enth chapter. In these cases we must follow Naudin,[915]
and admit that the "essence" or "element" of the two species, which terms I
should translate into the gemmules, have an affinity for their own kind,
and thus separate themselves into distinct stripes or blotches; and reasons
were given, when discussing in the fifteenth chapter the incompatibility of
certain characters to unite, for believing in such mutual affinity. When
two forms are crossed, one is not rarely found to be prepotent in the
transmission of character over the other; and this we can explain only by
again assuming that the one form has some advantage in the number, vigour,
or affinity of its gemmules, except in those cases, where certain
characters are present in the one form and latent in the other. For
instance, there is a latent tendency in all pigeons to become blue, and,
when a blue pigeon is crossed with one of any other colour, the blue tint
is generally prepotent. When we consider latent characters, the explanation
of this form of prepotency will be obvious.

When one species is crossed with another it is notorious that they do not
yield the full or proper number of offspring; and we can only say on this
head that, as the development of each organism depends on such
nicely-balanced affinities between a host of gemmules and developing cells
or units, we need not feel at all surprised that the commixture of gemmules
derived from two distinct species should lead to a partial or complete
failure of development. With respect to the sterility of hybrids produced
from the union of two distinct species, it was shown in the nineteenth
chapter that this depends exclusively on the reproductive organs being
specially affected; but why these organs should be thus affected we do not
know, any more than {387} why unnatural conditions of life, though
compatible with health, should cause sterility; or why continued close
interbreeding, or the illegitimate unions of dimorphic and trimorphic
plants, induce the same result. The conclusion that the reproductive organs
alone are affected, and not the whole organisation, agrees perfectly with
the unimpaired or even increased capacity in hybrid plants for propagation
by buds; for this implies, according to our hypothesis, that the cells of
the hybrids throw off hybridised cell-gemmules, which become aggregated
into buds, but fail to become aggregated within the reproductive organs, so
as to form the sexual elements. In