18 BIONOMIC LAWS. 
ing group than between them and the members of some other group 
with which free crossing has been interrupted, there segregate breed- 
ing or segregation exists. If it were possible to divide the 600 bison 
into two completely equivalent groups, the isolation of these two 
groups would not involve segregation; but the indiscriminate division 
of any intergenerating group into two or more isolated groups 
usually involves more or less difference in the groups, and, therefore, 
more or less segregation. 
If we carefully consider the process by which the different domestic 
races of any species have been produced, we shall find that the isola- 
tion of each race from every other race has, in every case, been a 
prime factor. Until modern times the hostility of different tribes of 
men and the want of free commerce between nations have secured the 
isolated breeding of the domestic races under the care of different 
tribes and peoples. Now, according to the principle I have just 
pointed out, the initial differences between those portions make them 
more or less segregated groups whenever free crossing between the 
portions is cut off. But this initial segregation is soon intensified by 
the transforming influences to which the different portions are sub- 
jected. As each portion is subjected to the care of a separate tribe 
of men who preserve such individuals of the offspring as best suit 
their purposes or fancies,-and as the individuals thus preserved sel- 
dom represent the average form of the wild species, transformation 
is soon produced. Each isolated group, if it survives under its new 
conditions, must produce forms increasingly adapted to meet the 
desires of those who care for them; and even when those on whom they 
depend have no idea of developing new characteristics by selection, 
unconscious selection usually takes place. If the selected form is not 
the average form, transformation necessarily follows. But as we are 
now considering the process by which divergent evolution has been 
produced in domestic races, the important point is, not that trans- 
formation is usually produced by domestication, but that, when this 
cause of transformation modifies continuously isolated portions of 
the same species, the result is always divergent and not parallel. Ac- 
count for it as we may, when a domestic breed has been transformed 
during isolation, the transformed portions are always found to be 
more or less unlike; and this is so even when the physical conditions 
are the same, and when the persons on whom the selection has devolved 
are representatives of the same race. 
I think it will be found that independent transformation (that is, 
transformation during isolation), is always divergent and never com- 
pletely parallel; and I believe this to be so whether the transforma- 
