MULTIPLE FACTORS 205 



subsequent generations. Exactly the opposite results 

 are expected when the population is heterogeneous for 

 multiple factors at the beginning. 



On several occasions Castle has stated that the prac- 

 tical breeder is especially familiar with the effects of 

 selection because he has obtained most of his results by 

 this method. It is intimated not only that the breeder 

 is in a position to look favorably on the doctrine of po- 

 tencies, but that his famiharity with work of selec- 

 tion entitles his views to special consideration. But 

 no one has in recent years denied that selection of 

 mixed material will lead to the isolation of definite 

 types and even of new types. 



To what has been said one additional consideration 

 must be urged. Mutations may occur at any time 

 and will be quickly observed if they are in the direc- 

 tion in which a selective process is being carried out. 

 It may not be easy to recognize the first appearance 

 of a mutant and, in fact, its presence may be detected 

 only after the selection has gone so far that its origin 

 is lost. The breeder may, if he is not extremely 

 observant, infer that his selection is producing the 

 desired effect on the potency of the character, while 

 in reahty he is studying the influence of a new 

 factor on the character under selection. This possi- 

 biUty may be illustrated by two cases. In Castle's 

 experiments two rats appeared that behaved like a 

 new type. In fact he gives them the value of mutants. 

 In Drosophila, Morgan carried out a selection experi- 

 ment for three years, involving upward of 75 gen- 

 erations. The character selected was a dark " trident'' 



