Given the facts of heredity and variability, the whole theory of organic evolution becomes neither more nor less than a theory of the causes which determine the breeding of like with like, to the exclusion of unlike. For the more firmly that we believe in heredity with variability as the fundamental principle of organic evolution, the stronger must become our persuasion that segregate breeding (or exclusive mating of like with like) must lead to divergence, while indiscriminate breeding (or free intercrossing of all varieties) must lead to uniformity. So long as there is free intercrossing, heredity makes in favor of fixity of type—or, at most, can permit change only in a single line, where successive generations undergo a continuous improvement, which may give rise to a ladder-like series of species in time. But in order that there should be a tree-like multiplication of species in space, or a simultaneous divergence of type, it is essential that free intercrossing be prevented at the origin, and throughout the development, of each branch. In other words, it is only when assisted by some form of segregation—which determines exclusive breeding of like with like—that heredity can effect arborescent or polytypic, as distinguished from catenated or monotypic, evolution. For the sake of greater clearness, I will call segregation in this sense _homogamy_, or the exclusive mating of individuals which belong to the same variety. Now homogamy may be secured in a very great number of different ways. Of these the most important, from every point of view, is natural selection. Here the exclusive breeding of like with like is determined by general fitness, and is effected by extermination of the unlike—i. e., the comparatively unfit. Moreover, this process leads to a continuous improvement in the way of adaptation, and in this important respect it stands alone among all the forms of homogamy. Nevertheless, we must note that, unless assisted by some other form of homogamy, natural selection can only produce monotypic evolution; never polytypic. Successive generations may thus continuously mount to higher stages of adaptation on the steps supplied by their own dead selves; but although they may thus give rise to a linear series of species in time, they can never thus give rise to a multiplication of species in space. In order to effect such multiplication, or _divergence_ of types, natural selection must be supplemented by some other form of homogamy, which can prevent intercrossing between the equally fit at the origin, and throughout the development, of every separate branch.