Man isolates varieties, and breeds from them, and continuing to separate those that vary in the direction he wishes to follow, a very great difference is, in a comparatively short time, produced. But these artificial varieties, though often more different from each other than some natural species, readily interbreed, and if left to themselves rapidly revert to a common type. In natural selection there is a great and fundamental difference. The varieties that arise can seldom be separated from the parent form and from other varieties until they vary also in the elements of reproduction. Thousands of varieties probably revert to the parent type, but if at last one is produced that breeds only with its own form, we can easily see how a new species might be segregated. As long as varieties interbreed together and with the parent form, it does not seem possible that a new species could be formed by natural selection, excepting in cases of geographical isolation. All the individuals might vary in some one direction, but they could not split up into distinct species whilst they occupied the same area and interbred without difficulty. Before a variety can become permanent, it must be either separated from the others or have acquired some disinclination or inability to interbreed with them. So long as they interbreed together, the possible divergence is kept within narrow limits, but whenever a variety is produced, the individuals of which have a partiality for interbreeding, and some amount of sterility when crossed with the parent form, the tie that bound it to the central stock is loosened, and the foundation is laid for the formation of a new species. Further divergence would be unchecked, or only slightly checked, and the elements of reproduction having begun to vary, would probably continue to diverge from the parent form, for Darwin has shown that any organ in which a species has begun to vary is liable to further change in the same direction.* (* "See Animals and Plants under Domestication" volume 2 page 241.) Thus one of the best tests of the specific difference of two allied forms living together is their sterility when crossed, and nearly allied species separated by geographical barriers are more likely to interbreed than those inhabiting the same area. Artificial selection is more rapid in its results, but less stable than that of nature, because the barriers that man raises to prevent intermingling of varieties are temporary and partial, whilst that which nature fixes when sterility arises is permanent and complete.