d to size and quality seems to me the inevitable consequence of
their unequal nutrition; for although the germ-cell as a whole usually
receives sufficient nutriment, minute fluctuations in the amount
carried to different parts within the germ-plasm cannot fail to occur.

Now, if a determinant, for instance of a sensory cell, receives for a
considerable time more abundant nutriment than before, it will grow
more rapidly--become bigger, and divide more quickly, and, later, when
the id concerned develops into an embryo, this sensory cell will
become stronger than in the parents, possibly even twice as strong.
This is an instance of a _hereditary individual variation_, arising
from the germ.

The nutritive stream which, according to our hypothesis, favours the
determinant _N_ by chance, that is, for reasons unknown to us, may
remain strong for a considerable time, or may decrease again; but even
in the latter case it is conceivable that the ascending movement of
the determinant may continue, because the strengthened determinant now
_actively_ nourishes itself more abundantly,--that is to say, it
attracts the nutriment to itself, and to a certain extent withdraws it
from its fellow-determinants. In this way, it may--as it seems to
me--get into _permanent upward movement, and attain a degree of
strength from which there is no falling back_. Then positive or
negative selection sets in, favouring the variations which are
advantageous, setting aside those which are disadvantageous.

In a similar manner a _downward_ variation of the determinants may
take place, if its progress be started by a diminished flow of
nutriment. The determinants which are weakened by this diminished flow
will have less affinity for attracting nutriment because of their
diminished strength, and they will assimilate more feebly and grow
more slowly, unless chance streams of nutriment help them to recover
themselves. But, as will presently be shown, a change of direction
cannot take place at _every_ stage of the degenerative process. If a
certain critical stage of downward progress be passed, even favourable
conditions of food-supply will no longer suffice permanently to change
the direction of the variation. Only two cases are conceivable; if the
determinant corresponds to a _useful_ organ, only its removal can
bring back the germ-plasm to its former level; therefore personal
selection removes the id in question, with its determinants, from the
germ-pl