Hitherto very few experiments have been made on these comparative degrees of fertility, yet it is by such alone, it seems to me, that physiological selections can be tested. Thus, _e.g._, my point about the 'interlocking' species (p. 392) is that in such cases I should expect a higher degree of fertility in A × A and B × B than crosswise. Indeed, my fear is that when I shall have proved by experiment that such is the general rule in such cases, naturalists will turn round and say: 'Well, of course, on merely _a priori_ grounds you might have known that such _must_ have been the case; for otherwise the two interlocking species could never have existed as separate species, they would have hybridised freely along the whole frontier line and eventually blended over the whole area.' And still more may this be said in the case of allied species, not merely interlocking, but intermixed through common areas. Therefore, as a believing F.R.S. said to me the other day, 'Your letters in "Nature" will at least have the effect of blunting the edge of such possible criticism in the future.' Of course you will laugh at the robustness of my faith in thus forecasting the line of future opposition, but I would like to ask you this much: Supposing, for the sake of argument, that twenty years hence I publish one hundred instances of allied species which grow intermixed in common areas, proving by experiment that in all the cases there is _some_ comparative degree of sterility between them (if only due to pre-potency of their own pollen), would you regard this as making in favour of physiological selection? Or are you already prepared to admit that such _must_ be the case, since otherwise the species A and B could not exist without fusion into one? If you say that you are prepared to admit this, it seems to me that you have already accepted the theory of physiological selection on _a priori_ grounds. Again, if I should publish one hundred other instances of allied species topographically isolated from one another, all of which were proved by experiment to present no degree at all of mutual infertility (so that A × A and B × B are not more fertile than crosswise), would you allow that, taken in conjunction with the previous set of experiments, these finally prove the theory of physiological selection to be true? If not, I do not see how it is possible to verify the theory at all: it is only by means of these two complementary lines of research that, as it seems to me, the theory can be experimentally tested.