Although he now represents that the _instability_ of germ-plasm is such that in _no_ case can amphimixis have had anything to do with the _origin_ of congenital variations, he continues to regard the _stability_ of germ-plasm sufficiently great to necessitate, in _all_ cases, the occurrence of amphimixis in order to promote the _development_ of congenital variations. In other words, notwithstanding that he now thinks all congenital variations must be begun by external conditions acting directly on an unstable germ-plasm, he also thinks that the _amount_ of variation thus produced is likely to be exceedingly minute, and must therefore be increased by subsequent amphimixis in order to fall within the range of natural selection. So that, although powerless to initiate congenital variation, amphimixis must still play an indispensable part in the process of evolution, as in all cases a necessary condition to the occurrence of natural selection. External conditions first cause slight changes in the determinants of a species; but these are so slight that they have to be augmented by amphimixis before they constitute material on which natural selection can act, and hence before they can become of any significance either in ontogeny or phylogeny. Such, I take it, is what Professor Weismann would now have us to understand; for otherwise I should have expected from him as frank a surrender of his theory of evolution (or the remnant thereof in his theory of variation) as he has made of its fundamental postulate. But, if such is his meaning, I may mention the reasons which appear to me to render it nugatory. In the first place, it is evident that in thus minimizing the possible range of congenital variation due to the action of external conditions on a non-absolutely stable substance of heredity, Weismann is making a wholly gratuitous assumption, for the sole purpose of saving what remains of his theory of evolution—i. e., the doctrine of the immense importance of amphimixis.