ge quantities in the blood and
therefore might affect the germ-cells directly. If the children
subsequently born are consistently defective it is not an inheritance of
a body character but the result of a direct modification of the
germ-plasm. The inheritance of an acquired modification of the body can
only be proved if some particular change made in the parent is inherited
as such by the child.

(7) There is often a failure to distinguish between the possible
inheritance of a particular modification, and the possible inheritance
of indirect results of that modification, or of changes correlated with
it. This is a nice but crucial point on which most popular writers are
confused. Let us examine it through a hypothetical case. A woman, not
herself strong, bears a child that is weak. The woman then goes in for
athletics, in order better to fit herself for motherhood; she
specializes on tennis. After a few years she bears another child, which
is much stronger and better developed than the first. "Look," some one
will say, "how the mother has transmitted her acquirement to her
offspring." We grant that her improved general health will probably
result in a child that is better nourished than the first; but that is a
very different thing from heredity. If, however, the mother had played
tennis until her right arm was over-developed, and her spine bent; if
these characteristics were nowhere present in the ancestry and not seen
in the first child; but if the second child were born with a bent spine
and a right arm of exaggerated musculature, we would be willing to
consider the case on the basis of the inheritance of an acquired
character. We are not likely to have such a case presented to us.

To put the matter more generally, it is not enough to show that _some_
modification in the parent results in _some_ modification in the child.
For the purposes of this argument there must be a similar modification.

(8) Finally, data are frequently presented, which cover only two
generations--parent and child. Indeed, almost all the data alleged to
show the inheritance of acquired characteristics are of this kind. They
are of little or no value as evidence. Cases covering a number of
generations, where a _cumulative_ change was visible, would be of
weight, but on the rare occasions when they are forthcoming, they can be
explained in some other way more satisfactorily than by an appeal to the
theory of Lamarck.[13]

If the evidence cur