72 



THE AMERICAN NATURALIST [Vol. XLIV 



should give rise to different lines of descent so that in the 

 F n generation when individuals of these different lines 

 are crossed, the character behaves as a di-hybrid instead 

 of as a mono-hybrid. In other words, it is more probable 

 that these units arise through variation in different in- 

 dividuals and are combined by hybridization, than that 

 actually different structures for forming the same char- 

 acter arise in the same individual. 



On the other hand, there is a possibility of an action just 

 the opposite of this. Several of these quantitative units 

 which produce the same character may become attached 

 like a chemical radical and again behave as a single pair. 

 Nilsson-Ehle gives one case which he does not attempt to 

 explain, where the same cross gave a 4:1 ratio in one 

 instance and 8.4 : 1 ratio in another instance. In his other 

 work characters always behaved the same way; that is, 

 either as one pair, two pairs, three pairs, etc. In my 

 work, the yellow endosperm of maize has behaved dif- 

 ferently in the same strain, but it is probably because the 

 yellow parent is homozygous for one yellow and heter- 

 ozygous for the other. They were known to be pure for 

 one yellow, but it would take a long series of crosses to 

 prove purity in two yellows. 



Let us now consider what is the concrete result of the 

 inter-action of several cumulative units affecting the same 

 character. Where there is simple presence dominant to 

 absence of a number n of such factors, in a cross where all 

 are present in one parent and all absent in the other 

 parent, there must be 4 n individuals to run an even chance 

 of obtaining a single F 2 individual in which the character 

 is absent. When four such units, A 1 A 2 A S A 4 are crossed 



