striped--bearing also in mind that the
act of crossing certainly causes the reappearance of long-lost
characters--it is a more probable view that the above-specified stripes
are due to reversion, not to the immediate wild parent-horse, but to
the striped progenitor of the whole genus.

I have discussed this subject of analogous variation at considerable
length, because, in a future work on natural species, it will be shown that
the varieties of one species frequently mock distinct species--a fact in
perfect harmony with the foregoing cases, and explicable only on the theory
of descent. Secondly, because these facts are important from showing, as
remarked in a former chapter, that each trifling variation is governed by
law, and is determined in a much higher degree by the nature of the
organisation, than by the nature of the conditions to which the varying
being has been exposed. Thirdly, because these facts are to a certain
extent related to a more general law, namely, that which Mr. B. D.
Walsh[873] has called the "Law of _Equable Variability_," or, as he
explains it, "if any given character is very variable in one species of a
group, it will tend to be variable in allied species; and if any given
character is perfectly constant in one species of a group, it will tend to
be constant in allied species."

This leads me to recall a discussion in the chapter on Selection, in which
it was shown that with domestic races, which are {352} now undergoing rapid
improvement, those parts or characters which are the most valued vary the
most. This naturally follows from recently selected characters continually
tending to revert to their former less improved standard, and from their
being still acted on by the same agencies, whatever these may be, which
first caused the characters in question to vary. The same principle is
applicable to natural species, for, as stated in my 'Origin of Species,'
generic characters are less variable than specific characters; and the
latter are those which have been modified by variation and natural
selection, since the period when all the species belonging to the same
genus branched off from a common progenitor, whilst generic characters are
those which have remained unaltered from a much more remote epoch, and
accordingly are now less variable. This statement makes a near approach to
Mr. Walsh's law of Equable Variability. Secondary sexual characters, it may
be added, rarely serve to