No. 594] BLENDING AND MENDELIAN INHERITANCE 325 



times little or no contamination is observable; sometimes 

 it is considerable. 



TABLE I 



Comparison of the Means of Families descended from the Same Pure- 

 line Grandparent, but from Parents of Unlike Character 

 Based on Hoshino's Tatle 6. 



Contamination sufficiently great would account for the 

 intermediate early as a modified early class and the inter- 

 mediate late as a modified late class. The matter of 

 coupling is unaffected by this hypothesis, since coupling 

 is shown with only one factor, Hoshino's factor A. The 

 observed variability and intergradation of the two inter- 

 mediate classes favors a hypothesis of contamination 

 rather than one of an independent modifying factor. 



If we suppose modification due to gametic contamina- 

 tion to occur in half of the gametes formed by F x indi- 

 viduals and that this contamination is definite in amount 

 (say equivalent to 5 days) the F 2 expectation would be 

 exactly the same as from a two-factor system such as 

 Hoshino adopts. As a matter of fact neither supposition 

 is exactly correct. If we adopt contamination as a suffi- 

 cient explanatory hypothesis, we must suppose the 

 amount of contamination to be variable; if we adopt a 

 definite modifying factor, we must suppose the amount of 

 modification to be variable. 



To make the matter clear, let us suppose the early race 

 to be stable at 35 days between sprouting and flowering 



