nd
with respect to mere cohesion there is probably always some degree of
fusion, at least near the surface. When two embryos during their early
development come into close contact, as both include corresponding
gemmules, which must be in all respects almost identical in nature, it is
not surprising that some derived from one embryo and some from the other
should unite at the point of contact with a single nascent cell or
aggregate of cells, and thus give rise to a single part or organ. For
instance, two embryos might thus come to have on their {394} adjoining
sides a single symmetrical arm, which in one sense will have been formed by
the fusion of the bones, muscles, &c., belonging to the arms of both
embryos. In the case of the fish described by Lereboullet, in which a
double head was seen gradually to fuse into a single one, the same process
must have taken place, together with the absorption of all the parts which
had been already formed. These cases are exactly the reverse of those in
which a part is doubled either spontaneously or after an injury; for in the
case of doubling, the superabundant gemmules of the same part are
separately developed in union with adjoining points; whilst in the case of
fusion the gemmules derived from two homologous parts become mingled and
form a single part; or it may be that the gemmules from one of two
adjoining embryos alone become developed.

* * * * *

Variability often depends, as I have attempted to show, on the reproductive
organs being injuriously affected by changed conditions; and in this case
the gemmules derived from the various parts of the body are probably
aggregated in an irregular manner, some superfluous and others deficient.
Whether a superabundance of gemmules, together with fusion during
development, would lead to the increased size of any part cannot be told;
but we can see that their partial deficiency, without necessarily leading
to the entire abortion of the part, might cause considerable modifications;
for in the same manner as a plant, if its own pollen be excluded, is easily
hybridised, so, in the case of a cell, if the properly succeeding gemmules
were absent, it would probably combine easily with other and allied
gemmules. We see this in the case of imperfect nails growing on the stumps
of amputated fingers,[924] for the gemmules of the nails have manifestly
been developed at the nearest point.

In variations caused by the