the retina may develop, and however complicated it may become, such
progress, instead of favoring vision, will probably hinder it if the
visual centres do not develop at the same time, as well as several parts
of the visual organ itself. If the variations are accidental, how can
they ever agree to arise in every part of the organ at the same time, in
such way that the organ will continue to perform its function? Darwin
quite understood this; it is one of the reasons why he regarded
variation as insensible.[29] For a difference which arises accidentally
at one point of the visual apparatus, if it be very slight, will not
hinder the functioning of the organ; and hence this first accidental
variation can, in a sense, _wait for_ complementary variations to
accumulate and raise vision to a higher degree of perfection. Granted;
but while the insensible variation does not hinder the functioning of
the eye, neither does it help it, so long as the variations that are
complementary do not occur. How, in that case, can the variation be
retained by natural selection? Unwittingly one will reason as if the
slight variation were a toothing stone set up by the organism and
reserved for a later construction. This hypothesis, so little
conformable to the Darwinian principle, is difficult enough to avoid
even in the case of an organ which has been developed along one single
main line of evolution, _e.g._ the vertebrate eye. But it is absolutely
forced upon us when we observe the likeness of structure of the
vertebrate eye and that of the molluscs. How could the same small
variations, incalculable in number, have ever occurred in the same
order on two independent lines of evolution, if they were purely
accidental? And how could they have been preserved by selection and
accumulated in both cases, the same in the same order, when each of
them, taken separately, was of no use?

Let us turn, then, to the hypothesis of sudden variations, and see
whether it will solve the problem. It certainly lessens the difficulty
on one point, but it makes it much worse on another. If the eye of the
mollusc and that of the vertebrate have both been raised to their
present form by a relatively small number of sudden leaps, I have less
difficulty in understanding the resemblance of the two organs than if
this resemblance were due to an incalculable number of infinitesimal
resemblances acquired successively: in both cases it is chance that
operates, but in the