For this reason I am disposed to attribute to saltatory variations a considerable share in the production of distinctive sexual characters. From saltatory variations in flowers, fruits, and leaves we know that these may be conspicuous enough even on their first appearance, and so we are justified in finding in such variations the first beginnings of many of the decorative distinguishing characters which occur in the males of so many animals, especially butterflies and birds. As soon as it is admitted that variations of considerable amount, which have been slowly prepared in the germ-plasm by means of germinal selection, can suddenly attain to expression, one of the objections against sexual selection is disposed of, for conspicuous variations are necessary for the operation of this kind of selection, since the changes in question must attract the attention of the females if they are to be preferred. Without such preference, even though it be not quite strict and consistent, a long-continued augmentation of the decorative characters is inconceivable. But as intra-germinal disturbances of the position of equilibrium in the determinant system is at the root of the saltatory variations of our cultivated plants, it must also have played a large share in the evolution of breeds among our domesticated animals, which is therefore by no means wholly due to artificial selection operating upon the variation of individual characters. In all breeds in the formation of which the production of more than a single definite character was concerned, as, for instance, in the broad-nosed breeds of dog--bull-dog and pug-dog--we may refer the peculiar variation of many parts to disturbances of the equilibrium of the determinant system, which bring to light, not suddenly as in the case of saltatory variations, but gradually and increasingly, the curious complex of characters. Darwin referred such transformations of the whole animal facies, where a single varying character is deliberately selected, to correlation, and by this he understood the mutual influence of the parts of an animal upon one another. Such correlation certainly exists, as we have already seen in discussing histonal selection, but here we have rather to do with the correlation of the parts of the germ-plasm, with the effects of germinal selection, which, affected by the artificial selection of particular characters, gradually brings about a more marked disturbance in the whole determinant system.