importance of experiments in comparison with reasoning, either inductive
or deductive. Bateson, however, has admitted that Mendelian experiments
and observations on mutation have not solved the problem of adaptation. It
seems to be demanded, nevertheless, that characters must be produced
experimentally and then inherited before the hereditary influence of
external stimuli can be accepted. Kammerer's experiments in this direction
have been sceptically criticised, and it must be granted that the evidence
he has published is not sufficient to produce complete conviction. But
experiments of this kind are from the nature of the case difficult if not
impossible. There is, however, another method--namely, to take a character
which is certainly to some extent hereditary, and then to ascertain by
experiment if it is 'acquired.' If it be proved that a hereditary
character was originally somatogenic, it follows that somatogenic
characters in time become hereditary. This is the reasoning I have used in
reference to my experiments on the production of pigment on the lower
sides of Flat-fishes, and I obtained similar evidence with regard to the
excessive growth of the tail feathers in the Japanese Tosa-fowls,
[Footnote: 'Observations and Experiments on Japanese Long-tailed Fowls,'
_Proc. Zool. Soc._, 1903.] which is a modification of a secondary sexual
character. In these fowls the feathers of the tail in the hens are only
slightly lengthened.

I learned from Mr. John Sparks, who himself brought specimens of the breed
from Japan, that the Japanese not only keep the birds separately on high
perches in special cages, but pull the tail feathers gently every morning
in order to cause them to grow longer. One question which I had to
investigate on my specimens, hatched from eggs obtained from Mr. Sparks,
was the relation of the growth of the feathers to the moult which occurs
in ordinary birds. My experiment consisted in keeping two cocks, A and B,
the first of which was left to itself, while in the second the feathers