According to Dr. Romanes, no organ is so subject to variation as the reproductive, and this in no respect so much as in degrees and kinds of fertility--we might almost say so subject to freaks of cross-sterility. Now, suppose we start with any well-defined species in a state of nature. With every generation there are many slightly divergent individual varieties, some greater and some less; but these are all immediately swamped by crossing with one another and with the parent stock, and the species remains unchanged. But suppose among these divergent variations there arise, from time to time, some which affect the reproductive organs in such wise that the variety, though perfectly fertile with its own kind, is infertile, or imperfectly fertile, with other varieties, and especially with the parent stock. The change may be only in the _time_ of flowering in plants, or season of heat in animals, or it may be actual infertility in sexual union. Right here we have the beginning of a new species. The variety is sexually isolated from the parent stock by cross-sterility, and therefore all its peculiarities, however trivial, are preserved by true breeding. Cross-breeding is necessary to make species, but true breeding preserves them. Cross-breeding tends ever to make varieties, but immediately destroys them again. This constant forming and swamping, separating and again merging of varieties, like mixing of dough, makes the whole mass (stock) more and more plastic and subject to variety. This plasticity finally gives rise to varieties of the kind which produces species by sexual isolation. By continued merging the centrifugal forces continually increase, but are continually repressed by crossing, until finally varieties break away to form species.
Now it is easy to see, from this point of view, why artificial varieties are cross-fertile. It is because in artificial breeding we are intent only on making varieties in form, size, color, etc., and not at all on making any characterized by cross-sterility with the parent stock. Cross-sterility with the parent stock, or with other varieties, would be of no advantage, because we control the breeding, and can breed true if we desire. Sexual isolation is not necessary, because we can use physical isolation. On the contrary, such cross-sterility would be a positive disadvantage to the breeder, by limiting the range of his experiments just where they would be most prolific in making new varieties. Hence, as might be expected, all domestic varieties are cross-fertile, unless it be the extreme varieties, which may, in some instances, have passed the limit of greatest fertility.