inter-specific sterility cannot be
increased by natural or any other known form of selection, but that it
contains within itself its own principle of destruction. If it is
proposed to get over the difficulty by postulating a larger percentage
of the variety annually arising within the species, we shall not affect
the law of decrease until we approach equality in the numbers of the two
varieties. But with any such increase of the physiological variety the
species itself would inevitably suffer by the large proportion of
sterile unions in its midst, and would thus be at a great disadvantage
in competition with other species which were fertile throughout. Thus,
natural selection will always tend to weed out any species with too
great a tendency to sterility among its own members, and will therefore
prevent such sterility from becoming the general characteristic of
varying species, which this theory demands should be the case.

On the whole, then, it appears clear that no form of infertility or
sterility between the individuals of a species, can be increased by
natural selection unless correlated with some useful variation, while
all infertility not so correlated has a constant tendency to effect its
own elimination. But the opposite property, fertility, is of vital
importance to every species, and gives the offspring of the individuals
which possess it, in consequence of their superior numbers, a greater
chance of survival in the battle of life. It is, therefore, directly
under the control of natural selection, which acts both by the
self-preservation of fertile and the self-destruction of infertile
stocks--except always where correlated as above, when they become
useful, and therefore subject to be increased by natural selection.

_Summary and Concluding Remarks on Hybridity._

The facts which are of the greatest importance to a comprehension of
this very difficult subject are those which show the extreme
susceptibility of the reproductive system both in plants and animals. We
have seen how both these classes of organisms may be rendered infertile,
by a change of conditions which does not affect their general health, by
captivity, or by too close interbreeding. We have seen, also, that
infertility is frequently correlated with a difference of colour, or
with other characters; that it is not proportionate to divergence of
structure; that it varies in reciprocal crosses between pairs of the
same species; while in the c