UNIVERSITY OF CALIFORNIA PUBLICATIONS IN AGRICULTURAL SCIENCES Vol. 2, No. 9, pp. 249-296, plates 45-52 December 31, 1924 INHERITANCE IN CREPIS CAPILLAEIS (L.) WALLR. III. NINETEEN MORPHOLOGICAL AND THREE PHYSIOLOGICAL CHARACTERS 1 BY J. L. COLLINS INTRODUCTION For several years variations in Crepis capillaris have been studied genetically. The study was commenced 2 in the hope of being able to determine whether the extensions of the Mendelian theory of heredity which were based on breeding data from Drosophila melano- g aster would hold for higher plants. For this purpose it was necessary to know the mode of inheritance of a number of characters. This paper is concerned with the description and mode of inheritance of a number of variations found in Crepis capillaris (L.) Wallr. It is evident that the material chosen for such a purpose should show variation of a hereditary nature and should also contain a low number of chromosomes. Crepis capillaris seemed to fulfil these requirements, for its chromosome number, 3 pairs, is the lowest reported for the higher plants, and the species is known as a variable one. Linkage has been demonstrated in a number of plants and in some of the higher animals. Unfortunately, the chromosome number in those species in which linkage has been observed is relatively high, and in no case is the number of groups of linked genes equal to the haploid number of the chromosomes. 1 This is a report on a part of a project supported by appropriations from the Adams Fund. 2 Studies commenced by Professor E. B. Babcock in 1915 and carried on by the writer under his direction since 1918; published as nos. 6 and 7 of vol. 2 in the present series. 250 University of California Publications in Agricultural Sciences [Vol. 2 Material and Methods The genus Crepis, containing over 150 species, is a member of the Cichorieae or chicory tribe of the Compositae, the best known related genera being Hieracium, Lactuca, Sonchus, and Taraxacum. Crepis capillaris (L.) Wallr. is an annual, but under certain cir- cumstances may assume the biennial habit. The plant first produces a rosette of radical leaves which have been found to vary in different plants from entire to bipinnately compound. The stem is usually single with paniculate branching above and varies from a few inches to four feet in height, largely depending upon conditions of growth. The cauline leaves are sessile, amplexicaul, clasping, the lower ones more or less lobed or pinnatifid, while the upper ones are slender and entire. The underside of the midribs of the rosette leaves, and to some extent the upper side, and the lower cauline leaves are more or less covered with bristly hairs. In many, but not all, plants the involucre and peduncle are glandular pubescent in addition to the fine gray tomentum which is always present. The brown terete achenes vary in length from 2 to 3 mm., are attenuate at both apex and base, and usually 10-ribbed. The yellow flower heads vary from 17 to 25 mm. in diameter. During the course of the investigations, achenes of C. capUlaris have been obtained from many localities of the temperate and sub- tropical zones of both the old and the new world. The species is apparently a native of Europe, but is now disseminated throughout the world. The methods used in growing experimental cultures of Crepis have been previously published (Collins, 1922). In presenting data from hybrid populations, the degree of corre- spondence of observed with calculated distribution has been deter- mined by use of tables of probable errors of Mendelian ratios prepared by the Department of Plant Breeding of Cornell University. In the case of some dihybrid populations the method suggested by Harris ( C )2 (1912) has been used. This formula is X 2 = 2- , in which o c is the observed frequency of any class, c, the calculated frequency for ( c )2 that class, and 2 indicates that all the values of the type ■ are c added together. From Elderton's 3 tables for calculating the goodness of fit, the probability for the chance occurrence of the deviations in the observed classes has been obtained from the calculated value of X 2 . 3 Given in Pearson, K., Tables for Statisticians and Biometricians, Cambridge Univ. Press, 1914. 1924] Collins: Inheritance in Crepis capillaris (L.) Wallr. 251 VARIATIONS IN CREPIS CAPILLARIS Observations upon cultures grown from the achenes obtained from localities in many different regions have resulted in the discovery of a number of variations. Those which have been studied sufficiently to show their method of inheritance are described below. In assign- ing symbols to serve as genetic representatives of particular char- acters, the system in general use has been followed, namely, the use of the initial letter (or letters) of the name given to the character, small letters indicating a recessive, and capital letters a dominant condition. BALD (b) On August 17, 1918, a single plant (19.18P 23 ) in a culture of 47 plants grown from achenes sent from Copenhagen was found to be devoid of glandular pubescence on the involucre and peduncle. This variation has been named ' bald. ' The second instance of this variation was in the same race but appeared only after two generations of inbreeding. Bald plants later appeared in cultures from other locali- ties as follows: Sweden, England, France, Chile, and the Azores. It was of importance to know whether the same or different genes were responsible for the appearance of 'bald' in cultures from such widely separated sources. This could be determined by crossing the different races. If a single gene were involved, then bald F x plants should result, while if, on the other hand, glandular plants resulted in the F x , this variation appearing in the different stocks would be the similar expression of different genes. As is shown in table 1, the same gene is present in each case. TABLE 1 The Fi Results of Crossing Different Geographical Races of Bald Character of Fi Culture No. Bald Glandular Total F 2 Copenhagen X Sweden (20.130) X Chile (21.23).. Sweden (19.235) X Cambridge (19.66) 9 4 56 1 7 5 9 4 Copenhagen (18.75) X Sweden (19.235) (19.H1, 20.57-8, 21.101) 56 Chile (20.36) X Azores (20.40), (21.25). 1 Sweden (19.H3) X Azores (20.40), (21.117) 12 252 University of California Publications in Agricultural Sciences [Vol. 2 In the last item in table 1 both bald and glandular plants are recorded. This is as it should be, for the 19. H3 plant was an F x glandular plant produced by crossing the Swedish bald race (19.235) with a Eureka glandular race (19.224). If the bald gene in the TABLE 2 F, Eesults from Crosses of BB X bb Character of Fi plants Pedigree No. Glandular (B) Bald (b) 19.H3 20.59 21.21 21.28 2 10 7 7 1 2 Total Expected 1 : 26 29 3 TABLE 3 Back Crosses of the F n Bb to bb Pedigree No. Progeny segregation B b 21.17 21.18 21.19 21.24 21.117 21.126 7 5 6 4 12 4 3 6 7 2 13 2 Total Calculated 1 :1 39 38.5 38 38.5 Deviation 0.5 ± 3.84 cultures from Sweden and from the Azores were identical, we should expect to obtain from such a back cross 50 per cent glandular and 50 per cent bald plants. The 5 to 7 segregation obtained is a close approximation to the expected 1 to 1 ratio. While the Copenhagen race has not been crossed with the Cambridge race, nor the Chilean 1924 Collins: Inheritance in Crepis capillaris (L.) Wallr. 253 race with any except that from the Azores, we have evidence of their identity, since they have each been crossed with the Swedish race, which in turn was proved to be identical with the others. The bald plants from France have not been tested. Bald is inherited as a simple monohybrid recessive, as is shown by the results obtained from crossing with glandular plants. Table 2 presents F x data from crosses of bald X glandular. The one bald plant in culture 20.59 probably resulted from the failure to remove a single pollen grain during emasculation and represents an error in technique. The two bald TABLE 4 F 2 Results prom the Cross BB X bb Progeny segregation Pedigree No. B b 20.59 20.60 20.141 20.142 20.118 10 2 56 16 74 1 3 17 7 23 Total Calculated 3:1 158 156.75 51 52.25 Deviation 1.25 db 4.22 plants in culture 21.21 may be ascribed to this same cause or to errors at time of transplanting, since culture 21.23, containing only bald plants, grew adjacent to 21.21 in the flat before transplanting to the field. Table 3 shows that 39 glandular to 38 bald plants were obtained when the F x (bald X glandular) were backcrossed to the recessive parent strain. The expected 1 to 1 ratio was therefore realized. The results from F 2 cultures confirm the conclusion regarding a single recessive factor conditioning the appearance of bald. While in almost all cases involving bald the glandular hairs are completely absent, in culture 20.141 some plants appeared to be somewhat inter- mediate, inasmuch as they developed a few small scattered gland hairs on the involucre. They were easily distinguishable from glandular plants. In table 4 these intermediates have been classified as bald, 254 University of California Publications in Agricultural Sciences [Vol. 2 but in the original records they were designated as intermediates. If the culture 20.141, containing the intermediate-bald plants, is removed from the table, the remaining cultures give an exact ratio of 3 glandular to 1 bald; when the intermediates are classified as bald, the deviation from a 3 to 1 ratio is less than the probable error. The progeny of two bald and two glandular F 2 plants were grown. Both of the former gave, as expected, only bald offspring, while the two glandular F 2 plants produced both types in F 3 . The nature of the intermediate plants has not been definitely determined. The selfed progeny from one plant (18.dlP 76 ) gave a culture (20.55) of 18 bald, 3 intermediate, and 3 glandular plants. That they were not due to the incomplete dominance of the hybrid produced by crossing bald with glandular is certain, for in the F x cultures (table 2) all plants were fully glandular. Another inter- mediate bald plant (22.153P 18 ) produced 5 glandular and 5 bald plants from selfed seed but none that could be classified as inter- mediate. SMOOTH MIDEIBS (s) The midribs of the rosette leaves usually have a hairy pubescence. From sporadically appearing plants, races have been obtained which do not show these rib hairs; such plants have been designated as * smooth' (s). The F x resulting from a cross between these two types of plants were all rib-haired, and in the F 2 there appeared 556 rib- haired to 40 smooth plants. This is approximately a 15 to 1 ratio and suggests the operation of two independent genes, each producing the same somatic effect. Duplicate genes are by no means unknown, having been reported a number of times in the literature of genetics. If two independent genes were operating in the cultures 21.140 and 21.141, the F x of this same cross when backcrossed to smooth should give a 3S to Is ratio and some F 3 populations should give a 3 to 1 segregation. Evidence from cultures of these two types has been obtained; the data from them together with data from other crosses involving this character are given in table 5. The F 3 culture 21.189 was grown from one plant of an F 2 culture containing 58 rib-haired and no smooth plants. Such a deviation is, however, only three times the probable error and may well be due to errors of random sampling. The culture Fi 19. HI was originally made to determine the relation of the gene for bald of the English race of Crepis to that in the Danish race and 1924 Collins: Inheritance in Crcpis capillaris (L.) Wallr. 255 was the hybrid between these two races. The parent plant from the English race was smooth, while the parent from the Danish race had rib hairs. TABLE 5 Showing F, and F 3 Results from the Cross SSS'S', SSs's', and SsS's' WITH SSS'S' Progeny segregal ion Pedigree No. S s F 2 21.140 237 17 F 2 21.141 319 23 F 3 22.189 Total 189 9 743 49 Cal( mlated 15:1 Deviation 742.5 49.5 0.5 ± 4.59 F 2 22.55 25 12 F 2 22.56 5 2 F 2 22.60 22 6 F 2 22.61 4 2 F 2 22.62 9 1 F 2 22.63 34 8 F 2 22.41 Total 66 24 165 55 Ca Iculated 3:1 Deviation 165 55 0.0 ± 4.52 Back cross 19. HI 55 17 Ca Iculated 3:1 Deviation 54 18 1.0 db 3.83 The 3 to 1 ratio obtained in 19. HI indicates that the rib-haired 2 used was heterozygous for the duplicate genes for rib hairs. This cross, as regards these characters, was a back cross of a heterozygote to the recessive parent, and constitutes additional evidence to sub- stantiate the duplicate gene interpretation given above for the inherit- ance of rib hairs in these cultures. 256 University of California Publications in Agricultural Sciences [Vol. 2 LEAF VARIATIONS From the very first acquaintance with C. capillaris, the different forms in the rosette leaves constituted the most striking and out- standing variations. They have proved equally as difficult to study genetically, due, first, to the difficulty in evaluating non-genetic variability resulting from age of plant and from environmental causes, and, second, to the complex heterozygotic nature of the material in the wild condition. Sears (1921) found in T araxacum that the degree of leaf dissection is correlated with the age of a given rosette. The leaves of a very young rosette are almost entire, becoming progres- sively more dissected as the rosette becomes older. Stork (1920), also working with Taraxacum, found that in very young plants the rosette leaves ranged in form from entire to deeply pinnatifid- runcinate, but became more, instead of less uniform, as they grew older. Neither condition can therefore be taken as typical for that species. In Crepis, a closely related genus, there is a more regular sequence of development of leaf shape for a particular rosette. The juvenile leaves are usually entire or nearly so, and assume their typical forms gradually as the plant reaches the mature rosette stage just preceding the appearance of the flowering stalk. At this time there exist individual differences which range in form from entire to deeply pinnatifid or compound pinnatifid. That these differences are genetic is shown, first, by the fact that inbreeding has resulted in the isolation of races of the different types which breed true when grown side by side under similar conditions, thus to a large degree eliminating the effect of the non-genetic factors, and, second, that the forms when crossed give a fairly uniform F x and segregate into the parental and F x forms in the second generation. By means of inbreeding and selection, a number of distinctive, uniform races have been obtained in almost homozygous condition. A brief description of each is given below. VIEIDIS Plate 45, figure 1 This form was isolated in 1919 from the Eureka (California) stock. The rosettes are small, 4 to 10 inches in diameter. The leaves are deeply lobed or pinnately parted, and are lacking in anthocyanin. 1924] Collins: Inheritance in Crepis capillaris (L.) Wallr. 257 The blade of the leaf is of a darker color than the midrib. The color of the blade is Ridgway's varleys green, 31' m. The midrib is covered on both upper and lower surfaces with hairy pubescence. The lobes are usually widest at the base, often having a minor lobe attached to the proximal edge of the base of the major lobe. Attached to the midrib between the lobes is a narrow wing. The lobes are usually close together, with the terminal lobe slender and pointed. H6 RACE Plate 45, figure 2 The H6 race was isolated in 1919 from a Berkeley Crepis stock. The size of the rosettes is more variable than in viridis, the rosettes ranging from 8 to 12 inches in diameter. The leaves are pinnately and bipinnately lobed ; the lobes are constricted at the base and rounded at the tip, and inclined to twist, so that the plane of the lobe is not in the same plane with the midrib. Anthocyanin is conspicuously present. There are no hairs on the midrib. The lobes, usually six in number, are widely spaced. The terminal lobe is large and blunt- tipped. The narrow wing on the midrib is crimped, presenting a ruffled effect. The wing and edges of the lobes contain a blackish purple coloring which appears very early in the development of the plant. The leaf color, according to Ridgway's Standard, is cedar green, 31m. The characters which make up this type are dominant, excepting smooth ribs, when crossed with viridis. PALLID Plate 45, figure 1 This race was obtained in 1919 by inbreeding in the same Eureka stock that produced the viridis race. The rosettes are from 6 to 10 inches in diameter. This race produces more leaves in the rosette than do the preceding races, giving the rosette a thick mat-like appear- ance. Pallid lacks anthocyanin and is a much paler green (Ridgway's forest green, 29'm.) than the two races described above. The lobes are broadest at the base, are set closely together, and have pronounced, pointed teeth. This race does not grow so rapidly as the darker green races. Rib hairs are present on the midrib. 258 University of California Publications in Agricultural Sciences [Vol. 2 SIMPLEX Z9 Plate 46, figure 1 Simplex Z9 was isolated in 1920 from a stock originating from seed collected at Quy Fen, England. The original culture consisted of plants ranging from entire to pinnatifid. The simplex Z9 race was obtained by inbreeding plants with entire leaves. Although inbreed- ing has reduced the amount of variation, there still appears in this supposedly homozygous race a small percentage of semi-pinnatifid- leaved plants (pi. 46, fig. 1). Anthocyanin and rib hairs are present. SCALAEIS e29 Plate 46, figure 2 This race was isolated in 1919 from the Eureka stock of Crepis which produced the viridis and the pallid races. It is characterized chiefly by long, simple, pinnately-divided leaves with pointed lobes. The terminal lobe is slender and elongated, often curved to one side near the tip. Both anthocyanin and rib hairs are present. The average number of lobes per leaf is 10. It is clominent when crossed with simplex Z9 or with viridis. Typical leaves of the scalaris e29 and the simplex Z9 races are shown in plate 52, together with the F x and F 2 types obtained when these two races are crossed. In the F 1 a few extreme variants occur which approach the simplex form, but the majority are more nearly like the scalaris and constitute a fairly uniform intermediate type. In the F 2 , three types are dis- tinguishable (see pi. 51, fig. 2), the two grandparental forms and an intermediate scalaris form similar to the F 1 . When the intermediate- scalaris and the scalaris are grouped together a 3 to 1 ratio is obtained (see table 6).. The intermediate forms differ from the scalaris in having the lobes less deeply incised, some more so than others, but still classifiable as intermediate. (See third and fourth leaves in P 2 , pi. 52.) From the results of breeding it appears that there is present one main gene for lobing and that dominant modifying genes are involved which act cumulatively, thus producing intermediates of different grades of pinnate lobing. As a corollary to this hypothesis races breeding true for different grades of intermediate pinnatifid lobing should be possible. There is evidence that such races occur. Several intermediate forms have been tested and found to be fairly constant. 1924 j Collins: Inheritance in Crcpis capillaris (L.) Wallr. 259 A race obtained from Seattle, Washington (named "Seattle") appears to be such a homozygous intermediate form. Races of Crcpis capillaris also differ in number of lobes per leaf and in length of leaf (Rau, 1923). The scalaris race shown in plate 52 has a large number of lobes. The two races differ, however, in length of leaf. The leaves of the scalaris parent shown in plate 52 are shorter, and of the simplex parent larger, than the mean size typical for each race. The F x is usually larger than either parent. The F 2 in the same figure shows the segregation for size which appears to be due to multiple genes. The inheritance of pinnatifid and entire leaf forms in capillaris conforms in general to the type of inheritance of corresponding forms in a number of other plants. Rasmusen (1916) found in species crosses in grapes that differences in leaf form behaved in a very similar way. The F 1 appeared to be intermediate between the shapes of the parent leaves. In the F 2 , a series was produced which included the grandparental forms, the F x type and different grades of inter- mediates. If the deeply toothed and intermediate toothed classes were grouped together, a ratio of 3 toothed to 1 non-toothed resulted. Shull (1918) found four different leaf forms of the shepherd's purse to be caused by two pairs of factors. As in Crcpis, the deeply pinnatifid forms were dominant. The plants were also subject to considerable fluctuating variation. Two races of Urtica, one having deeply serrated leaves, the other, leaves with entire edges, gave serrated leaves in F x and a ratio of 3 serrated to 1 entire leaf in the F 2 generation (Correns, 1912). In cotton, however, the deeply palmately parted leaf form is not dominant when crossed with the five-pointed upland type, but produces an intermediate type in F t with a ratio of 1:2:1 in the F 2 generation (Shoemaker, 1909). Kristofferson (1923) found that the difference in lobing of the leaves of two species of Malva was brought about through a single genetic factor, and resulted in a somewhat intermediate condition in F x and a 3 lobed to 1 non-lobed condition in the F 2 , although considerable variation in the degree of lobing in the pinnatifid class was recognized. Tedin (1923), on the other hand, found that pinnatifid and entire leaved plants differed genetically by two factors. 260 University of California Publications in Agricultural Sciences [Vol. 2 TABLE 6 The Eesults from the Cross of Leaf Forms. Sc X sc Progeny segregation Pedigree No. Sc sc 21.140 22.7 22.10 22.14 22.17 22.19 22.22 22.24 22.25 . 22.26 177 99 50 14 48 92 167 51 37 29 75 24 17 6 15 19 52 14 13 2 Total Calculated 3:1 764 750.75 237 250.25 Deviation 13.25 ± 9.24 SCALARIS e28 (Sc) Plate 47, figure 1 This pinnatifid leaf form was isolated in 1919 ; it originated from a single plant which was a sib to the one producing the scalar is e29 race. These two forms have much in common, but are different in size, e28 being smaller and not so vigorous as e29, and having shorter and blunter lobes. Two races of the pinnatifid leaf forms isolated from the Berkeley (H6) race of plants and from the Eureka population (e28), respec- tively, differ in a number of minor characters, as shown in the follow- ing comparative list : H6 (Berkeley) dark green dark green to blackish pronounced pronounced none pronounced blunt and rounded rounded wide (very) pronounced large Characters color of leaf color of midrib anthocyanin crimping of rib-wing rib hairs black edge on leaf terminal lobe lateral lobe lobe spacing Constricted base of lobes secondary lobes e28 (Eureka) dark green light green none or trace none present trace only narrow — pointed slender — more pointed wide (medium) none or trace none 1924] Collins: Inheritance in Crepis capillaris (L.) Wallr. 261 Plants of these two races when crossed showed almost the entire group of H6 characters (rib hairs excepted) in the ¥ 1 , while in P 2 (21.141) there appeared the parental types and in addition some composite types that showed some characters from each parent. When each character pair was considered separately, however, a peculiar sit- uation was presented. Six of the character pairs gave 9 to 7 ratios, and a seventh pair, rib hairs vs. smooth ribs, gave a 15 to 1 ratio. The data for these characters are included in table 7. It is quite probable that these six character pairs as given are the result of not more than three sets of genes, since the two characters, black edging of the leaves and anthocyanin of the midribs, are both concerned with the distribution of anthocyanin pigment in the plant. The shape of the terminal and of the lateral lobes is probably conditioned by the same pairs of genes, while the crimping of the wing of the midrib and the constriction of the base of the lobes also probably result from the action of the same gene. The Berkeley plants were evidently homozygous for the dominant complementary genes of all three character couples. This genotype may be expressed as AA'BB'CC, the simultaneous presence of both the primed and unprimed dominant genes being necessary to cause the development of the respective characters. The Eureka race would then have the genotype aa'bb'cc' with respect to these characters. TABLE 7 Segregation of Six Pairs of Characters in the F 2 from the Cross H6 X Scalaris e28. (Culture 21.141) Segregation Calculated 9 :7 Deviation 162 black edge : 103 green edge 149.06 : 115.93 154.71 : 120.33 143.1 : 111.3 143.1 : 111.3 143.1 : 111.3 149.58 : 116.34 12.94 ± 5.45 166 anthocyanin : 109 none 11.29 ± 5.55 142 angular lobes : 112 round 1.1 ± 5.33 150 narrow lobes : 104 broad lobes 6.9 ± 5.33 135 constricted lobes : 118 non-constricted 165 crimped wing : 101 flat wing 8.1 ± 5.32 15.42 ± 5.49 BEVOLUTE (r) Plate 47, figure 2 This race appeared in 1919 among offspring of a plant of the Eureka stock, which had been self -pollinated. The plants are char- acterized by a definite downward curling of the edge of the leaf 262 University of California Publications in Agricultural Sciences [Vol. 2 toward the midrib. It occurs in both entire and pinnatifid types, though it is more conspicuous in the former. In appearance much like the fwmfolia mutant of Oenothera Lamarckiana described by Shull (1921), in which both rosette and cauline leaves have edges curled under. The knowledge of the genetic basis for this character has been obtained incidentally in experiments designed to show in- heritance of other characters. The data thus obtained indicate that revoluteness is conditioned by complementary recessive genes. TABLE 8 Showing the Segregation of Eevolute Leaves in Two Cultures Progeny segregation Pedigree No. R r 19.e5 Calculated 3:1 62 59.25 17 19.75 Deviation 2.75 ± 2.60 21.140 Calculated 15:1 233 237.19 20 15.81 Deviation 4.19 ±2.60 It is significant that revolute appeared only in these two cultures, which were derived from a common source, because it indicates that the genes were present in the wild plants from which the starting point of these cultures was obtained. The 15 to 1 ratio made its appearance in the sixth generation from the wild plants (some out-crossing occurs in this pedigree), while the 3 to 1 ratio appeared in the second gen- eration. BICEPHALIC (bi) Plate 48, figure 1 This character designates a type of fasciation in which the buds are more or less joined together in twos. The peduncle is also fre- quently flattened. This variation was first found in 1920 on a single plant (20.30) which was grown from achenes obtained from Chile. This original plant was crossed with 20.130P 19 , which produced an F x culture of 9 normal plants. The F 2 , consisting of 81 plants, segre- gated into 60 normal to 21 bicephalic, clearly a monofactorial ratio. 1924J Collins: Inheritance in Crepis capillaris (L.) Wallr. 263 In no case were all the buds of a plant of the bicephalic kind. Some plants indeed produced only a few double buds. F 2 bicephalic plants of both types were selfed and F 3 cultures produced. The data from F 3 cultures are shown in table 9. TABLE 9 Type of Plants Produced by Selfing F 2 Bicephalic Plants F 2 Plant No. Progeny F 3 23.283 Bicephalic Normal *P 6 8 + 6 1 P70 + 2 6 P 9 6 + 8 P« + + 6 P 10 + + 6 P 2 3 + + 5 1 P24+ + 2 P30+ + 1 P44+ + 20 P46+ + 8 P48+ + 5 2 P57+ + 2 P 8l + + 5 (2?) * The single + indicates an F 2 plant on which but few bicephalic buds appeared. The ++ indicates plants having many such buds. It appears that F 2 bicephalic plants breed true in F 3 . Plant 70 which had only a few double buds, was apparently a heterozygote, for it gave a 3 to 1 ratio in F 3 . The other F 3 plants listed as normal may have been genetically bicephalic, since they showed some evi- dences of f asciation in the stems and malformation of buds ; but no doubling or cohesion of the buds was found. ANTHOCYANIN This pigment is distributed to many parts of the plant, but is most noticeable in the midribs of the leaves and on the lower portions of the stems. Culture 19.e8 segregated into 94 plants with antho- cyanin to 39 with none or developed only to a slight degree. The ratio in this case is 2.82 to 1.17, in which the deviation is less than twice the probable error. This segregation can be considered only as sug- gestive because of the difficulty of accurately classifying this character 264 University of California Publications in Agricultural Sciences [Vol. 2 in Crepis. The appearance of purple anthocyanin color depends upon a certain amount of sunshine and exposure to light. Plants known to be capable of producing the color will show it to only a small degree if conditions for anthocyanin development are adverse, while, on the other hand, races in which it does not normally appear conspicuously will produce it under conditions of sudden exposure to direct sun- shine or sometimes as a result of mutilation caused by animals or insects. The development of anthocyanin is a matter of degree, for the potentiality for its development is not entirely absent from any race so far obtained. In the viridis race we have it in its lowest and in the H6 race in its highest development. Crosses between high and low anthocyanin races (other than 19. e8 mentioned above) in general produced F x plants showing the darker anthocyanin of the H6 race, but in F 2 produced a series of forms showing a gradation in pigment from one parent to the other. In most cases the parental types were also duplicated. One such cross, H6 X viridis e33, gave an F 1 more nearly like the H6, but in F 2 the types were distributed as follows: 9 of H6, 3 of viridis, and 3 distinctly between these two parental types. The segregation of anthO'Cyanin has been observed in other cultures (e26= 3 to 1), but has not, in general, given sufficiently regular results to warrant the drawing of conclusions regarding its genetic basis. The analysis can only proceed when facilities are available to control more accurately the environmental factors which alter its development. DWAEF II (dll) Plate 48, figure 2 This variation first appeared in culture 21.99, which was the second selfed generation from achenes obtained from Lyons, France. It is characterized by a very small rosette of slender semi-scalaris leaves which are blotched with yellow and yellowish red coloration, giving them the appearance of being about half -dead. Due to their peculiar appearance the first plants were thought to be suffering from poor environment, although adjacent plants were healthy. The plants when mature are very small (3-6 inches in height), the stems very fine and spreading. In the first culture the dwarf effect appeared to be recessive (5 dwarfs in 16 plants) and bred true in the next gene- ration. Culture 22.159 from 21.99P 7S , a normal plant, contained 51 plants, 3 of which were dwarf II and 3 somewhat dwarfish but not typical for dwarf II. This is approximately a 15 to 1 ratio, and 1924] Collins: Inheritance in Crepis capillaris (L.) Wallr. 265 indicates that there may be duplicate genes for dwarf II ; sufficient data are not at hand to establish the hypothesis. Culture 22.160 (from 21.99P ir „ a normal plant) gave 84 normal plants. The yellow appearance of the leaves in dwarf II seems to be a dominant character from its appearance in 22.407, F x of the cross 22.169P 22 X 22.261P 4 , the male parent being a dwarf II plant from a pure culture. Inasmuch as the F x plants are not dwarfish, it appears that the yellowing and dwarfing may be due to separate but probably linked genes. All the dwarf II plants which have appeared were yellowish, and we may therefore assume that, instead of linkage, the appearance of dwarf II is dependent on the presence in the zygote of the dominant gene causing yellowing. DWARF III (dill) Plate 49, figure 1 This variation first appeared in 1919 culture e5. It reappeared in 1921 in a culture (21.76) which came from the same source as e5. The ratio of normal to dwarf III in 21.76 was 15 to 1, and in the progeny of 21.76?! (culture 22.117) 3 to 1. (See table 10 for data.) Dwarf III was at first called 'semi-lethal, ' because of the high mortality in this class of plants. These plants remain very much smaller than their normal sibs during the rosette stage and reach maturity much later. A large percentage die after they have formed a rosette and before they reach the flowering stage. This variation appeared in several members of the same stock which produced revolute, viridis, and pallid. SPREADING (sp) Plate 49, figure 2 A lax, open-branching habit which appeared in 20.37, the French stock of Crepis. The stems and branches are long and slender, appear- ing to be so weak they cannot support themselves in upright position. Dwarf II appeared in this race and all have this spreading habit. Data from crosses (21.26 and 22.173, table 10) show that it is a reces- sive character. When the same plant (20.37P 3 ) was crossed to another erect plant (19.in.Pu), it behaved as a dominant (21.28, 22.41, and 22.43, table 10). Of the F 2 cultures, only 22.173 was grown under desirable conditions; the others were overcrowded in greenhouse and lath house, which interfered with proper development of this char- acter. 266 University of California Publications in Agricultural Sciences [Vol. 2 PROCUMBENT (p) This variation is similar in appearance to spreading. It first appeared in culture 20.40, which came from achenes sent from the Azores Islands. Unlike spreading, it seems to be dominant, the F x plants, 21.28 (from 20.40P X 20.111PJ, being of the procumbent type. The F 2 cultures were grown under crowded and unfavorable TABLE 10 Segregation of Plant Characters Segregation Culture No. Normal Variant 21.76 Calculated 15:1 57 57.19 4 dwarf III 3.81 Deviation 0.19 db 1.28 22.159 Calculated 15:1 48 47.81 3 dwarf II 3.19 Deviation 0.19 ± 1.17 22.117 Calculated 3:1 12 12 4 dwarf III 4 Deviation 0.0 ± 1.17 22.99 Calculated 3:1 11 12 5 dwarf II 4 Deviation 1.0 ± 1.17 22.173 Calculated 3:1 70 erect 72 erect 26 spreading 24 spreading Deviation 2.0 ± 2.86 22.41 22.43 Total Calculated 1 :3 18 5 23 19.2 39 spreading 15 spreading 54 57.8 Deviation 3.8 ± 2.56 1924] Collins: Inheritance in Crepis capillaris (L.) Wallr. 267 conditions which made accurate classification difficult and uncertain. One F 2 gave a 1 to 1 ratio and another the ratio 2 procumbent to 1 normal. ERECT (e) Plate 50, figure 1 A strain characterized by erect habit of growth, large stiff lateral branches, and a thick rigid central axis. The branches make an acute angle with the axis, the whole plant having the form of an inverted cone. This form was selected from the F 2 of a cross between the Danish and Swedish stocks. PALE A (p) Plate 51, figure 1 The nature of this character has previously been discussed (Collins, 1921). It originally appeared in an F x hybrid and was considered a reversion to a possible, pre-composite, ancestral condition. It has appeared in every case in hybrids, never in inbred races, and was probably introduced with the Danish stock, since the same plant (17.198P 2 ) of that stock is in the pedigree of all the hybrids which have produced palea. Races homozygous for palea have been obtained. Preliminary data show palea to be conditioned by a single recessive gene. Linkage In a species having only three pairs of chromosomes, it would seem fairly easy to establish groups of linked genes, especially when the species was known to be more or less polymorphic. However, it has not yet been possible to realize this end, due to the unexpected relations of some of the genes in this species. For instance, there are four cases of complementary recessive genes, and three characters dependent upon duplicate dominant genes. The determination of linkage groups under such conditions is complicated because it re- quires a longer time to obtain races with a known and tested genotype. The gene for bald involucre appears from data in tables 12 and 13 not to be linked with the gene for smooth ribs nor with the gene for procumbent, since the ratios show independent segregation. It is of course obvious that linkage must occur between one pair of complementary genes for smooth ribs and one pair of complemen- tary genes for revolute leaves, since there are four pairs of genes and 268 University of California Publications in Agricultural Sciences [Vol. 2 only three pairs of chromosomes. A cross involving these two char- acters gave the following results (+ indicates the presence and — the absence of the character named) : TABLE 11 Dihybrid Segregation of Smooth X Revolute in a Culture which Gave a 15: 1 Ratio for Each Character Separately Culture 21.140 Smooth ribs Revolute leaves : + + + + Total Observed Calculated 57 : 3 : 3 : 1 : 202 224 16 11.79 32 11.79 2 3.93 252 252 (o-c) 2 c 1.98 1.50 34.64 0.12 X 2 = 38.24 P =.0000 The calculated numbers agree fairly well with those obtained except in the third class where the observed numbers are more than twice as large as the calculated number. This class may have been increased at the expense of the first class by placing in it some plants which genetically belonged in the latter. The observed number in the first class is considerably less than the calculated number for that class. These figures indicate that the genes are arranged in the three pairs of chromosomes as follows: R, s, — (R'S') (rV) — r, S, where primed genes are the complements of the unprimed genes. Were the linkages as follows (R's) and (r'S), the F 2 population should consist of three classes in the proportion of 14 :1 :1, assuming that little or no crossing over occurs. A high percentage of crossing over in the latter type of linkage would give approximately the results obtained. It appears, therefore, that either the dominants are linked, as stated above, or that there is a high percentage of crossing over between the linked genes. This inference can be tested experimentally, for races have been obtained which gave 3 to 1 ratios for both of the characters. Effects of Inbreeding The flowers of Crepis are perfect and, although self-fertilization can take place, the arrangement of the stigmas in respect to the stamens is such as to permit cross-pollination before self-pollination can be naturally effected. The stamens are united into a tube sur- rounding the style, and the pollen is shed on the inside of this tube. 1924] Collins: Inheritance in Crcpis capillaris (L.) Wallr. 269 TABLE 12 F 2 Eesults from the Dihybrid Cross, Glandular and Hairy Midrib X Bald and Smooth Kibs, Showing Independent Segregation Culture 22.41* Observed segregation Calculated segregation 9:3:3:1 (o-c) 2 c Glandular and 36 41.01 0.61 Rib Hairs Glandular and 11 13.68 0.52 Smooth Bald and 20 13.68 2.84 Rib Hairs Bald and 6 4.56 0.42 Smooth 73 72.96 X 2 = 4.39 P =0.2264 *Rib hairs vs. smooth in this culture show a 3 : 1 ratio. TABLE 13 Showing Independent Segregation in F 2 of Dihybrid Cross, Glandular-Erect X Bald-Procumbent Culture No. 22.41 Observed segregation Calculated segregation 9:3:3:1 (o-c) 2 c Glandular — procumbent 17 20.25 0.37 Glandular — erect 10 6.75 1.56 Bald- procumbent 7 6.75 0.01 Bald- erect 2 2.25 0.03 36 36.00 X 2 = 1.97 P = .5773 270 University of California Publications in Agricultural Sciences [Vol. 2 The style is bifid with the stigmatic surface on the adjacent faces of the lobes. With the beginning of anthesis the style elongates, pushing the upper end out from the stamen tube and sweeping the pollen out with it on its outer surface. The stigmatic lobes then separate and assume a position at right angles to the style. The pollen at this stage is below the receptive surface of the stigma, which is, however, exposed to insects, the means by which cross-pollination is effected. Later the stigma lobes curl into a short spiral which brings the receptive surface of the stigma in contact with its own pollen or that of an adjacent floret of the same head. Under natural conditions Crepis is often cross-pollinated by insects, and this preserves a heterozygosity of the germinal material. A similarity of the effects of continued inbreeding in Crepis to the effects of inbreeding in maize has been noted (Collins, 1920). It was shown that inbreeding caused a reduction in the size of the plants and increased the length of the vegetative period. Other data are now available which show in another way the general heterozygosity of Crepis capillaris as it occurs in a wild state. Thus the seed collected from a few wild plants near Eureka, California, has been the source of the following races: viridis, scalaris-e28, pallid, and revolute (leaf form variations) ; of three types of partial albinos (chlorophyll development) ; and of the variations, dwarf III and fasciation (the plant as a whole). From the Berkeley wild plants we have obtained plants with smooth ribs and the leaf form H6 ; from England, the leaf form simplex-Z9 ; from France, dwarf II, spreading, chlorina, and tubular flowers. Palea probably came from the Danish material. As mentioned in another section, bald has appeared independently in the cultures from six different geographical regions. The Eureka stock has produced the greater number of new races. This is not taken to mean that it is necessarily more heterozygous but that many more plants from this source have been under observation. We have presented here an instance of a remarkable germinal diversity in locally developed strains of a single species. Although many of the characters appeared only after hybridization between local races or stocks, the evidence does not, except in a few cases, show these characters to be due to complementary factors. The appearance of bald from such widely separated localities as Chile and Sweden and from other less widely separated localities is of particular interest, for it shows that either a certain locus of the germinal material mutates more readily than others or that all these local races have originated from a single stock 1924] Collins: Inheritance in Crepis capillaris (L.) Wallr. 271 in which this gene was present ; the former is, however, more probable, for it has been shown in Drosophila (Sturtevant, 1921) that certain loci are more mutable than others. Additional evidence that this is the case is found in the fact that a similar variation, bald, has been found to occur in at least four other species, C. bursifolia, C. biennis, C. aspera, and C. dioscoridis. A similar germinal diversity among local races of Drosophila m-elanog aster from equally widely separated localities has not been found, and Sturtevant suggests that this may be due to a frequent transportation of individuals from one locality to another. The chances are probably as great for transportation of Crepis seeds along with agricultural seeds as for the transportation of Drosophila among fruits. It is possible that some of these variations might have arisen from mutations occurring in the cultures under observation. A study of the wild plants in the fields about Eureka, however, disclosed the fact that some of the forms obtained in the greenhouse by inbreeding were also appearing there among wild plants. In this material it is impossible to say whether any new recessive variation appeared as the result of a recent gene mutation or the segregation of a recessive from a heterozygous parent stock. Variations in Chlorophyll A number of different variations involving a loss of chlorophyll have appeared. These variations are evident in the seedling stage, but, unlike the usual albinic condition in seedling plants, most of these albino types develop sufficient chlorophyll as the plant grows to enable the plant to live. One type of pure white seedling always dies in the seedling stage. The other types are either pure yellow or yellowish green. The percentage of seedling mortality in these classes is higher than in pure green seedlings. A complete analysis of the genetic relations of these different types has not yet been possible, but a sufficient study has been made to warrant a preliminary report in this general account of variations in Crepis capillaris. CHLOEINA (C) Chlorina signifies a chlorophyll deficiency in seedling and mature plants. The middle portion of the leaves of chlorina plants is yellow- ish, but both tip and base contain more or less chlorophyll and thus it is possible for the plant to function. This character first appeared 272 University of California Publications in Agricultural Sciences [Vol. 2 in culture 21.99. In 1922 a culture of six chlorina plants was obtained. When these chlorina plants were crossed with normal green plants, the two classes of plants — normal and chlorina — appeared in the progeny in equal numbers, thus indicating that the chlorina plants were heterozygous for green. Self-fertilization of the green resulted in only green progeny. The seedling progeny from self-fertilized chlorina plants consisted of three classes : pure yellow, pale green, and normal green, in the ratio 1 to 2 to 1. The yellow seedlings died, the pale green ones developed into chlorina plants, and the green seed- lings produced only green plants. The gene for chlorina is therefore dominant and has a lethal action when homozygous. TABLE 14 Segregation of Seedling Progeny of Self-fertilized Chlorina Plants Culture No. Green Pale green Yellow 24.171 24.173 24 . 174 46 13 66 60 17 ? 26 6 33 Total 125 77 65 Observed Calculated 3:1 202 200.25 65 66.75 Deviation 1.25 ±4. 77 In table 14 the seedlings in culture 24.174 intergraded in such a way that it was impossible to make an accurate segregation of pale green from green; consequently the two classes are combined in the table. Separation of the two green types in other cultures was less difficult, although it is apparent that some pale green plants have been included in the green class. GOLDEN YELLOW (y) The type known as golden yellow behaves as a monohybrid reces- sive as shown by data in table 15. These golden yellow seedlings gradually develop chlorophyll and finally reach maturity, although growing much more slowly than their green sibs. These plants can, however, be distinguished in the mature stage, due both to size and to the peculiar distribution of the chlorophyll. They produce mature rosettes that show a mottling 1924 Collins: Inheritance in Crepis capillaris (L.) Wallr. 273 of yellow and green through the leaves, which looks much like the plant disease known as ' mosaic, ' or rosettes on which the central and thus younger leaves of the plant are a clear yellow. These yellow leaves later develop chlorophyll and become normally green. It would appear from table 15 that the golden yellows would be homozygous recessives ; but this is not the case, for the seedlings from selfed 'yellow center' and from 'mottled' plants show some of them to be heterozygotes. Only one plant has yet been found which was homozygous for yellow. TABLE 15 monohybrid segregation of golden yellow in the progeny of Green Plants Culture No. Progeny segregation of seedlings 1921 Green Yellow 177P 13 10 3 177P 16 12 3 177P 17 278 84 177P 38 13 5 177P 40 15 3 177P 78 36 10 177P 124 23 6 Total 387 114 Calculated 3:1 375.75 125.25 Deviation 11.25 ± 6.54 That there are other genes which also produce yellow seedlings is evident from table 16. The three plants P 39 , 66 , and 76 were green as seedlings and normal green in the mature stage. They apparently were heterozygous for two recessive genes which produced the same or a very similar type of yellow. The progeny of P 25 indicate still another type of yellow indistinguishable phenotypically from those already mentioned. Here the production of chlorophyll in the seed- ling stage is dependent on the simultaneous presence of two dominant genes, and the absence of either one results in a yellow type of seedling. Trow (1916) reports a similar case of complementary recessive genes in the production of albino seedlings in Senecio, another genus of the Compositae. 274 University of California Publications in Agricultural Sciences [Vol. 2 TABLE 16 Showing Seedling Segregation in the Progeny of Green Plants Indicating Complementary Eecessive Genes for Golden Yellow and Duplicate Genes for Chlorophyll Progeny segregation of seedlings Culture No. Green Yellow 21.177P 39 21.177P 66 22.177P 76 44 13 45 3 1 3 Total 102 7 Calculated 15:1 102.19 6.81 Deviation 0.19 ± 1.70 21.177P 25 Calculated 9:7 22 19.687 13 15.312 Deviation 2.312 ± 1.98 VIEESCENT YELLOW (v) A third type of seedling called virescent yellow has a small amount of green color in addition to the yellow. These seedlings, like the yellow ones, may produce two types of mature plants, namely, pure green plants and green plants with pale green younger leaves at the center of the rosette. The data at present indicate that virescent plants are produced when a gene dominant to yellow but recessive to green is present with the gene for yellow, which changes yellow seed- lings to virescent and yellow-center rosettes to pale green centers. When virescent plants are self ed, then green, virescent, and yellow are obtained, but no virescent plants have appeared in the progeny of yellow plants. It is hoped that in another place it will be possible to publish more extensive data and a complete discussion of the inheritance of chlorophyll deficient characters in Crepis which cannot be given at this time. 1924] Collins: Inheritance in Crepis capillaris (L.) Wallr. 275 GENERAL DISCUSSION In order to establish and preserve true breeding strains of the different types observed in the eultures, type plants were self- pollinated in successive generations. This most intense type of inbreeding affected these cultures in very much the same way as inbreeding has affected maize. Reduction in size and a slower rate of growth were the most noticeable results of inbreeding together with a slight increase in sterility. Most of the experiments to show the effect of inbreeding in plants have been with domesticated forms in which it is possible to have a genotypic constitution that might not exist in a wild state, because characteristics which would unfit the individual for survival in natural conditions are often preserved under the artificial conditions of cultivation. The inference is that wild species would differ in fewer genes than their cultivated relatives. However, the inbreeding experiments on Drosophila (Castle, 1906) produced no bad effects. Collins (1919) states that self-fertilization in teosinte, a wild relative of maize, causes no loss of vigor such as is known to occur in maize. On the other hand, Darwin (1876) concluded that wild species which are naturally cross-pollinated are, on the whole, adversely affected by inbreeding. It appears then that the results of inbreeding any race, cultivated or wild, would be an index to its genotypic heterozygosity or homozygosity. With this as a criterion, there is indicated a condition of germinal heterozygosity in Crepis capillaris. There appears to be a certain similarity between wild heterozygous species of Crepis and the cultivated races of maize in the type of recessive genes which persist in the genotype. In maize, a number of genes are present which produce characters that are so abnormal (sterility, extreme dwarfs, albinos) that they are propa- gated only with difficulty and would seldom be found under natural conditions. Examples of similar forms have appeared in inbred strains of Crepis. It may therefore be considered that natural selec- tion has not eliminated these genes from the germinal material of the wild species. The genes in Crepis which affect vigor also produce results comparable to similarly acting genes in maize. Evidence of the genotypic heterozygosity of capillaris has also been gained from another source. Seeds have been obtained from widely separated localities and grown side hy side in the greenhouse 276 University of California Publications in Agricultural Sciences [Vol. 2 and garden. The number of different forms resulting either in the first or later generations and as a result of controlled cross-pollinations show that the germinal material was indeed far from homozygous. It is of importance, because of some current theories regarding the influence of the habitat upon the genotype of a local species (Tures- son, 1922), to observe the behavior of these various forms when grown in as nearly identical conditions as can ordinarily be furnished in a greenhouse or garden. Plants belonging to many different genera were collected by Turesson from contrasted habitat localities in Sweden and grown together in a common garden. He found that in general each particular type of a species found in each of several different habitats maintained its characteristics in the absence of the habitat to which it seemed especially modified. He sees in such phenomena a refutation of the theory, now generally held, that the form predominating in a given locality occurred as a chance mutation or recombination and was preserved through natural selection. The theory substituted for this is Lamarckianism expressed in modern terminology, namely, habitat causes a change in the fundamental genotype of the species such that a phenotype is developed which permits the plant to nourish in a specialized habitat. His report deals principally with three types of plants in all his species, viz., dwarf forms, upright or erect forms, and spreading or procumbent forms, each of which was found in a location favorable to the existence of that type while unfavorable to the other types; and each thus becomes a demonstration of the effects of natural selection. In our study of Crepis forms we have not been fortunate enough to study wild populations of Crepis in all of the localities from which we have obtained seed, but we have produced hereditary strains of erect forms, spreading forms, and dwarf forms from the same habitat at Eureka, a fact which does not especially favor the existence of any one type. Dwarf forms have also appeared in cultures from other places (France and Denmark), whose definite habitat characteristics are unknown to us. Similar plant forms are well known to occur sporadically in many wild and domesticated species. Mutations giving rise to pros- trate and dwarf types in plants are not infrequent when compared to other types of change. If we accept the idea of a genoiypic response of the species to the habitat, are we not also admitting the inconstancy of the gene, a theory which is no longer tenable? Continuing the assumption, it is not clear why these different hereditary types, such as we have in Crepis, remain constant in a single unvarying habitat. 1924] Collins: Inheritance in Crepis capillaris (L.) Wallr. 277 The very fact that they do not approach a common type under culti- vated conditions supports the theory of the constancy of the gene and is evidence of the inability of the habitat to induce genotypic changes. The occurrence of duplicate genes in other plants has brought forth the opinion that they may indicate the presence of duplicated chromosomes. Three cases of duplicate genes have been found in Bursa (Shull, 1920), a plant having 32 chromosomes (4 X 8), while a case of triplicate genes is reported in a wheat (Nilsson-Ehle, 1909) which has 42 chromosomes. This number is three times the number (14) found in several species of Triticum (Sax, 1921). Several pairs of duplicate genes have been found in Crepis capillaris. No plants producing such ratios have been examined cytologically, but in no visible way do they differ from plants which give 3 to 1 ratios for the same characters. From what is known regarding the effect of duplication of single chromosomes or of whole sets of chromosomes in Datura (Blakeslee, 1922) and in Nicotiana (Clausen and Good- speed, 1924), it is difficult to suppose duplication of chromosomes has occurred here. That we have parallel mutations in identical loci of two chromosomes of the same kind derived from a form with a different number by some meiotic irregularity is equally improbable, for capillaris has but three pairs of chromosomes, no two similar enough in size to be construed as duplicates. There are several other ways to account for- the appearance of duplicate genes, some of which have been discussed by Shull (1918). Four of these possi- bilities are (a) the occurrence of similar gene mutations in different chromosome pairs; (6) the mating of non-homologous chromosomes; (c) duplication of entire chromosomes; and (d) duplication of sections of chromosomes. The possibility of a chromosomal dupli- cation as the cause of the origin of duplicate genes in Crepis is very unlikely, as has been shown above. The other possibilities cannot be dealt with so readily. It would appear, however, that, had duplica- tion of a section of a chromosome taken place, other characters, the genes for which were located in the duplicated section, should show similar inheritance ratios. As a matter of fact, two other characters in Crepis capillaris give ratios of 15 to 1, but in the one case tested (revolute X smooth ribs) the type of linkage demanded by such an hypothesis was not obtained. Mating of non-homologous chromosomes should also result in duplication of other genes which should show linkage relations. Although only a small amount of critical data is as yet available, no confirmation of the linkage relations demanded 278 University of California Publications in Agricultural Sciences [Vol. 2 by these two methods of gene duplication has been obtained. Shull rejected the idea of the occurrence of two independent mutations as a cause of duplication of genes in Bursa on the ground that the char- acters were of such a complex nature that the occurrence of two independent mutations producing identically the same somatic results was on the verge of impossibility. The characters in Crepis for which there are duplicate genes cannot be considered as complex, and the occurrence of similar mutations in non-homologous chromosomes therefore seems at the present time to be the more reasonable explana- tion of the origin of duplicate genes in this species. Sturtevant (1921) has shown that some points in the germinal material of a given species are more susceptible to mutations than others. There is evidence that such a mutating locus occurs in capillaris, for the same character, bald, has appeared in a number of strains derived from widely separated localities. The identity of these genes for bald has been proved in all cases except one (France) by crosses in which they proved to be allelomorphic. That a certain locus may mutate in the same way in other species is at least indicated by the fact that this character is now known to occur in four other species, none of which has been grown extensively among our cultures. The gene for bald is recessive in capillaris and is also recessive in the species cross, setosa X capillaris. No less interesting and unique is the group of complementary genes found in C. capillaris where the appearance of three such pairs of genes are concerned with the inheritance of leaf characters and a fourth with chlorophyll. It is not strange, however, that a greater number of complex gene relations should be encountered in a species containing a low number of chromosome pairs than in species having a larger number, unless the larger number results from reduplication. There is probably a minimum number of genes which is necessary in any species, and there is no reason to believe, a priori, that a species with a larger number of chromosomes need have a correspondingly larger number of genes. There is also evidence from Drosophila that the genes are distributed at random in each chromosome (except in cases of multiple allelomorphs) and among the chromosomes. When this basic number of genes is distributed among a large number of chromosomes, more characters will show simple types of inheritance. When this basic number is distributed in a fewer number of chromo- somes, there will necessarily result more complex types of inheritance. 1924] Collins: Inheritance in Crepis capillaris (L.) Wallr. SUMMARY 1. Plants of Crepis capillaris are largely cross-fertilized, and this mode of reproduction operates to maintain a condition of genotypic heterozygosity. 2. Inbreeding wild plants thus produced results in the production of a number of pure races which show loss of vigor and reduction in size similar to the effects produced by inbreeding maize. 3. Four sets of duplicate genes are found to be responsible for the inheritance of four different characters. Two of these characters are shown not to be linked. Duplicated genes do not indicate dupli- cated chromosomes, for each pair is morphologically different from the others. 4. The recessive character 'bald' has appeared in a number of unrelated strains. This is evidence that a certain locus in one chromosome pair mutates more frequently in the same way than do other loci. The appearance of bald in other species may be due to a similar gene in each of these four species. 5. Several types of chlorophyll variations have appeared. Some show monohybrid recessive relations when contrasted with the normal condition, while others show more complex relations. 6. The different forms from widely separated localities show no tendency to approach a common type when grown continuously in the same place. It is with pleasure that the author acknowledges the helpful advice given by Professor Babcock and Professor Clausen throughout the progress of the work. 280 University of California Publications in Agricultural Sciences [Vol. 2 LITERATURE CITED Babccck, E. B., and Collins, J. L. 1920. Interspecific hybrids in Crepis. I. Crepis capillaris (L.) Wallr. X C. , tectorum L. Univ. Calif. Publ. Agr. Sci., vol. 2, pp. 191-204. Blakeslee, A. F. 1922. Variations in Datura due to changes in chromosome number. Am. Nat., vol. 61, pp. 16-31. Castle, W. E., Carpenter, F. W., et al. 1906. The effects of inbreeding, cross-breeding, and selection upon the fer- tility and variability of Drosophila. Proc. Am. Acad. Arts and Sci., vol. 41, pp. 731-786. Clausen, E. E., and Goodspeed, T. H. 1924. Inheritance in Nicotiana tahacum. IV. The trisomic character ' ' en- larged. " Genetics, vol. 9, pp. 181-197. Collins, G. N. 1919. Intolerance in maize to self-fertilization. Jour. Wash. Acad. Sci., vol. 9, pp. 309-312. Collins, J. L. 1920. Inbreeding and cross-breeding in Crepis capillaris (L.) Wallr. Univ. Calif. Publ. Agr. Sci., vol. 2, pp. 205-216. 1921. Eeversion in composites. Jour. Hered., vol. 12 ; pp. 129-133. 1922. Culture of Crepis for genetic investigations. Jour. Hered., vol. 13, pp. 329-355. CORRENS, C. 1912. Die neuen Vererbungsgesetze (Berlin), 75 pp. Darwin, C. 1876. The effects of cross- and self-fertilization in the vegetable kingdom (London), 482 pp. Harris, J. A. 1912. A simple test of the goodness of fit of Mendelian ratios. Am. Nat., vol. 46, pp. 741-745. Kristofferson, Karl B. 1923. Monohybrid segregation in Malva species. Hereditas, vol. 4, pp. 44-54. Nilsson-Ehle, H. 1909. Kreuzungsuntersuchungen an Hafer und Weizen. Lund's Univ. Ars- skrift, vol. 5, pp. 1-122. Basmuson, Hans. 1916. Kreuzungsuntersuchungen bei Beben. Zeitschr. f. Indukt. Abstamm. Vererb., vol. 17, pp. 1-52. Rau, Venkata 1923. Inheritance of some morphological characters in Crepis capillaris (L.) Wallr. Univ. Calif. Publ. Agr. Sci., vol. 2, pp. 217-242. Sax, Karl 1921. Chromosome relationships in wheat. Science, n.s., vol. 54, pp. 413-415. 1924] Collins: Inheritance in Crepis capillaris (L.) Wallr. 281 Sears, Paul B. 1922. Variation in cytology and gross morphology of Taraxacum. Bot. Gaz., vol. 73, pp. 425-446. Shoemaker, D. N. 1919. A study of leaf characters in cotton hybrids. Am. Breed. Assoc, vol. 5, pp. 110-110. Shull, G. II. 1914. Tiber die Vererbung der Blattfarbe bei Melandrium. Ber. Dent. Bot. Gesellschaft, vol. 31, pp. 41-80. 1918. Duplication of leaf lobe factor in Bursa. Brooklyn Bot. Garden, Mem., vol. 1, pp. 427-443. 1920. A third duplication of genetic factors in shepherds purse. Science, n.s., vol. 51, pp. 590. 1921. Three new mutations in Oenothera LamarcMana. Jour. Hered., vol. 12, pp. 354-363. Stork, Harvey E. 1920. Studies in the genus Taraxacum. Torr. Bot. Club Bull. 47, pp. 199-210. Sturtevant, A. H. 1921. Genetic studies on Drosopliiln simulans III. Genetics, vol. 6, pp. 179- 207. Tedin, Olcf 1923. The inheritance of pinnatifid leaves in Camelina. Hereditas, vol. 4, pp. 59-64. Trow, A. H. 1916. On "albinism" in Senecio vulgaris L. Jour. Gen., vol. 6, pp. 65-74. TURESSON, GOTE 1922. The genotypical response of the plant species to the habitat. Hereditas, vol. 3, pp. 211-350. EXPLANATION OF PLATES PLATE 45 Fig. 1. A rosette of the viridis race on the left with a pallid rosette on the right. Fig. 2. A typical rosette of the scalaris H6 race, showing blunt lobes, raffled Aving on midrib, constricted base of lateral lobes, and a twisting of the lateral lobes. [282] UNIV. CALIF. PUBL. AGRI. SCI. VOL. 2 COLLINSI PLATE 45 Fig. 1 Fig. 2 PLATE 46 Fig. 1. A rosette of simplex Z9 on the left, and at the right the aberrant pinnatifid type which appears in all cultures. Fig. 2. A rosette of the scalaris e29 race. [284] UNIV. CALIF. PUBL. AGRI. SCI. VOL. 2 [COLLINS] PLATE 46 Fie. 1 Fie. 2 PLATE 47 Fig. 1. A typical rosette of the pinnatifid leaf, scalaris e28. Fig. 2. A rosette showing revolute leaves. [286] UNIV. CALIF. PUBL. AGRI. SCI. VOL. 2 [COLLINSI PLATE 47 Fig. 1 Fig. PLATE 48 Fig. 1. The bicephalic type of faseiation. Fig. 2. A mature dwarf II plant. [288] UNIV. CALIF. PUBL. AGRI. SCI. VOL. 2 ICOLLINS] PLATE 48 % Fig. 1 V IT Fig. 2 PLATE 49 Fig. 1. Two dwarf III plants with two normal sibs. Fig. 2. A typical plant from the race with the spreading habit. [290] UNIV. CALIF. PUBL. AGRI. SCI. VOL. 2 ICOLLINSI PLATE 49 KB w Vr.i>. feS* ft&ft /% wfc* Fig. 1 fc *A-A'\ Fig. PLATE 50 Fig. 1. A typical plant of the erect growth habit. [292] UNIV. CALIF. PUBL. AGRI. SCI. VOL. 2 (COLLINSI PLATE 50 PLATE 51 Fig. 1. Palea on the left with a receptacle of a normal plant on the right. Fig. 2. Three F, rosettes from the cross, scalaris X simplex. [294J UNiV. CALIF. PUBL. AGRI. SCI. VOl . 2 [COLLINSI PLATE 51 Fig. 1 I Fig. PLATE 52 Fig. 1. Typical leaves from two plants of each of the parent strains and of the F 1} together with one leaf from each of eight F 2 plants, which show the results obtained when scalaris and simplex plants are crossed. Note the appearance in F 2 of the curved terminal lobe typical of the scalaris grandparent. [296] UNIV. CALIF, PUBL. AGRI. SCI. VOL. 2 (COLLINS, PLATE 52