ancestry, and therefore in cases in which evolution has been rapid, as with race-horses, this reversion produces wide divergences of structure from the parent, but in cases in which the evolution has been slow, as with lamp-shells, this reversion produces very slight divergences of structure. For instance, to take an extreme example, that amount of reversion which in the offspring of a man would result in a fish, would in the offspring of a lamp-shell result in another lamp-shell, only a little modified. For this reason it is that characters long present in a race are, as is well known, little capable of undergoing retrogression. Since racehorses have been rapidly evolved under a process of extremely stringent selection, reversion of the offspring to a comparatively recent ancestor causes a considerable amount of divergence from the parent. Let us try to imagine what would happen among race-horses if the stringency of selection were relaxed, or rather if selection in this case were altogether abolished, so that the inferior animals were allowed to propagate the species as well as the superior. I think we may prophesy with tolerable confidence. The selective action being withdrawn, the tendency towards reversion, towards atavism, would operate uncontrolled, and the race would rapidly revert to its ancestor, the ordinary horse.
We are now in a position to understand why evolution caused by natural selection is so extremely slow. In every species natural selection as a cause of evolution, and atavism as a cause of retrogression, are constantly at war. If the one force predominate we have evolution; if the other, retrogression; if the two forces balance one another, the species in which this occurs undergoes for the time neither evolution nor retrogression. Rapid evolution is accompanied by an increasing tendency to retrogression, because in such a case any
