326 



THE AMERICAN NATURALIST 



[Vol. L 



and the late race to be stable at 55 days. According to 

 Hoshino's hypothesis aabb stands for a 35-day period, 

 and AABB for a 55-day period. If we assign to the 

 assumed modifying factor B a delaying effect of five days, 

 then the class aaBB (early intermediate) will have a 

 value of 40 days, and the other homozygous class AAbb 

 (late intermediate) will have a value of 50 days. Hetero- 

 zygous classes will be intermediate as follows : 



TABLE II 



Composition of F, on Hoshino's Two-factor Hypothesis, with Eqtjiva- 

 nts in Days from Sprouting to Flow 



If, however, we replace the modifying factor B in this 

 scheme by a modification in A amounting to 5 days, then 

 we can dispense with B and yet obtain exactly the same 

 classes and in the same numerical proportions and with 

 nearly the same expectations as regards their breeding 

 capacity. Let us assume that a stands for a 35-day 

 period, A for a 55-day period, and that modified a, which 

 we will call a', stands for a 40-day period, and modified 

 A, which we will call A', stands for a 50-day period and 

 that all are allelomorphs of each other. Then the F t 

 gametes will be a -f a' + A' + A and F 2 will contain the 

 classes shown in Table III. 



It is evident that both schemes fit the observed facts 

 fairly well. Either one will explain the decreased varia- 

 bility of F 3 as compared with F 2 and the production of 

 several different types of F 3 families differing in the 

 amount of their variability, some of which are relatively 

 constant. But the former scheme will not answer with- 



