ontinued within a period not excessively remote, we might, as a
general rule, still expect to find more variability in such parts than
in other parts of the organisation which have remained for a much longer
period nearly constant. And this, I am convinced, is the case. That the
struggle between natural selection on the one hand, and the tendency to
reversion and variability on the other hand, will in the course of
time cease; and that the most abnormally developed organs may be made
constant, I see no reason to doubt. Hence, when an organ, however
abnormal it may be, has been transmitted in approximately the same
condition to many modified descendants, as in the case of the wing of
the bat, it must have existed, according to our theory, for an immense
period in nearly the same state; and thus it has come not to be more
variable than any other structure. It is only in those cases in which
the modification has been comparatively recent and extraordinarily great
that we ought to find the GENERATIVE VARIABILITY, as it may be called,
still present in a high degree. For in this case the variability
will seldom as yet have been fixed by the continued selection of the
individuals varying in the required manner and degree, and by the
continued rejection of those tending to revert to a former and less
modified condition.

SPECIFIC CHARACTERS MORE VARIABLE THAN GENERIC CHARACTERS.

The principle discussed under the last heading may be applied to our
present subject. It is notorious that specific characters are more
variable than generic. To explain by a simple example what is meant: if
in a large genus of plants some species had blue flowers and some had
red, the colour would be only a specific character, and no one would be
surprised at one of the blue species varying into red, or conversely;
but if all the species had blue flowers, the colour would become
a generic character, and its variation would be a more unusual
circumstance. I have chosen this example because the explanation which
most naturalists would advance is not here applicable, namely, that
specific characters are more variable than generic, because they are
taken from parts of less physiological importance than those commonly
used for classing genera. I believe this explanation is partly, yet only
indirectly, true; I shall, however, have to return to this point in
the chapter on Classification. It would be almost superfluous to adduce
evidence in support of the s