-^m 'A, '>:V i 1 i \ 1 1 MARCH, 1921 MEMOIR 39 CORNELL UNIVERSITY AGRICULTURAL EXPERIMENT STATION THE GENETIC RELATIONS OF PLANT COLORS IN MAIZE R. A. EMERSON ITHACA, NEW YORK PUBLISHED BY THE UNIVERSITY MARCH, 1921 MEMOIR 39 CORNELL UNIVERSITY AGRICULTURAL EXPERIMENT STATION THE GENETIC RELATIONS OF PLANT COLORS IN MAIZE R. A. EMERSON ITHACA, NEW YORK PUBLISHED BY THE UNIVERSITY CONTENTS PAGE Previous investigations 8 Source and description of materials used 9 Purple, type 1 9 Sun red, type II 1 1 Dilute purple, type III ll Dilute sun red, type IV 12 Brown, type V 14 Green, type VI 14 Relation of plant colors to environment 15 Sunlight a factor in color development 16 Moisture in relation to color 18 Temperature in relation to color 19 Soil fertility and color development 21 Rich compared with poor soil 21 Lack of particular nutrient elements 23 Relation of carbohydrates to color 26 Summary 28 Genetic analysis of color types 29 Crosses invohong the factor pairs A a, Bb, PI pi 29 Purple la x green VIc 29 Dilute sun red I\'a x brown V 32 Backcrosses of la x \'Ic and YVa, x V with IVa 34 Behavior of F2 color types in later generations 35 Intercrosses of F2 color types 48 Evidence from aleurone-color and linkage relations 58 Summary of results invohdng A a, Bb, PI pi 64 Crosses involving the multiple allelomorphs B, B"', b^,b 65 Interrelations of sun red I la, weak sun red lib, and dilute sun red IVa 67 Relation of weak purple lb to purple la, dilute purple Illa, and weak sun red lib 69 Crosses involving the multiple allelomorphs R'^, R^, R''3^ r*", r", r-*^'* 73 Green IVg x brown V 74 Intercrosses of F2 color types 86 Purple la x green -ant her ed dilute sun red Ill Summary of results involving the allelomorphic series R^, Ro, R^, r'', ro, r^^ 112 Relation of aleurone factors C c and Prpr to plant color 113 Ex-pression of plant-color and aleurone-color factors 114 Summary 118 Literature cited 120 Appendix (containing tables) 121 THE GENETIC RELATIONS OF PLANT COLORS IN MAIZE THE GENETIC RELATIONS OF PLANT COLORS IN MAIZE^ R. A. Emerson Under the designation " plant colors " are included the colors other than those related to chlorophyll, commonly seen in, but not limited to, such external plant parts of maize as the culm, the staminate inflorescence, the husks, the leaf sheaths, and to some extent the leaf blades. In con- trast to this group are colors and color patterns related to chlorophyll or associated with the pericarp and the cob, the silks, the endosperm, the alem'one. The colors included in the group considered here are due to water-soluble pigments, but the same is true of some of the other color groups named above. Moreover, colors of the chlorophyll group (Lind- strom, 1918) are found in the same plant parts as are the " plant " colors considered in this account. The plant colors as a whole are closely interrelated, but they are closely related also to aleurone colors and to certain of the silk and pericarp colors. It is obvious, therefore, that, while this classification is a more or less natural one, it is based primarily on convenience. The term " genetic relations " in the title to this memoir is to include not merely an account of the genetic analysis of the material at hand by means of hybridization experiments — tho that constitutes the greater part of the paper — but also some consideration of the variations of the several color types induced by or associated with environmental diver- sities. Some httle attention to matters of this kind was made necessary by the fact that presmnably homozygous material exhibited marked variations in extent and intensity of pigmentation when grown under diverse conditions. Since, as will be apparent later, the principal differences between certain of the color types under investigation are apparently quantitative ones, and since the materials at best exhibit no little complex- ity with respect to factorial interrelations of a genetic nature, little progress could have been made without some notion of the response of particular color types to certain factors of the environment. But this study has ' Paper No. 78, Department of Plant Breeding, Cornell University, Ithaca, New York. 7 8 R. A. Emerson been wholly subsidiary to the main purpose, namely, a genotypic analysis of the color types under observation. The writer's realization of the superficial nature of the environmental studies reported in this account in no way weakens his belief in the importance of acquiring an accurate knowledge of the chemistry of the pigments concerned and of instituting fundamental investigations into the physiology of their development — problems that must await the interest and effort of other workers. The studies reported here were begun in a small way in 1909 and have been continued, along with other problems in the genetics of maize, to the present time. The work was conducted at the University of Nebraska and supported by funds of that institution from 1909 to 1914. During 1911 facilities for growing and studying a considerable part of the cultures then in hand were generously afforded the writer by the Bussey Institu- tion at Harvard University. Since 1914 the work has been conducted at Cornell University. During these years, the v/riter has been assisted by a number of persons, among whom he desires to mention particularly Dr. E. W. Lindstrom and Dr. E. G. Anderson. Some data from the records of students associated with the writer are included in this account. The cultures giving these borrowed data are indicated in the tables by initial letters preceding the pedigree numbers, as follows: A = E. G. Anderson, L = E. W. Lind- strom, and S = Sterling H. Emerson. The illustrations are from water-color drawings by C. W. Redwood, Miss Carrie M. Preston, and Miss Bernice M. Branson. PREVIOUS INVESTIGATIONS So far as the writer is aware, little work with the plant colors of maize has been reported previous to this time. Webber (1906) reported the results of studies of the interrelations of aleurone, silk, anther, and glume colors, with the conclusion that color in all these parts is closely correlated but that there are definite breaks in the correlation. This conclusion, in terms of present-day usage, is apparently equivalent to the idea of close linkage with some crossing-over. East and Hayes (1911) identified certain aleurone-color genes, which are shown in the present account to be related to plant colors as well as to aleurone colors, and reported data concerning the inheritance of silk and anther colors. The writer (Emerson, 1918) added another aleurone-color pair also known to Plant Colors in Maize 9 be concerned in plant-color development. He had earlier (191 1) announced some of the plant colois discussed in the present paper and placed on record some evidence as to their genetic behavior. Gcrnert (1912) described types of maize that differ widely in color of anthers, glumes, silks, sheaths, and husks, and reported simple mendelian behavior in Fi and Fo of certain crosses. With this exception, Gernert's extensive investi- gation of plant-color types has not been reported, but the writer has been able, thru an exchange of material, to compare some of Gernert's types with those in his own cultures. SOURCE AND DESCRIPTION OF MATERIALS USED The plant-color types discussed in this paper came in the main from the crossing of two little-known varieties, one of which was obtained at a national corn exposition and the other from an exhibit at a local agri- cultural fair. One of the color types produced by this cross is the same as that of the dent varieties generally grown thi-uout the Corn Belt; a second is not infrequently seen in certain pop, flint, and sweet corn varieties; and a third occurs in the fields of flour corn of certain Indian tribes of the Southwest. One of the color types produced by the cross had no existence, so far as the writer knows, until it appeared in his cultures. Modifications of several of the six color types noted above have been produced by crossing with a color type common in a few varieties of sweet corn and closely related to the type most common in field maize. The principal color types concerned in this account are discussed in some detail in the descriptive notes below. They are: I — Purple II — Sun red III — Dilute purple IV — Dilute sun red V — Brown VI — Green PURPLE, TYPE I Material of the purple type was first obtained as a single ear from a local agricultural fair at Nehawka, Nebraska, in 190G. The varietal name is unknown. The uncrossed stock was a smooth-seeded pop corn 10 R. A. Emerson of medium size. No other stock of purple has been used in the crosses described later in this account, and the writer has never seen this color type in cultivation outside his own cultures. A sample of dent corn of apparently the same color tj^e was seen at a national corn exposition in 1909. A stock of purple was obtained from Dr. Gernert in 1914 but was not used in genetic studies. Another stock of purple was received more recently (1919) from Messrs. Collins and Kempton, the seed having come originally from Bolivia. Seedlings of the purple type are usually indistinguishable from those of types II, III, and IV (described more fully under type IVa, page 12),altho, unUke the other types, they develop some color when grown in darkness. Half-grown plants of type I usually have the lower sheaths prominently colored, in which respect they exceed type II plants in intensity of pig- mentation and are sharply differentiated from types III and IV. At the flowering stage, plants of type la have much purple color in nearly all parts, such as the cuhn, the brace roots, the leaf sheaths, the husks — even the inner ones — the cob, and the staminate inflorescence including the rachis, the spikelets, and the anthers (Plates I, 1, and V, 1). In some cases the color extends over the whole leaf, and it is always seen in the midrib. The purple pigment of type la develops in local darkness, as has been shown by covering various parts of growing plants with several thicknesses of heavy black paper (Plate VIII, 1). The color persists in mature plants with slight fading in the outer parts due to weathering (Plate VII, 1). The pericarp of type la is either colorless, red, or cherry, and the aleurone is either purple, red, or colorless. With red aleurone the anthers are reddish purple, and with cherry pericarp they are usually very dark purple, almost black (Plate I. 2 and 3). A subtype of purple known as weak purple, or type lb, is similar to la but the pigmentation is less intense, particularly in the cuhn and the inner husks (Plate V, 2). In early stages of growth it is often difficult to distinguish lb from Ha. The anthers of lb are usually deep purple, as are those of la, and the pericarp is the same as for la. Another sub- class of purple, Ig, is like la except that the anthers are green (Plate I, 4) and the pericarp is red or colorless, never cherry. The aleurone color is the same as in la. Plant Colors in Maize 11 sun red, type ii Sun red, the not a common color type, is encountered in a few varieties of sweet corn and pop corn. It is always produced in F2 of certain crosses, notably in purple x green. While this type is less highly colored than la, it has such strong color that it is not easily distinguished from the latter in early stages of growth. At the flowering stage, type Ila is sharply differentiated from type la in several respects. The staminate inflorescence of Ila is lighter than that of la, and the anthers are deep pink instead of purple (Plate III, 1). In type Ila, pigmentation of the cuhn, the leaf sheaths, and the husks is limited almost wholly to parts exposed to sunlight, hence the name sun red. The inner husks are therefore without red color, and rarely does much color develop in any but the outer layer of husks (Plate V, 3) not- withstanding the fact that sufficient light penetrates to the inner husks to induce the development of some chlorophyll in them. A tassel inclosed in a black paper bag produces no red color in either glumes or anthers (Plate VIII, 4). Since the color of sun red plants is so largely superficial, it disappears almost wholly from mature plants thru weathering (Plate VII, 2). Sun red plants have either red or colorless, but never cherry, pericarp, and either purple, red, or colorless aleurone. Sun red of type Ilg differs from Ila merely in having green instead of pink anthers. Type lib, known as weak sun red, differs from Ila in the lesser intensity and extent of its pigmentation. Particularly the leaf sheaths and the husks are less highl}^ colored than in type Ila. Often the color of the husks develops in alternate dark and light bars parallel to the upper margins of the overlapping husks (Plate V, 4). Types lib and llg have the same pericarp and aleurone colors as Ila. DILUTE PURPLE, TYPE III The dilute purple type, as well as the sun red, occurs regularly in F2 of purple X green, and most of the dilute purple material in the writer's cultures came originally from this and other cro.sses. It was first observ'ed in the progeny of such crosses in 1909. Recently two stocks of this color type have been received from G. N. Collins, one obtained from the Hopi Indians of southwestern United States and the other from Bolivia. 12 R. A. Emerson Seedlings and young plants of type Ilia show no more color than do those of type IVa, and apparently do not develop color in darkness. As the plants approach the flowering stage, they usually show somewhat more color than do plants of type IVa, particularly at the base of the culm and in the brace roots, and sometimes in the leaf sheaths. The staminate inflorescence is usually, tho not always, somewhat more highly colored than that of type IVa. The anthers are deep purple, like those of type la (Plate II, 1). With red aleurone the anthers are usually reddish purple, and with cherry pericarp they are dark purple, sometimes appearing nearly black (Plate II, 2 and 3). The anther color develops fully in dark- ness, but the glumes are slightly if at all colored when protected from light by black paper bags (Plate VIII, 3). As the plants mature, considerable color develops in the inner husks (Plate VII, 3), on the leaf sheaths, and particularly in the culm even where it is protected from strong Ught by the sheaths. In some cases the culm and the sheaths ultimately become nearly as strongly pigmented as type la, but ordinarily the mature plant is considerably less highly colored than the purple type (Plate VII, 4). The color seen in mature plants develops well in local darkness, in which respect also type Ilia is hke la. Dilute purple differs from purple, there- fore, mainly in a less intense pigmentation and in a delayed development of pigment. The pericarp of type Ilia is either red, cherry, or colorless, and the aleurone is either purple, red, or colorless, just as in type la. There exists a type of plant color which is closely related genetically to type Ilia, but which lacks red or purple color in culm, sheaths, silks, glumes, and anthers and is consequently known as Green, type Illg (Plate II, 4). The aleurone of this type is either purple, red, or colorless, and the pericarp is either red or colorless, never cherry. With respect to aleurone and pericarp, therefore, type Illg is like type Ig. DILUTE SUN RED, TYPE IV Dilute sun red is the commonest color type of maize in cultivation. It is practically the only color type seen in the dent varieties grown in the Corn Belt of the United States, and is common in flint, flour, sweet, and pop corns. Like the sun red and the dilute purple types, it always appears in crosses of purple la with green Vic. The seedlings of type IVa usually show more or less sun red pigment in the coleoptile, the leaf sheath, and the leaf margins. The young Plant Colors in Maize 13 plants ordinarily have considerable color at the base of the lower sheaths, but little or no color except green in other parts except in the margins of the leaves (Plate IX. 1). When the plants are grown on infertile soil, much bright red color develops in all parts exposed to light except the youngest leaves (Plate IX, 2). The seedlings and the very young plants are not ordinarily distinguishable from those of types la, Ila, and Ilia. Some time before the flowering stage, the plants of this type are sharply differentiated from those of types la and Ila, and are usually somewhat less highly colored than those of type Ilia. In normally grown plants, the color is confined mostly to the brace roots, and to the sheaths and the exposed parts of the culm at the base of the plants. Even at the flowering stage almost no color is seen in the upper sheaths or the upper part of the culm, and very Uttle in the husks (Plate VI, 1). The staminate inflores- cense is colored much as is that of the sun red type, tho the glumes are lighter than those of type Ila and the rachis is usually nearly devoid of color. The anthers show more or less pink, as do those of type Ila. There is much variation in the extent and intensity of pigmentation of glumes and anthers (Plate III, 2, 3, and 4), due in part to genetic differ- ences and in part probably to environmental influences. Late in the life of the plant, type IVa usually shows some color in the outer husks and also in exposed parts of the culm. Different strains show considerable variation in this respect (Plate VI, 1 and 2). Due to the slight develop- ment of pigment and because of weathering, the dry parts of mature plants show little red color (Plate VII, 6). Light is essential to the development of color in dilute sun red, IVa, just as in sun red, Ila. The aleurone and pericarp colors of dilute sun red, IVa, are the same as those of sun red, ITa. A wholly green type, that is, one devoid of pigment other than green in the plant parts here under consideration, is closely related genetically to type IVa and is therefore known as type IVg (Plate II, 4). Phenotypi- cally it is the same as type Illg. Just as in case of types Ig, II, Illg, and IVa, the pericarp of IVg is either red or colorless, never cherry, and the aleurone is either purple, red, or colorless. Genotypic diversities in the amount of color are noted for type IVa above. The lightest types of dilute sun red show no color except mere traces of red in the staminate spikelets. This condition is found in most plants of at least two varieties of sweet corn, Black Mexican and Crosby. From these varieties there 14 R. A. Emerson have been isolated strains that lack even this minimum of color. These strains furnished the original stock of type IVg. In no environment as yet encountered has any red or purple plant color developed in type IVg. BROWN. TYPE V The brown type was first seen in 1912, when it occurred in F2 of the cross purple la x green Vic. So far as the writer has been able to learn, brown plant color had not been reported previously, and he is unaware of its existence outside of his own cultures or of stocks grown from them. Seedlings and young plants of type V are wholly green. Before the flowering period is reached, a brown pigment begins to appear in the lower sheaths. At the time of flowering^ the culm, the sheaths, the husks (Plate VI, 3), and the staminate inflorescence (Plate IV, 1 and 2) are brown. The anthers are usually green. The brown color extends to the inner husks, to the culm beneath the leaf sheaths, and to the cob (Plate VII, 5). That light is not essential to the development of brown is shown further by the fact that the color appears under several thick- nesses* of black paper (Plate VIII, 2). It is not uncommon to find traces of purple associated with the brown in the brace roots and at the base of the inner husks (Plate VI, 3). Abnormally developed tassels, not infre- quently seen on plants grown in small pots in the greenhouse, in some cases show a little purple (Plate XI). The aleurone of brown plants is always colorless, except for xenia grains, and the pericarp is either brown, brownish, or colorless, never red nor cherry. Brown pericarp color of type V corresponds to red of types I, II, III, and IV, and brownish to cherry of types I and III. GREEN, TYPE VI The writer's stock of the green type originated from a single ear obtained at a national corn exposition held at Omaha in 1909. The corn was exhibited from southern Missouri, where it is grown locally. It is a large dent variety, rather late in season. Cultures of type Vic, derived from this stock, show no plant color other than green at any stage of development or under any environmental conditions to which they have as yet been subjected (Plates IV, 3, and VI, 4). Plant Colors in Maize 15 • Three subclasses of type VI are recognized. One of these, Via, is Hke Vic in every respect except that a shght amount of brown is sometimes seen in the outer husks and sheaths (Plate VI, 5). The second, VIb. is green except for a slight tinge of brown in the spikelets of the staminate inflorescence (Plate IV, 4). As a rule, the development of brown pigment in Via and VIb is not sufficient to differentiate with certainty the one from the other, or either from Vic. The three subclasses, a, b, and c, are therefore usually classed together as type VI. Both Via and VIb have been isolated from crosses involving Vic. The aleurone of all type VI plants, just as in those of type V, is colorless, except for such color as may be due to xenia. The pericarp of Via and Vic is either brown or colorless, never brownish, while that of VIb is brown, brownish, or color- less, as in the case of type V. With brownish pericarp,, type VIb usually shows unmistakable brown color in the staminate spikelets. RELATION OF PLANT COLORS TO ENVIRONMENT From the preceding descriptive notes and accompanying illustrations, it is clear that many of the differences separating the six major color types and their several subclasses are quantitative. Purple plants are more strongly colored than are sun red or dilute purple plants. Dilute sun red plants have less color than sun red or purple plants. Weak purple plants have less color than purple ones, but more than dilute purple ones, and weak sun reds are intermediate between sun reds and dilute sun reds. Dilute sun red plants vary, from those showing considerable color to those which, except for green, are nearly colorless. Wholly green plants are classed as subgioups of both dilute purple and dilute sun red. The subclasses of type VI differ so little with respect to color that they are ordinarily thrown together as one green type. Heterozygous brown plants are lighter than homozygous ones, and, since more than one factor pair is concerned; there is a fairly smooth gradation from the darkest to the lightest browns. Plants of types Via and VIb, when they show any brown, differ in the parts colored. The color of the staminate inflores- cence, and even of other parts, of purples, dilute purples, browns, and greens of type VIb is darker when the pericarp is cherry or brownish than when it is red, brown, or colorless. The natural intergrading of genetic types in this somewhat complex series is often made still more confusing by the variations accompanying 16 R. A. Emerson environmental diversities. A prominent geneticist, on observing some of the writer's cultures, was led to say that there were no sharply differ- entiating characteristics by which other than an arbitrary classification could be made, and asserted that he could select from a single progeny a series grading from the darkest to the lightest colors. The writer has some doubt that this could have been done, but the instance illustrates well the difficulties that confront one unacquainted with the materials. It is fortunate that some enviromnental influences which increase the difficulty of assorting certain color types make other types stand out more sharply than they otherwise would. Without some notion of these envi- ronmental effects, a genetic analysis of the material would indeed be difficult. SUNLIGHT A FACTOR IN COLOR DEVELOPMENT The relation of sunlight to the development of color has been noted briefly in the descriptions of some of the color types. The effects of sunlight or of local darkness, instead of adding to the confusion of color types, afford a means of sharp differentiation between certain types. So far as is known at present, no color develops in sun red or dilute sun red plants, or in the early stages of growth of dilute purple plants, except under the influence of fairly strong light. In the case of purple and of the later stages of growth of dilute pm-ple, there is no doubt that the color develops more rapidly at first in light than in darkness, but ulti- mately color develops fully, or apparently so, even in local darkness (Plate VIII). The seedlings of purple plants develop some color when germinated and grown in a dark chamber where no part of the plant receives light. There is some, tho very little, evidence that the development of brown pigment of type V is hastened by the influence of light, and what little brown color ever develops in type Via is confined to parts exposed to sunlight (Plate VI,, 5). It would not be surprising to find that the pigments seen in the purple, dilute purple, sun red, and dilute sun red types are the same chemically. In fact they look alike in water solution and apparently react in the same way to simple chemical tests. If they prove to be identical, it would seem to follow that purple and dilute purple plants have some inherent mechanism, perhaps an organic catalyzer, capable of initiating or hasten- ing chemical reactions, and that this mechanism is lacking in sun red Plant Colors in Maize 17 and diluto sun rod plants, in which the same reactions may possibly be brought about thru the action of sunlight. Usually a single thickness of black paper, such as is employed to pro- tect photographic plates from light, is sufficient to prevent the develop- ment of color in sun red plants (Plate VIII, 4). That more intense light is necessary for the production of sun red pigment than for the production of chlorophyll is shown by the ahnost entire absence of red color in all but the outer husks, while even the innermost husks are somewhat green (Plate V, 3). The pigments of purple and brown plants, on the contrary, develop well even when there is too little light for the formation of chloro- phyll (Plate VIII, 1 and 2). That the effect of light on color development is a definitely local one is shown by the sharp fine of demarcation between colored and colorless areas in culms, husks, and sheaths partly exposed and partly protected by overlapping sheaths or husks (Plate V, 3). Even a single piece of wrapping cord tied closely about a young ear, sheath, or culm of a sun red plant is sufficient to prevent the development of color beneath it. Evidently sun red pigment does not diffuse appreciably from the cells in which it forms. It is not meant to suggest by these observations that sunlight has no effect other than a local one on color development. On the contrary, there is evidence that the development of sun red color is influenced bj^ the presence of an abundance of carbohydrates which in turn are dependent on sunlight for their foraiation. A striking example of the relation of sunlight to color development is afforded by the barred pattern seen in the husks of some weak sun red plants (Plate V, 4). The pattern consists of alternate bars of red and green parallel to the upper margin of the overlapping husk next below them. By tracing in pencil on each exposed husk of a rapidly growing ear the margin of the husk overlapping it, it has been ascertained with certainty that the red bars correspond to the areas that are pushed out from under the over- lapping husk between early morning and late afternoon; while the green bars correspond to the areas pushed ,out during the late afternoon and night. Why color develops in only those parts of the husk that receive the sunlight when first exposed to the air, and not in the parts exposed some hours previously, is not known. Another illustration of the effect of sunlight on freshly exposed husks was seen in a very light type of weak sun red (Plate V, 5). Of two ears on the same cuhn, both very lightly 18 R. A. Emerson and about equally colored, the lower had its husks torn apart in the early forenoon so that the fresh inner husks were exposed at once to direct sun- Ught. In a few hours some red color began to show, and in a few days all the newly exposed husks were brilliantly colored, while the undisturbed upper ear remained only slightly colored. Similar results followed in repeated trials, and, in fact, failed only when the atmospheric conditions were such as to cause the newly exposed husks to wither during the first day. It is of interest to note also that similarly treated ears of dilute sun red plants, which rarely show any red color in the outer husks of j^oung ears, failed to develop color when the husks were torn apart, even tho they remained fresh for some daj^s. It is evident from all this, that, with respect to their relation to sunlight, there exists a series of color t^q^es varjdng more or less abruptly from dilute sun red, in which little or no sun red develops in even freshly exposed husks, thru weak sun red, in which color fomis in only freshly exposed husks, and strong sun red, in which much color develops in all exposed parts of the husks but not in parts protected from light, to strong purple, in which, tho sunlight may hasten color development, it is not essential to its foraiation. Tests of the influence on color development of light of different wave lengths have not been uniformly successful. Cramer photographic color screens were placed in partial contact with the uncolored inner husks of sun red plants, and the entrance of light otherwise than thru the screens was prevented by means of strips of black paper. These screens, by cutting out light of certain wave lengths, not only change the quality of Ught passing thru them but lessen the intensitj'' of the light. While the results, therefore, can have Uttle value, it may be of interest to physiolo- gists to note that considerable sun red formed under the orange and the bright red screens, and little or none under the green and the blue screens. MOISTURE IX RELATION TO COLOR It is well known that under field conditions maize does not grow well in wet soil. In such situations, not onl}'^ are the plants small, with their leaves pale green, but they often develop much red pigment. The writer has repeatedly observed that young plants, in flooded parts of fields where the soil had been covered with water for some days, were brilliantly red in all parts except the youngest leaves, while near-by plants on slightly Plant Colors ix Maize 19 higher land showed only the slight red at the base of the culms character- istic of young dilute sun red plants. For a study of the effect of soil moisture on color development under controlled conditions, plants of well-known stocks of purple la, sun red Ila, dilute purple Ilia, dilute sun red IVa, brown V, and green Vic and IVg, were grown in rich soil in earthen jars in the greenhouse during the summer of 1914. When the plants had reached a height of from 10 to 15 centimeters, the jars were separated into three lots — one with dry soil, another with moist soil, and a third with wet soil. The dry-soil lot received only sufficient water to keep the plants growing slowly and not enough to prevent wilting during the hotter part of the day. The moist- soil lot received just sufficient water to insure normal growth. The wet- soil lot was kept constantly in saturated soil with some free water above the soil surface. The test was continued until the plants of all lots reached the flowering stage. The plants in moist soil made the most rapid growth and flowered some- what earlier than the plants of the other lots. Their leaves were of normal green color and they showed the colors characteristic of the several color types. The plants in dry soil were smaller and very dark green. The development of purple, red, and l^rown color was practically the same as with the plants in moist soil. The plants in wet soil grew less rapidly than those in moist soil, but more rapidly than those in dry soil. Their leaves were somewhat lighter green than those of the moist-soil lot, but they showed practically the same amount of purple, red, and brown color. In fact the only differences between the three lots with respect to color at any time during the test were such as might well be related to the stage of development of the plants. All color types show more color in the later stages of growth. The moist-soil lot developed somewhat more rapidly than did the others and for a time showed slightly more color, but ulti- mately all lots had practically the same amount of color. Evidently the reddening of plants iri flooded fields is not due directly to the excess of soil moisture. TEMPERATURE IN RELATION TO COLOR Since moisture is not the direct cause of the reddening of maize plants in flooded fields, tho certainly connected with the phenomenon in some way, it follows that the effect must be produced by some indirect action 20 R. A. Emerson of the excess of water. Wet soils in spring are cold soils, and if the wet areas are of considerable extent the air above them is doubtless somewhat cooler than that above drier soil. It has been frequently observed that young plants which show much color during a cold spring show considerably less in the leaves developed after the weather has become wanner. Young plants of early-planted maize sometimes have more color than plants that are started later. Moreover, full-grown plants from late plantings often develop more color in the cool weather of autumn than similar plants that mature in the warm weather of late smiimer. It seemed important, therefore, to study the effects of various tempera- tures on color development. The same color types and the same stocks — in one test the identical plants — used in the soil-moisture test were grown in the greenhouse under diverse temperatures. Altho both rich and poor soils of diverse water content were used, the comparisons noted here were made between plants in the same kind of soil and with practically the same soil-moisture con- ditions. Two lots were grown during the winter of 1913-14 and two dur- ing the following summer. During the winter, one lot was kept in a warm house at temperatures varying from about 18° to 26° C, and one was kept in a cool house at temperatures varying normally from about 7° to 15° C. but during a part of the test dropping at night to l°or 2° C. Both lots were exposed to the full winter sunlight of the houses. During the sunmier test, one lot was kept as cool as possible by partial shading and free ventilation, the temperatures ranging from about 15° to 40° G. but occasionally exceeding these limits, and the other lot was kept in an unshaded house the ventilators of which were never opened. The night temperatures of the closed house averaged not more than one degree higher than those of the open house, but the maximum day tempera- tures in the closed house varied usually from about 44° to 50° G. and on three consecutive days reached 55° G. This extreme heat killed most of the plants grown in rich soil but did not seriously injure those in poor soil. Of course the relative humidity, as well as the intensity of the light, was materially different for the closed and the open house. As a result of these tests, no final differences in the development of color in any of the color types were observed between the lots grown at the very diverse temperatures. Of course differences were observed at certain times, but they are readily accounted for by the facts that the Plant Colors ix IMaize 21 plants developed less rapidly at both excessively high and excessively low temperatures than at more moderate temperatures, and that color shows less during the early stages of development than during later stages. It may be safely concluded, therefore, that color development in maize is not notably influenced, except perhaps indirectly, by diverse temper- atures. SOIL FERTILITY AND COLOR DEVELOPMENT There is still another way in which it was thought the excess of water might indirectly affect the development of color in maize plants in flooded fields. Not only may nutrient salts be removed in part by an excess of water, but certain of these salts — nitrates — are not fomied normail}^ in very wet soils. Tests were made, therefore, of the relation of soil fertihty to color development. Rich compared with poor soil The same plant-color types as were employed in the soil-moisture and temperature tests were included in these soil-fertility tests. In fact, for one of the tests the same plants were used as in the moisture and temperature studies. One lot of plants was grown in rich soil and a duplicate lot in poor soil. Field soil furnished the basis of both soils. To one lot was added about 50 per cent by measure of thoroly decayed stable manure, and to the other about 50 per cent of clean sand. The effect of soil fertility on color development of certain color types was strikingly apparent from the time the seedlings were two or three weeks old. At this age and for some time later, there was no appreciable difference in color between purples, sun reds, dilute purples, and dilute sun reds. In the rich soil all these color tjqjes had very little red color. There was some color in the coleoptile and the lower leaf sheath, but none in the leaf blades except for a slight amount in their margins. The same color types in poor soil had considerable color in the leaf blades and much color in the leaf sheaths. The plants in rich soil grew rapidly and were dark green, even the lower leaves remaining healthy. The plants in i)oor soil, on the contrary, grew less rapidly and were lighter green, and their lower leaves soon became yellow and died. In all cases the leaf blades became brilliantly red before they died. This is in strong contrast with the condition of the lower leaves of plants m diy, rich soil. When the 22 R- A. Emerson death of the lower leaves is caused by drouth, there is no corresponding development of red color. At the age of six weeks, the plants in rich soil were beginning to show slightly the color differences that in later stages are characteristic of purples, sun reds, dilute purples, and dilute sun reds. In poor soil, on the contrary, no color differences were seen. All the four types were highly colored thruout except for the youngest leaves (Plate IX, 1 and 2). At the flowering period, the plants in rich soil exhibited all the peculiarities of color by which purples, sun reds, dilute purples, and dilute sun reds are normally differentiated. Even in the poor soil something of the same color differences were discernible between the purples and sun reds on the one hand and the dilute purples and dilute sun reds on the other, but it is doubtful whether these two groups could have been separated accurately from a mixed culture. It would have been very difficult also to separate with certainty the purples from the sun reds or the dilute purples from the dilute sun reds, except by differences in anther color and by an examination of the inner husks and other parts protected from sunlight. Differences between the plants in rich and in poor soil were still pronounced in the case of dilute purples and dilute sun reds, but were scarcely discernible in the case of purples and sun reds except that the leaf blades were somewhat more highly colored with poor than with rich soil and that thruout the plants the colors appeared brighter in the former case owing to the less intense green of the poor- soil lots. The seedUngs of both brown and green color types showed no brown nor red color in either the rich or the poor soil. At the age of two months, some brown pigment began to show in the lower sheaths of the brown type, and at the flowering stage the plants had the typical coloration of brown plants. The difference in the development of brown between rich and poor soil was at no tune very noticeable. The color showed perhaps slightly earlier, and was perhaps slightly more intense, with the poor soil. Even this apparent difference, however, may have been due merely to the fact that the plants in poor soil were fighter and more yellowish green than those in rich soil. Dark green might readily mask the brown color somewhat. Green plants of both type Vic and type IVg exhibited no red nor brown color at any stage of development in either rich soil or poor soil. Plant Colors in Maize 23 From these observations it is apparent that variations in soil fertihty may effectively obscure genetic differences. A knowledge of the influence of soil fertility on color development is therefore essential to careful genetic work with the plant colors of maize. IMoreover, since soil fertility is subject to control thru cultural methods, different degrees of fertility can. be used as an aid to the sharp differentiation of certain genetic types. If, for instance, it is desired to separate, in the seedling stage, greens and browns on the one hand from the red-purple series on the other, this can he accomplished most readily in poor soil. In fact, the writer's practice, in studies requiring this separation, is to grow the seedlings in pure sand. In this medium seedlings of the purple-red series of color types become highly colored at a very early age, while seedUngs of the green and brown types show absolutely no red color. If, however, it is desired to distinguish sharply between purple and dilute puiple or between sun red and dilute sun red, fairly fertile soil is essential, and, usually, the more fertile it is, the more easily can the separation be made. The stronger colors develop almost as well in rich as in poor soil, while the weaker colors develop much less intensely in rich soils than in poor ones. On very poor soils, it is difficult to separate sun reds from dilute sun reds, and almost if not quite impossible to distinguish with certainty between sun reds and weak sun reds or between weak sun reds and dilute sun reds. Lack of particular ivdrient elements It having been established that differences in soil fertility result in marked differences in the development of red color in maize plants, it seemed important to determine whether particular nutrient salts are more con- cerned than others. Accordingly, plants of all the color types included in the tests previously reported were grown in glazed earthen jars in clean quartz sand and watered with nutrient solutions. The quartz sand was obtained from the Department of Agronomy of the University of Nebraska, and was known to be practically free from nutrient elements except iron. The nutrient salts and distilled water were obtained from the Department of Agricultural Chemistry of the same institution. The nutrient solution employed was one that had given good results with maize in certain experiments conducted previously by the Department of Agronomy. The complete nutrient solution, 0.2 j^er cent strength, contained per liter of water the following salts: 1 gram Ca (N03)2, 0.25 24 R. A. Emerson gram KNO3, 0.25 gram K2HPO4,. 0.25 gram MgS04, and 0.25 gram NaCl. Other solutions of approximately equivalent molecular strength, but each lacking one of the nutrient elements of the complete solution, were used. In the nitrogen-free solution, 0.7 gram CaCl2 and 0.22 gram K2SO4 were substituted for Ca(N03)2 and KNO3, respectively; in the phosphorus-free solution, 0.25 gram K2SO4 for K2HPO4; in the potassium-free solution, 0.2 gram NaNOs and 0.2 gram Na2HP04 for KNO3 and K2HPO4, respectively; in the calcium-free solution, 1 gram NaNOs for Ca(N03)2; in the magnesium-free solution, 0.3 gram Na2S04 for MgS04; and in the sulfur-free solution, 0.2 gram MgCl2 for MgS04. A complete nutrient solution of four times the strength indicated above, 0.8 per cent, was also used, and one lot was given water without the addition of nutrients. After the first three weeks, the nutrient solutions were all used at double strength, 0.4 and 1.6 per cent, and clear water was occasionally given. This treatment, owing to considerable evaporation of water, doubtless resulted in a gradual increase in the strength of the solutions. The tests were carried on at the same time with one of the tests of rich and poor soil, so that the latter might serve as a check on the nutrient-solution tests. At first the seedlings given 0.2-per-cent complete nutrient solution reacted about as did those in poor soil, while those given 0.8-per-cent nutrient solution were no more highly colored than those in rich soil. At one month of age, the plants watered for three weeks with 0.2-per-cent and one week with 0.4-per-cent complete solution were growing rapidly and were no more highly colored than those in rich soil, while the plants in the very strong solutions (0.8 and 1.6 per cent) were begimiing to wilt, perhaps from the toxic effect of the solutions. Thruout the remainder of the test, the plants given 0.4-per cent solution, alternated occasionally with clear water, were practically like those growing in rich soil both as respects vigor of growth and color development. In striking contrast to the plants given complete nutrient solution were the ones given clear water and those in nitrogen-free nutrient solution. Both these lots showed much color even at two weeks after germination, and soon thereafter the seedlings were red to the tips of their leaves. At the age of six weeks the plants of these two lots were much shorter and slenderer than those given complete nutrient solution. Their upper leaves were pale yellowish green, with much red, and the lower leaves were dead but still showing the red color that had developed earlier. Plant Colors in Maize 25 Next in point of coloration to the seedlings given nitrogen-froe nutrient solution and those given water alone, were the ones grown in phosphorus- free nutrient solution. The latter did not show red color so quickly as did the nitrogen-free lot, and at no time did they develop quite so much color. They showed, however, considerably more color at the age of one month than did seedlings in the complete nutrient solution. When six weeks old the plants of the phosphorus-free lot were relatively small, and had pale green upper leaves with little red color and dead lower leaves which still retained much red pigment. While somewhat larger than the plants in nitrogen-free solution and those in clear water, the phosphorus-free lot began wilting when about six weeks old and died considerably in advance of the nitrogen-free lot. Their roots showed early indications of injury, perhaps from toxic effects of the solution. Plants of all the other lots, in which one or another nutrient element had been omitted from the solution, exhibited little or no color reaction to the lack of a particular element. All of them were more vigorous in growth than the nitrogen-free and phosphorus-free lots, but much less so than the lot given complete nutrient solution. The sulfur-free lot for a time seemed to be developing more red, but later showed perhaps even less red, than the lot with complete nutrient solution. The mag- nesium-free lot showed prominent dark and light green stripes in the leaves similar to the green-striped chlorophyll pattern (Lindstrom, 1918). In some cases the tissue of the lighter stripes died and there was often some red coloration next to the dead tissue. The potassium-free lot had about the same amount of red color as the lot given complete nutrient solution, while the calcium-free lot showed less red color than any other lot in the test. It is perhaps noteworthy that in the nitrogen-free lot, and to some extent in the phosphorus-free lot. the new growth seemed to take place at the expense of the older leaves. The lower leaves first became light or yellowish green, then red, and finally died. That the development of red pigment is not necessarily connected, however, with the breaking down of the protoplasm, is seen in the failure of seedlings to develop red color in the older dying leaves of the lot in complete nutrient solution and of the potassium-free, magnesium-free, and calcium-free lots. In the calcium-free lot, growth was stopped by the death of the youngest parts, including the partly unrolled upper leaves, and yet these parts showed 26 R. A. Emerson no red. Moreover, the dying of the lower leaves due to excessively dry soil, or of the upper leaves from intense heat, is not accompanied by the development of red pigment. In similar tests with cuttings of Tradescantia viridis and T. lockensis grown in distilled water, in complete nutrient solutions, and in solutions each lacking one nutrient element, namely, N, P, K, Ca, Mg, or S, Czartkowski (1914) found that after five weeks red color appeared in the newly developed leaves in the cases of only distilled water and nitrogen- free solutions. He states, however, that Susuki reported a similar effect on plants of Hordeum from a lack of phosphorus. It will be recalled that in the writer's tests with maize, lack of nitrogen gave the most pro- nounced effect and lack of phosphorus induced considerable color develop- ment, while lack of sulfur seemed for a time to have an effect but no effect was apparent later. Frorh the results of the tests reported above, it is apparent that the reddening of young plants in flooded fields, as well as the intensification of color in older plants grown on poorly drained heavy soils, is not due to any direct effect of the excess of water in the soil or to a direct effect of the somewhat lower temperatures accompanying such conditions, but rather, perhaps, to the lessened fertility of cold, wet soils or to inability of the plant to obtain adequate nutrients under such conditions. An excess of water not only may remove certain nutrient salts from the soil, but also may prevent or greatly check nitrification. Moreover, under these conditions the soil solution is probably less concentrated. The reddening of young plants in cold, wet soils in spring, the greater develop- ment of color in plants maturing in the cool weather of late autumn, and the excessive development of red in plants on very light sandy soils, are possibly all due to the plants' inability to get from such soils an adequate supply of nutrient salts, particularly of nitrates. RELATION OF CARBOHYDRATES TO COLOR Several authors, notably Wheldale (1911), have discussed the relation of sugars to the production of anthocyanins in plants. Knudson (1916: 24, 62) found that maize and vetch grown in nutrient solutions containing certain sugars developed markedly more red color than did plants grown in sugar-free solutions. The writer has observed repeatedly an apparent relation between an excess of carbohydrates and the development of red Plant Colors in Maize 27 color in maize leaves. Of course the relation has been observed only in typos that normally produce some rod pigment. Neither brown, type V, nor green of either type IVg or type VI, has ever been observed with red color in the leaves, no matter what treatment has been given the plants. When loaves are folded at right angles to the midrib and the margin of the fold is creased sufficiently to break the softer tissues but not enough to break the water-conducting vessels, the part beyond the crease does not wilt, but within a few days it begins to lose some of its chlorophyll and within a week it becomes highly colored red (Plate X, 1). When leaves are similarly treated late in the afternoon of a bright day and the plants are kept in a dark room until the following day, the starch is, of course, found to have disappeared by translocation from the part of the leaves below the crease, while the cells of the bundle sheaths of the part beyond the crease are found to be packed with starch. There is so much starch in this part of a creased leaf that, on extraction of the chlorophyll with alcohol and treatment with iodin, the whole end of the leaf becomes almost black. While this does not prove a direct relation between an excess of carbohydrates and the development of red pigment, taken in connection with all the other observations it strongly suggests such a relation. It has been observed repeatedly that sweet-corn plants from which the ears have been removed in the edible stage develop within a week or two much more color than do neighboring plants that still retain their ears. Barren stalks also frequently show more color than do their ear- bearing neighbors. While no direct determination of the matter has been made it seems likely that barren plants, as well as plants from which the immature ears have been removed, may carry, in their leaves, husks, and cuhns, an excess of carbohydrates which would normally have been deposited in the developing seeds. The strong development of red pigment in the white, chlorophyll-free stripes of the japonica-striped type, when leaves are creased or when plants are grown in poor soil, may well be due to the passage of sugars from the green to the white parts. In some instances the red color seems to develop more quickly in the white stripes than in the green (Plate X, 2). Whether this difference is a real one, due perhaps to the readier access of light to the white parts, or is only an apparent difference due to the 28 R. A. Emerson masking effect of the green color, is not known. Certainly red pigments develop first in the chlorophyll-free epidermal cells.^ Czartkowski (1914) suggested, in connection with the account of his study of the relation of nutrient elements to color development, that lack of nitrogen may check protein synthesis, thus leaving unused the carbohydrates that would otherwise be used in growth, and that the excess of carbohydrates may favor anthocyanin formation. He was unable to understand why a lack of phosphorus or of sulfur did not like- wise influence color development, since these elements also are necessary to protein synthesis. Lack of phosphorus does apparently bear some relation to color development in maize, but the v/riter's tests afforded little or no evidence of such a relation between a lack of sulfur and pigment formation. If lack of nitrogen induces anthocyanin formation thru the checking of growth, thus allowing an accumulation of carbohydrates, it is not clear why other means of checking growth, such, for instance, as dry soil, do not also favor pigment formation, unless these other growth- checking factors at the same time limit photosynthetic activity. It is of interest to recall in this connection that plant colors of maize — brown no less than the red-purple series — develop first in the older parts where growth first ceases, such as the lower sheaths and the upper parts of the internodes of the culm. SUMMARY Whatever is the final outcome of studies of the relation of environmental factors to plant-color development in maize, enough has been noted to indicate a very complex relation. What is more complex than this chain of events — a chain that lacks many links in the way of particular chemical reactions: cold, wet soil checks or inhibits nitrification; lack of nitrogen in available fonn limits protein synthesis, which in turn allows an accumu- lation of carbohydrates; an excess of carbohydrates favors anthocyanin formation. The result is that young maize plants in cold, wet soil become highly colored. But to all this must be added the factor of sunlight, without which no red color develops in the leaves of young plants. And not the least consideration is the important fact that only plants of certain genetic constitutions show this color reaction to wet soils. It is to be hoped that some day, thru the coordinated efforts of 2 The histology of color development of the several plant-color tvpes has been investigated by Dr. E. G. Anderson, but the observations have not been published.^" Plant Colors in Maize 29 biochemists, physiologists, and geneticists, it may be possible to reach conclusions in this field of quite as fundamental importance to biology as the recent results of similar efforts of cytologists and geneticists. GENETIC ANALYSIS OF COLOR TYPES In the preceding parts of this paper the several plant-color types of maize are described and the variations induced in them l)y diversities of environment are discussed. The remainder of the paper is devoted to a presentation of data of a more distinctly genetic nature, and to an attempt at a factorial analysis of these data. The data are presented as if the F2 generation of the more complex crosses were the first which were obtained and on which hypotheses were formulated and appropriate tests made. As a matter of fact, this was not in all cases the actual procedure. In several instances the results of some of the simpler crosses were at hand and were used as an aid to the interpretation of the more complex ones when the latter were obtained. Moreover, the hypothesis presented here was not the only one, nor indeed the first one, formulated. As is usual in such work, various hypotheses were devised, tested, and discarded, until finally a factorial interpretation was found that fitted fairly well all the facts known. Many results with a bearing on plant color were obtained in other studies extending over a period of some eight or nine years. Since the practice of the writer is to number his pedigrees consecutively from year to year, an inspection of the pedigree numbers, as listed in the tables, suggests at once that some of the data presented as checks on other results could not have been obtained after these other results. Any data applicable as a test have been so used whether obtained for that purpose or in connection with other studies. Whether this mode of presentation is the best one must be left to the judgment of others. This at any rate is certain: the data could not have been presented chronologically and discussed in relation to such hypotheses as happened to be under test at the time any particular results were obtained, without adding unnecessarily to the complexity of the paper. CROSSES INVOLVING THE FACTOR PAIRS A a, B b, PI pi Purple la x green Vic Generations Fi and F^. — When purple plants with purple anthers (type' la) are crossed with plants lacking all red, purple, or brown 30 R. A. Emerson pigment, commonly known as green (type Vic), the Fi offspring are full purple. Whether or not a quantitative determination of purple pigment might reveal a difference, no dilution of the purple color is apparent to the eye in the Fi plants. Four crosses of this sort with a total Fi progeny of 111 purple plants are listed in table 1 (appendix, page 121). Seven Fo progenies of the Fi plants recorded in table 1 are hsted in group 1 of table 2. Fourteen other similar F2 progenies are shown in group 2 of the same table. The Fi plants from which these fourteen F2 progenies came are not recorded in table 1 because their purple parents were not homozygous. Some of the purple plants used as parents in these crosses were Fi's of the original cross of purple with green. Others were from Fi or some later generation of other crosses having the purple type as one parent. In every case the other parent was a green plant of type Vic. Since the purple Fi plants of these crosses were presumably the same genotypically as the Fi's shown in table 1, their F2 progenies may well be included tentatively with those of group 1 of table 2. Each of the twenty-one F2 lots exhibited six distinct classes of plants with respect to color. The 2117 plants were distributed among the six classes as follows: Purple ®",^ ^^^^^e Dilute g ^ ^ ^ ^ red purple sun red 952 305 275 91 278 216 2,117 Obviously no simple 3 : 1 mendelian behavior is in evidence here. More- over, only four classes are expected in dihybrids where dominance is exhibited. With dominance trihybrids ordinarily give eight classes in F2 in the well-known numerical relation of 27:9:9:3:9:3:3:1, while only six classes were observed. Inspection of the distribution of the 2117 individuals given above, however, suggests the possibility of a 27:9:9: 3:9:7 relation, which should be realized in a trihybrid if the last three classes were indistinguishable. A comparison of observed numbers with those expected on this hypothesis follows: Color types Purple ^un ™u}: ™>i^. Brown Green Total "Observed 952 305 275 91 278 216 2,117 Calculated^ 893 298 298 99 298 232 2,118 Difference -j-59 +7 —23 —8 —20 —16 —1 3 In this and most of the following comparisons, the theoretical distributions are calculated to the nearest whole number. Plant Colors m Maize 31 There are rather large differences between observed and expected numbers. The purples are considerably, and the sun reds slightly, in excess of expectation, while each of the other four classes has too few individuals. The probability that these deviations may be due to chance is approximately 0.11. One might expect, therefore, to encounter chance deviations of the magnitude observed here about once in nine such trials. This, of course, does not substantiate the three-factor hypothesis, but merely indicates that it is not necessarily out of keeping with the ol)servetl facts. Backcrosses with green Vic. — A better criterion perhaps is afforded by the backcross of Fi purples with the green parent type. Records of such crosses are shown in table 3. The backcrosses with F/s of table 1 are listed in group 1, and backcrosses with similar Fi purples of other lots in group 2. The same six phenotypcs o])served in the regular Fo generation occurred here also. On the basis of the three-factor hypothesis and with the assumption that there are three sorts of greens indistinguishable from one another, the individuals of this backcross should be distril^utcd equally to five classes with the sixth class containing three times as many indi- viduals as any other class. The observed distribution of the 1317 indi- viduals of the fourteen progenies is here compared with the expected distribution: ^ , , T^ 1 Sun Dilute Dilute -o r< t. x i Color types Purple ^.^^j purple sun red ^^'°^^ ^'»'^^" Total Observed 170 160 176 160 172 479 1,317 Calculated 165 165 165 165 165 495 1,320 Difference +5 — 5 +11 — 5 +7 ■ — 16 ■ — 3 While a few of the backcross progenies listed in table 3 exhibit con- siderable deviations from the expected distribution, the fourteen lots taken together approximate it closely. The probal:)ility that the observed deviations may be due to chance in random sampling is about 0.85. Deviations as great as these are to be expected thru chance alone, there- fore, in about six out of seven trials. Workiiig hypothesis. — To the three factor pairs used to interpret the results here reported, the symbols A a, B h, and PI pi have been assigned. The gene A is an anthocyanin factor. In the presence of a a ordinarily no anthocyanic pigment develops, tho brownish, or flavonol (Sando and Bartlett, 1921), pigment may be formed. The pair Bb is named for its 2 32 R. A. Emerson connection with the development of brown pigment, tho when both A and B are present, sun red pigment is produced. The pair PI pi is so teraied because of its relation to purple pigment. The phenotypic formulae as- signed to the several classes of plant color under consideration here are as follows: ABPl — la, purple AB pi — Ila, sun red Ah PI — Ilia, dilute purple A b pi — IVa, dilute sun red aBPl— V, brown aBpl —Via] ah PI — VIb !• green ah pi — VIcJ Obviously the hypothesis in accordance with which the above factorial assignments have been made is subject to several genetic tests. Naturally the first tests to suggest themselves are studies of the behavior of the several F2 types in F3 and later generations. Next in order are inter- crosses between the several classes. For reasons that will appear shortly, one of these intercrosses is here dealt with before consideration is given to F3 generations from the several F2 classes. Dilute sun red IVa x brown V From an examination of the factorial assignments listed above, it is evident that crosses of dilute sun red, A h pi, with brown, a B PI, should produce purple Fi plants, A B PI. Moreover, these Fi purples should be heterozygous for all three factors, AaBh PI pi, just as was assumed for the original cross of purple, A B PI, with green, a h pi. The F2 and later behavior of this cross should also, barring linkage, be hke that of the original cross, so that the two can most conveniently be considered together. Generations Fi and Fo. — The Fi generation of twenty-six crosses of dilute sun red with brown plants is given in table 4 (page 123). The dilute sun red parent plants were chosen from any convenient lots known to be homozygous with respect to ^, b, and pi. The brown parent plants, on the other hand, were from the F2 and later generations of the original cross of purple and green or from other crosses. It was to be expected, therefore, that some of the biown plants would be homozygous for both B and PI, and some would be heterozygous for B, some for Pi, and some Plant Colors in Maize 33 for both B and PI. This expectation was fully reaUzed. In group 1 of table 4 are recorded the progenies of nine crosses with a total of 263 indi- viduals. All but one plant of the lot were purple. The one dilute sun red plant was presumably due to accidental pollination of the dilute sun red mother plant. Since the dilute sun red parents of all these crosses were A A hh pi pi, the brown parents of the crosses hsted in group 1 must presumably have been aa B B PI PI. Similarly, the seven crosses listed in group 2 gave purple and sun red plants only, 143 of the former and 147 of the latter. Evidently the brown parents of these crosses were aa B B PI pi. Again, the six crosses shown in group 3 gave 105 purple, 123 dilute purple, and no other plants. The brown parents of the crosses were therefore, presumably, aaBh PI PI. Finally, the four crosses listed in group 4 gave 9 purple, 11 sun red, 19 dilute purple, and 17 dilute sun red. The brown parents of these four crosses are assumed, consequently, to have been a a B b PI pi. *' The F2 results from the purple Fi plants of these crosses of dilute sun red with brown are recorded in table 5. Fourteen progenies of the Fi plants hsted in table 4 are shown in group 1 of table 5, and five progenies from similar Fi plants not listed in talkie 4 are entered in group 2. Here, just as with the results of the cross of purple with green (table 2), fairly marked discrepancies between theory and observation appear when the several progenies are taken separately. When, however, the nineteen progenies are considered together, very close agreement is found between observation and exjiectation, as is shown by the comparison below. The probabihty that such deviations as are observed may be due to chance is approximately 0.88, which means that only about once in eight trials would as good a fit be expected. The comparison follows: Color types Purple la Observed 847 Calculated. . . . 845 DiiTerence .... +2 Sim red Ila Dilute purple Ilia Dilute sun red IVa Brown V Green Via, b, c Total 282 281 94 267 233 2,004 282 282 94 282 219 2,004 0—1 —15 +14 Backcrosses with green Vic. — In addition to the F2 results noted above as derived from self-poUinated Fi purple plants, a few F, purples were back- 34 R. A. Emerson crossed with the triple recessive green, type Vic. The records of these crosses, seven in all, are presented in table 6. The results are. as expected, in close agreement with the Imckcross data from the cross of purple with green. The comparison below indicates a good fit of (calculated to observed frequencies for the lot as a whole. The probability that such deviations as are observed may be due to mere chance is about 0.82, indicating that as great departures from expectation as these might be expected about four times in five trials. The comparison follows: Color types Purple Sun red p^j.p]g gun^red ^^'^^^ ^'^'^^^^ Total la Ila Ilia IVa V Via, b, e Observed 84 72 78 72 79 249 634 Calculated.... 79 79 79 79 79 237 632 Difference.... +5 —7 —1 —7 +12 +2 Backcrosses of la x Vic and IVa x V with IVa Purple plants of Fi of the crosses purple x green and dilute sun red x brown were crossed with homozygous dilute sun red stocks. On the basis of the hypothesis used above, the Fi plants are assumed to be ^ a B b Plpl and the dilute sun red plants A A hb pi pi. Four classes of plants, purple, sun red, dilute purple, and dilute sun red, should be pro- duced in equal numbers by this cross. The data are presented in table 7 (page 125). Progenies of Fi plants from the cross purple x green are listed in group 1 and those from the cross dilute sun red x brown in group 2. As will be seen from the comparison below, the observed numbers are in fair agreement with the hypothesis. The probability that such deviations as occur may be due to chance is approximately 0.67. In other words, there are two chances in three that deviations of this sort are due to errors of random sampling alone. The comparison follows: Color types Purple Sun red , , Total ^^ ^ purple sun red la Ila Ilia IVa Observed 299 270 288 291 1 ,148 Calculated 287 287 287 287 1 ,148 Difference +12 —17 +1 +4 Plant Colors in Maize 35 Behavior of Fz color types in later generations From all the foregoing it appears that the results obtained are in close accord with the proposed three-factor hypothesis in the case of both the cross purple x green and the cross dilute sun red x brown, and not alone for the Fi and Fo generations but also for l)ackcrosses with green and with, dilute sun red. It is now in order to inquire into the behavior of these crosses in F.-s and later generations. In the presentation of the additional data, the two crosses purple x green and dilute sun red x brown will be considered together. Later behavior of F2 purple la. — Purple plants of the F2 generation of the crosses under consideration are expected to be of eight genotypes. The expected F2 genetic formulae and the F3 color classes, together with the relative numbers of each, are as follows: F3 color types F2 genotypes Purple la Smi red Ila Dilute purple Ilia Dilute sun red IVa Brown Green VI \-~ A AB B PlPl 1 3 3 3 9 9 9 27 1 3 3 9 I 3 3 9 i 3 I 3 3 9 2 — A ABB PI pi 2 — A A Bb PI PI 2 — AaBBPlPl 4 — A A Bb PI pi A — AaBB Plpl 1 A — A a Bb PI PI 1 8 — A a Bb PI pi 7 If, instead of being selfed, the F2 purple plants are backcrossed to green of type Vic, the same F3 color classes are expected but the several classes should, of course, be equally frequent except in case of the F2 triple heterozygotes, which should thi'ow three times as many greens as of each of the other five types. The F3 data from thirty-five F2 plants are recorded in table 8 (page 125). In group 1 of the table are listed the progenies of eight selfed and one backcrossed F2 plants. From the backcross six color types appeared in frequencies of 4:4:11:4:4:18. The theoretical number for the first 36 R. A. Emerson five classes is 5.6 and for the sixth class is 17. The probability that such deviations as occur are due to chance is approximately 0.35, or more than one in three. The eight self-polHnated plants gave together the six types in frequencies as follows: Color types Purple Sun red J^J|.^|^ ^^jjf^^^ Brown Green Total la Ila Ilia IVa V Via, b, c Observed 193 66 60 16 57 34 426 Calculated 180 60 60 20 60 46 426 Difference +13+6 0—4—3 —12 The probability that such deviations as occur may be due to errors of random sampling is practically 0.27. Similar deviations might there- fore be expected somewhat more than once in four trials. It will be noted that two progenies lacking class IV are included in this lot (group 1, table 8). The total number of plants in these progenies were 37 and 17, respectively, and they should therefore have had, respectively, two and one plants in class IV. Five Fo purple plants (group 2, table 8) gave four color types (la, Ila, Ilia, and IVa) in F^, with total frequencies as shown below. Here the probability, P, equals 0.75, indicating that deviations of this magnitude might be expected thru chance in three out of four trials. The com- parison of observed with theoretical distributions follows: Color types Purple Sun red ^'^^^^ ^^^^^*^. Total "^^ ^ purple sun red la Ila Ilia IVa Observed 102 36 29 13 180 Calculated 101 34 34 11 180 Difference +1 +2 —5 +2 Progenies of seven other purple F2 plants (group 3, table 8) consisted cf the four color types la, Ila, V, and Via. Four of these F2 plants were self-pollinated and gave a total of 164 F3 plants. Four, including one that was also selfed, were backcrossed to green and yielded a total of 209 F3 plants. For the progenies from selfed F2 plants P = 0.20, and for those from backcrossed plants P =^ 0.57. There is, therefore, one chance in five Plant Colors l\ Maize 37 in the one case and considerably more than an even chance in the other case that deviations of the kind noted may have been due to errors of random sampUng. The comparisons follow: Color types Purple Sun red Brown Green Total la Ila V Via c. ,. J /Observed 95 31 23 15 164 ^^^^^"^ \ Calculated 92 31 31 10 164 Difference +3 —8 +5 Backcrossed<;^^''^''^'^^--- 54 58 44 53 209 ^^^^"^"^^''^"^^ Calculated... 52 52 52 52 208 Difference.... +2 +6 —8 +1 +1 Seven self-pollinated Ff purple plants gave progenies consisting of the four color types la, Ilia, V, and VIb (group 4, table 8). Here P = 0.75, indicating that there are three chances in four that such deviations as are shown are due to chance. The comparison follows: Color types Purple IJjj.p|g Brown Green Total la Ilia V VIb Observed 318 114 111 42 585 Calculated 329 110 110 37 586 Difference —11 +4 +1 +5 —1 Five F2 purple plants from self-pollination gave only two color types (la and Ila) in F3 (group 5, table 8). The total number of F3 individuals was 183, of which 139 were of color type la and 44 were of color type Ila, the expected numbers being, respectively, 137 and 46, and the deviation equaling 2 ± 4. One of these F> plants was also backcrossed to two greens, resulting in 12 purple and 9 sun red F3 plants where equality of the two classes was expected. The deviation here is 1.5 ± 1.5. Finally, two self-pollinated Fo purple plants produced 217 F3 individuals (group 6, table 8) of color types la and Ilia. There were 168 purple 38 R. A. Emerson and 49 dilute piirple where the expected numbers were 163 and 54, respec- tively — a deviation of 5 ± 4.3. It is seen, then, that in every case the F3 progenies of F2 purple plants were of color types expected on the basis of the three-factor hj^pothesis, and that the F3 distributions within any group were in close agreement with expectation. It is particularly noteworthy, however, that not all types of F3 behavior were observed, and that the distribution of the progenies of the thirty-five F2 plants tested was in rather imperfect agree- ment with expectation. Thus, no Fo purple plant bred true in F3 where one such plant was expected, and none gave progenies of pui'ple and brown only where at least two with such behavior were expected. It has already been pointed out (page 35) that eight classes of behavior of Fo purples are looked for, and that any twenty-seven F2 piu'ple plants should be distributed with respect to their F3 behavior in the relation 1:2:2:2:4:4:4:8. The actual and theoretical distributions are compared as follows: Observed 05205 7 79 35 Calculated.., 1.3 2.6 2.6 2.6 5.2 5.2 5.2 10.4 35.1 Difference —1.3 +2A —0.6 —2.6 —0.2 +1.8 +1.8 —1.4 —0.1 Wliile mere inspection of the above comparison might suggest poor agree-. ment between theory and observation, nevertheless P = 0.36, indicat- ing that such deviations as occur might be expected in more than one out of three trials, which is not a bad fit. So far, therefore, the avail- able data are in fair accord with the three-factor hypothesis. Before taking up a consideration of the F3 behavior of other F2 color types, it will be well to consider briefl}'^ the F4 behavior of F3 purple plants. Only one F3 pm-ple of the lot having all six color types (table 8, group 1), comparable to F2 purples, was tested in F4. This one plant gave an F4 with the four color types la, Ila, V, and Via. Onty eight other F3 purple plants were tested in F4. All these belonged to the lot consisting of color types la. Ilia, V, and VIb (group 4, table 8). The F2 purple plants giving rise to this group are assumed to have been of the genotype AaBb PI PL The Fj purple plants should therefore have been of four genotypes and should have given F4 behavior as follows: Plant Colors in Maize 39 Fi color types F3 genotypes Purple la Dilute purple Ilia • Brown V Green Vib 1 — A ABB PI PI 1 3 3 9 i 3 1 3 2~AABb PI PI 2 — AaBBPlPl A~AaBb PlPl 1 The data are presented in table 9. Four F4 progenies (group 1) were made up of the four color types la, Ilia, V, and VIb. The total numbers of plants of each of the four types, as seen below, were in close accord with expectation, P equaling 0.57. There is more than an even chance that such deviations as those observed may have been due to errors of random sampling. The comparison of observed with calculated results follows : Color types Purple ^ .'^^ j^ Brown Green Total la Illa V VIb Observed 185 68 74 20 347 Calculated 195 65 65 22 347 Difference —10 +3 +9 —2 Three of the eight purple Fs's (group 2) gave in F4 only purple and dilute purple plants, 88 of the former and 28 of the latter. The expected numbers were 87 and 29, respectively, showing a deviation of 1 ±3.1. One of the eight F3 purples (group 3) gave 67 purple and 21. brown plants in Fj, while the expected niunbers were 66 and 22, respectively. The deviation here is only 1 ± 2.7. None of the eight F.-^ purples Ijred true, but only one in nine was expected to do so. As already indicated, the theoretical distribution of nine F3 purples of the sort here under consideration, with respect to the four kinds of behavior in F4, is 1:2:2:4. The observed distril)Ution was 0:3:1:4. There is more than an even chance that these deviations may have been due to errors of random sampling, P equaling 0.57. 40 R. A. Emerson It should not be forgotten that, while a very poor fit of observation to hypothesis, as measured by values of P, throws doubt upon the correct- ness of the hypothesis, it does not follow that a good fit proves the hypothesis to be true. This is particularly true where small numbers are dealt with. It will be recalled in this connection that, owing prol^ably to the small numbers tested, no F2 purple has been found to breed true in F3 and none has been found to give only purple and brown offspring. It has been shown, however, that purple plants of the genotype A a B B PI PI exist, since one F3 purple threw only purple and brown plants in F4. Moreover, one of these F4 purples repeated this behavior in F5. Similarly it can be said that purples of the genotype A A B B PI PI have been recovered from the crosses under consideration, for two F4 purple plants of the lot composed of purples and dilute purples (group 2, table 9), when backcrossed to green, gave 18 purple plants and no other types in the next generation, and one of these two F4 purples, when crossed back to dilute sun red, gave 34 purple plants. Two other purples of the same F4 lot, when similarly crossed, gave both purple and dilute pm*ple, 23 of the former and 18 of the latter. Pm'ple plants of all the expected genotypes have therefore been recovered in one or another generation from F2 to F4 from the original crosses of purple x green and dilute sun red X brown. Moreover, these genotypes have been found in numbers not far from what might reasonably be expected considering the relatively small numbers tested. It now remains to inquire into the F3 and later behavior of F2 color types other than purple. Later behavior of F2 sun red Ila. — Sun red plants of F2 of the crosses purple x green and dilute sun red x brown are expected, in accordance with the three-factor hypothesis, to be of four sorts with respect to their behavior in F3, as follows: F2 genotypes F3 color types Siin red Ila Dilute sun red IVa Green Via, c 1 — AABBplpl. 2 — AABbplpL 2 — AaBBplpl. 4 — A a B bplpl. Plant Colors in Maize 41 Only nine Fo sun red plants wore tested by their F3 behavior, and no later generations were si'own. All the available data ara {jfiven in table 10 (page 128). Five F2 plants, when self-pollinated (group 1 of the table), gave the expected three classes of progeny, sun red, dilute sun red, and green, with a distribution of the F3 plants as given below, and in addition a single brown plant. To include this unexpected plant in the compari- son with the calculated distril)ution would give zero as the value of P, which is equivalent to saying that even in an infinite number of trials there is no chance of finding such a plant thru errors of random sampling. The single off-type plant is readily accounted for by supposing that a grain of foreign pollen was accidentally admitted in the pollination of the parent plant. Tho it is realized that, with such a convenient supposition always at hand, almost any result can be made to fit a theory, the reality of just such accidental pollinations will not be questioned ,by any one who has had experience in the technique of maize pollination. With the elimination of this one plant, the fit of observation to hypothesis is almost perfect. The comparison follows: Color types Sun red ^^^^^^ Green Total Ila IVa VIa,c Observed 126 42 55 223 Calculated 125 42 56 223 Difference +1 " —1 Three Fo sun red plants, including one of the five in the former test, were crossed back to green (group 1, table 10). The same three color types w^ere observed as in the self-pollinated plants, with the addition again of a single off-type plant, this time a purple one. Even if this plant is left out of consideration as due to an accidental pollination, the fit of observed with calculated numbers is not very good. Such deviations from theoretical behavior are to be expected thru chance alone only once in eight trials, P equaling 0.12. The comparison follows: Color types Sun red ^^^^ ^.^^ Green Total Ila IVa Via, c Observed 14 18 50 82 Calculated... 20.5 20.5 41 82 Difference —6.5 —2.5 +9 42 R. A. Emerson A single F2 sun red plant (group 2, table 10) gave, from self-pollination. 23 sun red and 9 dilute sun red F3 plants, a deviation fronri expectation of 1 ± 1.7. A single Fo sun red plant (group 3, table 10), when crossed with green Vic, gave 50 sun reds and 43 greens where equality was expected, a devia- tion of 3.5 ± 3.3. By way of summary of the behavior of Fo sun red plants, it must be noted that, while four sorts of behavior were expected, only three sorts were observed. While any nine such F2 plants should be distributed with respect to the four kinds of behavior in the relation 1:2:2:4, the observed relation was 0:1:1:7. Wliile mathematically this is not a very bad fit considering the small numbers involved, P equaling 0.24, it is inadequate for a deteiTQination of the possible genotypes of F2 sun red plants. Fortunately, certain crosses considered later (page 51) involving the sun red type, with presumably the same genetic constitutions as the F2 sun reds of this cross, afford a more nearly adequate test of the matter. Later behavior of Fo dilute purple Ilia. — F2 dilute purple plants should present the same types of behavior in F3 as F2 sun reds, but, of course, with somewhat different color types appearing, as follows: F2 genotypes 1— A Abb PI PI 2~AAbbPlpl. 2 — A abb PI PL 4~ A abb PI pL. F3 color types Dilute purple Ilia Dilute sun red IVa Green VIb, c The available data from this test are given in table 11 (page 129). Four F2 dilute purples (group 1) yielded the three color types expected, dilute purple, dilute sun red, and green, in the numbers shown below. There is considerably more than an even chance that the deviations from expecta- tion may be due to errors of random sampling, P equaling 0.58. The comparison follows: Plant Colors in Maize 43 r^ 1 J. Dilute Dilute ri n. . ■, C«^o^-*yP^^ purple sun red ^'''''' ^^^^^ Ilia IVa VIb, c Observed 95 31 50 176 Calculated 99 33 44 176 Difference —4 —2 +6 One of the dilute purple Fo plants used in this test was backcrossed with green VIc (group 1, table 11), with the result shown below. There is practically an even chance that the observed deviations may be due to errors of random sampling, P equaling 0.49. The comparison follows: ^ 1 , Dilute Dilute ^ t- ^ i Color t\T3es i , ureen Total ■^^ purple sun red Illa IVa VIb, c Observed 21 25 57 103 Calculated 26 26 52 104 Difference — 5 — 1 +5 — 1 One F2 dilute purple gave 57 dilute purple and 21 dilute sun red plants in Fa (group 2, table 11). The expected numbers were 58.5 and 19.5, respectively, the deviation being 1.5 ± 2.6. Three Fo dilute purples gave a total of 85 dilute purple and 20 green plants (group 3, table 11), the theoretical numbers being 79 and 26, respectively. The deviation from expectation, 6 plants, is just twice the probable error. One F2 dilute purple bred true in F3, producing 21 dilute purple plants and no other types (group 4, table 11). Thus, all the sorts of behavior expected of F2 dilute purples were reahzed in F3. The distribution of the F2 plants with respect to the four sorts of behavior was 1:1:3:4, instead of the theoretical distribution 1:2:2:4. Differences of this sort might be expected thru chance in four out of five trials, P equaling 0.80. Only three plants of these lots were tested in F4. One was a dilute sun red of the lot made up of dilute purples and dilute sun reds, and this one bred true in Fi as was expected of it, producing 34 dilute sun red plants. The other two plants tested further were dihit(,' purples of the lot contain- ing the three color types III, IV, and VI. Both again gave these three 44 R. A. Emerson types, the total numbers of the respective classes being 29, 5, and 18. The expected numbers, 29, 10, and 13, show a deviation from expecta- tion which might result thru chance about once in nine trials, P equaUng 0.11. Later behavior of F^ dilute sun red IVa.—z Dilute sun red plants of F2 should be of two sorts, A Ahhplpl and Aahhpl j)l. Five such plants were tested, with results as shown in table 12 (page 129). Of these five, two bred true, producing a total of 92 dilute sun red plants (group 2). One of these two, when backcrossed with green, gave 69 dilute sun red plants. Three of the five F2's gave in F3 dilute sun reds and greens, 62 of the former and 17 of the latter (group 1). The theoretical numbers were 59 and 20, respectively. The deviation of 3 plants is only a httle greater than the probable error, ± 2.6. With two of the F2 dilute sun red plants breeding true and three again throwing segregates, expectation was very nearly reahzed. Later behavior of F2 brown V. — Brown plants of Fo are expected to be of four genotypes and to show consequent differences in behavior in F3 as follows: F3 color types F2 genotypes 1 — aaBBPlPl 2 — aaBB Plpl. 2 — aaBbPlPl. 4 — aaBbPlpl.. Data for F3 from fourteen F2 brown plants are presented in table 13 (page 130). Five self-pollinated F2 browns (group 1) gave, in addition to one sun red prcsumabl}'- due to accidental pollination, 96 browns and 74 greens in F3, which is almost exactly a 9 : 7 relation, the deviation being 0.4 ± 44. Nine other selfed F2 browns (group 2) gave in F3 a total of 354 brown and 104 green plants. An exact 3 : 1 ratio for the total of 458 would be 343.5 and 114.5, respectively, the deviation being 10.5 ± 6.3. Such a deviation might be expected thru chance alone about once in four Plant Colors in Maize 45 trials. One of the F2 brown plants that, when solfod, gave a 3:1 ratio in F3, when crossed with green gave 34 brown and 41 green plants where equal numbers were expected, the deviation being 3.5 ± 2.9. None of the fourteen F2 brown plants bred true in F3. The fourteen plants should theoreticall}^ have given F^ ratios of 1:0, 3:1, and 9:7 in approximately the respective numbers of 1.6, 6.2, 6.2, while the observed numbers were 0, 9, 5. Such deviations might occur by chance once in five trials, P equaUng 0.22. It is often difficult and sometimes practically impossible from ordinary F3 progenies to distinguish between the two genotypes of brown which throw 3:1 progenies,, namely, a a B B PI pi and ao B b PI PI. The green plants thrown b}^ the former often show some brown pigment in the exposed parts of the sheaths and husks (type Via), a condition not seen in the greens (\Tb) thrown ])y the latter. In some lots the brown pigment is fairly conspicuous but in others it is very weak or is absent. Again, the greens of type VIb thrown by l^rowns of the genotype aaBb PI PI show considerable brown in the glumes of the staminate flowers. This is particularly pronounced when r<^'' (a gene for cherry pericarp which is effective only in the presence of PI) is present, but when this factor is lacking the brown color is often so faint that it is impossible to distinguish between a green plant carrying PI and one lacking it. If r'''' is present, the greeu plants carrjnng PI develop a Ught brownish pericarp at maturity while those lacking PI never show this pericarp color whether or not B is present. Here again, however, the hght brownish pericarp due to r^^, PI, and a a ma}^ be wholly masked if there happens to be present another pericarp color gene, P, which with a a brings about a strong brown color of the pericarp whether or not PI or B is present.^ On the whole, therefore, it is difficult, and often impossible, to determine the genotype to which a brown plant belongs, by an inspection of the green plants occurring in its progen3^ Because of this, the 3 : 1 lots of F3 progenies of Fo brown plants are lumped together in group 2 of table 13 without any attempt to separate them into the two classes expected. Fortunately, it is readily possible to distinguish between brown plants of the two genotypes under consideration here by means of appropriate crosses. * An account of these pericarp-color factors is to be published later by Dr. E. G. Anderson, who is making a study of the pericarp colors of maize. 46 R. A. Emersox When brown plants of all the genotypes expected in F2 of the crosses of purple X green or dilute sun red x brown are crossed with homozygous dilute sun red plants, the following behavior is expected in the next generation : F2 genotypes 1—aaBB PlPl 2~aaB B Plpl. 2^aaBb PlPl. 4 — aaBb Plpl. Purple la F2 X A Abb pi pi Sun red Ila Dilute purple Ilia Dilute sun red IVa A few such tests of F2 brown plants are recorded in table 14. Two plants (group 1), on being crossed with dilute sun reds, gave purples and sun reds only, 38 of the former to 45 of the latter, where equality was expected, the deviation being 3.5 ± 3.1. One of these plants has progeny from self- pollination listed in table 13, in group 2, the 3:1 lot. This plant was expected, of course, to throw only two color types from the cross with dilute sun reds, for otherwise it should not have given a 3 : 1 progeny on being selfed. The two brown plants in group 1 of table 14 must have been aa B B Plpl. Two other F2 brown plants (group 2) gave 32 purple and 38 dilute purple instead of the equal numbers expected, the deviation being 3.0 ± 2.8. These plants are assumed to have been a a B b PI PL A single F2 brown plant (group 3) when crossed with dilute sun red gave 15 purple plants, and is therefore assumed to have been a a B B PI PI. The behavior of several F3 brown plants when crossed with dilute sun reds is also shown in table 14. Three of these plants were from 9:7 F3 lots and therefore are presumably comparable with F2 browns. One of these three (group 4) gave the four color types I to IV in the numbers 1:2:6:3. It was probably a a Bh Plpl and should have given a 9:7 progeny if it had been selfed. The other two F3 browns of the 9:7 lot gave 49 purple plants (group 7) and are consequently regarded as aaB B PI PL All the other F3 brown plants tested were from the Plant Colors in Maize 47 3:1 lot listed in table 13, group 2. None of these should give more than two types when crossed with dilute sun red. One gave 46 purple and 1 dilute sun red (group 7), the latter doubtless from an accidental poUination of the dilute sun red mother plant. Two F3 browns gave 22 purple and 24 sun red plants (group 5), and four produced 73 purple and 85 dilute purple plants (group 6). To summarize, all the theoretically possible genotypes of brown plants have been found either in F2 or in such F3 lots as showed a 9:7 ratio of brown to green. Since these Fs's are comparable with F2 browns, thej^ maj^ be added to the F2's in this summary. Of the twenty-one brown plants thus grouped, the numbers found to belong to each geno- tj^pe are compared below with the calculated numbers. The deviations are such as might be expected to occur once in three trials, P equaling 0.34. The comparison follows: aciBBPl j)l aaB B PI PI or a a B h PI pi Total aaBbPl PI Observed 3 12 6 21 Calculated 2.3f+) 9.3(+) 9.3(+) 21 Difference +0.7(-) +2.7(-) — 3.3(+) Later behavior of F2 green VI. — All Fo green plants should breed true phenotypically in F3. Data from eight such F3 progenies are given in table 15, group 1 (page 132). There were observed a total of 179 green plants, and no other tjq^es. Progenies of sixteen green plants of the F2 lots listed in tables 3 and 6 (pages 122 and 124), produced l^y backcrossing Fi purples to greens, are given in table 15, groups 2 to 5. The total number of green plants in these progenies is 311. A single brown plant found in one of these progenies is assumed to have been due to acci- dental pollination. Green plants are thert'fore found to breed true green as expected, but there is nothing in this fact to indicate that green plants of the crosses under consideration are genotypically alike. That the five genotypes expected on the basis of the three-factor hypothesis were present among the progenies hsted in table 15 is demonstrated in the next section of this paper. 48 R,. A. Emerson Intercrosses of F^ color types It has been shown in the preceding pages that all the six color tj^pes occurring in F2 of a cross between purple and green behave in F3 and later generations as is expected on the basis of the three-factor hypothesis suggested to account for the F2 results. It remains to determine whether the several color types behave in accordance with the hypothesis when intercrossed one with another. Of the fifteen possible intercrosses between phenotypically different types, two have already been discussed. The cross of purple with green has formed the basis of the whole discussion. The cross of dilute sun red with brown, since it was expected to give the same results as the original cross of purple with green, was most conven- iently considered with that cross in generations later than F2. The results of this second cross have been in accord with expectation. The other thirteen intercrosses are now to be considered, together with intercrosses of some types that are phenotypically alike. Dilute sun red IVa x green Via. VIb, Vic. — The progenies of self- pollinated green plants were listed in table 15 in several groups in accordance with what was learned of their genotypic constitution by the crosses to be considered here. The regular F3 lots, from self-pollinated F2 greens of self-pollinated Fi purples, were put in group 1 of table 15. Only one of the same F2 greens (table 16, group 2) was crossed with homo- zygous dilute sun red, A Ahhpl pi. The result was 67 dilute purple plants. Another green plant, an F3 from a self-pollinated F2 green, gave, when similarly crossed, 9 dilute purple plants (group 2). Evidently both these green plants were a abb PI PI. Four other F3 green plants, when crossed with dilute sun red, gave a total of 148 sun red plants (group 1, table 16). One of these four belonged to an F3 lot containing browns and greens in a 3:1 relation, and could not, theoretically, have done other than give all sun red or all dilute purple when crossed with dilute sun red. Two of the four were from greens of an F3 lot made up of purples, sun reds, browns, and greens, and were therefore assumed to be aa B B pl pl, as the crosses with dilute sun red showed them to be. One of the four green plants, however, belonged to an F3 lot of browns and greens in a 9:7 relation and was consequently comparable to an F2 green. A sixth F3 green also belonged to a 9 : 7 lot, comparable to an F2 lot. When crossed with dilute sun red (group 3, table 16), it gave 24 dilute sun red plants, Plant Colors in Maize 49 and is therefore assumed to have been a abb yl pi. All three of the theoretically possible homozygous genotypes have therefore been demon- strated among the Fo greens or among Fs's comparable to F2's. In addition to the green plants of the direct Fo and F.s generations, noted above, fifteen other greens were crossed with dilute sun red. All these greens belonged to a single progeny, 2019, which was the result of a backcross of an Fi purple with a green, a abb pi pi (table 3, group 1). All of them should therefore have been heterozygous for B or PI, or have lacked these dominant genes. Seven of the fifteen, when crossed with dilute sun red, gave 110 sun red and 85 dilute sun red plants (group 4, table 16), a deviation from equality of 12.5 ± 4.7. The green parent plants are consequently regarded as a a B b pi pi. Five others of the fifteen green plants (group 5) gave a total of 50 dilute purple and 65 dilute sun red, a deviation from equality of 4.5 ± 3.7, and hence are assmned to have been a abb PI pi. Three of the fifteen (group 6) gave a total of 106 dilute sun red plants. These three must, it is supposed, have been a abb pi pi. Naturally, in the course of the writer's maize studies, many other crosses between green and dilute sun red have been observed. But no purpose can be served by presenting here all this mass of data. Much of it has accumulated in connection with a study of the interrelations of plant and aleurone color, and will find its appropriate place in a later publication on that topic. A few Fo and backcross progenies of dilute sun red Fi's of such ciosses are, however, listed in table 17 (page 134), to serve as an indication of the behavior of all. Three F.; progenies (group 1, table 17) contained 269 dilute sun reds and 99 greens, a deviation from the expected 3: 1 ratio of 7 ± 5.6. Five progenies of Fi dilute sun reds back- crossed to green Vic (group 2) included 357 dilute sun reds and 358 greens, a deviation from the expected 1 : 1 ratio of only 0.5 ± 9.0. The behavior of a number of the sun red and dilute^ purple plants listed in table 16 has been studied in F2 and later generations. Consideration of this later behavior is conveniently deferred to a later section of this paper (pages 51 and 53), where it is taken up with other crosses which should theoretically give similar results. Green x green. Via, VIb, Vic. — A number of green plants of progcMiy 2019, discussed above, were intercrossed. That these green plants l)rcd true green when selfed was shown by the records of table 15 (groups 3 50 R. A. Emerson to 5) . That they were of three distinct genotypes was shown by the data recorded in table 16 (groups 4 to 6). The behavior of random intercrosses of the same green plants is now to be considered. The data are given in table 18. The green plants that served as parents of the crosses listed in group 6 of table 16, it was decided, must have been a a bbpl jd. When such plants are crossed with green plants of any of the other genotypes, nothing but green plants should result. A single cross of one of these greens with a green of the constitution a a B b pi pi (table 16, group 4) gave 23 green plants (table 18, group 1) as expected. Another cross of one of these greens with a green of the genotype a abb PI pi (table 16, group 5) gave 22 green plants (table 18, group 2). Crosses of green plants belonging to like genotypes should, of course, give only green plants. Three crosses of plants shown to be a a 5 6 pi pi (table 16, group 4) gave 72 green plants (table 18, group 3) . A single cross between plants shown to be a ah b Pi pi (table 16, group 5) gave 24 green plants (table 18, group 4). Five crosses of plants of genotype a a B b pi pi with plants of genotype a abb PI pi gave a total of 40 brown and 105 green plants (table 18, group 5). Here a 1:3 ratio of brown to green is to be expected. The theoretical numbers are therefore 36 and 109, respectively, and the devia- tion is 4.0 ± 3.5. The important fact here is that all these intercrosses of greens gave the color types expected on the basis of the results of crosses of the same individual green plants with dilute sun reds. The writer deems himself fortunate in having been able to obtain results approxunat- ing so closely a complete demonstration of the several genotypes of green, since the selfing, the crossing with dilute sun reds, and the inter- crossing of greens, were made at the same time, with the green plants chosen wholly at random. Brown V x green Vic. — When brown plants are crossed with green plants of type Vic, the Fi plants arc brown, and browns and greens alone appear in F2. Since brown is supposed to be a B PI and type Vic green a b pi, the F2 progenies should exhibit 9 : 7 ratios. Eleven F2 progenies are listed in table 19 (page 135), with a total of 317 brown and 223 green plants. The theoretical numbers are 304 and 236, respectively, showing a deviation of 13 ± 7.8. There is more than one chance in four that such a deviation is due to errors of random sampling, P equaling 0.27. Plant Colors in Maize 51 Of any nine F2 brown plants of this cross, theoretically one should breed true in F3, four should give a 3:1 ratio, and four should give a 9:7 ratio. Six F^'s were tested, with the results shown in table 20. Two bred true, with a total of 29 brown plants (group 1). Two gave ratios classed as 3:1, the totals (group 2) being 100 brown to 40 green, a devia- tion of 5.0 ± 3.5. Two gave progenies interpreted as 9:7 (group 3), totaling 39 l)rown and 39 green, the deviation being 5.0 ± 3.9. Of the 3:1 Fs lot, two browns bred true in F4, producing 59 brown plants, and one green l^red true, producing 56 green plants. The distribution of the F2 brown plants with respect to their F3 behavior — two breeding true, two throwing a 3:1 ratio, and two a 9:7 ratio — was as near expectation, 1:4:4 in nine, as could perhaps be expected from such small nimibers. If these six F2 browns are combined with the four- teen F2 browns of the original cross of purple x green noted earlier in this paper (page 44), a very good fit of the hypothesis and observation is found (z2 = 0.88). Theoretically these two lots of F2 browns should be of the same genotypes, so that they may well be so combined. The comparison follows : F3 ratios 1:0 3:1 9:7 Total Observed 2 11 7 20 Calculated 2 9 9 20 Difference 0+2 —2 Sun red II a x green Vic— When both parents are homozygous, the cross of green of type Vic with sun red results in sun red plants only. Three such crosses gave 112 sun red plants. Crosses with heterozygous sun red plants gave Fi progenies of sun red together with dilute sun red or green or both, depending presumably upon whether one or the other or both of the factors A and B were heterozygous. Fi sun red plants of such crosses are presumed to have the formula AaBhpl-pl, and should therefore produce in F2 the three color types sim red, dilute sun red, and green, in the relation 9:3:4. Sixteen F2 progenies of such crosses are listed in table 21, group 1 (page 136). It has already been shown (page 48) that crosses of some green plants, a B pi, with dilute sun reds, A h pi, give sun red Fi offspring, which are also assumed to be A a Bh pi pi. Five F2 progenies of such crosses are, for convenience, considered here 52 R. A. Emerson (group 2, table 21) with the crosses of sun red and green. While certain of the individual progenies, due perhaps to the small numbers concerned, deviate considerabl}^ from the expected results, the twenty-one progenies (groups 1 and 2, table 21) taken together approach so closely to expectation that there is more than one chance in four that the observed deviations may be due to errors of random sampling, P equaling 0.28. The com- parison of observed with expected numbers follows: Color types Sun red g^^j^^Vg^i Green Total Ila IVa Via, c Observed : Ila xVIc 827 268 383 1,478 IVa X Via 343 120 179 642 Total 1,170 388 562 2,120 Calculated 1,193 398 530 2,121 Difference —23 —10 +32 —1 Fi sun red plants, A a B h pi pi, were also backcrossed with green plants of type Vic, a b pi. Fifteen progenies of these backcrosses are listed in table 21, the progenies from the cross Ila x Vic in group 3 and those from the cross IVa x Via in group 4. The expected relation of 1:1:2 was realized fairly well in the results, the odds against the observed devia- tions' being due to chance being about three to two, P equaling 0.39. The observed and expected results are compared as follows: . Color types Sun red ^^"^^^^ Green Total Ila IVa Via, c Observed : (Ila x Vic) X Vic 134 123 267 524 (IVa X Via) X Vic 442 465 962 1 ,869 Total 576 588 1 ,229 2 ,393 Calculated 598 598 1 .196 2 ,392 Difference —22 —10 +33 +1 Plant Colors in Maize 53 Dilute purple Ilia x green Vic. — Since dilute purple differs from sun red merely in having the dominant PI factor instead of B, crosses of dilute purple with green of type Vic should behave just as did the crosses con- sidered in the preceding section, except that dilute purples take the place of sun reds in the progeny. Eight crosses of dilute purple with green of type Vic resulted in 91 dilute purple plants. The F2 results of these crosses are given in table 22, group 1. Since the Fi plants of these crosses are assumed to have been A ah b PI pi, the F2 results should be the same as those expected from crosses of greens of type VIb with dilute sun reds. The Fi's of the latter crosses have already been discussed (page 48). The F2 results, six progenies, are for convenience considered here (group 2, table 22) . "Wliile the expectation of a 9 : 3 : 4 relation was not very closely realized in the ol)served results, such deviations as those found might be expected thru chance about once in eight trials, P equaling 0.13. The comparison of observed and expected distributions follows: Color types ^^^ ^ G.een Total Ilia IVa VIb, c Obsei'ved : Ilia X Vic 416 149 173 738 IVa x VIb 274 102 107 483 Total 690 251 280 1 ,221 Calculated 687 229 305 1,221 Difference +3 +22 —25 A single Fi plant backcrossed with green gave the same three color types in the relation 26:20:56. The theoretical distribution is 25.5:25.5:51.0. Deviations of the observed order might be expected somewhat more than twice in five trials, P equaling 0.44. Seven F2 greens bred true in F3 with a total of 359 individuals. One dilute sun red F2 plant bred true with a progeny of 156 dilute sun red plants. Of the F2 -dilute purples, some bred true, some threw the three types seen in F2, some gave only dilute purple and dilute sun red, and some gave only dilute purple and green. Notwithstanding the rather poor fit in F2, therefore, the fact that practically all the expected .classes 54 R. A. Emerson of behavior were exhibited in F3 makes it seem Ukely that the deviations in F2 were due mainly to chance. Sun red II a x brown V. — A single cross of brown with sun red gave purple plants only, as was expected. Since both parents were homozygous, all the Fi plants should have been of the genotype Aa B B Plpl and should have produced in F2 the four types purple, sun red, brown, and green, in the relation 9:3:3:1. The three F2 progenies of this cross are recorded in table 23 (page 137). The expected color types were produced in approximately the expected numbers. The odds against the observed deviations' being due to chance are three to two, P equaling 0.40. A comparison of observed with expected distributions follows: Color types Purple Sun red Brown Green Total la Ila V Via Observed 120 29 37 10 196 Calculated 110 37 37 12 196 Difference +10 —8 —2 Purple la x brown V . — Crosses of brown with purple gave purple Fi's, and four F2 progenies gave a total of 116 purple and 38 brown plants, which is very near the 3:1 ratio expected from Fi plants of the genotype A a B B PI PI, the deviation being 0.5 ± 3.6. Nine Fi purples backcrossed to browns gave progenies totaling 484 purple and 477 brown plants, a deviation from the expected equality of 3.5 ± 10.5. Purple la x sun red Ila. — Purples and sun reds should differ by a single factor pair, PI pi. The Fi purples backcrossed to sun red should give a 1 : 1 ratio of the parental types. Five such backcrosses gave 47 purple and 57 sun red plants, a deviation from expectation of 5 ± 3.4. No progenies of selfed Fi's were observed. Purple la x dilute purple Ilia. — Purples are assumed to differ from dilute purples by the factor pair B b. Six Fi purples backcrossed with dilute purple gave 40 purple and 52 dilute purple plants. This is a deviation from the expected equaUty of 6 i 3.2. No other tests of the cross of purple x dilute purple were made. Sun red Ila x dilute sun red IVa. — Sun reds and dilute sun reds should differ in one factor pair, B b, and should therefore give a simple 3 : 1 result in F2. The Fi generation of six crosses of these color types consisted of 135 sun red plants. Sixteen F2 progenies Hsted in group 1 of table 24 Plant Colors in Maize 55 (page 138) totaled 998 sun rod and 314 dilute sun red, a deviation from the 3:1 ratio of 14 db 10.6. Fourteen backcrosses of Fi sun red plants with dilute sun reds (group 2, table 24) resulted in 811 sun reds and 742 dilute sun reds, a deviation from the expected equality of 34.5 13 ±.3. Two F2 dilute sun reds bred true in F3 as expected (table 25, group 1), with a total of 50 dilute sun red offspring. Two F2 sun red plants (group 2) gave a total of 19 sun reds in F3, and a third F2 plant, on backcrossing with dilute sun red, gave 101 sun reds. Four other F2 sun red plants gave both sun reds and dilute sun reds in their F3 progenies (group 3), the respective numbers being 373 and 127; the calculated numbers are 375 and 125, respectively, showing a deviation of 2 ± 6.5. Of the seven F2 sun reds tested, four were heterozygous and three apparently homozygous for the B factor. On the whole, therefore, the crosses of sun red with dilute sun red behaved approximately as expected. Dilute purple Ilia x dilute sun red IV a. — Five crosses of dilute sun red with dilute purple gave a total of 344 Fi plants, all dilute purple. Since these Fi's are supposed to be heterozygous for the PI factor only, a 3:1 Fo distribution of color types should result. Seven F2 progenies Usted in group 1 of table 28 (page 139) had a total of 261 dilute purple and 87 dilute sun red plants, exactly a 3:1 relation. Five Fi plants were backcrossed with dilute sun red (group 2) and resulted in 275 dilute purples and 263 dilute sun reds. The deviation from the theoretical 1 : 1 relation is 6 ± 7.8. Only two F2 dilute purples were tested by their F3 behavior. Neither bred true, the total produced being 38 dilute purples to 17 dilute sun reds, a deviation from the 3:1 ratio of 3.3 ± 2.2. As far as they go, then, the results are in close agreement with what is expected of the crosses here under consideration. Sun red Ila x dilute purple Ilia. — Theoretically, crosses of sun red, A B pi, with dilute purple, A b PI, should give purple, A B PI, in Fi. Two crosses, as shown in group 1 of table 27 (page 140), gave a total of 24 purple and no other types. Here the parents were doubtless homozygous. If one or the other of the panMits is heterozygous^ two color types are to be expected in Fx. A single cross (group 2. table 27) gave 74 purple and 75 sun red plants. Such a result is to l)e expected when the sun red parent is homozygous, A A B B pl pi, and the dilute purple parent is heterozygous, A Abb PI pl. Two other crosses (group 3) gave 56 R. A. Emerson a total of 28 purple and 29 dilute purple plants. The parents are therefore assumed to have been A A B h pi pi and A A hb PI PI, tho the same results should have been obtained if one or the other, but not both, of the parents had been A a. The important point here is that purple plants were produced in all crosses, showing that sun red and dilute purple carry complementary factors for purple. The factors are assumed, in keeping with the hypothesis under test, to be B and PI. In accordance with this hypothesis, the Fi purple plants should be A A B b PI pi and should throw four color types in F2. No direct F2 progenies have been observed^ but seven progenies from backcrosses of Fi purples with dilute sun reds are recorded in table 28. While the deviations from the expected equality among the four classes are rather large, they are not greater than might occur by chance about once in four trials, P equaling 0.26. The comparison follows: Color types Purple Sun red ^^^"l"" ^'^^'^^ Total ''^ ^ purple sun red la Ila Ilia IVa Observed 99 110 104 83 396 Calculated 99 99 99 99 396 Difference +11 +5 —16 Purple la x dilute sun red IVa.— Crosses of purple with dilute sun red should give purple Fi plants, A A B b PI pi, and 9:3:3:1 F2 progenies. Four such crosses resulted in 65 purple plants in Fi. The F2 results are reported in table 29, group 1. The distribution of the individuals of the twenty-six progenies taken together is shown below in comparison with the calculated distribution. The four color types expected were observed in approximately the expected numbers. Deviations such as shown might be expected thru chance about twice in eleven times, P equaling 0.18. Color types Purple Sun red ^'*"f^ ^^^"*^, Total •^ ^ ^ purple sun red la Ila Ilia IVa Observed 1 ,013 316 • 296 100 1 ,725 Calculated 970 323 323 108 1 ,724 Difference +43 —7 —27 —8 +1 Plant Colors in Maize 57 Some of the Fi purple plants were crossed back to dilute sun red, with results as given in group 2 of table 29 and sumnmrized below. The seventeen progenies together approached the expected equality of the four color types so closely that the observed deviations might be expected thru chance more than twice in five trials, P equaling 0.44. Color types Purple Sun red ^^^^^^ ^^^"^^, Total •^ ^ ^ purple sun red la Ila Ilia IVa Observed 323 306 325 289 1 ,243 Calculated 311 311 311 311 1,^4 Difference +12 —5 +14 —22 —1 Sixteen Fo purple plants were tested b}^ their Fs progenies (table 30). Seven F2 purples (group 1) gave again the four color types purple, sun red, dilute purple, and dilute sun red, the several classes being I'epresented by 268, 105, 78, and 28 individuals, respectively, while the calculated numbers were 269, 90, 93, and 30. The odds against such deviations being due to chance are about three to one, P equaling 0.24. One of the seven F2 purple plants was crossed with green a abb pi pi and gave the same four classes of progeny, represented by 26, 25, 24, and 21 plants, respectively. Evidently these F2 purples were like the Fi's, A A Bb PI pi. Four other F2 purples (group 2, table 30) gave only purple and sun red progenies. Three of these when sclfed gave 60 purple and 22 sun red. Two of these three and one other, when backcrossed with dilute sun red or green, gave 32 purples and 31 sun reds. The four F2's are therefore regarded asAABBFl pl. Five F2 purples (group 3) gave purples and dilute purples only. Four of these, which were selfed, gave 162 purples and 48 dilute purples, while the fifth, which was backcrossed to dilute sun red, gave 17 purples and 15 dilute purples. These five F2's are consequently regarded as AABb PI PI. None of the sixteen F2 purples tested bred true in F.-), A A B B PI PI. A single? F3 purple (group 6), however, which occurred in the ¥i lot showing the four color tj^pes (group 1) and which was therefore comparable to the F2 purples, Ijrcd true in Fi, pi-oducing 69 purples on being selfed and 18 on being backcrossed to green. Of three other F3 purples of the same 58 R. A. Emerson Fs lot, two (group 4) gave only purples and sun reds, and one (group 5) gave only purples and dilute purples. The twenty .Fo and Fs purples tested, therefore, were distributed with respect to the four kinds of behavior in the relation 7:6:6:1, in contrast to the calculated distribution of approximately 8.9:4.4:4.4:2.2. There is more than an even chance that such a difference may be due to errors of random sampling, P equaling 0.53. On the whole, therefore, the F2 purples of this cross behaved in later generations as was expected of them. F2 sun red plants of the cross purple x dilute sun red showed two types of behavior in F3 (table 31, group 1). Three Fo's bred true, with 53 sun red plants in F3. Four gave a total of 70 sun red and 24 dilute sun red plants. Where an expected ratio of one true breeding to two segregating progenies was expected, the observed relation of three to four is not a bad fit. F2 dilute purples also showed the two types of behavior expected in F3 (group 2, table 31). Three bred true, with a total of 97 dilute purple plants, and six gave a total of 217 dilute purple and 86 dilute sun red plants. The 1:2 ratio was therefore exactly realized. Three F2 dilute sun reds bred true in F3 (group 3) as was expected of them, producing a total of 72 dilute sun red plants. Numerous F3 plants of the several color types of the cross under con- sideration here were tested by F4 and F5 progenies, with results wholly consistent with expectation. It is deemed unnecessary to give the records of these later generations in detail. Evidence from aleurone-color and linkage relations The evidence presented up to this point in support of the three-factor hypothesis, involving A a, Bh, PI pi, has had to do .with the behavior of the several F2 color types in later generations and in intercrosses. There remain to be discassed some bits of evidence which, while less direct, are perhaps no less trustworthy. This evidence deals with (1) the relation of aleurone color to plant-color types, (2) the linkage of certain plant-color types with endosperm color, and (3) the Hnkage of other color types with the liguleless leaf. Relation of aleurone color to plant color. — Of the plant-color factors considered in this section of the paper, the pair A a is concerned also in the development of alem'one color. It has been shown by the writer Plant Colors in Maize , 59 ill a previous paper (Emerson, 1918) that the presence of three dominant factorS; A, C, and R, is necessary for the development of aleurone color. It is assumed that the factor pair A.a for aleurone color is identical with the pair A a for plant color. Some of the evidence on which this assumption is based may well be considered at this point in order to justify the use of the same symbols for both plant and aleurone color. After the identity oi Aa has been estal)lished, certain relations of aleurone color to plant color can be used to check up some of the conclusions previously drawn with respect to the genetic interrelations of the several plant-color types. It will be recalled that dilute sun red crossed with green gave dilute sun red in Fi and a 3:1 ratio of the two types in F2 (table 17, group 1, page 134), and that backcrosses of Fi with green gave a 1 : 1 ratio (group 2). The F2 seeds of these Fi plants also exhibited a 3: 1 relation — 424 colored and 127 colorless, deviation 10.8 ± 6.9 — thus showing that only one factor pair, A a, C c, or R r, was heterozygous. The colorless seeds produced 98 green plants, and the colored ones produced 269 dilute sun reds and 1 weak plant, recorded as green, which died in the seedling stage. Obviously the factor that differentiates dilute sun red from green is the same as the one that in these cases differentiated the colored from the colorless seeds, or some factor very closely linked with it. Fortunately, Fi plants closely related to the ones which when selfed showed the behavior noted above, were backcrossed with green, colorless-seeded A testers (Emerson, 1918). Of the resulting seeds 632 were colored and 590 were colorless, evidently a 1:1 relation — the deviation being 21 ± 11.8 — showing that the Fi plants were, with respect to aleurone color, A a C C R R. The colored seeds gave rise to 357 dilute sun red plants and the colorless seeds to 358 green plants. Evidently, therefore, it is the Aa pair that differentiates dilute sun red from green. This is in support of the assumed genotypes A h pi and a b pi for dilute sun red and green, x^spectivel}'. The single progeny recorded in group 3 of table 9 (page 127) came from a plant known to be A a with respect to aleurone color and pro- ducing 130 colored and 41 colorless seeds. The 3:1 aleurone-color relation shows it to have been heterozygous in only one aleurone-color factor, and therefore AaCCRR. The colored seeds, ACR, produced 67 purple plants, and the colorless ones, aC R, produced 21 brown plants. 60 R. A. Emerson Evidently, purples are differentiated from browns by the A a pair alone, just as dilute sun reds are differentiated from greens. This is quite in keeping with the assmned genotypes, A B PI and a B PI, for purple and brown, respectively. Two of the progenies recorded in group 3 of table 8 (page 126) involved both aleurone and plant color. The heterozygous parents were back- crossed with green A testers and produced 125 colored and 127 colorless seeds. The factor pair differentiating these two seed classes was therefore Aa. The colored seeds, A C R, produced 15 purple and 14 sun red plants, while the colorless seeds, aC R, gave 9 brown and 14 green plants. Since it is shown in the preceding paragraph that purples and browns differ with respect to the pair A a alone, it may be inferred that the sun reds and the greens of these lots also differed with respect to A a alone. The assumption heretofore made with respect to the genotypes of these color classes, A B PI, A B pi, a B PI, and a B pi, for pm"ple, sun red, brown, and green, respectively, is given support by this relation of aleurone color to plant color. Two of the progenies recorded in group 1 of table 9 (page 127), and one in group 4 of table 8 (page 126), were grown from self-polhnated plants known to be A a with respect to aleurone color and found to have 644 colored and 228 colorless seeds. The 3 : 1 seed-color relation shows them to have been AaC C RR. The colored seeds, A C R, gave 294 purples and 113 dilute purples, while the colorless seeds, aC R, gave 119 browns and 40 greens. If purples and brov/ns differ with respect to A a alone, as they have been shown to do, presumably the dilute purples and the greens of these lots also differ in the same way. This is in keeping with the assumption that the genotypes of the color classes are A B PI, A b PI, a B PI, and a bPl, for purple, dilute purple, brown, and green, respectively. These comparisons of the relations of aleurone color to plant color have confirmed definitely the supposition that purples, sun reds, dilute purples, and dilute sun reds have the dominant factor A, and browns and greens the recessive factor a. The comparisons have also afforded some support for the assumed genetic constitution of the several color types with regard to B b and PI pi. More definite evidence for" the latter, however, is afforded by the linkage relations now to be discussed. Liyikage of plant color with endosperm color. — It has been known since 1942 that a Unkage exists between the factor pau* PI pi and endosperm Plant Colors in Maize 61 color. The data siigsost irregularities or complexities which cannot be straightened out until more definite information is at hand with regard to the two or more factor pairs concerned in the development of yellow endosperm.^ Only such data are presented here as are necessary to show the relations of the several plant-color types to endosperm color, f A single progeny recorded in table 27, group 2 (page 140), was made up of 74 purple and 75 sun red plants. The lot resulted from a cross of a white-seeded sun red plant with a dilute purple plant which was heteroz3'-gous with respect to both yellow endosperm and plant color. The j^ellow seeds produced 58 purple and 20 sun red plants, and the white seeds produced 16 purple and 55 sun red plants. The yellow-seeded sun reds and the white-seeded dilute purples are known to be the crossover classes. The ratio of non-crossovers to crossoveis is 113:36, and the percentage of crossing-over, therefore, is 24.2. Evidently a factor pair for yellow endosperm, Y y, is linked with the factor pair that differentiates purple from sun red. In accordance with the hypothesis under test, this plant-color factor pair is PI pi — purple = ABPl, and sun red = ABpl. Two other progenies (table 26, group 1, page 139) had a total of 116 dilute purple and 42 dilute sun red plants. The selfed parent plants were heterozj^gous for yellow endosperm as well as for plant color. The yellow seeds gave 99 dilute purple and 17 dilute sun red plants, and the white seeds gave 17 dilute purple and 25 dilute sun rod plants. This F2 distribution, as shown below, is very close to expectation ( z^ = 0.26) on the basis of 25 per cent of crossing-ovei between the factor pair Yy and the pair that differentiates dilute purple from dilute sun red. It seems likel}^ therefore, that the same plant-color factors, PI pi, are concerned here as in the progeny consisting of purples and sun reds. This is in keeping with the theoretical genotypes, A b PI and A b pi. assmned for dilute purple and dilute sun red, respectively. The comparison between the observed Fo distribution and that calculated on the basis of 25 per cent of crossing-over follows: Observed 99 17 17 25 = 158 Calculated, y 102 17 17 23 = 159 Difference —3 +2 —1 ' This problem is being investigated by Dr. E. G. Anderson. 62 R. A. Emerson A single progeny (table 8, group 3, page 126) from a selfed parent heterozygous for yellow endosperm, contained purple, sun red, brown, and green plants, totaling 63, in the relation 35:15:6:7. These four color types are expected to occur in a total of 64 in the relation 36: 12: 12:4 from a selfed plant of the genotype AaB B Plpl. The observed deviation from expectation might occur by chance once in nine trials, P equaling 0.11. Theoretically, the green plants of this lot, aB pi, are differentiated from the browns, a B PI, by the same factor pair, Plpl, that differentiates the sun reds, A B pi, from the purples, A B PI. If this is true, the same linkage relations should exist for yellow endosperm with the brown-green lot as with the purple-sun-red lot. From yellow seeds there came 29 purples and 8 sun reds, and from white seeds 6 purples and 7 sun reds. Such a distribution should be very closely realized ( /.^ = 0.97) from 30 per cent crossing-over between Y y and PI pl. The yellow seeds produced also 5 brown and 3 green plants, and the white seeds 1 brown and 4 green plants. While the number of individuals is too small to give a reliable indication, it is of interest to note that the coefficient of asso- ciation (Collins, 1912) calculated from the series 5:3:1:4, or 0.739, is practically that calculated from 26 per cent of crossing-over. In so fai" as these records go, therefore, they support the assumption that brown and green in this lot are differentiated by the same factor pair as are purple and sun red, and thereby support the hypothesis under test. A plant heterozygous for the three plant-color pairs A a, B b, PI pl, and for Yy, backcrossed with a white-seeded green plant of type Vic, a h pl y, gave the six color types, purple, sun red, dilute purple, dilute sun red, brown, and green, in the numerical relation 10:13:17:11:9:33 (ta])le 6, page 124), which is a close fit (P = 0.61) to the expected relation, 1:1:1:1:1:3. From yellow seeds the resulting series was 8:6:13:2:7 :.17, and from white seeds it was 2:7:4:9:2:16. When the classes having A Pl, purple and dilute purple, were lumped together, and similarly those having A pl, sun red and dilute sun red, the yellow seeds gave 21 plants wifh Pl and 8 with pl, while the white seeds gave 6 with Pl and 16 with pl. Of these 51 plants, there were 14 in the crossover classes, or a percentage of crossing-over of about 27.5 ±4.1, approximately the same as in the cases cited above. In this lot there are theoretically three kinds of greens, a B pl, ah Pl, and a h pl, one of which has Pl and two of which have pl, while all the browns, a B Pl, have Pl. If there be Plant Colors in Maize 63 assumed 25 per cent of crossing-over between Fy and PI pi, equivalent to a 3:1:1:3 gametic series, yellow seeds should give 3 brown to 5 green, and white seeds 1 brown to 7 green, as shown below: Yellow White Brown, a B PI 3 1 Green, a B pi 1 3 Green, ab PI 3 1 Green, abpl 1 3 The yellow seeds actually gave 7 brown to 17 green and the white seeds 2 brown to 16 green, which is a close fit to the calculated relation, 3:5: 1:7 (P = 0.59). In this case as in the others, then, the linkage relations between Y y and PI pl afford additional support for the belief that the several color types actually bear to one another the relation assumed in the assignment of hypothetical genetic formulae (page 32). Linkage of plant color icith leaf type. — It has been known for some years that a leaf type termed liguleless (Emerson, 1912) is linked with the factor pair that differentiates sun red from dilute sun red. As an illustration of this, two backcross progenies, 8250 and 8253, with a total of 145 sun red and 147 dilute sun red plants, may be cited. These progenies came from a cross of normal-leaved sun red, A B pl Lg, with liguleless- leaved dilute sun red, A b pl lg, backcrossed with liguleless dilute sun red. Of the normal-leaved plants 104 were sun red and 41 were dilute sun red, while of the liguleless-leaved plants 48 were sun red and 99 were dilute sun red. The non-crossovers were to the crossovers as 203:89, or a per- centage of crossing-over of 30.5. Since the factor pair that differentiates sun red from dilute sun red has been assigned the symbol B b, the linkage noted here is evidently between B b and Lg lg. Six progenies from backcrosses of heterozygous normal-leaved purples with liguleless dilute sun reds gave purples, sun reds, dilute purples, and dilute sun reds in the relation 197:177:178:167, which is not far from the equality expected, P equaling 0.46. Among the normal-leaved plants, the four color types occurred in the relation 123:117:47:55, and among the liguleless-leaved plants in the relation 74:60: 131: 112. Evidently the purples bear the same relation to the dilute purples as the sun reds do to 3 64 R. A. Emerson the dilute sun reds. For sun reds and dilute sun reds, the non-crossovers are to the crossovers as 229:115, or a crossover percentage of 33.4 ± 1.7. For purples and dilute purples, the relation is 254:121, or a crossover percentage of 32.3 ± 1.5. It follows from this that the factor pair, B h, which differentiates sun red, A B -pi, from dilute sun red, A b pi, is the same as that which differentiates purple from dilute purple. And this is in keeping with the hypothesis under test, in accordance with which purple and dilute purple have been assigned the genotypes A B PI and A b PI, respectively. In a single progeny resulting from a backcross of a heterozygous normal- leaved purple plant with a liguleless-leaved green plant, greens occurred, as expected, with about three times the frequency of the average of the other five color classes. The progeny included 14 browns and 49 greens. Of the normal-leaved plants there were 10 browns and 19 greens, and of the liguleless-leaved plants 4 browns and 30 greens. On the basis of the hypothetical genotypes assigned to browns and greens, and with the assumption of 33 per cent of crossing-over between B b and Lg Ig, the four classes, normal brown, normal green, liguleless brown, and hguleless green, should bear the relation 2:4:1:5. For a total of 63 plants, the relation would be approximately 11:21:5:26, whereas the observed relation was 10:19:4:30. The deviations from expectation are such as might occur by chance in more than three out of four trials, P equaling 0.78. In this case, as in the others reported, the linkage relations between B b and Lg lg afford support for the view that the several color types bear the relation to one another inferred from the hypothetical genotypes assigned them. Summary of results involving A a, Bb, PI pi The results of the cross of purple with green — which gave in F2 six color types, namely, purple, sun red, dilute purple, dilute sun red, brown, and green, with a numerical relation of approximately 27:9:9:3:9:7 from selfed Fi's and about 1:1:1:1:1:3 from Fi's backcrossed to green — have been interpreted on the basis of the interaction of three factor pairs, A a, Bb, and PI pl. This hypothesis has been subjected to practically every genetic test available, as summarized below. Each of the six Fo color types has in turn been tested by its behavior in F3, and in several cases behavior in F4 and even in later 'generations Plant Colors in Maize 65 has been noted. All the possible combinations of intercrosses between the several types have been studied, except dilute purple x brown. In most cases these intercrosses have been carried to the F2 generation, and in several instances to F3 and F4. Thruout the tests, the results have been in close agreement with those expected from the hypothesis. In almost every instance all the color types expected in each generation of the several crosses, and no others, have appeared. Moreover, the numerical relations found to exist between the several color types and also between the several classes of behavior have been reasonably close to expectation. It is true that in some instances the fit of observation to hypothesis has not been particularly good, but even here the observed deviations have been of such an order as might be expected to occur occasionall}^ thru the chance errors of random sampling. In addition to the tests afforded by the behavior of the several Fj color types in later generations and in intercrosses, the relations of aleurone color involving the factor pair A a to tlie several plant colors, and the linkage relations of the plant-color factors PI pi with the endosperm-color factors Y ij and of the plant-color factors B h with the leaf-type factors Lg Ig, have been included in the investigation. These tests have shown that the several color types bear to one another the relations to be deduced from the hypothetical genotypes assigned them. The conclusion seems justified, therefore, that the three-factor hypoth- esis proposed as an interpretation of the F2 results obtained in crosses of purple with green has been substantiated, in so far as it is possible to substantiate any hypothesis. CROSSES INVOLVING THE MULTIPLE ALLELOMORPHS B, B'^ , b\ b In the preceding section of this account, six color phenot3'pes of maize have been discussed, namely, purple, sun red, dilute purple, dilute sun red, brown, and green. In addition to these six phenotypes, green plants have been shown to consist of three genotypes, which in some instances are slightly different phenotypically. Besides these six sharply separable phenotypes, there exist certain intermediate forms. The constancy of the.se types from year to year, under fairly uniform environmental conditions, leaves no doubt that they are genotypically as well as pheno- typically distinct from the types considered heretofore. 66 R A. Emerson One of these forms, known as weak purple, type lb, is intermediate in certain respects between purple and sun red, and in other respects between purple and dilute purple. Plants of this type, prior to the flowering stage, frequently resemble sun reds more than purples. The pigmentation of the sheaths is less intense than with purples, and in some instances less than with strong sun reds. There is, however, sooner or later a tendency for pigment to develop on the stalk beneath the sheaths (Plate V, 2) , In this respect weak purples resemble dilute purples as the latter often appear in a' late stage of their development. The anthers of weak purples are usually full purple, like those of purples and dilute purples, in which respect they show no resemblance to sun reds. A second intermediate form, known as weak sun red, type lib, stands between sun red and dilute sun red. The sheaths and husks are less extensively and less intensely pigmented than is true of full sun red, and yet exhibit much more color than in dilute sun red (Plate V, 4). The anther color of weak sun red is like that of both sun red and dilute sun red. While the difference between the extreme sun-color types, sun red and dilute sun red, is probably only a quantitative one — as is also presumably true of the difference between purple and dilute purple — little difficulty is experienced in separating sun red from dilute sun red plants on the one hand, or purple from dilute purple plants on the other. Frequently, however, it is difficult, or even impossible, at early stages of plant growth, to separate sun reds from purples. The existence of such intermediate forms as weak purple and weak sun red adds materially to the difficulties of classification. In fact, correct classification of all these types by inspec- tion alone is possible only at the flowering stage. For certainty in classi- fication, even at the flowering stage, environmental conditions, particularly soil fertility, must have been favorable thruout the growing period of the plants. While infertile soil exaggerates the difference between dilute sun red and green, by bringing about an excessive development of red ■ pigment in the one type while no color develops in the other, on fertile soil only are revealed the finer distinctions between sun red, weak sun red, and dilute sun red. It is perhaps fortunate that the genetic relations of these several types are such that ordinarily not all of them occur in a single progeny. Plant Colors in Maize 67 Interrelations of sun red Ila, weak sun red lib, and dilute sun red IVa Numerous crosses of weak sun reds, lib, with dilute sun reds, IVa, have given weak sun reds in Fi and approximately three weak sun reds to one dilute sun red in F2, just as crosses of strong sun red with dilute sun red give three strong to one dilute sun red (table 24, group 1, page 138). Records of such crosses are given in table 32 (page 144). Twelve F2 progenies, totaling 1729 individuals, showed the two types in the relation 1300:429, almost exactly a 3:1 ratio, the deviation being 3.3 ± 12.1. The data for F3 of these crosses are like those for crosses of strong sun red with dilute sun red (table 25). One weak sun red F2 bred true in F3 with a total of 77 weak sun red offspring (table 33, group 1). Four others gave both weak and dilute sun reds (group 2), in the relation 128:54, a deviation of 8.5 ± 3.9 from a 3:1 ratio. One dilute sun red bred true (group 3), with 95 dilute sun red plants in F3. A cross of weak sun red, lib, with strong sun red, Ila, gave strong sun red in Fi and the two parent types in F2 in the relation 71 : 16, a deviation from the 3: 1 ratio of 5.75 ± 2.72. There is, therefore, nearly one chance in six that the observed deviation may be due to errors of random sampling, P equaling 0.16. In none of these crosses, strong with weak, weak with dilute, and strong with dilute sun red, have other than the parent types appeared in F2. If weak sun red is due to the action of some additional modifying factor, not heretofore considered, types other than those of the parents should have occurred in some of the crosses. The natural conclusion, therefore, is that weak sun red, lib, is due to an allelomorph of B and b, the pair concerned with the difference between sun red, I la, and dilute sun red, IVa. This third allelomorph, responsible for weak sun red, may well be designated B^. Further evidence in support of the assumption that an allelomorph of B and b is concerned with weak sun red is afforded by linkage studies involving strong, weak, and dilute sun red with loaf type. Evidence has been offered (page 63) to show that B b and Lg Ig are linked with about 30 to 33 per cent of crossing-over. A single progeny, 8252, from a sun hkI plant heterozygous for leaf type and plant color backcro.ssed to liguleless weak sun red, contained 108 sun red and 109 weak sun red plants. Of the normal-leaved plants 80 68 R- A. Emekson were sun red and 38 were weak sun red, while of the Hguleless-leaved plants 28 were sun red and 71 were weak sun red. The ratio of non-crossovers to crossovers is 151:66, or 30.4 ± 2.1 per cent of crossing-over. The percentage of crossing-over between Lg Ig and the factor pair differentiating sun red and weak sun red, B B"", is, therefore, practically the same as the linkage between Lg lg and B b. Four backcross progenies, 8246-8249, involving sun red, contained 469 weak sun red and 396 dilute sun red plants. Of the normal-leaved plants 153 were weak sun red and 261 were dilute sun red, while of the liguleless- leaved plants 316 were weak sun red and 135 were dilute sun red. The non-crossovers are to the crossovers as 577:288, or 33.3 ±1.1 per cent of crossing-over. Here again, therefore, the linkage between Lg lg and the factor pair differentiating weak sun red from dilute sun red, B^ b, is practically the same as that between Lg lg and B b or between Lg lg and BB"". From the facts (1) that in crosses between any two of the three types sun red, weak sun red, and dilute sun red, the third type is not produced, and (2) that the linkage value between Lg lg and the factor pairs differen- tiating weak sun red from sun red and from dilute sun red is approxi- mately the same as that between Lg ly and B b, it seems evident that weak sun red is due to a factor B^ belonging to the triple allelomorphic series B, 5"', b. It seems probable that this series of allelomorphs contains other members in addition to the three listed above, but there is at present little conclu- sive evidence in support of the idea. There are certainly several forms, commonly classed as dilute sun red, that differ considerably in the amount of red pigment developed, and certainly some of these differences are genetic. As is shown in the next section of this account, some of these differences, particularly with respect to silk, anther, and leaf-blade color, are due to the effect of the aleurone-color factors R r. Environmental conditions, particularly soil fertility, influence the development of this pigment so greatly that the prol^lem becomes a difficult one. There is, however, some evidence that at least two forms of dilute sun red are differentiated by a factor pair belonging to the series B, B^, b. These forms differ principally in the amount of color in the fresh husks (Plate VI, 1 and 2), and to some extent in the sheaths, which arc the plant parts most strikingly different in sun red, weak sun red, and dilute sun red. Plant Colors in Maize 69 A type of dilute sun red with stronger husk pigmentation than ordinary dikite sun red shows was crossed with an ordinary dihite purple. Leaf type also was involved in the cross. The Fi plants were dilute purples with somewhat more pigment in the husks of young ears than is usual with that type. A single progeny, grown from an Fi backcrossed with liguleless dilute sun red of a light type, consisted of 25 dilute purples and 18 dilute sun reds. Each of these classes was sorted with some difficulty into light and more strongly colored subclasses, in accordance with the amount of color on the husks of the young ears. Of the more strongly pigmented dilute sun reds 4 had normal and 6 had liguleless leaves, while of the lighter dilute sun rods 6 had normal and 2 had liguleless leaves. Of the more strongly colored dilute purples 4 had normal and 13 had liguleless leaves, while of the lighter ones 4 had normal and 4 had liguleless leaves. While these numbers are small and the behavior was somewhat irregular, it is perhaps noteworthy that the factor pair differentiating the lighter from the more strongly colored plants, of both the dilute sun red and the dilute purple classes, exhibited an apparent linkage with Lq Ig of a value not far from that observed between Lg Ig and B b, B B^, and B^ b. The observed percentages of crossing-over were 32.0 for the dilute purples, 33.3 for the dilute sun reds, and 32.6 for the entire lot. This evidence, slight as it is, plainly suggests a fourth member, b% of the B series of allelomorphs, which may be stated tentatively as B, B^, 6*, b. Relation of weak purple lb to purple la, dilute purple Ilia, and weak sun red lib By methods similar in the main to those outlined above, Dr. E. O. Anderson has been able to show that weak purple is differentiated from purple on the one hand and from dilute purple on the other by the same factor, B^, that differentiates weak sun red from sun rod and from dilute sun red. At the time when Dr. Anderson undertook to determine the genetic relations of weak purple, nothing was known of the relation of weak sun red to sun red and dilute sun rod as presented above. Further- more, there was no indication as to whether weak purple was dififerentiated from purple and dilute purple by an allelomorph of /^ 6 or of PI pi, or by some distinct factor pair that might modify the ordinary result of the interaction of the pairs A a, B h, and PI pi then known to be concerned in the production of plant colors. The evidence to be presented here 70 R- A. Emerson is taken almost wholly from Dr. Anderson's records, and the conclusions derived from it are his. It is with Dr. Anderson's permission and at his suggestion that, for the sake of completeness of this account of the inheritance of plant colors, his results are here presented. A cross of a weak purple lb with a homozygous dilute purple Ilia resulted in 25 weak purples only, while a cross of another weak purple with a homozygous dilute purple, a sib of the plant used in the first cross, gave 63 weak purples and 53 dilute purples. Two of the Fi weak purples were backcrossed to dilute purples, and a third to dilute sun red. The result (table 34, group 1, page 145) was 141 weak purples and 163 dilute purples, a deviation of 11 ±5.9 from equality.' Five crosses of weak purples with dilute sun reds gave a total of 32 weak purples and 25 dilute purples, a deviation from equality of 3.5 ± 2.5, while two other such crosses gave 29 weak purples only. Evidently these weak purple plants differed from dilute purples by a single factor pair. This pair could not have been PI pi, for the crosses of weak purple with dilute purple, A b PI, gave the same results as those with dilute sun red, Ab pi. This leaves the ■possibility that B b or some unknown factor pair was concerned. Three crosses of weak purple lb with purple la resulted in 52 purple plants. A single cross of weak purple with sun red Ila gave 18 purples. Evidently both purple and sun red carry some factor that acts to change weak purple fco purple. Unfortunately, no later generations of any of these crosses were grown, but it is evident from the Fi results and from what is known of the interrelations of purple, sun red, and dilute purple that the dominant factor B, common to both purple and sun red, is con- cerned in the change from weak purple to purple. Since the crosses of weak purple with dilute purple, A b PI, and with dilute sun red, A b pi, gave no purples, while crosses of weak purple with purple, A B PI, and with sun red, A B pi, gave purple, the Pljjl pair is not concerned in the difference between weak purple and purple any more than in that between weak purple and dilute purple. These results, however, do not exclude the possibility that weak purple may be Ab PI, like dilute purple, with the addition of some unknown dominant modifying factor. A single weak purple plant, which was, so far as known, imrelated to the weak purples considered above, when crossed with two unrelated dilute sun reds gave progenies consisting of 15 weak purples and 13 weak sun reds. Seven progenies of these Fi weak purple plants backcrossed Weak Dilute Dilute T'^ + r.l sun red purple sun red iotai lib Ilia IVa 526 460 537 2,004 501 501 501 2,004 Plant Colors ix Maize 71 with dilute sun reds are listed in table 34, ^mnp 2. These progenies consisted of four color types, weak purple, weak sun red, dilute purple, and dilute sun red, in the numerical relations given below: Color types ^^'^f ''^ purple lb Observed 481 Calculated 501 Difference —20 +25 — 4i +36 The deviations from equality of the four classes expected of a dihybrid are so great that they would not occur by chance alone more than once in twenty trials, P equaling 0.05. Dr. Anderson's notes indicate that there was considerable difficulty, in the case of two of the cultures, in dis- tinguishing dilute purple from dilute sun red. Whether this difficulty may account in part for the poor fit is not known. The outstanding fact, however, is the appearance of the four classes and no others. Since weak sun red is known to differ from dilute sun red by the factor pai' B^' b, the inference is clear that weak purple differs from dilute purple by Ihe same pair and by no others. The formvdae assumed for the four color types are, therefore, A B^ PI, A B"' pi, A b PI, and A b pi, respectively. If the foregoing conclusion is correct, crosses of weak sun reds with dilute purples should give weak purples in Fi and the same four color classes in F2 as are noted above for crosses of weak purple with dilute sun red. A single cross of a dilute purple with a homozygous weak sun red resulted in 18 weak purple plants. Two crosses of dilute purples with weak sun reds heterozygous for B^ b gave 12 weak purples and 11 dilute purples. That the production of weak purples in these crosses was not due to the b or PI factors of the dilute purple parents is evidenced by the fact that crosses of the same dilute purple individuals witli sun reds gave full purples in Fi. One of the Fi weak purples, A A B'^ b PI pl, of the above crosses was backcrossed with dilute sun red, A bpl, with the result (table 34, group 3) showa below. The expected equality of the Weak Dilute Dilute sun red purple sun red iotai lib Ilia lYa 28 22 27 98 24.5 24.5 24.5 98 72 R. A. Emerson four color types was closely approached in the results, x^ equalmg 0.80. The comparison of observed with expected results follows: ! Color tj^es "^^^^^ I '' ^ purple ' lb Observed 21 Calculated 24 . 5 Difference..- —3.5 +3.5 —2.5 +2.5 The progeny of a purple plant heterozygous for B B"^, PI pi, and the endosperm color pair Y ij, backcrossed with a white-seeded weak sun red plant, A B^ pi y, affords evidence of another kind with respect to the interrelations of strong and weak purple and of strong and weak sun red. It has been noted previously (page 60) that PI pi and Yy are Imked, with a somewhat irregular percentage of crossing-over. The backcross gave the four color types purple, weak purple, sun red, and weak sun red, in the numerical relation 60:48:59:62. The observed deviations from the equahty expected of a dihybrid are such as might occur by chance more than once in two trials, P equaling 0.54. The distribution of these 229 plants to the four color types when the progeny of yellow seeds and that of white seeds are considered separately is as follows: Color types Purple vv eaK purple Sun red la lb Ila Yellow seeds 48 36 12 8 White seeds 12 51 Total Weak sun red lib 17 109 45 120 Evidently weak purple, assumed to be A B"" Pi, here bears the same relation to weak sun red, A B"' pi, that purple, A B Pi, ]& known to bear to sun red, A B pi. In case of the purples and the sun reds alone, the linkage of PI pi with Y y is, shown by 99 non-crossovers to 20 crossovers, or 16.8 ± 2.7 per cent of crossing-over. When the weak purples and the weak sun reds are alone considered, the non-crossovers are to the crossovers as 81:29, a crossover percentage of 26.4 i 2.8. While the Weak sun red lib Dilute purple Ilia Dilute sun red IVa Total 315 125 119 280 164 317 894 830 Plant Colors in Maize 73 difference between these two percentages of crossing-over, 9.G ± 3.9, is consiclera])le, it is probably not statistically significant, P equaling 0.09. Still further evidence in favor of the assumption that weak purple is differentiated from dilute purple by the factor pair /i"' b, just as weak sun red is differentiated from dilute sun red, is afforded by data from six of the progenies recorded in group 2 of table 34. These data, it will be recalled, were obtained from Fi's of weak purple x dilute sun rod back- crossed to dilute sun red. The Fi weak purples were heterozygous for liguleless leaf as well as for plant color, AAB^bPlplLglg, and the dilute sun reds with which they were backcrossed were liguleless, A b yl Ig. The 1724 plants were distributed as follows: Color types ^\"^^^^ lb Normal leaves 296 Liguleless leaves 108 Evidently the linkage relations of liguleless with weak purple and dilute purple are similar to those already known for liguleless with weak sun red and dilute sun red (page 67). Of the 921 weak svm reds, A B^ pi, and dilute sun reds, A b pi, 632 belong to the non-crossover and 289 to the crossover class, a percentage of crossing-over of 31.4 rb 1.0. Similarly, of the 803 weak purples and dilute purples, the non-crossovers are to the crossovers as 576:227, a percentage of crossing-over of 28.3 ± 1.1. The difference between these two percentages of crossing-over, 3.1 rb 1.5, is such as might occur by chance once in six trials, P equaling 0.16. By way of summary, it may be noted that, from appropriate intercrosses of the several color typos and from determinations of the linkage relations, of these types with liguleless leaf and with yellow endosperm, weak purple and weak sun red have been shown to have the genotypes A W" PI and A B^ pi, respectively. This establishes the existence of the triple alle- lomorphs, B, B'" , b. There is some evidence in favor of the occurrence of a fourth member of this series, b". CROSSES INVOLVING THE MULTIPLE ALLELOMORPHS /?'', Hf, W^ , /, 1^ , r*^ In an earlier section of this account (page 29) dealing with crosses involv- ing only A a, B b, and PI pi, three types of green plants were reported, 74 R. A. Emerson namely, a B pi (Via), ah PI (VIb), abpl (Vic). Still another type of green — a type wholly devoid of purple, red, or brown pigment — has been used in several crosses, with results quite unlike those obtained from corre- sponding crosses with the other green types. For reasons that })ecome apparent later, this fourth type of green is regarded as genetically similar to dilute sun red and is known as type IVg. Green IVg x brown V Generations F\ and F^. — When brown, a B PI, is crossed with green of any of the three types previously studied, brown appears in Fi and brown and green in F2. If green VTc, a b pi, is used in the cross, the F2 ratio approaches 9 : 7, while if green Via, a B pi, or VIb, a b PI, is used, 3:1 F2 ratios are of course expected (tables 19 and 20, page 135). In striking contrast with such results are those obtained from a cross of brown with green IVg. Two such crosses gave 78 purple plants in Fi, and a third cross resulted in 72 purple and 63 sun red plants. It will be ' recalled that just such results as these were obtained from crosses of dilute sun red with brown (tables 4 and 14, pages 123 and 131). The brown plant, 2031-20, which gave purple and sun red Fi plants when crossed with green IVg, was the identical plant previously reported (table 4, group 2) to have given 55 purples and 55 sun reds when crossed with a dilute sun red plant. Moreover, this same brown plant was shown (table 20, group 2, page 135) to have produced from self-pollination 82 browns and 34 greens. Evi- dently it was aa B B PI pl. The important point here is that crosses of brown with green IVg give exactly the same results in Fi as if green IVg were a dilute sun red, A A bb pl pl. There are other reasons, in addition to the Fi results of crosses with brown, for supposing that green IVg has the factor A. When the pericarp- color gene P occurs together with A, the resulting pericarp color is always red, but when P and a a are associated the pericarp color is brown. When green IVg plants have pericarp color it is red rather than brown, while that of greens Via, VIb, and Vic is always brown. Again, the A factor is known to be essential to the production of aleurone color (Emerson, 1918), and the stock of IVg green plants used in these crosses, a strain of the variety Black Mexican sweet corn, was homozygous for purple aleurone. It is noteworthy in this connection that many, perhaps most, plants of this variety show very slight traces of sun red, and these traces are Plant Colors in Maize 75 limited commonly to the glumes of the staminatc inflorescence. Appar- ently the stock of green IVg, which under no environmental conditions to which it has been subjected has ever been observed to produce the slightest trace of sun red, is merely an extreme minus variation of dilute sun red. Not only were the Fi results of the cross of brown with green IVg like those of the cross of brown with dilute sun red, but the same major color types appeared in F2 (table 35, page 145). The distribution of all the indi- viduals of six F2 progenies to the six major color types heretofore recognized is compared below with the theoretical distribution calculated on the assumption that the green IVg parent was genotj^pically a dilute sun red, A Ahb pi pi: Color types Purple Sun red ^jj.^*^ ^"^"^^^^ Brown Green Total Observed 309 100 67 19 88 98 681 Calculated 287 96 96 32 96 74 681 Difference +22 +4 —29 —13 —8 +24 The outstanding features of this comparison are the relatively small deviations, in comparison with the number of individuals, for the purple, sun red, and brown types, and the relatively large deviations for the dilute purple, dilute sun red, and green classes. The relative importance of the several deviations is best seen by a comparison of the quotients of calculated frequencies into the squares of corresponding deviations, from which z2 and P are derived (Elderton's and Pearson's tables). These quotients for the several classes are: T^ 1 a J Dilute Dihite t> n Purple Sun red 1 ■, Brown Green ^ purple sun red 1.69 0.17 8.76 5.28 0.67 7.78 If these quotients were no greater in the case of dilute purple, dilute sun red, and green than for purple, sun red, and brown, there would be about two chances in five that the observed deviations might be due merely to errors of random sampling, a fairly good fit being shown — /^ = 5.06, P = 0.41. But as they stand, these deviations could be expected to occur thru chance alone not more than once in five thousand similar trials, a 76 'R. A. Emerson very poor fit being shown — z^ = 24.35, P = 0.0002. Evidently, green IVg does not give the same results in F2 of this cross as does dilute sun red. It is to be supposed, of course, that green IVg differs in some essential genetic way from dilute sun red, else it would not remain true green for generation after generation while the typical dilute sun red constantly produces a conspicuous amount of sun red pigment. It was therefore to be expected that the dilute sun red class would be deficient in F2 while the green class would show a corresponding excess. But if the 24 green plants in excess of the calculated number be added to the dilute sun red class, that class becomes too large by eleven individuals, the excess now becoming almost as great as the observed deficiency. Moreover, the dilute purple class, it must be remembered, remains greatly deficient. If it be supposed that the excess of greens came about at the expense of dilute purples as well as of dilute sun reds, a very good fit of observation to theory is obtained'. On redistribution of the 24 greens in excess of expectation to the dilute purples and dilute sun reds in the 3 : 1 relation usually existing between these classes, the corrected distribution for the six classes is as shown below. There are almost two chances in five that the deviations may be due to random sampling, P equaling 0.38. Color types Purple Sun red ^"^"^^^^ g^'^''^!^^^ Brown Green Total Corrected distribu- tion 309 100 85 25 88 74 681 Calculated 287 96 96 32 96 74 681 Difference +22 +4 —11—7—8 Mere closeness of fit cannot, of course, be regarded as proof of the supposition on which the corrected distribution was made. But there are other considerations which greatly strengthen the hypothesis. In the case of all the F2 progenies listed in table 35, it was o]:)served that some of the purple plants, altho quite as strongly colored otherwise as normal purples, had wholly green anthers in place of the usual dark purple ones (Plate I, 4). Likewise some of the sun red plants had green instead of pink anthers. In striking contrast to this, not a single dilute purple or dilute sun red plant with green anthers was seen hi the whole lot, the dilute Plant Colors in Maize 77 purples, so far as observed, having dark purple anthers and the dikite sun reds pink anthers, just as in the lots considered in the first section of this paper. Counts of the purple and the sun red plants with different anther colors were made for only three of the six F2 progenies (table 36), and for these lots not every plant was noted at the time when it was possible to determine the anther color positively. When some anthers have become dry and weathered, it is impossible to tell whether they were pink or green when fresh. Less difficulty is experienced with purple anthers, which hold their color much longer. Unfortunately, the records of the three F2 families were not made early enough for positive identification of anther color of all plants. Of 1G2 purple plants, 117 had purple anthers and 33 had green anthers, while 12 were not recorded. Of 50 sun red plants, 21 had pink anthers and 12 had green anthers, with 17 not recorded. In these two lots the plants with purple and pink anthers were together about three times as numerous as those with green anthers, thus suggest- ing a simple monohybrid relation between colored and green anthers. Working hypothesis. — If the genetic factor which is responsible for green anthers of purple and sun red plants be assumed to cause, in the case of dilute purples and dilute sun reds, not merely the anthers but the whole plant — leaves, sheaths, husks, glumes, stalk, and so forth — to be green, a satisfactoiy working hypothesis is afforded. The factor concerned here has been found to be the well-known aleurone-color factor R, or else some factor very closely linked with it. Some of the evidence on which this statement is based is presented later in this paper (pages 80, 98). It may be pointed out in passing that the relation between anther color and aleurone color here noted was studied by Webber (1906) some years before the several aleurone-color and plant-color factors had been determined. Since aleurone color is not primarily concerned in the present account, it might be less confusing if the case were regarded as one of complete hnkage, and if some other symbol for anther color were used and all reference to the R factor omitted in this paper. Until recently there was nothing known of aleurone-color })ehavior that made necessary the assumption of more than the simple factor pair, R r. The plant-color behavior, on the other hand, as becomes apparent later, necessitates the assumption of a group of multiple allelomorphs responsible in turn for diverse combinations of colors of leaves, sheaths, anthers, silks, and other plant parts. The commonest combinations in the writer's cultures are 78 R. A. Emerson strong pink anthers with deep red silks, hghter pink anthers with red- dish or pinkish silks, green anthers with green silks, and so on, but there exist also such combinations as strong pink anthers with green silks, green anthers with reddish silks, and the like. Moreover, different inten- sities of dilute sun red in leaf blades, glumes, and other parts are sometimes combined with various silk-color and anther-color combinations. There is evidence that at least several of these combinations behave as would be expected if each were a definite unit allelomorphic to any one of the others. Perhaps the most remarkable feature of this series of allelomorphs — or supposed allelomorphs — is the fact that a single unit behaves as a dominant with respect to the color of one plant part and as a recessive with respect to that of another part. Thus, a combination of dominant pink anthers with recessive green siUvS is common in the writer's cultures. The wholly green plants used in the crosses here under consideration are recessive for green silks, anthers, glumes, sheaths, husks, and other parts, and dominant for colored aleurone. Since the aleurone-color symbols R r have long been employed in the usual way, R as the dominant and r as the recessive allelomorph, this usage is adhered to in this paper. The effect of these factors on plant color is indicated by superscripts. Thus, both R^ and / are dominant allelomorphs with respect to pink anthers and reddish silks, while both R^ and r^ are recessive for green anthers, silks, and so on. In the crosses here considered it is known that / and R^ are the pair concerned. With respect to plant color, therefore, as contrasted with aleurone color, / is dominant and R^ is recessive. While it is realized that this usage may tend to confuse the hasty reader, the use of any other symbols that have so far suggested themselves would result in greater confusion ultimately, particularly when the interrelations of plant color and aleurone color are taken up. To return to the F2 behavior of crosses of green IVg with brown, the following notation should express the F2 results obtained, provided the proposed hypothesis is tenable: Phenotypes Plant color Anther col 81- -ABPlr"-— la Purple Purple 27- -ABPIR"— Ig Purple Green 27- -AB'plr'- — Ila Sun red Pink 9- ~ABplR"~ Ilg Sun red Green 27- -Ab PI r'— Ilia Dilute purple Purple Plant Colors in Maize 79 Phenotypes Plant color Anther color Q—AhPlR"— Illg Green Chven 9 — Ah pi r'' ■ — / Va Dilute sun red Pink S — AbpIR" —I Vg Green Green 27 — a B PI r'-— V Brown Green Q — aBPIR"— V Brown Green 9 — aBplr'' — Via Green Green 3 — aBplR" — Via Green Green 9 — abPlr'- ^Vlb Green Green Z — abPlR" —VIb Green Green 3 — abj)l r'' — Vic Green Green 1 — abplR" — Vic Green Green 256 The theoretical numerical relation between the several color combi- nations, in the order given above except that all greens are included in the last class, is 81:27:27:9:27:9:36:40, total 256. The distribution of the 353 individuals of the three F2 progenies for which anther records were made (table 36, page 146) is compared below with the theoretical distribution. In order that all plants may be included, the few purple and sun red plants whose anther colors were not noted are arl^itrarily distributed to the colored-anther and green-anther classes in a 3 : 1 ratio. The fit of observation to hypothesis is so good that there are three chances in five that the deviations may be due to errors of random sampling, P equaling 0.60. Plant color Purple Purple Sun red Sun red ^^[."^^j^ ^^^^^^ Brown Green Total Anther color Purple Green Pink Green Purple Pink Green Green la Ig Ila Ilg Ilia Wa V Illg, IVg, VI Observed 126 36 M 16 39 10 42 50 353 Calculated 112 37 37 12 37 12 50 55 352 Difference -\-U —1 —3 +4 -^2 —2 —8 —5 -|-1 TMien the six Fo progenies listed in table 35, for three of which no records of anther color were made, are grouped without reference to anther color, the comparison of observed and calculated numbers are as given below. For the six progenies there is practically an even chance that the devia- tions ma}^ be due to errors of random sampling, P equaling 0.48. It will be recalled that when these same progenies were compared with the dis- Purple Sun red Dilute purple Dilute sun red BrowTi Green Total la Ila Ilia IVa V Illg, IVg, VI 309 100 67 19 88 98 681 287 96 72 24 96 106 681 80 R. A. Emerson tribution calculated on the basis of the three-factor hypothesis, the fit was very poor, P equaling 0.0002 (page 76). Comparison of the observed distribution with the distribution calculated on the four-factor basis follows : Color types Observed Calculated Difference -^22 +4 —5 —5 —8 —8 Relation of aleurone color to plant color. — It is evident from the compari- sons already given that the four-factor hypothesis fits well the F2 data, which is of course to be expected since it was invented for that purpose. But this fact alone is far from a substantiation of the hypothesis. The genetic tests ordinarily available are the behavior of the several F2 types in later generations and in intercrosses. Since aleurone color as well as plant color is involved in these crosses, still another test can be employed. The six Fi plants whose Fo -progenies are recorded in table 35 produced from self-pollination a total of 955 seeds, of which 388 had colored and 567 colorless aleurone. This obviously approaches closely a 27:37 ratio, the percentage of colorless seeds being 59.4 ±1.1 while the theoretical per- centage is 57.8 (Emerson, 1918). The deviation from expectation, 1.6 ± 1.1 per cent, is such as might be expected by chance once in three trials, P equaling 0.33. Evidently, therefore, the aleurone factors A, C, and R are concerned in these crosses. Since A and R are assumed by the hypothesis to be plant-color factors also, there is afforded opportunity of comparing the plant-color classes fi'om colored with those from color- less seeds. Since colored aleurone requires the interaction oi A, C, and R, colored seeds should never produce brown j^lants nor green plants of type VI, both of which are a a. As seen from the data given below, no brown plants came from colored seeds but a few wholly green plants appeared. Greens of type IVg are of course to be expected from seeds homozygous for R^. Owing to the fact that a larger percentage of colorless than of colored seeds produced plants, the theoretical distribution with respect to plant color, given below, was calculated separately for colored and for colorless seeds. J^or the colored seeds there are nearly two chances in Plant Colors in Maizk 81 five (P equaling 0.58), and for the colorless seeds only about one chance in fovn-teen (P equaling 0.07), that the observed deviations may be due to errors of random sampling. The comparisons follow: Color types Purple Sun red ^^'[.'J^J*^ ^^''^^^^ Brown Green Total la Ila Ilia R'a V Illg, IVg, VI Colored seeds: Observed 14S .51 25 8 23 255 Calculated 143 48 32 11 21 255 Difference +5 +3 —7 —3 +2 Colorless seeds: Observed 161 49 42 11 88 75 420 Calculated 136 45 39 13 104 89 426 Difference +25 +4 +3 —2 —16 —14 It is noteworth}^ that the ratio of purples and sun reds to dilute purples and dilute sun reds is considerably greater for plants grown from colored seeds than for those from colorless seeds. This is to be expected from the fact that R must be present in all colored seeds, while some of the colorless seeds here concerned were doubtless r r. Hence, R^ R^ should have occurred more frequently in the colored than in the colorless seeds, and should, by the hypothesis here under test, have reduced the numbers of dilute purples and dilute sun reds, causing these plants to appear as greens, types Illg and IVg. If the 23 green plants grown from colored seeds are added to the dilute purples and dilute sun reds, the ratio of strong to dilute purples and sun reds approaches closely the ratio observed for the plants from colorless seeds. It is even more instructive to note the relation of aleurone color to plant color in the case of the three F2 lots for which anther colors were recorded (table 36, page 146). For this comparison the few purple and sun red plants whose anther colors were not recorded have been distributed to the colored-anther and green-anther classes in approximately the ratio in which these anther colors were found to occur in the cases in which anther colors were recorded. Since a larger proportion of colorless than of colored seeds produced plants, the theoretical (list rilnit ion has i)cen calculated separately for the two classes of seeds. The comparisons follow: 82 * R. A. Emerson Plant color Purple Purple Sun red Sun red ^pjg guli red ^^^"^ ^'"^^^ ^^^^^ Anther color Purple Green Pink Green Purple Pink Green Green la Ig Ila Ilg Ilia IVa V Illg, IVg, VI Colored aleurone: Observed 48 25 11 12 16 4 10 126 Calculated 47 24 16 8 16 5 10 126 Difference +1 +1—5+4 0—10 Colorless aleurone: Observed 78 11 23 4 23 6 42 40 227 Calculated 62 10 21 4 21 7 55 47 227 Difference +16 +1 +2 +2 —1 —13 —7 In view of the rather large number of plant-color classes and the com- paratively small number of individuals concerned here, the fit of the observed to the theoretical distribution is remarkably good. The devia- tions are such as might be expected by chance seven times in ten trials for the colored-seeded lot (P = 0.70), and about once in four trials for the colorless-seeded lot (P = 0.26). In addition to this comparison of the lot as a whole, it should be noted that, while among the purple and sun red plants as a whole the expected relation of colored (purple and pink) anthers to colorless (green) anthers i.s 3:1, for the colored-seeded lot it is 2:1 and for the colorless-seeded lot it is 6:1. The observed relations were 59:37 and 101:15, or about 1.6:1 and 6.7:1, respectively. On the whole, therefore, this comparison, involving aleurone color as well as plant color, supports the suggested factorial interpretation. Later behavior of F2 purple I. — Only three F2 purples with purple anthers were tested in F3. One of these, 2960-9, resulted in purple plants with purple, la, and green, Ig, anthers, and sun red plants with pink, Ila, and green, Ilg, anthers, in the respective numbers 14:9:6:3. A purple plant of the genotype A A B B PlplB!' / should give these four classes in the relation 18:6:6:2, The observed deviations might be expected twice in five trials, P equaling 0.41. Another F2 purple plant, 2958-8, gave F3 progeny consisting of the same eight color types as were seen in F2 in table 36 (page 146) . Evidently the F2 purple plant w^as A a B bPlpl R^ /. The deviations from expecta- tion are such as might occur by chance in about seventeen out of any twenty such trials, P equaling 0.86. The comparison follows: Plant Colors in Maize 83 Plant color Purple Purple Sun red Sun red Anther color Purple Green Pink Green la Ig Ila Ilg Observed 28 13 11 3 Calculated 29 10 10 3 Dilute purple sun red ^^°^'^ ^reen Total Purple Pink Green Green Ilia IVa V IIIr, IVg ,VI 7 3 IG 11 02 10 3 13 14 02 Difference —1 +3 +1 0—3 +3 —3 The third purple-ant hered Fo purple tested, 29G1-3, gave in Fs all the color types except dilute purple. Ilia, and dilute sun red, IVa. A purple plant of the genotype AaBBPlplR^f should give the color types observed. The observed deviations from expectation might occur by chance about once in seven trials, P equaling 0.15. The comparison follows: Plant color Purple Purple Sun red Sun red Brown Green Total Anther color Purple Green Pink Green Green Green la Ig Ila Ilg V VI Observed 37 11 5 2 12 3 70 Calculated 30 10 10 3 10 8 71 Difference +7 +1 —5 —1 +2 —5 — 1 A single green-anthered F2 purple, 2960-7, gave four F3 color types, purple, sun red, brown, and green, all with green anthers. This behavior is to be expected from an F2 genotype A a B B PI pi R^ R^. One of the F3 purples, 4956-1, repeated this l)ehavior in F.i. The F3 and F4 progenies are shown together in the following comparison, for which P = 0.60: Color types Purple Ig Observed 84 Calculated 86 Difference —2 —2 +6 —2 It is of interest to note in this connection that a plant of the genotype AaBB PlplR'^ R^ could not exhibit a 27:37 fatio of colored to color- less aleurone, as was the case for some of the plants dealt with earlier. Sun red Brown Green Total Ilg V . VI 27 35 7. 153 29 29 9 153 84 R. A. Emerson For A a fl" R" the aleurone-color ratio must be either 9 : 7 or 3 : 1, depending on whether the third aleurone-factor pair is C c or C C. The F2 purple plant 2960-7 showed a 9:7 aleurone-color ratio with 86 colored and 74 colorless seeds, AaCcRR, while the F3 plant 4956-1 exhibited a 3:1 ratio with 213 colored and 67 colorless seeds, A aC C R R. Another purple plant of the same F3 progeny, 4956-32, exhibited a 3 : 1 aleurone- color ratio and threw only green-anthered purple and sun red plants. Its genotype must have been A A B B C c Plpl R" R^. Thus it is often possible, from behavior in the following generation, to know the genotype not only with respect to plant color but for aleurone color as well. This is particularly true when the B factor is present. Of the twenty-four sorts of behavior possible, according to hypothesis, for F2 purples of the cross under consideration, four sorts have been exhibited in F3 and a fifth shown in F4. This is far from an adequate study of the F2 purples. All that can be claimed, therefore, is that, so far as they go, the results are in accord with the hypothesis. Behavior of other F2 color types. — Only oie F2 sun red plant with pink anthers, 2961^, was tested in F3. It produced sun reds with pink and sun reds with green anthers, dilute sun reds, and greens. Since anther color was noted for only a part of the plants, it has to be disregarded in classifying the F3 progeny. The color types sun red, dilute sun red, and green occurred in the numerical relation 114:23:57. Of the eight possible genotypes of pink-anthered sun red, only three could throw these three color classes — Aa B h/ /, A A B h R^ /, and Aa B h R" /. From the first genotype a 9:3:4, from the second a 12:3:1, and from the third a 36 : 9 : 19, relation should exist between the F3 classes. The poor fit of observed numbers to the 9:3:4 relation makes it improbable that the first genotype is concerned, there being only about one chance in twenty-two that the ol^served deviations are due to errors of random sampling, P equaling 0.045. The comparison follows: Color types Sun red Ila Observed. . 114 Calculated . 109 Difference. +5 Green Total Dilute sun red IVa Via, c 23 57 194 36 49 194 —13 +8 Plant Colors in Maize 85 A moro conclusive reason for throwing out the first genotype is the fact that the phmt had some seeds with colored aleurone, which would have been impossil)le if it were i- r. The second genotype is discarded because of the extremely poor fit of observed numl)ers to the 12:3:1 relation. There is an almost inconceivably small chance that thc^ observed deviations may be due to errors of random sampling, x- (equaling ISO. (When n' = 3 and x~ = 29, P = 0.000001. Higher values of /' when iV - 3 are not listed in Pearson's tables.) The comparis:)n follows: Color types Sun red ^'^^^*^ Green Total sun red Ila, g IVa IVg Observed 114 23 57 194 Calculated 146 36 12 194 Difference —32 —13 +45 The elimination of the first two genotypes leaves the third genotype as the only one that can be concerned here. The fit of observed numbere to the 36 : 9 : 19 relation is very close, x^ equaling 0.84. (Values of P are not listed in Pearson's tables for values of x- less than 1; when x^ = 1 and n' = 3, P = 0.6].) The comparison follows: Color types Sun red J^}}""}^^ Ha, g Observed 114 Calculated 109 Difference +5 — 4 — 1 This comparison leaves little doubt that the genotype of the F2 plant concerned is AaBbR^/. There are, moreover, other considerations which go far toward identifying the genotype as given here. The fact that some sun red plants of F3 had green and others pink anthere is evidence for the constitution R' /. Since dilute sun red plants appeared in F3, there can be no question as to B h. The F2 plant showed a 9:7 aleurone- color segregation, and therefore, in addition to R r, it must have been rired ^^^^" Total :Va IVg, Via, c 23 57 194 27 58 194 g6 R. A. Emerson either A a or C c. An F3 sun red plant with green anthers, R" R", had 97 colored and 20 colorless seeds, again indicating either A a or C c. If it was A A B b C c R^ R'^, both colored and colorless seeds should have given sun red and green plants in a 3:1 ratio; if it was A a B B C C R" R^, the colored seeds should have given sun red and the colorless ones green plants only, the plant-color ratio again being 3:1; but if it was A aBh C C R^ R^, the colored seeds should have produced sun red and green plants in a 3 : 1 ratio and the colorless seeds green plants only, the ratio of sun reds to greens in the two lots together being 9:7. Actually the colored seeds resulted in 23 sun red and 10 green plants and the colorless seeds in 10 green plants only, the ratio of sun reds to greens being 23 : 20, thus approaching 9:7. There is, therefore, considerable assurance that the F3 plant was AaBhC C R^ i^^- that the Fa plant was AaBhC C R^ /, and that the F3 numerical relation of plant colors was 36:9:19, as originally suggested by the closeness-of-fit test. A single dilute purple plant of F2, 2960-4, was tested in F3 and found to give 38 dilute purple and 39 green plants. Of the eight possible geno- types for F2 dilute purples, the only ones that could give only dilute purples and greens in F3 are A A h h PI PI R^ /, A a h h PI PI / /, and Aahh PIPIR^ /. The first two should give a 3:1, and the third a 9:7, F3 ratio. The plant had colored aleurone, which throws out of consideration the second genotype with r r. The F3 plant-color ratio fits fairly well a 9 : 7 but not at all a 3 : 1 expectation, the observed numbers being 38 : 39 and the calculated numbers 43 : 34 and 58 : 19, with deviations of 5 and 20, and probable errors of 2.6 and 2.9, respectively. The deviation from a 9:7 ratio might occur by chance once in five trials, P equaling 0.20, but that from a 3 : 1 ratio not more than twice in about a million trials, P equaling 0.000002. The genotype AahhPlPlR"/ is therefore decidedly favored by these results. The aleurone-color record shows that this genotype is possible, since there were 57 colored and 56 colorless seeds, a relation about halfway between the 9:7 and the 27:37 ratio due io A aC C Rr and A a C c Rr, respectively. Intercrosses of Fi color types It is realized that the tests of Fo types by studies of their behavior in later generations as reported above, are markedly inadequate to serve Plant Colors in Maize 87 as a demonstration of the hypothesis suggested to account for the F2 behavior of the cross of brown, type V, with green, type IVg. It is note- worthy, however, that no resuUs have been found that do not agree with the hypothesis. Fortunately, several intercrosses of the types found in F2 afford additional evidence. Purple Ig x green Vic. — Green-anthered purples, A B PI RF, crossed with greens of type Vic, a h pi f, should give F2 results identical with those found from the original cross of brown, a B PI f, with green of type IVg, A b pi R'', since Fi in either case should be A a B bPl pi R^ /. Two such crosses are recorded in table 37, group 1 (page 146). The Fi plants were both purple, with purple anthers. In F2 the same eight t,ypes were noted as in Fo of the cross of brown with green IVg (table 36). The anther color was not recorded, however, for many of the plants, so that only six color classes are shown, as in table 35. While all the expected color types are present, the fit of observed to calculated numbers is so poor that the observed deviations should not occur by chance more than once in thirty trials, P equaling 0.033. The comparison follows: Color types Purple Sun red ^ ^,^ ^^ , Brown Green Total -' ^ ^ purple sun red la. g Ila, g Illa IVa V Illg, IVg, VI Observed.... 80 13 9 9 20 27 158 Calculated... 66 22 17 6 22 25 158 Difference... +14 —9 —8 +3 —2 f2 If, notwithstanding the poor fit shown above, the Fi was A a B b PI pi R" /, a backcross of Fi with green of type Vic, a b pi /, should result in the same six major plant-color types, but no green-anthered purples or sun reds should occur. Such crosses are listed in group 2 of table 37. All the purple plants had purple anthers and all the sun red plants had pink anthers. Moreover, the six color classes appeared in so very nearly the expected relation of 1:1:1:1:1:3 that deviations as great as those observed might be expected to occur by chance perhaps ninety-nine times in one hundred trials, x^ equaling 0.85 (when x^ ~ 1 and n'- 6, P = 0.96). The comparison follows: 88 11. A. Emerson Color types Purple Sun red ^^^^^ g^^^ 'Jg^ Brown Green Total la Ila Ilia IVa- V VI Observed 36 29 31 31 31 95 253 Calculated 31.0 31.6 31.6 31.6 31.6 94.9 252.9 Difference +4.4 —2.6 —0.6 —0.6 —0.6 +0.1 +0.1 If an Fi supposedly A a B bPlpl R^ r-^, be backcrossed to dilute sun red, type IVa, Ab pi /, color types la, Ila, Ilia, and IVa should appear, none of them with green anthers. Such crosses are presented in group 3 of table 37. The anthers thruout were purple or pink, and the several color types appeared in approximately equal numbers, as expected, there being more than two chances in five that the observed deviations may have been due to errors of random sampling, P equaling 0.42. The comparison follows: Color types Purple Sun red ^'^^^^f^ ^^'^^""^^^ Total la Ila Ilia IVa Observed 115 97 95 111 418 Calculated 104.5 104.5 104.5 104.5 418 Difference +10.5 —7.5 —9.5 +6.5' If the same Fi genotype, A a B h PI pi R" /, he backcrossed with green of type IVg, A bplR", there should occur five major color types, brown not appearing, and both green and colored anthers should be found in ])oth the purple and the sun red plants. The records of such a cross are given in group 4 of table 37. The seven expected color types occurred in numljers near enough to expectation so that there are nearly three chances in ten that the deviations may have been due to errors of random sampling, P equaling 0.29. The most pronounced deviations are the excess of dilute sun reds and the deficiency of greens. The comparison follows: Plant color Purple Purple Sun red Sun red ^''"!® Dilute ^^^^ ,j,^^„j ^ ^ purple sun red Anther color Purple Green Pink Green Purple Pink Green la Ig Ila Ilg Ilia IVa Illg, IVg Observed 10 i:j 7 S 10 15 1.3 76 Calculated .• 9.5 9.5 9 5 9.5 9.5 9.5 19 76 Difference +0.5 +3.5 -2.5 —1.5 +0.5 +5.5 —6 Plant ( ulous in Maize 89 In conclusion it seems safe to say that the cross of gi'een-anthered purpl(\ Ig, with green of type Vic, havS given results similar to those yielded by the cross of brown. V, with green of type IVg. Since this was to have been expected from the hypothesis suggested by the F2 generation of the latter cross, the results just discussed lend support to that hypothesis. Purple Ig x dilute sun red IVa. — In accordance with the hypothesis under consideration, green-anthered purple is A B PI R^ and dilute sun red is A h pi /. Fi of the cross should he A A B b PI pi P!' /, and F2 should consist of the five major color types, purple, sun red, dilute purple, dilute sun red, and green of types Illg and IVg, with both green-anthered and colored-anthered subclasses of purples and sun reds. The Fi plants were purple-ant hered purples, as expected. Three F2 progenies are recorded in table 3$, group 1. Anther color could not be recorded in all cases, but in each of the three F2 progenies both green and colored anthers were noted for both purple and sun red plants. In one progeny, 5042-5045, of a total of 57 purples and sun reds, 41 had colored and 16 had green anthers, which is not far from the expected 3 : 1 relation. The 415 F2 plants w^re so distributed among the five color classes that the chances are nearly three in five that the deviations observed may have been due to errors of random sampling, P equaling 0.58. A comparison of observed and theoretical distributions follows: Color tvpes Purple Sun red 1 , Green Total ^ ^ ^ purple sun red la, g Ila, g Ilia IVa IHg, IVg Observed , 243 71 59 22 20 415 Calculated 234 78 58 19 26 415 Difference +9 —7 +1 +3 —6 An Fi of the cross here considered, 6557-12, A A Bh Pi pi R" /, was backcrossed to a dilute sun red, A b pi /. Four color types occurred in the progeny, as expected, and all the plants had colored anthers. The deviations from expectation were such as might occur by chance in consider- ably more than one out of any two such trials, P equaling 0.56. The comparison follows: R. A. Emerson Purple Sun red Dilute purple Dilute sun red Total la Ila Ilia IVa 43 43 35 48 169 42 42 42 42 168 90 Color types Observed " Calculated Difference +1 +1 —7 +6 +1 Purple la x green IV g. — The cross between purple la and green IVg should have given results identical with those expected from the cross of green-anthered purple with dilute sun red. Tho parents are supposed to have been A BPl/ and A b pi R\ and the Fi, therefore, A A B h PI pi R^ /. The Fi's were purple-anthered purples. Two r2 progenies are listed in table 38, group 2. All the expected color -types occurred, but the observed frequency distribution was such as might be expected to occur by chance only about once in eleven trials, P equaling 0.09. If these progenies are grouped into five classes, anther color being disregarded, the fit is somewhat better, P equaling 0.16. The comparison of observed and theoretical frequencies follows: Plant color Purple Purple Sun red Sun red ^^^^^f ''^'^"H Green Total ^ ^ purple sun red Anther color Purple Green Pink Green Purple Pink Green la Ig Ila Ilg Ilia IVa Illg, IVg Observed 26 14 17 3 9 2 1 72 Calculated: 31 10 10 3 10 3 5 72 Difference —5 +4 +7 — 1 —1 —4 The F2 of this cross exhibited, as expected, practically the same results as were obtained from the cross of green-anthered purple with dilute sun red. Unlike that cross, the one under consideration here was checked by the behavior of some of its F2 types in later generations. A single F2 purple-anthered purple produced in F3 16 plants (table 39, group 1), including only purple, sun red, and dilute purple in the relation 9:4:3. Of both the purples and the sun reds, some plants had colored and some had green anthers. Obviously two other types, dilute sun red and green, should occur in such an F3 and doubtless would have been found had a larger number of plants been grown, for the F2 plant, in order to have produced the color types recorded, must have been A A B b Plpl R' /. Plant Colors in Maize 91 Only one plant of each of the missing classes was to have been expected, and the distribution as a whole was not far from expectation, P equaling 0.59. Both the types lacking in F3 occurred in F4, a pink-anthered sun red F:i producing sun reds and dilute sun reds, while green-anthered purples produced in one instance purples, sun reds, and greens, and in another instance purples and greens only, all with green anthers. This F3 lot may consequently l)e regarded as A A B b PI pi R" /, and therefore equivalent to the F2 lot from which it came, and its F4 progenies equivalent to F3 progenies. A second F2 purple-anthered purple was backcrossed to green plants of types IVg and Vic (group 1, table 39). From the backcross with green of type IVg, Ab pi R^, five major color types appeared and both the purple and the sun red types contained subtypes with colored and with green anthers. While all the classes expected from an F2 of the genotype A A Bb PIpIR^ / occurred, the frequency distribution was so far from expectation that there is only one chance in five hundred that the observed deviations may have been due to errors of random sampling, P equaling 0.002. The expected and observed distributions are as follows: Color types Purple Sun red ^'^"t^ ^'^''^^ Green Total •^ ^ ^ purple sun red Ia,g Ila, g Ilia IVa Illg, IVg Observed 15 15 5 1 9 45 Calculated 9 9 9 9 9 45 Difference +6 +6 —4 —8 Whether the discrepancy is genetically significant or was due to some acci- dent of pollination cannot now be determined. A backcross of the same F2 plant with green of type Vic, ab pi /, yielded only four color types, as expected (group 1, table 39), the anthers being colored in all cases. The excess of purples and deficiency in two other classes makes the deviations from expectation fairly great, so that there is only about one chance in seven that they may have been due to errors of random sampling, P equaling 0.14. The comparison follows: 92 R. A. Emerson Color types Purple Sun red ^^^^ ^^ Total la Ila Ilia IVa Observed 27 19 15 14 75 Calculated 19 19 19 19 76 Difference +8 —4 —5 —1 A third purple-anthered purple, an F- plant of the lot regarded as equivalent to Fo's, gave in the next generation purple-anthered purples and pink-anthered sun reds in the relation 31:7 (group 2, table 39). From the genotype A A B B PI pi //, these two phenotypes should appear in a 3:1 ratio. The deviation from expectation was 2.5 ± 1.8, or only such as might be expected about once in three trials, P equaling 0.34. Two green-anthered purples of F2 and two of the equivalent F3 lot noted above were tested by a later generation. Two of the four yielded three color tj^jes, purple, sun red. and green, all with green anthers (group 3, table 39). Such behavior is expected from the genotype A A B b PI pi R^ R^. The 9:3:4 relation is approached so closely that the value of P cannot be determined from Pearson's tables, x^ equaling 0.36. The comparison follows: Color types Purple Sun red Green Total Purple Sun red Green Ig Ilg Illg, IVg 37 11 14 36 12 16 Observed 37 11 14 62 Calculated 36 12 16 64 Difference +1 —1 —2 —2 The same two .green-anthered purples were backcrossed with green of type IVg, and one of them and a sib of the other with green of tj-pe Vic, with results as shown in group 3 of table 39. The crosses with type IVg, AhplI^, gave the same three classes as did the self-pollinations, and the frequency distribution differed from expectation by values that might occur by chance about once in two trials, P equaling 0.49. The comparison follows: N Maize 93 Sun red Green Total Ilg nig, ivg 32 53 119 30 60 120 Color types Purple Ig Observed 34 Calculated 30 Difference +4 +2 —7 —1 The backcrosses of these green-anthered purples with green of type Vic, a b pi /, as was to be expected, gave very different results. There were produced four instead of three phenotypes, all with colored (purple or pink) instead of green anthers. The deviations from the theoretical frequency distribution are such as might be expected about once in five trials, P equaling 0.21. The comparison follows: Color types Observed Calculated Purple la Sun red Ila Dilute purple Ilia Dilute sun red IVa Total 44 44 48 44 33 44 52 44 177 176 Difference +4 —11 +8 +1 The other two green-anthered purples that were tested yielded only two phenotypes, green-anthered purple and green, in the relation 56:18 (group 4, table 39). The genotype AABhPlPlRUi' should give these two phenotypes in a 3 : 1 ratio. The deviation from expectation was therefore 0.5 ± 2.5. One of the same plants backci-ossed to green of type IVg gave 28 green-anthered purples and 27 greens where equality was expected. Of the twelve kinds of behavior expected of F2 purples of the cross of purple-anthered purple with green IVg, only four have been demonstrated. So far as they go, however, the results are quite in accord with the hypoth- esis under test. In addition to the F2 purples, sun reds and dilute purples also were tested by later generations, as detailed below. Three pink-anthered sun reds gave sun reds and dilute sun reds only, all with pink anthers (table 40, group 1). These three plants are therefore regarded as A A Bh pi pi//. The ratio observed was 97:26. The deviation from the expected 3:1 ratio was 4.75 it 3.24, or such as might 94 R. A. Emerson occur by chance once in three trials, P equahng 0.32. One of these three sun reds, when crossed with a dilute purple, Ah PI/, gave 71 purples and 77 dilute purples, all with purple anthers, where equal numbers were expected. Three other F2 pink-anthered sun reds produced nothing but sun red plants in F3, 228 in all (group 2, table 40). Some plants of each progeny had pink and some had green anthers. Small plantings of each lot were made in the garden and larger plantings in the field. Anther color was noted in the case of the garden plants only. The records show 44 with pink and 16 with green anthers, a deviation from a 3: 1 ratio of only 1.0 ± 2.3. The F2 sun reds are therefore assumed to have been A A B B plpl R^ /. One of these F2 plants was backcrossed to green, both of type IVg and of type Vic, resulting in a total of 108 sun red plants (group 2). Altho no counts were made for anther color, it was noted that the cross with green IVg, Ah pi R^, gave both pink- and green-anthered plants, while the cross with green Vic, ah pi /, gave pink anthers alone. Only t'^^o of the six possible genotypes of F2 sun reds were demonstrated. Only one dilute purple F2 plant was tested further (group 3, table 40). From self-pollination it yielded 46 dilute purple and 9 dilute sun red plants, all with colored (purple or pink) anthers. The deviation from a 3:1 ratio, 4.75 ± 2.17, is such as might be expected by chance about once in seven trials, P equaling 0.14. The same F2 plant when back- crossed to green of types IVg and Vic (group 3) gave 85 dilute purples and 82 dilute sun reds where equality was expected. Evidently this F2 was A A hh Plplf /. No F2 dilute sun red or green plants were tested further. One F3 dilute sun red, however, was found to breed true, producing an F4 of 30 pink-anthered dilute sun reds. Likewise, eight F3 and F4 greens gave a total of 126 green plants in the next generation. In so far as tests have been made, therefore, the cross of purple-anthered purple with green IVg has behaved as expected on the basis of the hypo- thetical genotype assigned to Fi, namely, A A B bPlplR^ /. Purple Ig x green I Vg. — Green-anthered purples are assumed to be A B PI R^, and green IVg to be A 6 pi R^ . The Fi genotype is therefore, theoretically, A A B b PlplR^ R^, and F2 should consist of the three color types purple, sun red, and green, all with green anthers. Eight such F2 progenies are recorded in table 41, group 1. The three types Plant Colors in Maize 95 occurred in so nearly the expected relation of 9:3:4 that the observed deviations might be expected by chance considerably more than once in three trials, P cquahng 0.37. The comparison follows: Color types Purple Sun red Green Total Ig Ilg Illg, IVg Observed 293 105 150 548 Calculated 308 103 137 548 Difference —15 +2+13 The F2 greens of this cross are assumed to consist of the genotypes Ah PI R' and Abpl R^, which, if / had been present instead of R'^, would have been dilute purples and dilute sun reds, respectively. In substantiation of this assumption, crosses of Fi's, all green-anthered purples, with dilute sun red, A b pi /, and with green Vic, ah pi /, are recorded in group 2 of table 41. As expected, the result was the four classes purple, sun red, dilute purple, and dilute sun red, all with colored anthers. The expected numerical equality of the four classes was so closely approached that deviations such as those observed might be expected by chance in nearly three out of four trials, P equahng 0.74. The comparison follows: Color types Purple Sun red ^^^ ^^"^"^^^ Total la Ila Ilia IVa Observed 58 61 62 70 251 Calculated 63 63 63 63 252 Difference —5 —2 —1 +7 — 1 Still another Fi was crossed with a pink-anthered sun red, A B pi /, and gave 68 purples and 67 sun reds, all with colored anthers, where equal numbers were expected. So far as tested, therefore, the cross of green-anthered purple with green IVg has given the results expected on the basis of the hypothesis under test. Purple Ig x brown V.— A cross of green-anthered purple, A B PIR^, with brown, a B PI /, gave in Fi 49 purple-anthered purples, presumably 4 96 R. A. Emerson A aB B PI PI R^ /. An r2 progeny was grown from only one Fi plant, 6653-6, resulting in two major color types, purple and brown, in approxi- mately a 3 : 1 ratio. The purples were, as expected, of two subtypes, one with purple and the other with green anthers. The theoretical rela- tion of 9:3:4 was realized so closely that the observed deviations might be expected by chance in at least two out of three trials, x^ equaling 0.76 (when z'=l and n' = 3, P = 0.61). The comparison follows: Color types. f^^'P}?' ^^'Plf Brown Total •^ ^ purple anthers green anthers la Ig V Observed 23 5 9 37 Calculated 21 7 9 37 Difference +2 —2 A second Fi plant, 6653-2, was backcrossed with green IVg, Abpl R^, resulting in 39 purple plants, 21 with purple and 18 with green anthers, where equal numbers were expected, the deviation from expectation being 1.5 ± 2.1. The same Fi plant was crossed with a heterozygous dilute sun red, A abb pi pi / /, resulting in 45 purple-anthered purples and 18 browns, the deviation from the expected 3:1 ratio being 2.25 db 2.32. Purple Ig x dilute purple Ilia. — Crosses of green-anthered purple, A B PI R^, with dilute purple, Ab PI /, gave in Fi purple-anthered purple, A A B b PI PI R^ /. The F2 should consist of purple-anthered and green-anthered purples, dilute purples, and greens, the three major color types appearing in the relation 12:3:1. In F2 from a single Fi plant, 5263-3, both purple-anthered and green-anthered purples were noted, but detailed counts based on anther color were not made. The deviations from the expected numbers for the three major types were such as might occur by chance in nine out of twenty such trials, P equaling 0.45. The comparison follows: Color types Purple ^^^\l Green Total la, g Ilia Illg Observed 36 11 5 52 Calculated 39 10 3 52 Difference —3 +1 +2 Plant Colors in Maize 97 A second Fi plant backcrossecl with green IVg, A b pi R", gave the expected four types. The deviations from the equal frequency expected for the several types was such as might occur by chance somewhat more than once in four trials, P equaling 0.27. The comparison follows: Color Purple, Purple, Dilute types purple anthers green anthers purple la Ig Ilia Observed... 59 67 80 Calculated.. 71 71 71 Green Total Illg 77 283 71 284 Difference.. —12 —4 +9 +6 — 1 Dilute purple Ilia x green IVg. — A single cross of dilute purple, Ah PI /, with green I\'g, A b pi K', gave dilute purple, A Abb PI pi R^ /, in Fi, and three phenotypes, dilute purple, dilute sun red, and green, in F2 (table 42, group 1, page 150). The observed frequencies were 23 :8: 10, which is the nearest possible approach to the expected 9:3:4 relation for a total of 41 individuals. One Fz- dilute purple gave similar results in F3, indicating the same genotype as the Fi dilute purples. The F4 progenies of this F3 lot may be regarded as equivalent to Fa's, and are, therefore grouped with the F3 in table 43. Three F3 and F4 progenies (table 43, group lA) approached the 9:3:4 relation so closely that the observed deviations might occur by chance in nearly three out of five trials, P equaling 0.59. The comparison follows: Color types ^^^^1^ ^^^t^. Green Total •^ ^ purple sun red Ilia IVa Illg, IVg Observed 143 48 73 264 Calculated 149 50 66 265 Difference —6 —2 +7 —1 The green plants of these F3 and F4 lots, as well as those of the F2 lot listed in group 1 of table 42, are assumed to be Ab Pi R" and Abpl R", and consequently to differ from the dilute purples and dilute sun reds only in having R'^ RF in place of R^ / or / /. That the R r pair is thus con- cerned in these results can be shown ])y a comparison between the plant- 98 R. A. Emerson color phenotypes resulting from seeds with colored aleurone and those from seeds with colorless aleurone. The Fo progeny came from a plant that produced from self-pollination colored and colorless seeds in the relation 60 : 24. This close approach to a 3:1 ratio indicates that the Fi plant could have been heterozj^gous for only one of the aleurone-factor pairs A a, C c, or Rr (Emerson, 1918). A cross with a C tester, A c R, resulted in 43 colored and no colorless seeds, while a cross with an R tester, A C r, gave 46 colored and 32 colorless seeds, thus indicating R r as the factor pair concerned. The colorless seeds must therefore have been r r, presumabl}^ / /, and in accordance with the hypothesis under test should have produced no gre^i plants. Some of the colored seeds, on the contrary, should have been R R, supposedly R^ R^, and these should have given green plants. For the most part, the colored and the colorless seeds were planted separately. The 9:3:4 relation of the three plant-color types is theoretically made up of a 6:2:4 relation from colored seeds and a 3:1:0 relation from colorless seeds. Actually, from colorless seeds there appeared dilute purple and dilute sun red plants in the ratio 69:15. The deviation from expectation, 6.0 ± 2.7, might be expected to occur about once in seven trials, P equaling 0.14. From colored seeds the deviation from the theoretical distribution was such as might occur thru errors of random samphng almost once in four trials, P equaling 0.23. The comparison follows: Dilute Dilute sun red Color types ^^^ *^ '' ^ purple Ilia IVa Observed , 92 42 Calculated 102 34 Green Total Illg, IVg 70 204 68 204 Difference —10 4-8 +2 Aleurone is in some cases self-colored and in some cases mottled. Mottled aleurone ordinarily occurs only when the R factor is heterozygous, but not all heterozygous individuals are mottled (Emerson, 1918). Mottled seeds of the cross under discussion, just as colorless ones, since they are presumably R^ /, should produce no green plants. In the case of some of the progenies noted above, the colored seeds were sorted into self-colored, mottled, and colorless. Since usually about one-third Plant Colors in Maize 99 of the colored seeds arc mottled, the 9:3:4 relation of plant-color types observed in this cross should be made up of a 3:1:0 relation from color- less seeds, 3:1:0 from mottled seeds, and 3:1:4 from self-colored seeds. Of the progenies for which the seeds were sorted in this way, the color- less seeds produced dilute purple and dilute sun red plants in the relation 60: 14, with a deviation from 3: 1 of 4.5 db 2.5, the mottled seeds gave the same plant-color types in the relation 30: 12, with a deviation of 1.5 ± 1.9, and the self-colored seeds yielded dilute purple, dilute sun red, and green in the relation 48:19:64 (the theoretical distribution for a total of 131 individuals is 49:16:66), the deviations being such as might occur by chance perhaps three times in four trials, z^ equaling 0.64. On the whole, therefore, these crosses, and particularly the interrelations of aleurone and plant colors, afford strong evidence in support of the hypothesis under test. Before presenting further F3 results from these crosses, it may be well to consider other crosses of dilute purple with green IVg which, so far as plant color alone is concerned, have given results quite like those pre- sented above but which exhibit a wholly different relation between plant color and aleurone color. The green plants concerned in these other crosses were C testers for aleurone color (Emerson, 1918), and were there- fore known to be A c R, presumably A c B?. The dilute purple plants concerned were homozygous for aleurone color, and were consequently A C R, presumably A c R^. These crosses differ, then, from the ones discussed above in having R^ in place of r" and c in place of C. Since the C c pair is supposed not to have any relation to plant color, the results for plant color should be quite like those for the other cross and there should be no relation between plant color and aleurone color. The results for F2 are presented in table 42, group 2, and the F3 results in table 43, group IB. The three plant-color types appeared in F2 in the relation 328:113:148, and in F3 in the relation 40:14:23. Considered together these lots deviated very slightly from expectation, x~ equaling 0.31. The comparison follows: Color tvDes ^^^"1^ ^'^"^'', Green Total »^oior Types purple sun red Ilia IVa Illg, IVg Observed 368 127 171 666 Calculated 375 125 166 666 Difference —7 +2 -\-d 100 R. A. Emerson The seeds from which these plants were grown consisted of colored and colorless in approximately a 3 : 1 ratio, as is expected when the C factor alone is heterozygous. The deviations from the expected 9:3:4 relation for plants from colored seeds was such as might occur by chance more than once in three trials, P equaling 0.36, and for plants from colorless seeds such as might occur once in six trials, P equaling 0.17. The comparisons follow : Plant-color types ^^^ ^J^^^ Green Total Ilia IVa Illg, IVg Colored seeds: Observed 215 58 89 362 Calculated 204 68 90 362 Difference +11 —10 —1 Colorless seeds: Observed 65 32 32 129 Calculated 73 24 32 129 Difference —8+8 The resvilts presented for plant color alone and in relation to aleurone color in these crosses are therefore quite in keeping with the hj'^pothetical constitution assigned to the Fi plants, namely, AAhhPlplKR'Cc, just as the results from the other crosses were in keeping with the assumed genotype AAbhPlplR'/CC for their Fi plants. A single Fi plant was backcrossed with green IVg, Ah pi R\ with results as shown in table 42, group 3. The three color types dilute purple, dilute sun red, and green, occurred in the relation 46:45 : 86. The expected distribution for a total of 177 individuals is 44:44:89, showing almost a perfect fit, x- equaling 0.21. For both the lots of crosses under discussion, further tests are afforded by the behavior in F3 and F4. As already shown, some of the F2 dilute purples had the same genetic constitution as the Fi plants (table 43, groups lA and IB). The progenies of two other dilute purples, one of F2 and the other of an equivalent F,-?, produced dilute purple and dilute sun red plants only (group 2, table 43), in the relation 82:23. The devia- Plant Colors in Maize 101 tion from a 3:1 ratio is 3.25 ± 2.99. From their behavior and in viow of the crosses in which they occurred, one of these plants is assumed to have been A Abb PI pi /V and the other A Abb PI pi R' R\ A single dilute purple of an F? lot equivalent to an F2 gave dilute purple and green plants only (group 3, table 43). The two color types appeared in the ratio 62:16, a deviation from 3:1 of 3.5 ± 2.6. The Fj plant is therefore assumed to have been A Abb PI PI R" /. Colorless and mottled seeds produced dilute purple plants only, as was expected. From self-colored seeds there resulted dilute purple and green plants in the relation 26:16, a deviation of 2.0 ± 2.0 from the expected 2:1 ratio. Two dilute sun red plants gave progenies of dilute sun reds and greens in the relation 63:22, a deviation from a 3:1 ratio of 0.75 ± 2.69 (group 4, table 43). Presumably these plants were A Abb pi pi R"/ and A Abb pi pi R'' R^. Four other dilute sun red plants bred true in the next generation (group 5, table 43), producing a total of 197 dilute sun red plants. These plants are therefore assigned the genotype A Abb pi pi / /. Seven green plants likewise bred true (group 6, table 43), producing a total of 130 green plants. These plants were presumably Abpl R^ and AbPl R\ To summarize, all types of behavior were observed in F3 and equivalent F4 generations of the cross of dilute purple with green IVg except true- breeding dilute purples. Only eight dilute purples were tested, and only one in nine is expected to breed true. Sun red 11 g and 11 a and dilute sun red IV a x green Illg and IVg. — Two crosses of green-anthered sun red with green IVg gave green-anthered sun red plants in Fi, theoretically A A B b j)l pi R' R^. The parent types only appeared in F2 (table 44, group 1). The observed numbers of green-anthered sun reds and greens were, respectively, 216 and 77. The deviation from the expected 3:1 ratio was 3.7 5zb 5.00. A cross of pink-anthered sun red with green IVg gave pink-anthered sun red in Fi, theoretically A A Bbplpl R^ /. Fi plants backcrossed with green IVg, A b pi R\ gave three major plant-color types (group 2, table 44) — sun red, dilute sun red, and green — with the sun reds appear- ing in two subtypes, one pink-anthered and the other green-anthered. Theoretically the four types should have been represented by an equal number of individuals. The deviations from this expectation were such 102 R.. A. Emerson that there is considerably more than an even chance that they might have been due to errors of random sampHng, P equaling 0.56. The comparison follows: ^ 1 , Sun red, Sun red, Dilute ^ rr„+„i Color types pi^k anthers green anthers sun red ^'^^^ ^^^^^ Ila Ilg IVa IVg Observed 105 90 105 109 409 Calculated 102 102 102 102 408 Difference +3 —12 +3 +7 +1 Crosses of dilute sun red with green IVg gave 54 dilute sun red plants in F], A A hhplplR^ /. In F2 (group 3, table 44) there resulted from a self-pollinated Fi, dilute sun red and green plants in the relation 55:22, a deviation from the expected 3:1 ratio of 2.75 ± 2.56. An Fi back- crossed with green IVg gave the same two color types in equal numbers, 30 each, exactly as expected. Numerous other crosses of this sort have been observed in connection with studies of the interrelations of aleurone- color and plant-color factors. Since these data are to be presented in a later paper and since they are wholly in accord with the data given in group 3 of table 44, they are not discussed here. In an earlier section of this paper dealing with the factor pairs A a, B b, and PI pi only (page 29), it was shown that the green plants there noted are of three kinds, namely, abpl, a B pi, and a b PL Thruout the present section of the paper, which deals with the relation of the multiple-allelomorph series containing R", r\ K^ r\ it has been assumed that plants which in the presence of / or R^ are dilute purple or dilute sun red, are green in the presence of homozygous R^. The data presented are wholly in accord with this interpretation, thereby giving considerable assurance of the probable correctness of the hypothesis. The reported interrelations of plant color and aleurone color when the latter was known to involve the R r pair, have still further strengthened this assurance. It remains now to present even more direct evidence, namely, that obtained from crosses of green plants encountered in this study, with sun red and dilute sun red plants. These green plants are assumed to be A 6 PI R^, type Illg, and Abpl R\ type IVg. Certain F3 and F4 progenies consisting of green-anthered purples and greens in a 3: 1 relation are listed in table 39, group 4. These green plants Plant Colors in Maize 103 were all. presumably, A b PI /?". Green plants of a later generation, grown from these greens, when crossed with sun red plants, type Ila, gave 64 purple-anthered purples and no other types (table 45, group 1). Another green crossed with dilute sun red resulted in 4 dilute purples. Obviously the same results would have been obtained had the green plants used in these crosses been ah Plr^, instead of A 6 P/ 7^" as they are sup- posed to have been. As a matter of fact, however, one of these green plants had homozygous colored aleurone, and therefore must have been A C R. The other two greens, while they had colorless aleurone, came from lots known, from their 3:1 aleurone-color ratios and from crosses with aleurone testers, to be heterozygous for C alone, and therefore A c R. Moreover, the green plants from lots consisting of purples and greens in a 3 : 1 relation could not have been a a, for the parents of such lots, if hetero- zygous for A, must have produced purples and browns rather than purples and greens. The green plants could therefore have been nothing other than A b PI R'. Similarly, progenies consisting of green-anthered purples and sun reds, and greens, in a 9:3:4 relation, are hsted in table 39, group 3. Green plants of these lots and their green descendants might be either Ab PIR^ or A b pi R", or might be heterozygous for PL Six such green plants were crossed with dilute sun reds (table 45, group 2). None of these greens could'have been of the types discussed in the earlier section of this paper, namely, ah PI/ and the like, for they were shown b}^ appropriate tests (Emerson, 1918) to he A c R and some of them have even been used as C testers for aleurone color. Two of these green plants crossed with dilute sun reds gave dilute sun reds only, 59 in all, and are consequently regarded as being Ah pi R^. Two others by sunilar crosses gave dilute purples and dilute sun reds in the relation 20:30, a deviation of 5.0 ±2.4 from the expected equality from plants of the genotype A Abb PI pi R" R". Two other greens were crossed with heterozygous dilute sun reds, A Abb pi pi R'^ /, and gave dilute purples, dilute sun reds, and greens in the relation 69:54:106. The theoretical distribution among these three classes for a total of 229 individuals, based on the assumption that the green parent plants were A Abb Pi pi R" R\ is 57:57:115, a devia- tion that might occur by chance about once in five trials, P equaUng 0.19. Progenies consisting of dilute purples, dilute sun reds, and greens in a 9:3:4 relation are listed in table 43, group lA. Descendants of one of 104 R. A, Emerson these green plants were crossed with dilute sun reds which were Fi's of crosses between dilute sun red and green IVg. The results were dilute purple and green plants in the relation 328:338 (table 45, group 3), a deviation from a 1 : 1 ratio of 5.0 ± 8.7. Since the heterozygous dilute sun red plants were A Ah h pi pi R^ /, the green plants crossed with them are assumed to have been A b PI R^. That this assumption is correct appears the more evident from the fact that the green plants were homo- zygous for colored aleurone, and hence A C R. Green IVg x green Vic. — Twelve crosses between green plants of type IVg and green plants of type Vic gave a total of 159 Fi plants, all dilute sun red. With respect to aleurone color, all the type IVg plants concerned in these crosses were known to be A c R, and, in fact, were in general use as C testers for aleurone color. With respect to plant color, therefore, they are assigned the constitution Ahpl R^. Of the type Vic greens, four were known to be A testers for aleurone color, and were therefore, with respect to aleurone color, a C R. Their plant-color constitution is accordingly set down as ahpl R^. Six of the type Vic greens had an aleurone-color constitution of aC r, their plant-color genotype being accordingly ahpl /. The other two Vic greens were certainly a b pi, but whether they were R'^ or / is unknown. In Fa, dilute sun red and green plants were present in the ratio 420:291 (table 46, group 1, page 154). From an Fi of the genotype A ahh pi pi plus* R^ / or R^ R^, a 9 : 7 ratio of dilute sun red to green is to be expected in F2, since both A and / or R^ are assumed to be necessary for the pro- duction of anthocyanic pigment, which distinguishes dilute sun red from green. The theoretical ratio for a total of 711 individuals is 400:311. The observed deviation from this ratio, 20.0 ± 8.9, is such as might occur by chance about once in eight trials, P equaling 0.13. Two Fi plants backcrossed to green Vic, a h pi R^, gave 66 dilute sun red and 58 green plants, and two backcrosses with green IVg, A h pi r^, gave 96 dilute sun reds and 96 greens, equality of the two classes being expected in the case of both crosses (group 2, table 46). That the two parent types of green occurred in F2 is shown by their relations to aleurone and pericarp color. In the case of every cross, green plants were produced from both colored and colorless seeds. Those from colored seeds could have been only A h pi R^. Since some seeds were colorless because of a a and some because of c c, both parent types of green should have been present in the lots grown from colorless seeds. Plant Colors in Maize 105 In one cross there was prosont tho pericarp factor P, which with A gives a red and with n a a l)rown pericarp. All the F2 green plants from colored seeds had red pericarp, and of those from colorless seeds the majority had brown pericarp. From the colorless seeds there should have occurred also a conil)ination type of green, a b pi W, but no tests were made for the identification of this type. Ten dilute sun reds of F- were tested by their F3 behavior. Three of these (table 47, group 1) gave dilute sun red and green plants in the relation 108 : 77, a deviation from a 9 : 7 ratio of 4.0 ± 4.6. Five other F2 plants (group 2) gave the two color types in the relation 187:66, a devia- tion from a 3 : 1 ratio of 3.0 ± 4.6. Two F2's (group 3) bred true dilute sun red, producing 78 dilute sun red and no green offspring. Theoretically, of 9 Fo dilute sun reds, there should occur in F3, true-breeding, 3:1, and 9:7 progenies in the numerical relation 1:4:4, The observed rela- tion between these three sorts of l^ehavior for the ten F2's tested was 2:5:3. Deviations such as these might occur by chance about once in two trials, P equaling 0.49. Green IVg x green Via. — Certain crosses of green IVg with green \I have given sun red plants in Fi. The t^^pe VI greens belonged to families in which the B factor was known to be present. They were therefore doubtless a B pi plus / or R\ and the Fi's were probably A a B b pi pi plus / R^ or R" R^. F2 consisted of the three major color t^'pes sun red, dilute sun red, and green (table 48, group 1) in the relation 586:161:348. Obviously this is not a 9 : 3 : 4 relation, for the deviations from such expecta- tion, -30, -44, +74, could not be expected to occur thru errors of random sampling once in a million such trials, z^ equaling 30.9 and P equaling .000000+. As a matter of fact, an Fi of the genotype suggested above should give in F2 the three color types observed in the relation 36:9:19. The observed frequencies of the several classes fit this expectation so closely that the deviations from it might occur by chance in about one out of five trials. P equaling 0.19. The comparison of observed and expected frequencies follows: Color types Sun red g^^^ ^.^^ Green Total Ila, g IVa IVg, Via, c Observed 586 161 348 1 ,095 Calculated 616 154 325 1 .095 Difference —30 +7 +23 106 R. A. Emerson Not only were the frequencies of the major color types fairly close to expectation, as indicated above, but the expected subclasses of sun red with pink anthers and with green anthers were observed. Counts of anther color were made -in the case of only 65 individuals. These plants were distributed to the four color classes, pink-anthered sun red, green-anthered sun red, dilute sun red, and green, in the order 24:9:10:22. The theoretical distribution of 64 individuals being 27:9:9:19, the deviations are such as might occur by chance perhaps twice in three trials, z^ equaling 0.91 (when x^ = 1 and n' = 3, P = 0.61). Only three Fa sun reds were tested in F3. One of them (group 2, table 48) bred true sun red, but segregated with respect to anther color. It was there- fore presumably A A B B pi pi/ R''. Two other F2 sun reds (group 3) gave sun red and green offspring in the ratio 229:71, a deviation of only 4.0 ± 5.1 from a 3:1 ratio. One of these two F2 plants was crossed with a dilute sun red, resulting in 55 sun red plants. The two F2 plants, therefore, were presumably A a B B pi pi. Anther color was not deter- mined, but the fact that the green plants of F3 all came from colorless seeds is conclusive evidence for the presence of 4 a and against the pres- ence of / R^. The genotype of the F2 plants is accordingly set down as AaBBplpl / /. Green Illg x green Vic. — Green plants known to be of type Vic, ah pi /, were crossed with greens which were known to be R^ R^ and which from their parentage might have had Pi. The result in Fi was dilute purple, supposedly A ah h PI pi f R^. Two F2 lots (table 49, group 1) consisted of dilute purples, dilute sun reds, and greens in the relation 109 : 37 : 135. From the assumed genotype of Fi, there should occur in F2 the observed color types in the relation 27:9:28. The observed frequencies deviated from the theoretical ones by amounts such as might occur by chance once in three trials. P equaling 0.33. The comparison follows: ri 1 2. Dilute Color types ^^^^^^^ Ilia Observed 109 Calculated 119 Difference —10 —3 +12 — 1 Dilute ^ sun red ^'^^^" Total IVa Illg, IVg, Ylb, c 37 135 281 40 123 282 Plant Colors in Maize 107 The dilute purples of F2 were presumably all A b PI / and the dilute sun reds all A b pi /. Of the Fo greens there should theoretically have been six types, namely, AbPlR^, AbplR", abPl/, ah pi/, ah PIP?, and a b pi R^. The relation of these plant colors to aleurone color and to a pericarp color known as cherry, present in these families, affords an opportunity of checking some of these hypothetical formulae. Cherry pericarp is a bright reddish purple, somewhat variable in intensity. In the parent of one of these F2 progenies it was sufficiently light to make possible the determination of the underlying aleurone color. The F2 seeds consisted of colored and colorless aleurone in the ratio 140:171, a devia- tion from a 27:37 ratio of 9.0 ± 5.9, or such a deviation as might occur by chance three times in ten trials, P equaling 0.30. The Fi plants were known to be AaRr, and in order to give a 27:37 ratio with respect to aleurone color they must have been also C c. Cherry pericarp is of such a nature that it never develops except in the presence of PI. With A and PI it is cherry, but with a and PI it is brownish. It had been regarded by the writer as due to a factor, Ch, but recently Dr. E. G. Anderson has shown (by unpublished data) that the writer's Ch is apparently another allelomorph of R, and at present it is known to exist only in the form /^. Since all dilute purples of the lots under consideration here are assumed to be A 6 PI f^, they should all have cherry pericarp. Again, since dilute sun reds are pi pl, they should all have colorless pericarp. Furthermore, since all green plants from colored seeds are supposed to be R^ R^, their pericarp should likewise be colorless. Finally, since the colorless seeds may lack color because of either a a, r r, or c c alone, or because of both a a and r r, some green plants from colorless seeds should have color- less pericarp, a R^ or A c R", and some should have brownish pericarp, a PI /^. Of course all green plants with pl pl also must have colorless pericarp. The observed results are whoUy in accord with these suppositions. In one F2 progeny, pericarp color was determined for all except a few plants. From seeds with colored aleurone, all the dilute purples had cherry peri- carp and all the dilute sun reds and greens had colorless pericarp. These three classes of plant and pericarp color showed frequencies deviating from the theoretical 27:9:18 relation by quantities such as might occur by chance almost once in four trials, P equaling 0.23. From seeds with colorless aleurone, all dilute purples had cherry pericarp, all dilute sun 108 R. A. Emerson reds had colorless pericarp, and greens had in part brownish and in part colorless pericarp. The deviations from the expected 27 : 9 : 18 : 20 relation of these four color classes were such as might occur thru errors of random sampling in more than seven out of any ten such trials, P equaling 0.72. The comparisons follow: Plant color Dilute purple Dilute sun red Green Green Total Pericarp color Cherry Colorless Brownish Colorless Ilia IVa VIb Illg, IVg, Vic Colored aleurone: Observed. . . . 43 10 35 88 Calculated . . . 44 15 29 88 Difference.-. . —1 —5 +6 Colorless aleurone: Observed. . . . 38 11 32 28 109 Calculated . . . 40 13 27 29 109 Difference . . . —2 —2 +5 —1 Further tests of the factorial composition, with respect to PI, of some F2 green plants of this cross are afforded by crosses between them and sun red and dilute sun red plants. One F2 green crossed with sun red gave 27 purple plants (table 49, group 2). Since the green parent plant came from a colored seed, it is assumed to have been PI PI R^ BF plus A A ov A a. Two other greens crossed with dilute sun red gave 39 dilute purple plants, and were therefore PI PI (group 2, table 49). Since one of these green plants had brownish and the other had colorless pericarp, they are assumed to have been also f^ and R' W, respectively. A fourth F2 green crossed with sun red gave purple and sun red plants, and a fifth green crossed with dilute sun red gave dilute purple and dilute sun red plants, indicating Plyl (group 3, table 49). The first of these two had brownish and the second had colorless pericarp. They must therefore have been r'^'' and I^ W, respectively. A sixth F2 green crossed with dilute sun red gave only dilute sun red plants, and so must have been yl'pl (group 4). Plant Colors in Maize 109 Green Illg x green Via. — In the soctioiis immodiatoly preceding this, it has been shown that intercrosses of greens may give dilute sun reds (page 104), dikite purples (page 106), or sun reds (page 105) in Fi, the particular color type depending on the genotypes of the greens chosen for crossing. It remains to be shown that purple la can be produced ])y intercrosses of gre'ens. A cross of green Via, aBplY, with green Illg, Ah PI R\ should give this result, Fi being AaBb PlplR' /. Such a cross has been made, with results as expected. A stock of green plants was isolated from a cross of brown V, a B PI /, with green VIc, a h pi /, and was shown, by crosses with aleurone testers and with dilute sun red IVa, to be type Via, a B pi /. Another lot of greens arose from a cross of purple Ig with green IVg. The purple Ig parent was. from a lot consisting of purple la, purple Ig, dilute purple Ilia, and green Illg, coming from a cross of purple Ig with dilute purple Ilia heterozygous for R^ /. It was therefore AABhPlPlR^ W. The green IVg plant with which it was crossed was known to be A 6 pi RF. The Fi of this cross consisted, as was expected, of purples and greens only. The purples were type Ig and must have been heterozygous for B b and PI pi, and the greens must have been type Illg and heterozygous for PI pl, or A Abb PI pi R^ RF. Two of these Fi greens were crossed with one of the greens of tj^pe Via mentioned above. The two crosses, 9659 and 9660, resulted as expected in purple-anthered purples, type la, and pink-anthered sun reds, type Ila, in the relation 18:20. It has been demonstrated, therefore, that by crossing wholly green plants of appro- priate genotypes it is possible to produce purple-anthered purples, the most highly colored type known, a type that is dominant to all other types. Green Illg x purple la. — A green plant with homozygous purple aleurone and belonging to a family (table 39, group 4) consisting of green-anthered purples and greens only, and therefore theoretically being A b PI R^, was crossed with a purple-anthered purple, A B PI /. A purple-anthered purple Fi, A A Bb PI PI f R", 5350-9, was backcrossed with green IVg of the genotype A b pl ?■", with the result that in the next generation there appeared four color types, purple-anthered purple, green-anthered purple, dilute purple, and green, in the relation 28:22:21:29. The deviations from the expected equal distribution of the 100 individuals were such as might occur by chance in considerably more than half of 110 R. A. Emerson such trials, P equaling 0.57. It will be recalled that results like these were obtained from a cross of green-anthered purple with dilute purple (page 96), and of course the same results were to be expected since the Fi in both cases is supposed to have been A A B b PI PI /R^. The cross now under consideration has interest from the standpoint of the relation of aleurone color to plant color, and also for certain hnkage relations. The Fi was known to be, with respect to aleurone color, A A Rr. Whether it was C C or C c was not known, since a strong red pericarp made aleurone counts impracticable. The green plant on which the Fi was backcrossed, was determined by appropriate tests to be C C, so that the relation of the Fi purple to C is immaterial. The backcross resulted in approximatelj'- equal numbers of seeds with and without aleurone color, there being 109 colored and 110 colorless seeds. The colorless seeds must have been A B C PI/ t^ and Ah C PI/ 7^, and should therefore have produced purple-anthered purples and dilute purples only; while the colored seeds must have been A B C PI R" / and Ah C PI R^ r^, and should correspondingly have produced green-anthered purples and greens only. The results were quite in accord with expecta- tion, as is shown in the following comparison: Green Total Illg 29 51 49 It has been shown earlier in this paper (page 63) that a linkage exists between the factor pair B h and a factor pair, Lg Ig, for normal or ligule- less leaf, the percentage of crossing-over being about 30. It happens that the Fi of this cross was Lg lg as well as B h, B lg having come from one parent and h Lg from the other, and that the green plant used in the backcross was h lg. There is no question here that the purple-anthered purples and dilute purples produced from colorless seeds differed with respect to the B h pair only. Their linkage with liguleless leaf, as indi- cated by the percentage of crossing-over, was 29.4, or a deviation from 30 of 0.6 ± 2.0. Practically the same linkage relation was found for the plants from colored seeds, green-anthered purples and greens. In this case the percentage of crossing-over was 27.5, a deviation from 30 of Color types Purple, purple anthers Purple, green anthers Dilute purple la lg Ilia Colored seeds. . . . 22 Colorless seeds . . . 28 21 Plant Colors in Maize 111 2.5 ±2.1, or such as might occur by chance about twice in five trials, P equahng 0.42. It is to be assumed, therefore, that the same difference exists between green-anthered purples and greens as between purple- anthered purples and dilute purples, namely, a difference with respect to the factor pair B b. This in turn is merely additional evidence that plants which in the presence of / are dilute purples, A b PI, appear as greens in the presence of R" r", which is the hypothesis under test thruout this section of the paper. Purple la x green-anthered dilute sun red A purple-anthered purple, known from appropriate aleurone-color tests to he R R and hence A B PI R^, was crossed with a dilute sun red which differed from most dilute sun reds in showing much less anthocyanic pigment, particularly in early stages of growth, than is usual in plants of that type, and in having little, if any, color in its anthers. The F/s, 2975, were purple-anthered purples. F2 was expected to show the four color types, purple, sun red, dilute purple, and dilute sun red, commonly found in crosses of purple la with dilute sun red IVa. As a matter of fact, the single F2 progeny grown was found to consist of these four color types as major classes, but each class was found to have colored-anthered (purple or pink) and green-anthered subclasses. The difference between the two subclasses for purple and sun red was sharp, just as is the case in crosses of purple la with green IVg, but it was often difficult to separate green-anthered dilute purples from green-anthered dilute sun reds. Ordinarily, anther color (purple or pink) is the surest means of distinguish- ing between dilute purple and dilute sun red. When both have green anthers the separation must be based on the relative amount of pigment in other plant parts — a difference that is usually not very marked until late in the life of the plants, when dilute purples usually show materially more pigment, especially in parts not exposed to the sun, than do dilute sun reds. It will be recalled that in crosses of purple la with green IVg, both colored and green-anthered purples and sun reds appear, but that all the dilute purples and dilute sun reds have colored anthers, the green- anthered individuals appearing as wholly green in all plant parts except perhaps the pericarp. But in the cross here considered, no wholly green plants were found. 112 R. A. Emerson The natural supposition is that there is here concerned still another form of the R factor, such that, while it does not allow color to develop in the anthers, does nevertheless result in the development of some antho- cyanic pigment in other parts of the plant. The dilute sun red plant used as one parent of this cross was found to he A c R with respect to alcurone. The factor particularly concerned in the behavior here reported is there- fore assigned the designation R''^. The Fi plants are accordingly assumed to have been A A Bb Plpl R'' R^^. The frequency distribution for the eight color types observed in F2 approached the theoretical distribution so closely that deviations of the magnitude observed might occur by chance nearly three times in any ten such trials, P equaling 0.72. The com- parison follows: x~\:i.,4-« T^cii,*-^ T^,*i,,+« T^;i,,4-« Total 491 491 Plant color Anther color Purple Purple Purple Green Sun red Pink Sun red Green Dilute purple Purple Dilute purple Green Dilute sun red Pink Dilute sun red Green Observed. . . Calculated . 212 207 77 69 66 69 22 23 60 09 23 23 22 23 3 8 Difference. . +5 +8 —3 —1 —3 One F2, a green-anthered purple, was tested in F3. This plant bred true, producing 128 green-anthered purples and no other types. It is unfortunate that the relation of aleurone color to plant color in this cross afforded no check on the assumption that the observed behavior with respect to anther color of dilute purples and reds was due to a factor belonging to the allelomorphic series R^, R^, /, r^. True, the Fi plant tested was heterozygous with respect to aleurone color, but this was known to be due to C c. Since no further tests have been made, the only evidence in support of the assumption of a factor R^^ is the very close fit of observed with theoretical frequency distributions, the fact that colored and green anthers in purple and sun red types of many other crosses have been found to be due to the R factor, and the demonstrated presence of R in the green-anthered sun red plant used in the cross. ,0 jch Summary of results involving the allelomorphic series R^, R^, R^^, 1,1,1 Crosses of brown with green of type IVg have been shown to result in purple Fi's, and in eight color types in F2 in a numerical relation approxi- mating 81:27:27:9:27:9:36:40, or in six major color types, anther color being disregarded, in approximately the relation 108:36:27:9:36:40. Plant Colors in Maize 113 It has been noted that these results are wholly unhke those for crosses of brown with green reported in an earher section of this paper, and are similar in general, tho with marked differences in detail, to previously discussed crosses of brown with dilute sun red. As an interpretation of these results, it has been assumed that, in addition to the three pairs A a, B b, PI 2)1, a fourth pair — members of a multiple-allelomorph series, such as R^ Bf, / R", or R^ i^ — is concerned. It has been assumed further that R^ or / is necessary ordinarily for the development of dilute purple and dilute sun red and for the appearance of purple and pink anthers in purples and sun reds, respectively, while /i" R° or r'' r^ is necessary for green anthers of purples and sun reds and for the con- version of dilute purples and dilute sun reds into wholly green plants. Smiilarly, the appearance of green-anthered dilute purples and dilute sun reds in a single cross has been ascribed to R^^ R^^. The relation of the R allelomorph to both aleurone color and plant color has afforded rehable tests of the hypothesis. Other tests have consisted of the behavior in later generations of the several F2 color types and the results of intercrosses between these types. Neither of these tests has been carried to the point of exhausting all the possibilities, but in all a considerable number of tests have been made and all have given results in support of the hypothesis. A single linkage test, involving the B b pair with leaf type, Lg Ig, has afforded added support. On the whole, therefore, the hypothesis haS been, if not substantiated, at least rendered highly probable. RELATION OF ALEURONE FACTORS C C AND Pr pr TO PLANT COLOR The relations of the aleurone factors A and R to plant color have been noted repeatedly in this account. A single observation suggests a rela- tion between the aleurone-f actor pair C c and leaf color. Culture 2909 came from colored seeds of a selfed ear showing a 3 : 1 ratio of colored to white seeds, and therefore heterozygous for a single pair of aleurone-color factors. Several ears in the resulting progeny also gave 3 : 1 ratios. Tests of four plants with aleurone testers gave conclusive evidence that the C c pair was the one concerned. One selfed plant of the lot, 2909-32, had 318 colored and 105 white seeds. Both the colored and the white seeds produced only sun red plants, some with green and some with pink anthers, indicating the genotype A A B BC cplpl W R^. All the plants showed strong sun red pigment in the sheaths and the outer husks, but 114 R. A. Emerson there was distinctly more red color in the leaves of the plants from colored seeds than in the leaves of the plants from white seeds. Particular atten- tion has not been given to a possible effect of the C factor on mature plant colors of other color types. Many cultures of dilute sun reds and greens have afforded opportunities for observing any effect of C and c on red color in the leaves of seedlings, but no effects have been noted. No particular attention was paid to the matter at the time when the seedlings were under observation, but if the C c pair had exerted any marked influence it would probably have been noted. Numerous cultures of dilute sun red seedlings have been noted with respect to possible effects of the aleurone-factor pair Pr pr, but no effect has been observed, the purple and the red seeds having produced seedlings with apparently the same intensity of red color. Likewise, no influence of Pr pr on mature plant color has ever been observed in the case of either sun red or dilute sun red. With purple and dilute purple plants, however, a distinct effect is noticeable. Purple and dilute purple plants from seeds with purple aleurone have purple anthers, while those from seeds witJi red aleurone have reddish purple anthers (Plate I, 1 and 3, and Plate II, 1 and 3). A similar effect is often seen also in the color of the inner husks. In neither the anthers nor the husks is the effect always suffi- ciently distinct to make possible an accurate separation of plants from purple and from red seeds if they are growing in mixed cultures. In some cases, however, the difference is very distinct. And when the seeds are separated with respect to purple and red aleurone, the two lo+s of plants resulting usually show fairly distinct differences in anther color and often in husk color as well. EXPRESSION OF PLANT-COLOR AND ALEURONE- COLOR FACTORS The mode of expression of the several plant-color factors has been dis- cussed in detail in this paper, and similar discussions of aleurone-color factors are available in numerous other papers. Since aleurone colors and certain plant colors — the purple-red series — are doubtless antho- cyanins, it seems natural to expect close interrelations between them. Many such relations have been noted in this account. There are certain matters, however, which need to be brought together in a summary discussion. Plant Colors in Maize 115 It will be recalled (Emerson, 1918) that for the development of any aleurone color, the presence of three dominant factors, A, C, and R, and also of a duplex recessive factor pair, i i, is necessary. The Pr pr pair has no visible expression except when associated with this combination of the other factors, and then it determines whether the color shall be purple or red. So far as is now known, the plant-color situation with respect to complementary factors is not quite so complex. Something of the same sort is seen, however, in the fact that no anthocyanic pigment ordinarily develops except either in the presence of A and R"", /, or r"'', or in the presence of A, B, and R^ R^ or r^ /. With the first of these combinations, the pairs B h and PI pi determine the particular color type of the purple-red series. Two of these types, purple and dilute purple, are modified further by Pr pr, and the intensity of their color depends also on the member of the R series present, r'^'^ exerting a more pronounced effect than R'' or /. One type at least, sun red, is influenced somewhat by C c. With the second combination. A, B, and R^ R'' or r^ r", the pair PI pi determines whether the type shall be purple or sun red. For the formation of the non-anthocyanic (flavonol) pigment, brown, the interaction of a a with" either B or PI is essential, and usually very little color develops except when both B and PI are present. Brown is made more intense by the presence of /*. Of the factors concerned with plant colors of maize, the A a pair is one of the most fundamental, since it differentiates sharply the antho- cyanins of the purple-red series, A B PI, A B pi, Ah PI, A b pi, from the non-anthocyanic brown, aBPl, and the slightly brown or green a B pi and ab PI and the wholly green a b pi. Without A no anthocyanin shows in either the aleurone or the other parts of the plant. A second fundamental pair is Pi pi, which differentiates the sun colors from those that develop in local darkness. Purple (A B PI), dilute purple (A b Pi), and brown (a B PI) are all able to .develop in darkness; while sun red (ABpl), dilute sun red (Abpl), and the slight brown sometimes seen in a B pi, do not develop except in the presence of light. Whether or not the slight brown sometimes present in a 6 Pi forms in darkness has not been determined. To the Pr pr pair is due a definite qualitative difference in the colors formed which is presumably associated with a difference in chemical composition of the pigments. In the presence of Pr aleurone color is purple, and with pr it is red, and a similar difference, tho not always 116 R. A. Emerson so sharp a one, is seen in the effects of Pr pr on the anther and husk color of purples and dilute purples. The factors R^ and r^ on the one hand, both recessive with respect to plant color, and R^ and / on the other hand, both dominant for plant color, apparently alwaj'S differentiate between colored and colorless anthers and silks in the purple-red series of plant colors, and, when B is absent, determine whether or not antho- cy.anin forms in any part of tlie plant. The pair B h influences mainly the intensity of pigmentation. Thus, purple, A B PI, is more strongly colored than is weak purple, A B^ PI, which in turn is more strongly colored than is dilute purple, A b PL The same relation holds between sun red, A B pi, weak sun red, A B^ pi, and dilute sun red, A b pi. Brown color shows very little in ab PI but is strongly developed in a 5 PL A similar difference, however, exists between the slight brown of a B pl and the full brown oi a B PL In the one case in which an effect of C c has been noted, C acted as an intensifier of color. There are som.ewhat marked differences between the several factor pairs with respect to the stage of plant development at which their influence is expressed. Seedlings of purple, sun red, dilute purple, and dilute sun red normally exhibit no characteristic differences in intensity or extent of pigmentation. The B b and PI pl pairs, which differentiate these color types so sharply at a later stage of growth, do not, therefore, come into expression early. All of these types are more highly colored late in their growth period than they are as seedlings, but the later changes are much more pronounced, for instance, in dilute purple than in dilute sun red, and somewhat more so in purple than in sun red. Apparently, Pl exerts its influence comparatively late, but under the intensifying influence of B, even Pl expresses itself fairly early. The several factor pairs differ more or less with respect to the particular plant parts affected. Differences in the expression of B, B^, and b are more apparent in the husks and the sheaths, particularly the upper sheaths, than elsewhere. When plants of the genotype a B pl, common^ classed as green, show any brown, the color is limited to the sheaths and the outer husks. The difference between purple (A B Pl) and sun red {A B pl) on the one hand, and dilute purple (A b Pl) and dilute sun red (A b pl) on the other, is more pronounced in the husks and the sheaths than elsewhere. Little difference is apparent between the two groups with respect to the color of anthers, glumes, silks, and the like. The pair Plant Colors in Maize 117 PI pi is perhaps expressed most definitely in the color of anthers, tho the expression is by no means limited to them. Puiple {A B PI) and dilute purple {A b PI) differ from sun red (A B pJ) and dilute sun red (A hpl), not merely in having purple rather than pink anthers, but also in the coloration of their inner husks, their culms, and the like. What little brown color is seen in ab PI is limited almost wholly to the staminate inflorescence. The staminate inflorescence of purples, A B PI, and of browns, a B PI, is strongly colored, but that of dilute purple, A b PI, except for anther color, is not very different from what is seen in dilute sun red, A b pi. The PI factor, when associated with r"^^, is expressed in the pericarp as cherry in purple and in dilute purple, and as brownish in brown and in green of the genotype a b PL Factors B b and PI pi are not known to be concerned with aleurone color. All the other factors affecting plant color are aleurone-color factors also. Of these the pair Pr pr influences anther color of purple and dilute purple, and to some degree the husk color as well. The pair C c has been observed to affect the leaf color of mature plants of the sun red type. The pair A a is expressed to some degree in all such parts as culms, sheaths, husks, glumes, anthers, and silks. The pericarp, if a pericarp factor P is present, is red with A and brown with a, or if r'^'^ and Pi are present, it is cherry with A and brownish with a. The R series of factors influences many plant parts. With duplex R^ or r^, no color develops in any part of the plant, except the aleurone, provided B is absent. With B these factors give colorless anthers and silks merely. Factors K^ and /, if A also is present, affect practically all plant parts in which anthocyanic pigments ever develop, but are not iknown to have any influence on the color of the pericarp. The factor r'^^ is, how- ever, concerned with pericarp color provided PI also is present. This factor has a marked influence on the amount of color that forms in the leaves, particularly of dilute purple and dilute sun red. It is of no little interest that the R series of factors, which behaves as a group of multiple allelomorphs with regard to plant color, usually acts as a simple pair in respect to aleurone color.* Moreover, some of these factors act as dominants with respect to aleurone color and as recessives with respect to plant color, while the dominance of others is « There is'^some evidence that at least one aleurone-color pattern is dependent on an allelomorph of R r, the three thus constituting a group of triple allelomorphs affecting aleurone-color development. 118 R. A. Emerson the reverse of this. For example, r" and r"'' are recessive for aleurone and dominant for plant color, and R^ is dominant for aleurone and recessive for plant color, while R'^ is dominant and r^ recessive for both alem-one and plant colors. SUMMARY In this account, six major plant-color types of maize, purple, sun red, dilute purple, dilute sun red, brown, and green (colorless), together with the subtj^pes, weak purple, weak sun red, green-anthered purple, green- anthered sun red, and five genotypes of green, are described and illustrated, and their environmental and genetic relations are discussed. The sun red and dilute sun red types are shown to be dependent on Hght for the development of their color, while the purple, dilute purple, and brown types develop their characteristic colors in darkness. Diversities of temperature and of soil moisture are shown to have no direct effect on the formation of maize plant colors but to have an indirect relation to them thru their influence on soil fertility, which in turn bears a definite relation to the development of the purple-red series of plant color, anrthocyanins, but little or no relation to brown. Sun colors particularly are shown to be markedly intensified by infertile soil. It is noted that the several types of the purple-red series are sharply differentiated when grown on fertile soil, but that their characteristic differences are largely masked by growth on infertile soil, while the brown-green* series is most readily distinguished from the purple-red series, especially in the seedling stage, if grown on infertile soil. It is suggested that the effect of infertile soil may be due to a deficiency of nitrogen, and perhaps of phosphorus. Observations indicating a close connection between the accumulation of carbohydrates and strong colora- tion are reported, and the inference that the effect of infertile soil is brought about thru checking growth without inhibiting photosynthesis, thus allowing an accumulation of carbohydrates, is discussed. In an attempt at a genetic analysis of the several plant-color types, data accumulated during a period of some ten years, and involving an examination of approximately 680 progenies and not less than 48,000 individual plants, are reported. As an interpretation of the results obtained from the more complex crosses, the allelomorphic pairs A a and PI j)l, and the multiple allelomorphs B, B"", b\ h, and R\ R\ R'\ Plant Colors in Maize 119 /, r*, r*^, are assumed and genetic formulae are assigned to the several color types as follows: purple, A B PI; sun red, A B pi; dilute purple, Ab PI; dilute sun red, Abpl; brown, a B PI; green, a B j^, ah PI, ah pi; all these having in addition R\ /, or r"^. The factor i^''" is assumed to be the causal factor for green anthers and silks in purple, sun red, dilute purple, and dilute sun red types, and W and r'' are assumed to have the same effect on purple and sun red and to insure colorlessness (green type) thruout in what would otherwise be dilute purple and dilute sun red, the R series having no effect on brown, except for r'^'', which intensifies brown as well as purple and dilute purple. Of the R scries, R^ is dominant and 1^ is recessive for both plant and aleurone color, f and r'^^ are dominant for plant and recessive for aleurone color, R^ is recessive for plant and dominant for aleurone color, and R''^ is dominant for aleurone color and also for plant color except of the anthers and the silks, for which it is recessive. The A a pair is concerned with both aleurone and plant color, and the aleurone pairs C c and Pr pr are assumed to exert a modifying effect on certain plant colors. The principal hypotheses involved are shown to be in keeping with observed facts when subjected to practically all the available genetic tests, such as backcrosses of Fi with multiple recessives, behavior of F2 types in later generations, intercrosses of the several F2 types, relation of aleurone color to plant color, linkage of certain plant-color types with normal- and liguleless-leaf types and of other plant-color types with yellow and white endosperm. Approximately 32 per cent of crossing- over is reported between B h and Lg Ig and about 20 to 30 per cent between PI pi and Y y. 120 R. A. Emerson LITERATURE CITED Collins, G. N. Gametic coupling as a cause of correlations. Amer. nat. 46:569-590. 1912. CzARTKOwsKi, Adam. Authocyanbildung und Aschenbestandteile. Deut. bot. Gesell. Ber. 32:407-410. 1914. East, E. M., and Hayes, H. K. Inheritance in maize. Connecticut Agr. Exp. Sta. Bui. 167: 1-142. 1911. Emerson, R. A. Genetic correlation and spurious allelomorphism in maize. Nebraska Agr. Exp. Sta. Ann. rept. 24:58-90. 1911. The inheritance of the ligule and auricles of corn leaves. Nebraska Agr. Exp. Sta. Ann. rept. 25:81-88. 1912. A fifth pair of factors, A a, for aleurone color in maize, and its relation to the C c and R r pairs. Cornell Univ. Agr. Exp. Sta. Memoir 16:225-289. 1918. Gernert, W. B. The analysis of characters in corn and their behavior in transmission, p. 1-58. (Published by the author, Champaign, Illinois.) 1912. Knudson, Lewis. Influence of certain carbohvdrates on green plants. Cornell Univ. Agr. Exp. Sta. Memou- 9:1-75. 1916. LiNDSTROM, E. W. Chlorophyll inheritance in maize. Cornell Univ. Agr. Exp. Sta. Memoir 13 : 1-68. 1918. Sando, Charles E., and Bartlett, H. H. The occurrence of quercetin in Emerson's brown-husked type of maize. Journ. agr. research. 1921. (In press.) Webber, Herbert J. Correlation of characters in plant breeding. Amer. Breeders' Assoc. 2:73-83. 1908. Wheldale, M. On the formation of anthocyanin. Journ. genetics 1:133-158. 1911. Memoir .3^, A Modified Babcock Method for Determining Fat in Butter, the second preceding number in this series of publications, was mailed on December 10, 1920. Plant Colors in Maize 121 APPENDIX TABLE 1. Fi Progenies of Purple la x Green VTc Pedigree nos. Number of Pi Fi 1 1 plants (Purple la) 724-1 X 722-1 857 IS 1121-8x1122-7 1420, 1512, 2022. 40 1122-5 X 1121-2 1419, 1511. . 36 1525-5 X 1546-5 20.56 17 Total, 4 progenies 111 TABLE 2. F2 Progenies of Purple la x Green Vie Pedigree nos. Number of F2 plants Group Fi Fo Purple la Sun red Ila Dilute purple Ilia Dilute smi red IVa Brown V Green Via, b, c 1 1419- 1.. 1511- 1.. 1512-12. 2022- 3.. 2056- 6.. -11.. -16.. 1513 2018 2020 4012,4013. 2415, 2416, 42.84 2417, 2418, 2553-2559, 4001-1007. 2412, 4066, 4067 94 61 54 7 39 96 17 22 19 16 6 13 22 3 26 13 23 6 17 24 11 12 4 7 3 4 3 1 20 13 21 4 16 26 8 23 9 7 1 10 8 7 Total, 7 progenies 36S 101 120 U 108 65 2 1514-24.. -31 2000- 8. . 2019-28. -34.. 2906- 1 . . 2907- 1 . . - 7. , 2981- 2. - 5 4020- 7 4032- 1 - 3 - 4. 20.54 2055 2419,4065 4281 4282 5303 5290-5293. 7050, 7051 5299, 5300, 70.54, 7055 .50)6, 5067 . 5068, 5069 .5712,6810 .5739 50!^ 5087 20 22 92 24 21 17 93 105 17 20 10<) 16 15 13 7 4 29 8 6 7 26 46 4 6 44 5 / 5 8 4 21 4 4 5 34 30 5 2 26 3 5 4 1 2 8 4 2 7 10 1 1 12 2 4 1 5 2 19 4 7 6 34 38 8 2 31 3 4 7 2 6 25 6 4 3 23 31 3 3 33 2 3 7 Total, 14 progenies .... 5^ 204 155 57 170 151 Total, 21 progenies 952 305 275 91 278 216 122 R. A. Emerson TABLE 3. F'' Progenies of Purple x Green Backcbossed with Geeen (la X Vic) X Vic Pedigree nos. Number of F2 plants Group Fi X Vic F2 Purple la Sun red Ila Dilute purple Ilia Dilute sun red IVa Brown V Green Via, b, c 1 1420- 1x1430- 3 1511- 1x1516- 1. 1512-12 X -14. 2056-16x1995- 6. 1514 2019 2021 2413, 4068 12 18 23 4 19 8 18 10 15 12 16 8 16 8 10 6 14 18 13 8 45 50 44 18 Total 4 progenies 57 55 51 40 53 157 2 2867-69x4032- 1. 2906- 1x2887-10. 2907- Ix -22 - 7x4032-41. 4020- 7 X 2888-13 . 4032- 2 X 2921- 4 . 3x2888- 5. 3x2922-16. 4x2888- 1. 4x2921- 4. 5740 5305 5296,7052, 7053... 5301,5302 5714 5094 5086 5085 5089 5090-5092 7 7 10 16 2 8 19 14 5 25 4 5 11 16 9 6 16 10 15 13 6 2 10 16 9 IS 21 22 12 19 3 8 11 19 4 12 12 16 17 18 4 3 9 25 4 15 18 8 18 15 10 9 26 47 18 33 46 34 45 54 Total, 10 progenies 113 105 125 120 119 322 Total, 14 progenies 170 160 176 160 172 479 Plant Colors in Maize 123 TABLE 4. Fi Progenies of Dilute Sun Red IVa x Brown V Pedigree nos. Number of Fi plants Group Pi Fi Purple la Sun red Ila Dilute purple Ilia Dilute sun red IVa 2025-23x2192-14.. 2029- 8 X 1945-11 . - 8x2013-19. - 8x2014- 8.. 2031-10x1945-10. -32x2012- 1.. 2948-16x4042- 2.. 4253- 2x4299- 2.. 4305- 5x4042- 2.. 2333, 4314 25 30 17 5 16 20 79 46 24 2304, 3596 2311 2310,4034 2309 1 2392 5168, A108, A120.... 5528, 6748A 5193, 5194 i Total, 9 progenies ... . .... 262 I 2018-69x2192-18.. 2030-13 X -14.. 2031-20x2012- 1.. 2043- 2x2026-17.. 2049-14x2192-14.. 2473- 3x2341- 1. 4370- 5 X 3000- 2 . 2386, 4301 30 7 55 15 24 4 8 35 6 55 18 21 2 10 4319 2 2325, 2326, 2543, 2544, 2950, 2951 . . 2347,4326 2336, 4327 4029 4746, 4747 Total, 7 progenies 143 147 2023-19x2192-12.. -23 X -12.. 2027- 9 X -14. 2410- 4x2417- 2. - 6x - 1. 5500- 5 X 5130- 1 . . 2332,4311 19 15 15 9 23 24 26 16 18 6 32 25 2330, 4310 2334, 4316 3 2993, 2994 2995-2998. . A65 Total, 6 progenies 105 123 2025-10x2192-14.. 2029-27x2012- 1.. -32 X - 1 . . -34x2014- 8.. 4315 1 4 3 1 2 3 5 1 6 5 6 2 3 2319,4055 4 2316,4318 3 2314, 4054 6 Total, 4 progenies - 9 11 19 17 124 R. A. Emerson TABLE 5. Fo Progenies of Dilute Sun Red IVa x Brown V Pedigree no.3. Number of F2 plants Group Fi F2 Purple la Sun red Ila Dilute purple Ilia Dilute sun red IVa Brown V Green Via, b, c 1 2310- 2.. 2332- 1 . . 2950- 1 . . - 4.. -17.. -19.. 2995- 7 . . 2996- 1 . . 4029- 2 . . 4034- 1 . . - 2.. 5193- 1 . . 5194- 5.. 5528- 8.. 4036, 4037 . 2999, 3000 . 5036,5037. 5030, 5031 . 5034,5035. 5032, 5033 . 5000-5007 5008, 5009 . 5095 5098,5099. 5104 A135 A136 6748B 15 31 36 37 32 39 75 150 61 46 42 20 10 49 7 9 12 15 5 12 24 50 23 12 20 5 3 11 6 8 12 13 14 10 20 58 11 19 17 4 12 14 3 1 2 3 6 3 5 20 5 7 8 3 1 4 3 15 10 8 13 12 21 48 22 17 13 4 4 12 7 7 9 13 9 5 17 45 11 7 21 1 7 18 Total, 14 progenies 643 208 218 71 202 177 2 2973- 5.. 2974- 9.. 4046- 3.. 5173- 4.. S17-19... 5056-5062. 5063-5065. 5157,5158. A128 7762 55 75 20 19 35 23 24 11 5 11 21 23 6 8 5 6 10 5 1 1 17 18 V 9 14 17 22 4 9 4 Total, 5 progenies . . . . 204 74 63 23 65 56 Total, 19 progenies .... 847 282 281 94 267 233 TABLE 6. F2 Progenies of Dilute Sun Red x Brown Backcrossed with Green (IVa X V) X Vic Pedigree nos. Number of F2 plants Fi X Vic F2 Purple la Sun red Ila Dilute purple Ilia Dilute sun red IVa Brown V Green Via, b, c 2310- 1x2411-6.. 2922-13x4029- 2.. 4029- 2x2921-10.. 4034- 1x2922-16.. - 2x2921-68.. 5813-25x5528- 8.. A129-12xA108-6.. 4035 5652,5653.. 5096 5100-5103.. 5105 6749 A243,A244. 3 22 9 10 12 3 25 4 13 18 13 5 19 3 19 12 17 5 2 20 6 24 8 11 4 4 15 2 27 13 9 4 1 23 17 75 51 33 16 9 48 To al, 7 progenies . . . . 84 72 78 72 79 249 Plant Colors in Maize 125 TABLE 7. Fj Progenies of Purple x Green and Dilute Sun Red x Brown Back- crossed WITH Dilute Sun Red (la X Vic) X IVa, and (IVa x V) x IVa Pedigree nos. Number of Fa plants Group Fi X IVa F2 Purple la Sun red Ila Dilute purple Ilia Dilute sun red IVa 1 2050-16 X 1992-13 2889-54 X 4032- 1 2414, 4069, 4070. . . 5741-5744 18 24 16 27 21 21 15 24 Total, 2 progenies 42 43 42 39 2 6730 - 9 X 6748A- 5 . , 6748A-16 X 6751 -22.. -18 X -22.. -19 X - 1 . . -20 X - 1 . . A121- Ox A108- 8.. L188- 1 x5528 - 8.. 7467,7828 7229 7230 7231 7232 A241, A242, A401, A462 6786, S2 87 40 28 40 30 28 4 79 32 28 33 25 25 5 75 42 26 30 32 38 3 71 41 35 36 20 45 4 Total, 7 progenies 257 227 246 252 Total, 9 progenies 299 270 288 291 TABLE 8. Fs Progenies of Selfed and Back crossed Fo Purple Plants of the Crosses Purple la x Green Vie and Dilute Sun Red IVa x Brown V Pedigree nos. Number of F3 plants Group F2 F3 Purple la Sun red Ila Dilute purple Ilia Dilute sun red IVa Brown V Green 1513-41 2015, 400S, 1009 204S, 2475, 4010,4011 4208 4271 4275 5210 5213 5214 32 01 15 13 10 9 25 22 14 14 7 8 8 3 7 5 8 21 5 6 6 3 (; 5 3 7 1 2 1 6 18 6 3 7 1 4 12 (VIa,b,c) 8 -68 1 2018- 2 - 9 10 4 3 2020- 1 3 40O5- 6 1 -62 -63 1 4 Total, 8 progenies 193 66 60 16 57 34 2020-117x2043-11 4279 4 4 11 4 4 18 126 R, A. Emerson TABLE 8 (continued) Pedigree nos. Number of F3 plants Group F2 F3 Purple la Sun red lla Dilute purple Ilia Dilute sun red IVa Brown V 1 Green 2 1513 - 35 -138 2018 - 6 4066 - 3 6748B- 41 2046 2052 4270 5216,5217. 7400 11 16 25 38 12 4 6 9 14 3 6 1 5. 13 4 1 2 5 5 Total, 5 progenies 102 36 29 13 1513- 59 - 92 -133 4037- 5 2047 2053 2049 5136,5137. 20 24 16 35 6 6 4 15 5 3 9 6 (Via) 3 3 2 7 3 Total, 4 progenies 95 31 23 15 2020-46 X 2200- 8 2411- 4x2412- 2 2443- 2x - 2 2922-12x4037- 5 4283 2981-2983. 2984-2986. 5138-5140. 5 8 7 34 1 6 8 43 3 2 7 32 3 10 4 36 Total, 4 progenies 54 58 44 53 2018-27 4280 4276 4277 5079 5010-5013. 5218 A78 19 19 29 11 195 29 16 5 3 7 4 77 12 6 9 9 6 4 71 6 6 (Vib) 1 2020-15 -30 3 4 4001-12 1 4 4005- 5 4066- 5 5099-22 29 3 1 Total, 7 progenies 318 114 111 42 (; 1513- 2 -110 2018- 92 -119 2412- 1 2050 2051 4273 4269 4033 19 16 31 33 40 3 6 13 9 13 Total, 5 progenies . 139 44 Plant Colors in Maize TABLE S (concluded) 127 Pedigree nos. Number of F3 plants Group F, Fa Purple la Suji red Ila Dilute purple Ilia Dilute sun red IVa Brown V Green 5 (con- tin- ued) 2411-5x2412-1 . 24^-1 X -1 . 4032 4019,4020 4 8 I S Tot^l, 2 progenies 12 9 6 4006- I 4065-14 5014,5015. 5209 126 42 37 12 Total, 2 progenies 168 49 TABLE 9. F4 Progenies of Self-pollinated Purple Plants of F3 Lots Consisting OF Color Types la, Ilia, V. and VIb Pedigree nos. Number of F4 plants Group F3 F4 Purple la Dilute purple Ilia Brown V - Green VIb 5010- 7 - 9 7020, 7021 7022 7023 51 53 46 35 15 19 17 17 12 22 26 14 8 6 1 -11 5011- 4 7024, 7025 5 7028. 7029 1 Total, 4 progenies . 185 68 74 20 4276-32 5010- 2 5011- 6 5181, A170 46 14 28 12 2 14 2 7092 7091,6837 Total, 3 progenies. 8- 5 -19 5049- -13 5052- -3 - -12 6829- 9 1 5243, 5244 . 5245, 5246 . 6913, 6914. 6833 S5 6832 7655-7657. Total, 7 progenies. (Illg, IVg) 5 13 11 24 15 32 30 130 152 R. A. Emerson TABLE 44. F2 Progenies of the Crosses Sun Red Ilg x Green IVg, Sun Red Ila X Green IVg, and Dilute Sun Red IVa x Green IVg Pedigree nos. Number of F2 plants Group Fi F2 Sun red Dilute sun red Green Pink anthers Ila Green anthers Ilg Pink anthers IVa Green anthers IVg 1 4787-6 52&4-3 6983,6984 7003-7006 122 94 52 25 Total, 2 progenies .... 216 77 2 Fi X IVg 7317-6 X 7318-4 7318-1 X 7317-4 -4x -6 8214-8217 .8222-8225 8218-8221 22 38 45 31 25 34 24 34 47 32 26 51 Total, 3 progenies 105 90 105 109 5267-3 6671-6674 .7725,7726 55 30 22 3 F, X IVg 7031-14x6857-5... 30 Plant Colors in Maize 153 TABLE 45. Fi Progeotes of Crosses of Sun Red Ila and Dilute Sun Red IVa with Green Illg and IVg Pedigree nos. Number of Fi plants Group Pi Fi Purple la Dilute purple Ilia Dilute sun red IVa Green Ila X Illg 7097-5 X A 159-25. .. 7357-3 X 7356- 1 . . . 7710 33 31 8151,8152 1 Total, 2 progenies &4 IVa X Illg A9-22X 7097-1 7709 4 IVa X IVg 6860- 8x6869-1.... A9-14X 7060-1.. .. 7713 28 31 7708, A283, A284 Total, 2 progenies 59 IVa X Illg 6860-13x6871-39... 6861- 2 X 6751- 3. . . 7714 11 9 12 18 2 7711 Total, 2 progenies 20 30 IVa X Illg 6861-4 X 6882-5 7039-3x7061-1 7512, 7513, 7716. 7727,7728 25 44 11 43 (Illg, IVg) 34 72 Total, 2 progenies 69 54 106 IVa X Illg 7312-8x7313-2 7313-2 x 7314-1 7314-1 X 7313-1 -6x -2 8183 86 31 126 85 (Illg) 92 8184 19 3 8200, 8201 8185 129 98 Total, 4 progenies 328 338 154 R, A. Emerson TABLE 46. F2 Phogeniks of the Cross Green IVg x Green Vie Pedigree nos. Number of F2 plants Group Fi F2 Dilute sun red IVa Green 5534-4 6791, 6792 7179,7180 7181, 7182 7177, 7178 7175,7176 7163,7164 7169, 7170 7171,7172 35 51 32 64 63 52 60 63 (IVg, Vic) 29 6530-1 43 -2 30 6531-1 42 1 -2 38 7032-1 39 7036-3 36 7037-2 34 420 291 Fi X Vic 7032-7 X 6878-42 7729,7730 7767,7768 24 42 iVIc) 24 7034-5 X -42 . 34 66 58 2 Fi X IVg 7037-4 X 7049-7 7173, 7174 7731,7732 48 48 (IVg) 50 7049-1x7037-4 46 Total 2 progenies 96 96 TABLE 47. F3 Progenies of F2 Dilute Sun I Green Vie Ied Plants of th E Cross Green xVg x Pedigree nos. Number of F3 plants Group F2 F3 Dilute sun red IVa Green IVg, Vic 6791- 3 7148,7149 7154,7155 7159 23 69 16 19 6792- 6 45 1 -11 13 Total, 3 progenies 108 77 6791-22 7150, 7151 7152,7153 7157 7158 7160 65 49 38 23 12 23 -23 18 6792- 7 12 2 -10 10 -13 3 Total, 5 progenies 187 66 6792- 5 -25 7156 7161 48 30 3 Total, 2 progenies 78 Plant Colors in Maize 155 TABLE 48. Fa and F3 Progenies of the Cross Green IVg x Green Via Pedigree nos. Number of Fa and F3 plants Group Fi Fa Sun red Dilute sun red IVa Green 2400- 2 2952- 5 -22 2902,2903 4838-4843 4830-4837 4810-4813 4814-4817 4818-4821 4930, 4931 (Ila, g) 7 36 99 88 111 92 153 3 3 15 32 20 30 58 (IVg,VIa,c) 5 18 57 1 2953- 4 - 7 -21 2957- 2 59 62 45 102 Total, 7 progenies 586 161 348 2 Fa 4930-31 F3 6991,6992 119 2903- 2 4930-22 4783-4786 6993, 6994 (Ila) 99 130 (Via) 32 39 3 Total, 2 progenies 229 71 Fa X IVa 2903-2x2889-38 4787-4790 55 156 R. A. Emerson TABLE 49. F2 Progenies of the Cross Green Illg x Green Vie, and Fi Progenies of Crosses of F2 Greens with Sun Red Ila and Dilute Sun Red IVa Pedigree nos. Number of F2 plants Group Fi F2 Purple la Sun red Ila Dilute purple Ilia Dilute sun red IVa Green Illg, IVg, VIb, c 2907-8 5297,5298.. 7085, 7086, 7722, 7723 28 81 11 26 38 5262-5 1 97 Total 2 Droeenies lotai,^ progenies 109 37 135 Pi Fi Number of Fi plants Illg X Ila 7085-10 XA159-24. 7717 27 2 Illg X IVa 7086-2 X 7102-7. . . . -3 X -8. . . . 7207 7719 . 14 25 Total, 2 progenies . . 39 3 Illg X Ila 708&-6X A159-17.. Illg X IVa 708&-4 X 7102-8. . . . 7718, A298, A299 7720 28 41 11 9 4 IVg X IVa 7086-8 X 7102-8. . . . 7721 22 Mkmoik 39 Plate I ANTHER, GLUME, AND RACIIIS COLOR OF PURPLE 1, Purple, type la, typical, anthers purple: 2, type la with »■'"'', anthers near- black ; 3, type la with i>r, anthers reddish ; 4, type Ij;. with Jiy or r'J, anthers yreen U'rawings by C. \V. Kedwood, somewhat diagrammatic) Mkmoir 39 Plate II CW.f^a.dv-/-ood AXTHER, OLT-ME, AXD RACHIS COLOR OF T^ILFTE PT'RPLE .\XT> OREEX 4. Groon, typos Ilii: a.ul IVf: with R, ^ H K r— ■*■ ^ K ^ fj !^ '" "tli iH o *^ M <*-4 ' tc H °.s a -2 '- L Memoir 39 Plate X COLOR DEVELOPMEXT IX BROKEN LEAVES 1. nihitp sun red. type IVa. about one week after the leaf was creased- 2. dilute purple, type Ilia with japc.nica white stripe.s, about three days after the leaf was creased (Drawings by Carrie M. I'reston) ilKMOIH :i9 Plate XI md ^\i'j y 1 ♦ r% r ABERRANT COLORATIOX OF BROWX TASSEL Toorly flfveloped tassels of brown, tvpo V soin(.fini..v; ,.vi,n,if ,, , • , developed parts ^oiiKtinns (.xinl.it purple m al)norman.v (Drawing by Carrie M. Preston) Syracuse, N Y P*I IAN. 21. 1308 , i - ,*A'