168 APPENDIX I—DIVERGENT EVOLUTION. 
continued to regard it as strengthened and developed through the 
action of natural selection.* 
(3) Reasons for believing that potential segregation can not be accu- 
mulated by natural selection.—Concerning this last point I wish to give 
reasons fora different opinion. I believe that qualities simply produc- 
ing segregation can never be accumulated by natural selection, for— 
First. When separate generation comes in between two sections 
of a species they cease to be one aggregate, subject to modification 
through the elimination of certain parts. Both will be subject to sim- 
ilar forms of natural selection only so long as the circumstances of 
both and the variations of both are nearly the same, but they will no 
longer be the members of one body between which the selecting pro- 
cess is carried out. On the contrary, if they occupy the same district 
each group will stand in the relation of environment to the other, mod- 
ifying it, and being modified by it, without mutually sharing in the 
same modification. 
Second. Though one may exterminate the other, the change that 
comes to the successful group through the contest is not due to its 
superiority over the other, but to the superiority of some of its own 
members over others. 
Third. Whenany segregate form begins toarise we can not attribute 
its success to the advantage of isolation, for it is not the success, but 
the separateness of the success, that is due to the isolation. 
Fourth. The power of migration, or any other power directly re- 
lated to the environment, may be accumulated by natural selection, 
and afterward lead to segregation, but, according to my method of 
judging, the advantage of segregation over intergeneration is not the 
cause of the preservation of forms endowed with segregative qualities, 
for they will certainly be preserved as long as they are able to wina 
bare existence, which is often a lower grade of success than the one 
from which they are passing. 
(4) How shall we explain the accumulation of potential segregation2— 
But if the accumulation of prepotential segregation is not due to nat- 
ural selection, how shall we explain it? The divergence of a group 
can not take place without its being segregated from the original stock 
as well as from other types; and the potency of the sexual elements of 
the new group will be maintained in their relations to each other by 
some form of reflexive selection; butas there can be no reflexive selec- 
tion between the segregated groups, the potency of the elements for 
crossing outside of the group will in time be impaired; and then we 

* Since my comments on this passage were written I have discovered that Dar 
win has omitted it from the sixth edition. 
