utations" but yielded only negative results.

Thus we come to the conclusion that Darwin ("Origin of Species" (6th
edition), pages 176 et seq.) was right in regarding transformations as
taking place by minute steps, which, if useful, are augmented in
the course of innumerable generations, because their possessors more
frequently survive in the struggle for existence.

(b) SELECTION-VALUE OF THE INITIAL STEPS.

Is it possible that the significant deviations which we know as
"individual variations" can form the beginning of a process of
selection? Can they decide which is to perish and which to survive? To
use a phrase of Romanes, can they have SELECTION-VALUE?

Darwin himself answered this question, and brought together many
excellent examples to show that differences, apparently insignificant
because very small, might be of decisive importance for the life of the
possessor. But it is by no means enough to bring forward cases of this
kind, for the question is not merely whether finished adaptations have
selection-value, but whether the first beginnings of these, and whether
the small, I might almost say minimal increments, which have led up
from these beginnings to the perfect adaptation, have also had
selection-value. To this question even one who, like myself, has been
for many years a convinced adherent of the theory of selection, can only
reply: WE MUST ASSUME SO, BUT WE CANNOT PROVE IT IN ANY CASE. It is not
upon demonstrative evidence that we rely when we champion the doctrine
of selection as a scientific truth; we base our argument on quite other
grounds. Undoubtedly there are many apparently insignificant features,
which can nevertheless be shown to be adaptations--for instance, the
thickness of the basin-shaped shell of the limpets that live among the
breakers on the shore. There can be no doubt that the thickness of these
shells, combined with their flat form, protects the animals from the
force of the waves breaking upon them,--but how have they become so
thick? What proportion of thickness was sufficient to decide that of two
variants of a limpet one should survive, the other be eliminated? We can
say nothing more than that we infer from the present state of the shell,
that it must have varied in regard to differences in shell-thickness,
and that these differences must have had selection-value,--no proof
therefore, but an assumption which we must show to be convincing.

For a long time the marvellously co