s depends
exclusively on differences in their sexual constitution.

* * * * *

On the principle which makes it necessary for man, whilst he is selecting
and improving his domestic varieties, to keep them separate, it would
clearly be advantageous to varieties in a state of nature, that is to
incipient species, if they could be kept from blending, either through
sexual aversion, or by becoming mutually sterile. Hence it at one time
appeared to me probable, as it has to others, that this sterility might
have been acquired through natural selection. On this view we must suppose
that a shade of lessened fertility first spontaneously appeared, like any
other modification, in certain individuals of a species when crossed with
other individuals of the same species; and that successive slight degrees
of infertility, from being advantageous, were slowly accumulated. This
appears all the more probable, if we admit that the structural differences
between the forms of dimorphic and trimorphic plants, as the length and
curvature of the pistil, &c., have been co-adapted through natural
selection; for if this be admitted, we can hardly avoid extending the same
conclusion to their mutual infertility. Sterility moreover has been
acquired through natural selection for other and widely different purposes,
as with neuter insects in reference to their social economy. In the case of
plants, the flowers on the circumference of the truss in the guelder-rose
(_Viburnum opulus_) and those on the summit of the spike in the
feather-hyacinth (_Muscari comosum_) have been rendered conspicuous, and
apparently in consequence sterile, in order that insects might easily
discover and visit the other flowers. But when we endeavour to apply the
principle of natural selection to the acquirement by distinct species of
mutual sterility, we meet with great difficulties. In the first place, it
may be remarked that separate regions are often inhabited by groups of
species or by single species, which when brought together and crossed are
found to be more or less sterile; now it could clearly have been of no
advantage to such separated species to have been rendered mutually sterile,
and consequently this could not have been effected through natural
selection; but it may perhaps be argued, that, if a species were rendered
sterile with {186} some one compatriot, sterility with other species would
follow as a necessary consequence. In t