NATURAL SELECTION 135 



the same chapter shows: "Selection as an agency in 

 evohition mnst then be restored to the important 

 place which it held in Darwin's estimation, an 

 agency capable of producing continuous and pro- 

 gressive racial changes." 



Now the only really critical piece of work done on 

 this subject, that of Johannsen, had already led to 

 the opposite result. It can hardly be said, therefore, 

 that the subject had not been studied closely enough. 

 The evidence from Castle's own experiments with 

 hooded rats when studied more criticallv has shown 

 that it still remains to be proven that genes are sub- 

 ject to quantitative variations and are amenable to 

 selection. 



Conclusions 



The evidence discussed in this chapter is consistent 

 with the view that the individual gene is not affected 

 by selection, and that the initial changes commonly 

 observed when selection is practised on a mixed 

 population are due to recombinations of the differ- 

 ent kinds of genes affecting the same character that 

 are present in most populations. Since these modify- 

 ing genes behave in inheritance strictly in accord- 

 ance with ^Mendel's laws there are no grounds for 

 assuming that they are different from other genes. 

 Selection ceases to produce any further effects after 

 these genes have been sorted out and the material 

 has become homozygous for them. 



