No better test case of this kind has ever been furnished than that of Mr. Gulick's land-shells, which Mr. Wallace is specially considering in the part of his book where the sentence above quoted occurs. How, then, does he meet this case? He meets it by assuming that in all the numerous adjacent valleys of a small island there must be as many differences of environment, each of which is competent to induce slight varietal changes on the part of its occupants by way of natural selection, although in no one case can the utility of these slight changes be surmised. Now, against this explanation there are three overwhelming considerations. In the first place, it is purely gratuitous, or offered merely in order to save the hypothesis that there _can_ be no other cause of even the most trivial change in species than that which is furnished by natural selection. In the second place, as Mr. Gulick writes to me in a private letter, "if the divergence of Sandwich Island land molluscs is wholly due to exposure to different environments, as Mr. Wallace argues on pages 147-150, then there must be completely occult influences in the environment that vary progressively with each successive mile. This is so violent an assumption that it throws doubt on any theory that requires such support." In the third place, the assumption that the changes in question must have been due to natural selection, is wholly incompatible with the facts of isolation elsewhere--namely, in those cases where (as in that of Ireland) a large section of species, instead of a small section, has been indiscriminately isolated. Mr. Wallace, as we have seen, inadvertently alludes to these "many other cases of isolation" as evidence against apogamy being _per se_ a cause of specific change. But although, for the reason above stated, they are without relevancy in this respect, they appear to me fatal to the explanation which he gives of specific changes under apogamy where only small sections of species are concerned. For example, can it be rationally maintained that there are more differences of environment between every two of the many contiguous valleys of a small island, such as Mr. Gulick describes, than there are in the incomparably larger area of the whole of Ireland? But, if not, and if natural selection is able to work such "occult" wonders in each successive mile on the Sandwich Islands, why has it so entirely lost this magic power in the case of Ireland--or in the "many other cases of isolation" to which Mr. Wallace refers? On his theory there is no coherent answer to be given to this question, while on our theory the answer is given in the very terms of the theory itself. The facts are plainly just what the theory requires that they should be; and therefore, if they were not as they are, the theory would be deprived of that confirmation which it now derives from them.