DIVERGENCE RESULTING FROM ISOLATION. 253 
selection, and that all diversity of selection is due to exposure to 
different environments. The divergences in the cases above referred 
to, it is said, are probably due to differences in the environment that 
are not easily recognized. This was the explanation suggested by 
Darwin when the facts were reported to himin 1872. The division of 
a species into isolated portions did not seem to him to furnish any 
factor that could produce divergence unless it was aided by exposure 
to different external conditions. The same view is expressed in his 
“Origin of Species,’’ sixth edition, page 319. 
My reply is twofold. (1) The theory that all divergences in Sand- 
wich Island land mollusks are due to differences in the environment 
requires us to believe that there are occult influences increasing in 
difference with each additional mile of separation, and that these 
influences control the natural selection of the mollusks, but have no 
influence on any other species occupying the same areas. A theory 
that involves so heavy an assumption can not be received when a 
simpler theory is open to us. (2) I believe I can entirely remove this 
objection, urged against my conclusion on these purely theoretical 
grounds, by showing that there are certain causes of divergence, not 
depending on exposure to different environments, that are necessarily 
introduced by the division of a species into isolated groups; and that, 
under the influence of these causes, diversity of habits may arise pro- 
ducing diversity of selection, even while the fragments are exposed to 
the same environment. 
I have elsewhere called attention to the fact that the independent 
breeding of separated groups, as far as we can judge, always tends to 
produce divergence; and I have shown that, when a species is indis- 
criminately broken into independent fragments, the tendency to diver- 
gence will, on the average, vary in direct proportion to the instability 
of the species and in inverse proportion to the size of the fragments; 
for on these factors depends the probable degree of departure of the 
average character of the fragment from the average character of the 
species previous to its being broken into fragments, and, therefore, 
the degree of segregation. 
I wish now to show that the maintenance of certain classes of char- 
acters always belonging to an unbroken species is due to a form of 
selection that can continue only so long, and so far, as free crossing 
continues. Reflexive selection is aformative principle, depending on 
the relations in which the members of an intergenerating group of 
organisms stand to each other, while they continue to intergenerate ; 
but when two portions of an original species have become so divergent 
as to compete with each other in the same area without crossing, they 
