But in a domestic bird like the fantail where Natural Selection does not come into play, the tail-feathers could hardly be limited by "utility for flight," yet two more tail-feathers could certainly exist in a fancy breed if "utility for flight" were the only obstacle. It seems probable that the real barrier is an _internal_ one in the nature of the organism, and the existence of such is just what is contended for in this chapter. As to{118} the differences between domestic races being greater than those between species or even genera, that is not enough for the argument. For upon the theory of "Natural Selection" all birds have a common origin, from which they diverged by infinitesimal changes, so that we ought to meet with sufficient changes to warrant the belief that a hornbill could be produced from a humming-bird, proportionate time being allowed. But not only does it appear that there are barriers which oppose change in certain directions, but that there are positive tendencies to development along certain special lines. In a bird which has been kept and studied like the pigeon, it is difficult to believe that any remarkable spontaneous variations would pass unnoticed by breeders, or that they would fail to be attended to and developed by some one fancier or other. On the hypothesis of _indefinite_ variability, it is then hard to say why pigeons with bills like toucans, or with certain feathers lengthened like those of trogans, or those of birds of paradise, have never been produced. This, however, is a question which may be settled by experiment. Let a pigeon be bred with a bill like a toucan's, and with the two middle tail-feathers lengthened like those of the king bird of paradise, or even let individuals be produced which exhibit any marked tendency of the kind, and indefinite variability shall be at once conceded. As yet all the changes which have taken place in pigeons are of a few definite kinds only, such as may be well conceived to be compatible with a species possessed of a certain inherent capacity for considerable yet definite variation, a capacity for the ready production of certain degrees of abnormality, which then cannot be further increased.