EVOLUTION 



447 



statistically investigated either by experiment or observa- 

 tion. Such investigation and not verbal discussion can 

 alone here, as at other points, throw light on the problem 

 of evolution. Difficulties there are, and I would not have 

 the reader overlook them. The hypothesis would be 

 much simpler if we could omit stage (c) altogether. But 

 I cannot put on one side the evidence for differential 

 fertility and the correlation of fertility with other char- 

 acters. I cannot doubt the existence of genetic 

 selection as a factor of evolution, but in the present 



D F 



a' c H b' t) 



Fig. 32. 



state of our knowledge it introduces a new series of 

 complex problems, although methods of solution are not 

 beyond reach. At the risk of wearying the reader I 

 must point out at least one of these. Let the figure 

 DCE represent the distribution of some character in the 

 manner indicated on p. 385. Thus Cc is the modal 

 frequency, and c gives the type or modal value of the 

 character. Suppose now that the most fertile character 

 is identical with the modal value c. Now let differentia- 

 tion take place, and let a and d give the types of the 

 two differentiated groups, Aa and Bb representing on 

 some scale (possibly different from that of the original 

 frequency of DCE) the modal frequencies and DAG, 

 HBJ the respective frequency distributions. Now under 



