228 University of California Puhlicatioiis in Botany [Vol.5 



Two points were emphasized in the pre\ious communications 

 dealing with the experiment on the inheritance of flower-size in 

 .Y. acuminata hybrids. First, it was demonstrated that, with 

 reference to the range of variation of the flowers borne on a single 

 plant of one of its parents, the F^ range was distinctly greater than 

 the parental range. Further, of all the flowers produced by all the 

 various hybrid plants, the largest flowers Avere as large as the largest 

 of all the flowers produced on plants of the large-flowered parent. 

 This was also true for the smallest flowers. Nothing more was claimed 

 as to the variability of flower-size in the F^ .Y. acuminata hybrids. 

 It was not stated that the range of the Fi populations was greater 

 than the range of the populations of the corresponding parents. In 

 respect to this particular point, however, the above fre(iuency distri- 

 butions are at least interesting. In variety I and variety III (table 3), 

 the large-flowered and small-flowered varieties and the Fj hybrids 

 between them in 1911, 1912, and 1913, the F^ range is throughout 

 greater than the range of either parent in corresponding years. The 

 Fi between varieties II and III in 1911, when compared with the 

 parents of the same year (table 5), serves to emphasize this point, as 

 does the Fj of the cross between varieties I and II (table 1). The 

 number of plants is obviously inadequate, yet there is here an indi- 

 cation again that the question of the extent of variation in F^ is still an 

 open one. It might be of some significance if, in place of examining 

 only one-eighth as many F^ plants as F, plants, an example of the 

 iLsual procedure, the number of individuals in these two generations 

 were made more nearlj^ equal in favor of the larger number (cf. Shull, 



1914, p. 131). 



The second point which was taken up in the earlier reports had 

 to do with the inadequate nature of the expanded Mendelian notation 

 as an explanation of the inheritance of quantitative characters and 

 other complex situations. An apparently widespread doubt as to its 

 adequacy in such connections makes it possible to note that nothing 

 during the two years since the publication of the original reports has 

 made untenable our position on this subject therein stated (cf. Castle. 



1915, p. 97). Shull (1914) feels that we have demonstrated a notable 

 increase in variability as to flower-size in the Fo generation previously 

 reported on (Goodspeed, 1913) and reports that we "refu.sed to as- 

 cribe this greater variability to Mendelian segregation." The actual 

 statement made was that the greater variability of F., as compared 

 w^th Fi "appears to make it plain that segregation does occur" {ihid., 



