Permanence and Evolution. 89 



conspicuous characters. But let alone that we 

 want positive evidence, it is difficult to say why 

 callosities should be less likely to be encouraged 

 under domestication than length of bill. Man- 

 kind, of course, could not select purely internal 

 characters ; but if correlation occupies such an 

 important place in the scheme of organised 

 nature as according to Darwinism it must do, 

 such points must often be correlated with those 

 man has selected, also points grossly conspicuous 

 to man must equally often be correlated with 

 others of importance to the being 'itself. The 

 whole idea rests on a misconception of what in 

 selection we do. When, e.g. t we get a white 

 horse by pairing two white horses, and so in a 

 few generations establish a white sub-breed, we 

 simply abstain from crossing ; we show the per- 

 manence, not the mutability of types. The only 

 case where a doubt occurs is where we get 

 offspring possessing qualities not possessed by 

 their parents, or possessed in a lesser degree; but 



