ancestor reverted to differs more in structure from the last member of the race than a corresponding ancestor in a race that has been more slowly evolved. I must, however, guard myself against possible misconception. When I speak of atavism as a cause of retrogression, I do not necessarily mean that in each case of atavism the whole organism in all its parts retrogresses. This must rarely happen, especially when evolution has been slow. What I do mean is that, generally speaking, atavism, by causing various small retrogressive changes in various parts of the organism, is by its cumulative effects a check to evolution, and even sometimes a cause of retrogression. An organism may undergo retrogressive changes in some of its parts owing to the severity of natural selection becoming relaxed as regards them, at the same time that, owing to increased severity of natural selection, it is undergoing evolution as regards other parts. Thus in the case of a species of bird, the habits of which are becoming less and less aerial, natural selection will cause the legs to undergo evolution, but atavism the wings to undergo continued retrogression backwards, successively from recent ancestors to remoter ancestors, till, if the power of flight no longer influences the survival rate, there will be an approximation in structure, as far as the wings are concerned, to the structure of a remote wingless or rather limbless ancestor; but an approximation only, for, as regards such complex organs as wings, we must not expect to find that a degenerate example in any stage of the retrogression resembles very closely an ancestral wing, in a corresponding stage of the evolution; for instance, we must not suppose that the wing of the apteryx is a close reproduction of a remote ancestral form, for the reason that though the wing in all its parts undergoes retrogression, yet, like the parts of a complex organism, all the parts of the wing do not
