The paper likewise indicates a point which, in studying Mr. Wallace's theory, I have missed. It will be remembered that the only apparent difference between his theory and mine has been shown to consist in this--that while I was satisfied to state, in a general way, that natural selection is probably able to increase a selective fertility which has already been begun by other causes, Mr. Wallace has sought to exhibit more in detail the precise conditions under which it can do so. Now, Mr. Gulick shows that the particular conditions which Mr. Wallace describes, even if they do serve to promote an increase of cross-infertility, are conditions which preclude the possibility of natural selection coming into play at all. So that if, under these particular conditions, a further increase of cross-infertility does take place, it does not take place in virtue of natural selection. To me it appears that this criticism is sound; and, if so, it disposes of even the one very subordinate addition to our theory which Mr. Wallace "claims" as the most "distinctive" part of his. The following is the criticism in question:-- On pages 173-186 Mr. Wallace maintains that "Natural selection is, in some probable cases at all events, able to accumulate variations in infertility between incipient species" (p. 174); but his reasoning does not seem to me conclusive. Even if we grant that the increase of this character [cross-infertility] occurs by the steps which he describes, _it is not a process of accumulation by natural selection_. In order to be a means of cumulative modification of varieties, races, or species, selection, whether artificial or adaptational [i.e. natural], must preserve certain forms of an intergenerating stock, to the exclusion of other forms of the same stock. Progressive change in the size of the occupants of a poultry-yard may be secured by raising only bantams the first, only common fowls the second, and only Shanghai fowls the third year; but this is not the form of selection that has produced the different races of fowls. So in nature, rats may drive out and supplant mice; but this kind of selection modifies neither rats nor mice. On the other hand, if certain variations of mice prevail over others, through their superior success in escaping their pursuers, then modification begins. Now, turning to page 175, we find that, in the illustrative case introduced by Mr. Wallace, the commencement of infertility between the incipient species is in the relations to each other of two portions of a species that are locally segregated from the rest of the species, and partially segregated from each other by different modes of life. These two local varieties, being by the terms of his supposition better adapted to the environment than the freely interbreeding forms in