interpreted in
corroboration of it. Selection can give rise only to what is of use to
the organism actually concerned, never to what is of use to some other
organism, and we must therefore expect to find that in flowers only
characters of use to _themselves_ have arisen, never characters which
are of use to insects only, and conversely that in the insects
characters useful to them and not merely to the plants would have
originated. For a long time it seemed as if an exception to this rule
existed in the case of the fertilisation of the yucca blossoms by a
little moth, _Pronuba yuccasella_. This little moth has a
sickle-shaped appendage to its mouth-parts which occurs hi no other
Lepidopteron, and which is used for pushing the yellow pollen into the
opening of the pistil, thus fertilising the flower. Thus it appears as
if a new structure, which is useful only to the plant, has arisen in
the insect. But the difficulty is solved as soon as we learn that the
moth lays its eggs in the fruit-buds of the Yucca, and that the
larvae, when they emerge, feed on the developing seeds. In effecting
the fertilisation of the flower the moth is at the same time making
provision for its own offspring, since it is only after fertilisation
that the seeds begin to develop. There is thus nothing to prevent our
referring this structural adaptation in _Pronuba yuccasella_ to
processes of selection, which have gradually transformed the maxillary
palps of the female into the sickle-shaped instrument for collecting
the pollen, and which have at the same time developed in the insect
the instinct to press the pollen into the pistil.

In this domain, then, the theory of selection finds nothing but
corroboration, and it would be impossible to substitute for it any
other explanation, which now that the facts are so well known, could
be regarded as a serious rival to it. That selection is a factor, and
a very powerful factor in the evolution of organisms, can no longer be
doubted. Even although we cannot bring forward formal proofs of it _in
detail_, cannot calculate definitely the size of the variations which
present themselves, and their selection-value, cannot, in short,
reduce the whole process to a mathematical formula, yet we must assume
selection, because it is the only possible explanation applicable to
whole classes of phenomena, and because, on the other hand, it is made
up of factors which we know can be proved actually to exist, and
which,