69 71 72 73 78 82 87 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY BACKGROUND GEOCHEMISTRY OF SOME ROCKS, SOILS, PLANTS, AND VEGETABLES IN THE CONTERMINOUS UNITED STATES BY JON J. CONNOR, HANSFORD T. SHACKLETTE, and Others ABSTRACT Geochemical summary statistics for 48 elements in natural materials from M7 landscape units have been compiled based on field and laboratory studies since 1958. Each landscape unit is briefly identified as to kind and location, and the expected concentration for one or more elements is given together with factors indicating the degree of observed variation in the study and the degree Of laboratory or “analytical" variation. Also listed are the observed range and the total number of element analyses made in each study. The data on which these summaries are based have three attributes in common: They represent “large-scale" or regional geochemical studies; they represent background or “ordinary" natural geochemical variation; and they were collected according to objective sampling designs. The summaries clearly demonstrate the wide diversity to be expected in elemental properties of landscape units and suggest that published element abundances for broad categories like “soil" or “carbonate rock" may be misleading. INTRODUCTION Increased public concern about real or suspected chemical deterioration of the environment, together with a growing awareness of the role of trace elements in health and nutrition, underscores the need for realistic data on the chemistry, particularly the trace element chemistry, of the natural environment. A moderately voluminous lit- erature exists in which geochemical abundances Of many trace elements are summarized, but it deals largely with rocks; less seems to be known (or at least less has been published) about the distribution Of trace elements in soil, plants, and waters. Some of the standard references on trace element abundances in these materials are Goldschmidt (1954), Rankama and Sahama (1955), Mason (1958), and Turekian and Wedepohl (1961). More recent summaries include Wedepohl (1967—1973), Shacklette, Hamilton, Boerngen, and Bowles (1971) and Durum, Hem, and Heidel (1971). In addition to these, the US. Geological Survey is currently revising Clarke’s (1924) Data of Geochemistry. The revision is being published as separate chapters of US. Geological Survey Professional Paper 440. Nearly a third of the projected chapters have been published to date, including chapters summarizing the composition Of rocks (Parker, 1967) and waters (White, Hem, and Waring, 1963; Livingstone, 1963). Because of its well-known toxicity, mercury in the environment has received as much or more attention than any Other trace metal. Three general references to mercury in the environment are US. Geological Survey (1970), Shacklette, Boerngen, and Turner (1971), and Jenne (1972). General summaries of trace elements in soils and in plant tissue are rare, although Shacklette, Sauer, and Miesch (1970) gave trace element averages for a variety Of both, including foodstuffs, in Georgia. Much of the extant information describing the “natural condition” Of the geochemical environment is of unknown reliability. Generally, many of the data (particularly the older data) on which the published summaries are based were not collected to serve as guides to background geochemistry. Rather, they may have been collected to study such things as specific geochemical pro- cesses, or to delineate mineralized areas, or perhaps to esti- mate the degree of chemical pollution; in short, samples of natural materials tend to be collected for a variety of scien- tific reasons but only rarely for the purpose of describing the ordinary properties Of the natural chemical environ- ment. One long-term US. Geological Survey effort, however, has for its goal precisely this purpose. For more than a decade now, personnel of the US. Geological Survey have been engaged in several field projects in regional geo- chemistry, the primary purpose of which has been to describe the geochemical variation Of broad natural units in the United States. Fundamental tO all these studies has been the conscious attempt to measure the geochemical variation as it occurs in nature. Some Of this work has been published, but most Of it has not. Many studies are curren- tly underway, and we anticipate that such studies will continue. Fl F2 STATISTICAL STUDIES IN FIELD GEOCHEMISTRY One such study merits special mention. The US. Geological Survey has recently completed a geochemical survey of the State of Missouri, many aspects of which are unique to environmental geochemistry. It not only is a survey of a broad, geologically diverse area, but it was undertaken partly in support of active, trace-element related epidemiologic studies sponsored by the University of Missouri. Moreover, we think it is the first study of its kind in which an attempt has been made to characterize rocks, waters, soils, and plants chemically by a unified team approach using (and in part testing) efficient and ob- jective sampling designs. Connor and others (1972) described this work in a preliminary way. Details of the study are available in a series of limited-distribution progress reports (US. Geological Survey, 1972a—f, 1973). The data tabulated in the present report represent the work of many people. Principal investigators are listed with a short description of each study, and authorship is cited for all published data. All unpublished data are pre- liminary. The reader is cautioned that some of the data summaries given here may be subject to minor revision. The sampling designs, data analyses, and geochemical summaries on which this report is based are statistical in nature; extended discussions of these subjects can be found in Miesch (1967a, b, 1972), Connor and others (1972), and Connor and Myers (1973). A proper list of acknowledgments for this report would comprise more than 100 people including computer programmers, specialists in data handling, and assistants in the field, laboratory, and office. Unquestionably, the most important contributors are the chemists, spectro- graphers, and other laboratory personnel who catalogued, prepared, and measured the concentrations of up to 69 elements in more than 8,000 samples of rocks, soils, and plant material over a period of more than 10 years. They are: Lowell Artis, Phillip Aruscavage, J.W. Baker, A.J. Bartel, S.D. Botts, L.A. Bradley, Floyd Brown, Mike Brown, J.W. Budinsky, G.T. Burrow, C.L. Burton, Alice Caemmerer, J.P. Cahill, E.Y. Campbell, G.W. Chloe, Don Cole, E.F. Cooley, N.M. Conklin, W.B. Crandell, Maurice DeValliere, J.I. Dinnin, P.L.D. Elmore, E.J. Fennelly, W.H. Ficklin, J.L. Finley, F.J. Flanagan, L.D. Forshey, I.C. Frost, Johnnie Gardner, J.L. Glenn, W.D. Goss, Frank Grimaldi, J.C. Hamilton, T.F. Harms, J.L. Harris, A.G. Haubert, R.G. Havens, R.H. Heidel, A.W. Helz, M.B. Hinkle, Claude Huffman, Jr., R.L. James, L.B. Jenkins, James Kelsey, Herbert Kirschenbaum, B.W. Lanthorn, L.M. Lee, K.W. Leong, H.H. Lipp, Irving May, B.A. McCall, RF. McGregor, J.B. McHugh, J.D. Mensik, V.M. Merritt, Leung Mei, H.T. Millard, Jr., D.E. Moore, Roosevelt Moore, John Moreland, Wayne Mountjoy, A.T. Myers, H.M. Nakagawa, H.G. Neiman, W.W. Niles, D.R. Norton, Uteana Oda, C.S.E. Papp, L.F. Rader, R.L. Rahill, L.B. Riley, E]. Rowe, J.J. Rowe, V.E. Shaw, G.D. Shipley, Leonard Shapiro, F.O. Simon, H. Smith, M.W. Solt, A.L. Sutton, Jr., D. Taylor, J.A. Thomas, Barbara Tobin, J.E. Troxel, J.H. Turner, R.L. Turner, R.E. Van Loenen, G.H. VanSickle, J.S. Wahlberg, F.N. Ward, E.F. Welsch, Roberta Wilkie, T.L. Yager, R.J. Young, and RA. Zielinski. METHODS OF STUDY OBJECTIVES The summary data listed herein have three common characteristics and only data consistent with these attri- butes have been included in the tables: 1. The data represent large-scale studies, in which an assessment of regional geochemical effects is one of the objects of study. The definition of large-scale or regional is somewhat arbitrary, but all studies listed herein involved a conceptual natural unit whose geographic extension in at least one direction is of the order of 80 km or more. 2. The data represent background concentrations. Data known or suspected to reflect epigenetic mineralization, or pollution and other man-induced effects have been inten- tionally excluded. Because of intimate contact with the atmosphere, plants may be more susceptible to the effects of low-level, broad-scale chemical pollution than are soils and rocks. The geochemistry of cultivated soils may, of course, reflect agricultural practices. Nevertheless, only data believed to be essentially free of unusual geochemical effects have been summarized in these tables. 3. The data were collected according to objective experimental designs in an attempt to insure that unbiased estimates of the “natural” variation were ob- tained. The suit of samples underlying each summary sta- tistic includes varietal samples in approximately the same proportion that these varieties occur in nature. Com— monly, such objectivity has been approached through in- troduction of randomization procedures into selection of the sampling sites. Nearly all of the sample suites were analyzed in a randomized sequence to circumvent any potential effects of systematic laboratory error. COLLECTION PROCEDURES The samples on which these tabulations are based were collected according to two general kinds of sampling design. The more conventional design is one in which a single sample of the material of interest was collected at each of numerous sites spread rather evenly over the area of study. A large number of studies, however, were based on a second, more complicated sample design, one in which an effort was made to quantify the effects of “regional” variation and the factors underlying such variation. Sample collection in these latter studies was based on hierarchical designs in which samples were ”nested” at various geographic scales in order to assess the proportion of geochemical variation exhibited at each scale. GEOCHEMISTRY OF SOME ROCKS, SOILS, AND PLANTS IN THE UNITED STATES F3 Krumbein and Slack (1956) discussed such designs and the requisite mathematics in detail. The particular sample design used for each field study is given under the descrip- tion of that study in the following section. Collection procedures in the field tended to be consistent from study to study. All rock samples were taken from outcrop as single samples of a few kilograms weight. They were collected from atrificial exposures as well as from natural outcrop. These samples were trimmed in the field or laboratory in an attempt to exclude all visible weathering rinds or surface effects. Admittedly, not all weathering effects are necessarily excluded thereby. For the most part, soils were collected at or near the surface as single samples weighing about 1 kg. They were commonly taken from shallow holes dug with spade, geologic pick, or other metal tools. Special collection procedures were used in Studies 16, 18, and 19. Soils collected from deeper parts of the soil horizon were commonly taken from sides of large pits dug specifically for sampling. Attempts were made to take soil that had not been in contact with metal collecting tools. In studies 16, 18, and 19 the samples of soil material were sieved to remove rock particles larger than 2 mm in diameter before the samples were pulverized. In the other soil studies the rocks were removed from the samples by hand sorting; therefore rock particles somewhat larger than 2 mm in diameter may have been pulverized with the soil material of some samples. Samples of native plants were commonly collected from the terminal parts of branches over as much of the plant as practical. In deciduous woody plants, stems of the previous few years’ growth were always collected; leaves were rarely collected. For conifers, leaves (needles) and stems were collected together. No attempt was made to wash or otherwise clean the collected materials prior to analysis. Elemental analysis of the cultivated plant samples was performed on the edible parts of the plant prepared as for eating but uncooked. DESCRIPTIONS OF FIELD STUDIES The locations of the individual field studies on which the data in this report are based are shown in figures 1 and 2 and briefly described below by the principal investigators who provided the data given in tables 5—53. STUDY NO. 1 [Granitic rocks of Precambrian age in the St. Francois Mountains, Missouri] By RICHARD J. EBENS Samples weighing a few kilograms each were collected from outcrops of granite and rhyolite of Precambrian age in the St. Francois Mountains of southeastern Missouri in the fall of 1970. Two samples were collected from each of 15 randomly selected sampling localities in the granites of the area, and a similar suite of samples was collected from the rhyolites of the area. Ten randomly selected samples were split into 2 parts and the entire suite of 70 was placed in a random sequence prior to analysis. Preliminary results of this study are given in U.S. Geological Survey (19726, p. 13). Study conducted by Richard J. Ebens. STUDY NO. 2 [Arkose of the Fountain Formation of Permian and Pennsylvanian age in central Colorado] By A. T. MIESCH Samples weighing a few kilograms each were collected during the summer of 1966 from outcrops of arkose 0f the Fountain Formation of Pennsylvanian and Permian age east of the Front Range in central Colorado. Ten randomly located samples were collected from each of 8 stratigraphic sections of the Fountain Formation. The length of the outcrop belt was divided into four approx- imately equal parts and two sections were randomly located in each part. Twenty samples were split and the total suite of 100 was placed in a random sequence prior to analysis. Study conducted by A. T. Miesch, Jon J. Connor, and Terry Quinlan. STUDY NO. 3 [Sandstone, shale, and carbonate rocks of the Sauk sequence of Precambrian, Cambrian, and Ordovician age on the cralonic part of the Western United States] By A. T. MIESCH Samples weighing a few kilograms each were collected from outcrops of the Sauk sequence of $1055 (1963) in the Western United States during 1962—66. The stratigraphic sequence consists largely of rocks of Cambrian and Early Ordovician age although along the western edge of the study area some rocks of late Precambrian age are included. Throughout the study area the interval typically comprises a basal transgressive sandstone that is typified by the Prospect Mountain Quartzite, Tapeats Sandstone, and Flathead Sandstone, a thin overlying shale typified by the Bright Angel and Pioche Shales, and a thick upper carbonate interval typified by the Pogonip Group. Each of the three lithic parts of this sequence exhibits a variety of formational names in different places, and each lithology was treated as a separate problem although the same sampling design was used for each part. Ten major sampling localities approximately 400 km from each other were distributed over the study area. In each locality, eight stratigraphic sections were located in the following manner. Two sections were located approximately 3 km apart, another pair of two sections (also 3 km apart) was located about 15 km from the first pair; another group of four sections was located in a similar manner 80 km away. Where possible within each section, 2 samples were randomly located in the basal sandstone, 2 more in the shale interval, and 2 in the overlying carbonate rocks for a maximum of 6 samples per section and 48 per sampling locality. In addition to the 10 major sampling localities, 10 minor localities were dispersed over the study area in an attempt to occupy wide gaps between the major sampling STATISTICAL STUDIES IN FIELD GEOCHEMISTRY F4 Ada—5:73 435m 2: .0 3,5,. 3. 3:5: 33 m 1.8.». E mcanamv 4X8 0:. E 35:36 v.3 :2: 3:53 o. 5%: 92: as. :0 2.55:2 .EENEO 82: 55% @523 £53 35; he cot-834'.— u::..:..— .E .2 .E .8 .Nm .5 .8 .mm .8 .mm .«m .8 .mm .8” .NS .on .on .3" .o: .N: .«S .0: .w: . , , . , . _ d _ _ _ _ _ _ _ _ _ . ‘ N N \ \ .NN \ Ioum w wMMhmZOJE 00m 0 I, .a \ $3.: 8? 1/ I...~ . fr I .8 \ /, 1.3 / \Irr Z z (ILK / a L L p .w~\ .