undergo equal retrogression. Various causes prevent the retrogression from being everywhere equal, since the conditions under which retrogression takes place cannot be an exact reversal of those under which evolution occurred; for instance, the retrogressing wing of the apteryx is attached to a body which has not undergone a pari passu retrogression. Moreover, even while the wing was undergoing evolution, parts of it must have undergone retrogression; this and that structure or part of a structure in it, useless in a wing, but useful in the organ which preceded it, must have tended to disappear at the same time that other structures or parts of structures, more useful in a wing than in the organ that preceded it, tended to increase; the two warring forces of natural selection and atavism thus operating to bring the wing and the organism to which it belongs into completer harmony with the environment. Thus all the parts of the degenerated wing of an apteryx cannot have undergone equal retrogression, and therefore the resemblance between the degenerated wing and the ancestral form in any of its stages cannot be very close.
It follows, if the above theory be correct, that while it is not possible by means of selection, natural or artificial, to bring about rapid and extensive evolution, since such evolution must soon be checked by the increasing tendency towards far-reaching reversion, it is possible by means of selection to bring about rapid and extensive, indeed unlimited, retrogression. This hypothesis, however, is at variance with accepted doctrines. Mr. Herbert Spencer says—
