greatly assisted by two other important factors in the problem. It has already been shown that, with each modification of form and habits, and especially with modifications of colour, there arises a disinclination of the two forms to pair together; and this would produce an amount of isolation which would greatly assist the specialisation of the forms in adaptation to their different conditions of life. Again, evidence has been adduced that change of conditions or of mode of life is a potent cause of disturbance of the reproductive system, and, consequently, of infertility. We may therefore assume that, as the two forms adopted more and more different modes of life, and perhaps acquired also decided peculiarities of form and coloration, the infertility between them would increase or become more general; and as we have seen that every such increase of infertility would give that portion of the species in which it arose an advantage over the remaining portions in which the two varieties were more fertile together, all this induced infertility would maintain itself, and still further increase the general infertility between the two forms of the species.
It follows, then, that specialisation to separate conditions of life, differentiation of external characters, disinclination to cross-unions, and the infertility of the hybrid produce of these unions, would all proceed pari passu, and would ultimately lead to the production of two distinct forms having all the characteristics, physiological as well as structural, of true species.
In the case now discussed it has been supposed, that some amount of general infertility might arise in correlation with the different modes of life of two varieties or incipient species. A considerable body of facts already adduced renders it probable that this is the mode in which any widespread infertility would arise; and, if so, it has been shown that, by the influence of natural selection and the known laws which affect varieties, the infertility would be gradually increased. But, if we suppose the infertility to arise sporadically within the two forms, and to affect only a small proportion of the individuals in any area, it will be difficult, if not impossible, to show that such infertility would have any tendency to increase, or would produce any but a
