toxin
on the germ. The lesion, however well localized it seems, is transmitted
by the same process as, for instance, the taint of alcoholism. But may
it not be the same in the case of every acquired peculiarity that has
become hereditary?

There is, indeed, one point on which both those who affirm and those who
deny the transmissibility of acquired characters are agreed, namely,
that certain influences, such as that of alcohol, can affect at the same
time both the living being and the germ-plasm it contains. In such case,
there is inheritance of a defect, and the result is _as if_ the soma of
the parent had acted on the germ-plasm, although in reality soma and
plasma have simply both suffered the action of the same cause. Now,
suppose that the soma can influence the germ-plasm, as those believe who
hold that acquired characters are transmissible. Is not the most natural
hypothesis to suppose that things happen in this second case as in the
first, and that the direct effect of the influence of the soma is a
_general_ alteration of the germ-plasm? If this is the case, it is by
exception, and in some sort by accident, that the modification of the
descendant is the same as that of the parent. It is like the
hereditability of the alcoholic taint: it passes from father to
children, but it may take a different form in each child, and in none of
them be like what it was in the father. Let the letter C represent the
change in the plasm, C being either positive or negative, that is to
say, showing either the gain or loss of certain substances. The effect
will not be an exact reproduction of the cause, nor will the change in
the germ-plasm, provoked by a certain modification of a certain part of
the soma, determine a similar modification of the corresponding part of
the new organism in process of formation, unless all the other nascent
parts of this organism enjoy a kind of immunity as regards C: the same
part will then undergo alteration in the new organism, because it
happens that the development of this part is alone subject to the new
influence. And, even then, the part might be altered in an entirely
different way from that in which the corresponding part was altered in
the generating organism.

We should propose, then, to introduce a distinction between the
hereditability of _deviation_ and that of _character_. An individual
which acquires a new character thereby _deviates_ from the form it
previously had, which form the g