feathers in the germ-plasm had already
struck out a path of ascending variation, and this movement was taken
advantage of by the breeder, who continually selected for reproduction
the individuals in which the ascending variation was most marked. For
all breeding depends upon the unconscious selection of germinal
variations.

Of course these germinal processes cannot be proved mathematically,
since we cannot actually see the play of forces of the passive
fluctuations and their causes. We cannot say how great these
fluctuations are, and how quickly or slowly, how regularly or
irregularly they change. Nor do we know how far a determinant must be
strengthened by the passive flow of the nutritive stream if it is to
be beyond the danger of unfavourable variations, or how far it must be
weakened passively before it loses the power of recovering itself by
its own strength. It is no more possible to bring forward actual
proofs in this case than it was in regard to the selection-value of
the initial stages of an adaptation. But if we consider that all
heritable variations must have their roots in the germ-plasm, and
further, that when personal selection does not intervene, that is to
say, in the case of parts which have become useless, a degeneration of
the part, and therefore also of its determinant must inevitably take
place; then we must conclude that processes such as I have assumed are
running their course within the germ-plasm, and we can do this with as
much certainty as we were able to infer, from the phenomena of
adaptation, the selection-value of their initial stages. The fact of
the degeneration of disused parts seems to me to afford irrefutable
proof that the fluctuations within the germ-plasm _are the real root
of all hereditary variation_, and the preliminary condition for the
occurrence of the Darwin-Wallace factor of selection. Germinal
selection supplies the stones out of which personal selection builds
her temples and palaces: _adaptations_. The importance for the theory
of the process of degeneration of disused parts cannot be
over-estimated, especially when it occurs in sterile animal forms,
where we are free from the doubt as to the alleged _Lamarckian factor_
which is apt to confuse our ideas in regard to other cases.

If we regard the variation of the many determinants concerned in the
transformation of the female into the sterile worker as having come
about through the gradual transformation of the ids i