122 Glume- Le7igth and Grain-Length in Wheat 



Tables XIX, XXI, and XXII show that the F^ . K type is not repro- 

 duced in i^a as does the equation F^. Mj^ —Fq. Mj^= 06S mm. An 

 upward "shift" in mean ghime-length appears to have occurred, and 

 the shifted value to have bred true, for despite the 1913-14 seasonal 

 difference, F^ . Mj^ and F^ . M^ differ by 014 mm. only. It will be recalled 

 that in the case of K glume-length the differences concerned were so 

 small that they could not safely be accepted as evidence of shift. 



Measurable Characteristics of tJie Polish Grain Type in F^, F^, and F^. 



The tables of frequency distribution and the values of M show that 

 F2 . P very closely resembles its F^ progeny, but that both of them are 

 definitely " shifted " down from Fq . P. The amount of the shift is : 

 F,.Mp-F.^.Mp= 1-36 mm. = 1334 % of F, . Mp, 

 Fo . Mp -F,.Mp= 1-45 mm. = 14-21 % of F^ . Mp. 



That there is a segregation of grain types in F.^ is shown by the 

 regularity and resemblance of the distributions etc. of F^ .P and F3 . P 

 (cf. also F, . K and F., . K). 



I. 

 Measurable Characteristics of the Heterozygous Grain Type in Fq, F2, and F.^. 



Thus the classification of the grains of the F.^ and F^ plants by 

 means of the glumes with which they are associated, has led to the 

 conclusion that three grain types occur in F2. These types naturally 

 have the same frequency relations as the glume types, and as pointed 

 out in § IV above the evidence indicates that the relations are actually 

 an experimental expression of a 1 : 2 : 1 ratio. A unimodal distribution 

 (Table XX, i.e. grain-lengths of the complete F^ has thus been shown 

 to consist of three overlapping separate constituents K, I, and P, the 

 overlap having been much accentuated by the " inwards " shift of 

 P and K from the Fq values. 



In the glume-length results of § IV (under " / ") it was shown that 

 the progeny of the F^ . I plants had a length distribution similar to 



