t all habitual actions
based on memory. Disturb the normal order and the performance is
arrested. The better we know "God save the Queen," the less easily can
we play or sing it backwards. The return of memory again depends on the
return of ideas associated with the particular thing that is
remembered--we remember nothing but for the presence of these, and when
enough of these are presented to us we remember everything. So, if the
development of an embryo is due to memory, we should suppose the memory
of the impregnate ovum to revert not to yesterday, when it was in the
persons of its parents, but to the last occasion on which it was an
impregnate ovum. The return of the old environment and the presence of
old associations would at once involve recollection of the course that
should be next taken, and the same should happen throughout the whole
course of development. The actual course of development presents
precisely the phenomena agreeable with this. For fuller treatment of
this point I must refer the reader to the chapter on the abeyance of
memory in my book "Life and Habit," already referred to.

Secondly, we remember best our last few performances of any given kind,
so our present performance will probably resemble some one or other of
these; we remember our earlier performances by way of residuum only, but
every now and then we revert to an earlier habit. This feature of memory
is manifested in heredity by the way in which offspring commonly
resembles most its nearer ancestors, but sometimes reverts to earlier
ones. Brothers and sisters, each as it were giving their own version of
the same story, but in different words, should generally resemble each
other more closely than more distant relations. And this is what
actually we find.

Thirdly, the introduction of slightly new elements into a method already
established varies it beneficially; the new is soon fused with the old,
and the monotony ceases to be oppressive. But if the new be too foreign,
we cannot fuse the old and the new--nature seeming to hate equally too
wide a deviation from ordinary practice and none at all. This fact
reappears in heredity as the beneficial effects of occasional crossing on
the one hand, and on the other, in the generally observed sterility of
hybrids. If heredity be an affair of memory, how can an embryo, say of a
mule, be expected to build up a mule on the strength of but two
mule-memories? Hybridism causes a fault i