r insects, in which exactly one half or one quarter of
the body resembles the male, and the other half or three quarters the
female; and when this occurs the opposite sides of the body, separated from
each other by a distinct line, sometimes differ in the most conspicuous
manner. Again, these same principles apply to the cases given in the
thirteenth chapter, in which the right and left sides of the body differ to
an extraordinary degree, as in the spiral winding of certain shells, and as
in the genus Verruca among cirripedes; for in these cases it is known that
either side indifferently may undergo the same remarkable change of
development.

Reversion, in the ordinary sense of the word, comes into action so
incessantly, that it evidently forms an essential part of the general law
of inheritance. It occurs with beings, however propagated, whether by buds
or seminal generation, and sometimes may even be observed in the same
individual as it advances in age. The tendency to reversion is often
induced by a change of conditions, and in the plainest manner by the act of
crossing. Crossed forms are generally at first nearly intermediate in
character between their two parents; but in the next generation the
offspring generally revert to one or both of their grandparents, and
occasionally to more remote ancestors. How can we account for these facts?
Each organic unit in a hybrid must throw off, according to the doctrine of
pangenesis, an abundance of hybridised gemmules, for crossed plants can be
readily and largely propagated by buds; but by the same hypothesis there
will likewise be present dormant gemmules derived from both pure
parent-forms; and as these latter retain their normal condition, they
would, it is probable, be enabled to multiply largely during the lifetime
of each hybrid. Consequently the sexual elements of a hybrid will include
both pure and hybridised gemmules; and when two hybrids pair, the
combination of pure gemmules derived from the one hybrid with the pure
gemmules of the same parts derived from the other would {401} necessarily
lead to complete reversion of character; and it is, perhaps, not too bold a
supposition that unmodified and undeteriorated gemmules of the same nature
would be especially apt to combine. Pure gemmules in combination with
hybridised gemmules would lead to partial reversion. And lastly, hybridised
gemmules derived from both parent-hybrids would simply reproduce the
original hybrid f