244 Gates. 



E. 3. The variability of the rubricdyx character, R. 



From table III we have found that different (otherwise identical) 

 individuals from the same cross are capable of producing R in different 

 ratios, and that when crossed back with grandiflora the latter exerts 

 a depressing effect on anthocyanin production. As already intimated, 

 this not only reduces the percentage of Rs but also causes wide 

 variations in the depth of red and, less frequently, e\-en in the extent 

 of the red colour pattern on the buds. Observations of this kind soon 

 raise the practical question, "Wliat is a unit-character, and when does 

 it cease to be such? In following the variability of such a character 

 it soon becomes obvious that it cannot be considered apart from the 

 organism as a whole. 



Let us, therefore, consider the whole bud, in which the character 

 R is most sharply developed. In doing so we discover that the cha- 

 racter R has (i) a morphological and (2) a physiological aspect. Mor- 

 phologically, the pigment is definitely distributed. In the vast majority 

 of cases the hypanthium and the median ridge of the sepals are either 

 definitely red throughout (R), or definitely green throughout (r ), even 

 when the colour is very pale. This definite pigment-distribution is 

 doubtless closely wrapped up with the morphogenesis of the organism, 

 and was used as the basis of bud classification. I have never found 

 intermediates as regards distribution of pigment in the buds, either 

 in pure or heterozygous rubricalyx or in F^ crosses with grandiflora. 

 But in (i) crosses between red and green buds of this Fj, (2) occa- 

 sional F2 cultures, {3) green-budded segregates from such crosses in 

 F2, and in (4) back-crosses with grandiflora, such intermediates do 

 sometimes occur. In table IV the twenty cases found among 2794 

 plants of my 1912 cultures are tabulated. In the column "pigmen- 

 tation of bud", the condition described was observed on all the buds 

 of a plant at a given tine. \Y\\\\q the buds of all 20 plants were ranged, 

 as will be observed, all the way between the typical rubricalyx and 

 rubrinervis conditions as regards bud-pigmentation, yet in only two 

 cases (marked inter) were they so perfectly intermediate that they 

 could not be classified as R or r. 



Contrasted with the distribution of pigment in R, which almost 

 invariably displays itself as a presence or absence of anthocyanin in 

 definite areas, is the quantity of pigment produced by any part of the 

 plant. As far as one can judge (which is not a very safe criterion), 

 rubricalyx appears to produce at least 5 or 6 times as much antho- 



