G. H. SHULL 97 



that the mere quantity of anthocyan which a plant produces does not 

 make the difference between the color-pattern of 0. rubricalyx and 

 that of 0. rubrinervis, for certainly those plants with red-spotted leaves 

 and dark-red or brilliant-red stems, produced many times as much 

 anthocyan as the greenish-stemmed plants, though only the latter had 

 the rubricalyx-type of buds. 



It is interesting to see that even Gates (1914) occasionally catches 

 a glimpse of the fact that the difference between 0. rubrinervis and 

 0. rubricalyx is something more than a mere quantitative difference in 

 the amount of red pigment produced by each. He says (1914, p. 246) 

 " It is then perfectly clear that, although extent of pigment on the buds 

 behaves, with very few exceptions, as a definite unit-character, showing 

 the phenomena of dominance and absence ; yet the amount of pigment 

 is very probably reduced to a half in the F l hybrids, and it is certainly 

 diluted very much (probably one-half) again on crossing back with 

 grandiflora....The cause of the definiteness of distribution of the 

 pigment in the buds is a problem in morphogenesis." After this 

 statement one may be surprised to read (p. 270) that " the most careful 

 observation and study shows that the red pigmentation-character R is 

 in the last analysis, inherited in quantitative fashion." 



The Mendelian behavior of the rubrica ^-character was inferred 

 by Gates from very insufficient evidence. When an attempt is made 

 to interpret my results on a Mendelian basis, difficulties are encountered 

 which seem at present nearly insurmountable. If we suppose that my 

 original r///>r<V'//y.-plant (No. 11410(1)) was a heterozygote in respect 

 to two independent determiners for the rubricalyx-iy^e of buds, the 

 progeny secured by self-fertilizing this plant was in as close agreement 

 with expectation, as the small number of plants would require, the 

 ratio being 107 : 1 instead of 15 : 1 ; but if there were no further 

 complications, such a constitution for the rubricalyx parent would lead 

 to the expectation of rubricalyx and nou-rubricalyx in the ratio 3:1 in 

 every cross between this plant and other plants which lacked both of 

 the supposed determiners for the rubricalyx-ch&r&cter. Instead of this, 

 the ratios in the four hybrid families here reported, were 0'91 : 1, T44 : 1, 

 1'80 : 1 and 2*65 : 1, or if we combine the two rubricalyx-rubrine i 

 families, the result is 165 : 183 or nearly 1 : 1, while the two Lamarckiana 

 crosses combined give 161:75, or about 2:1. Both of these ratios 

 actually occur in typical Mendelian inheritance, and could be accounted 

 for here by such additional assumptions as these ; namely, (a) that rny 

 rubrinerris plant, No. 1123(8), not only lacked the two independent 



