THE PRODUCTION OF DOMESTIC RACES. 17 



selection," in a very interesting chapter of his volume entitled "The 

 Chances of Death and Other Studies in Evolution." I fully agree 

 with this author in the emphasis he puts on the importance of a care- 

 ful statistical investigation of the subject; but I think the term he 

 has chosen would naturally apply to sexual and impregnational selec- 

 tion as well as to this form of selection which is dependent on degrees 

 of fertility. If these are all to be called forms of selection, I would 

 suggest that the special principle under consideration might be desig- 

 nated fecundal selection, and its effect on the group might be called 

 fecundal transformation. 



3. The Divergence o} Races. 



In order to produce two or more divergent breeds from one breed, 

 it is necessary, first, that the original stock should be divided into 

 separate portions that are prevented from crossing (this is conven- 

 iently called isolation) ; and, second, that there should be some cause 

 transforming one or more of these separated portions. Now, the 

 cause producing transformation in a separated portion may be the 

 incapacity of a given small fragment of a race or species to perpetuate 

 the original average character of that race or species ; or, after sepa- 

 ration, the portion may be subjected to some form of unbalanced 

 selection. 



Returning, for illustration, to the 600 bison, we may suppose them 

 to be indiscriminately divided into two isolated herds of 300 each, in 

 which case the average character of each herd would probably be 

 nearly the same as that of the other herd, and the descendants of the 

 two herds, if not subjected to different kinds of selection, would prob- 

 ably present no important differences, at least for many generations. 

 If, however, the whole herd were broken up into pairs, the offspring 

 of the pairs forming 300 isolated groups, each of which was never 

 allowed to cross with any other group, we should probably find that 

 perceptible divergences would soon present themselves in some of 

 the groups, even if they were not subjected to unlike selection. 



Another form of isolation, still more effective in producing diver- 

 gence, would be gained if the individuals presenting the highest de- 

 gree of some special kind of variation were brought together in one 

 group that was prevented from crossing with the others. The 

 division of organisms which have descended from one original inter- 

 generating group into two or more groups that do not possess the 

 same average character, and that lack either the capacity, inclination, 

 or opportunity for intergenerating, I call segregation. Whenever 

 there is a closer resemblance between the members of an intergenerat- 



