If we look into the matter more carefully, we see that it is not strictly true to say that germinal selection alone brings about the extinction of a species by cumulative augmentation of structures which are already excessive; _it is the incapacity of personal selection to keep pace with the more rapid changes in the conditions of life and to reduce excessive developments to any considerable extent in a short time_. This would always be possible in a long time, for the determinants of the excessive organ _E_ can never be equally strong in all the ids; they always fluctuate about a mean, however high this mean may be. Here again it must still be possible that reducing-divisions and amphimixis may lead to the formation of majorities of ids with weaker _E_-determinants, and if sufficient time be allowed, artificial selection could, by consistently selecting the individuals with, let us say, weaker antlers, give rise to a descending variation-movement. There are no variation-movements which cannot be checked; every direction can be reversed, but time and something to take hold of must be granted. That was wanting in the case of the giant stag, for it would not have been saved even if its antlers had at once become a couple of feet shorter, and germinal selection can hardly make so much difference as that. Analogous to hereditary deformities, and of special interest in connexion with the processes within the germ-plasm, are '_sports_' variations of considerable magnitude which suddenly appear without our being able to see any definite external reason for them. I have already discussed these in detail in my _Germ-plasm_, and have shown how simply these apparently capricious phenomena of heredity can be understood _in principle_ from the standpoint of the germ-plasm theory.