-forms, on their actual amount of difference,
and on the nature of the conditions of life to which the crossed offspring
are exposed. But we must be careful not to confound these cases of
reversion to characters gained from a cross, with those given under the
first class, in which characters originally common to _both_ parents, but
lost at some former period, reappear; for such characters may recur after
an almost indefinite number of generations.

{36}

The law of reversion is equally powerful with hybrids, when they are
sufficiently fertile to breed together, or when they are repeatedly crossed
with either pure parent-form, as with mongrels. It is not necessary to give
instances, for in the case of plants almost every one who has worked on
this subject from the time of Koelreuter to the present day has insisted on
this tendency. Gaertner has recorded some good instances; but no one has
given more striking cases than Naudin.[83] The tendency differs in degree
or strength in different groups, and partly depends, as we shall presently
see, on the fact of the parent-plants having been long cultivated. Although
the tendency to reversion is extremely general with nearly all mongrels and
hybrids, it cannot be considered as invariably characteristic of them;
there is, also, reason to believe that it may be mastered by long-continued
selection; but these subjects will more properly be discussed in a future
chapter on Crossing. From what we see of the power and scope of reversion,
both in pure races and when varieties or species are crossed, we may infer
that characters of almost every kind are capable of reappearance after
having been lost for a great length of time. But it does not follow from
this that in each particular case certain characters will reappear: for
instance, this will not occur when a race is crossed with another endowed
with prepotency of transmission. In some few cases the power of reversion
wholly fails, without our being able to assign any cause for the failure:
thus it has been stated that in a French family in which 85 out of above
600 members, during six generations, had been subject to night-blindness,
"there has not been a single example of this affection in the children of
parents who were themselves free from it."[84]

* * * * *

_Reversion through Bud-propagation--Partial Reversion, by segments in the
same flower or fruit, or in different parts of the {37} body in t