because, as we have seen, these coloured marks are
eminently liable to appear in the crossed offspring of two distinct and
differently coloured breeds; and in this case there is nothing in the
external conditions of life to cause the reappearance of the slaty-blue,
with the several marks, beyond the influence of the mere act of crossing on
the laws of inheritance.

No doubt it is a very surprising fact that characters should reappear after
having been lost for many, perhaps for hundreds of generations. But when a
breed has been crossed only once by some other breed, the offspring
occasionally show a tendency to revert in character to the foreign breed
for many generations--some say, for a dozen or even a score of generations.
After twelve generations, the proportion of blood, to use a common
expression, of any one ancestor, is only 1 in 2048; and yet, as we see, it
is generally believed that a tendency to reversion is retained by this very
small proportion of foreign blood. In a breed which has not been crossed,
but in which _both_ parents have lost some character which their progenitor
possessed, the tendency, whether strong or weak, to reproduce the lost
character might be, as was formerly remarked, for all that we can see to
the contrary, transmitted for almost any number of generations. When a
character which has been lost in a breed, reappears after a great number of
generations, the most probable hypothesis is, not that the offspring
suddenly takes after an ancestor some hundred generations {161} distant,
but that in each successive generation there has been a tendency to
reproduce the character in question, which at last, under unknown
favourable conditions, gains an ascendancy. For instance, it is probable
that in each generation of the barb-pigeon, which produces most rarely a
blue and black-barred bird, there has been a tendency in each generation in
the plumage to assume this colour. This view is hypothetical, but could be
supported by some facts; and I can see no more abstract improbability in a
tendency to produce any character being inherited for an endless number of
generations, than in quite useless or rudimentary organs being, as we all
know them to be, thus inherited. Indeed, we may sometimes observe a mere
tendency to produce a rudiment inherited: for instance, in the common
snapdragon (Antirrhinum) a rudiment of a fifth stamen so often appears,
that this plant must have an inherited tendency to produc