178 SELECTION. 



but as their breeding is uncontrolled, their get falls outside the 

 limits of the species. In respect to the effect of selection these 

 domestic species differ somewhat from species in a state of 

 nature, and it is necessary to consider the nature of the differ- 

 ence as well as the points where domestic and wild species 

 are affected by selection in the same way. This consideration 

 will be illuminating chiefly in the question, how far we can 

 generaUze the phenomena observed in the evolution of do- 

 mestic species. 



It is possible to develop a new breed of dogs and to bring it 

 to perfection in a country over-run with mongrel dogs of all 

 kinds, simply by choosing suitable males for certain females 

 and controUing their mating. Our choice of individuals can be 

 made according to any preconceived idea of what we want to 

 preserve in our dogs, what we want to combine, what should 

 be excluded. By controlling our animals at mating-time, the 

 fitness or survival- value of the characters selected by us need 

 not be taken into consideration at all. We are, moreover, not 

 concerned with the constitution of the dog-population at large, 

 in the region where we are perfecting our species of domestic 

 dogs. No group of animals or cross-fertihzed plants can ever 

 grow to the status of a species, unless it is either so constituted 

 or so situated, that matings within the group are far more 

 numerous than outcrosses. 



In the case of the domestic species, selection is a means of 

 strict isolation in addition to its influence on the constitution 

 of the selected group. Natural selection must necessarily act 

 qute differently in different situations, even if we can con- 

 ceive how in the long run, and speaking statistically, it will 

 tend on the average to act by favouring a certain type. But 

 artificial selection is selection according to one ideal, and it is 

 acting as a directing factor just as strongly and efficiently in 

 favour of an altogether irrelevant quality, or even of a decided- 

 ly harmful character, as in favour of a quality which makes for 

 the success of the individual showing it. 



If we compare two groups, which are each sufficiently effect- 



