The Method of Evolution 6i 



animals so extensively pigmented that they would readily 

 pass for the "Irish type," which has white on the belly 

 only, but which is known to be in crosses a Mendelian alter- 

 native to the hooded tj-pe. By selection we have practi- 

 cally obliterated the gap which originally separated these 

 types, though selected animals still give regression toward 

 the respective types from which they came. But this 

 regression grows less with each successive selection and ulti- 

 mately should vanish, if the story told by these statistics 

 is to be trusted. As yet there is no indication that a limit 

 to the effects of selection has been reached. 



From the e\'idence in hand we conclude that Darwin 

 was right in assigning great importance to selection in 

 evolution; that progress results not merely from sorting 

 our particular combinations of large and striking unit- 

 characters, but also from the selection of slight differences 

 in the potentiality of gametes representing the same unit- 

 character combinations. 



Accordingly we conclude that unit-characters are not 

 unchangeable. They can be modified, and these modifica- 

 tions come about in more than a single way. Occasionally 

 a unit-character is lost altogether or profoundly modified 

 at a single step. This is mutation. But more frequent 

 and more important, probably, are slight, scarcely notice- 

 able modifications of unit-characters that aiiford a basis for 

 a slow alteration of the race by selection. Mutation, then, 

 is true, but it is a half-truth; selection is the other and 

 equally important half of the truth of evolution, as Darwin 

 saw it and as we see it. 



