The question of the origin of the feathers can be discussed only from the speculative point of view, as they are fully developed in the Archaeopteryx, and there is no approach toward them in any other living or fossil organism. But a long discussion of the problem has convinced scientific men that the feathers are evolved from the scales of the reptile ancestor. The analogy between the shedding of the coat in a snake and the moulting of a bird is not uninstructive. In both cases the outer skin or epidermis is shedding an old growth, to be replaced by a new one. The covering or horny part of the scale and the feather are alike growths from the epidermis, and the initial stages of the growth have certain analogies. But beyond this general conviction that the feather is a development of the scale, we cannot proceed with any confidence. Nor need we linger in attempting to trace the gradual modification of the skeleton, owing to the material change in habits. The horny beak and the reduction of the toes are features we have already encountered in the reptile, and the modification of the pelvis, breast-bone, and clavicle are a natural outcome of flight. In the Chalk period we find a large number of bird remains, of about thirty different species, and in some respects they resume the story of the evolution of the bird. They are widely removed from our modern types of birds, and still have teeth in the jaws. They are of two leading types, of which the Ichthyornis and Hesperornis are the standard specimens. The Ichthyornis was a small, tern-like bird with the power of flight strongly developed, as we may gather from the frame of its wings and the keel-shaped structure of its breast-bone. Its legs and feet were small and slender, and its long, slender jaws had about twenty teeth on each side at the bottom. No modern bird has teeth; though the fact that in some modern species we find the teeth appearing in a rudimentary form is another illustration of the law that animals tend to reproduce ancestral features in their development. A more reptilian character in the Ichthyornis group is the fact that, unlike any modern bird, but like their reptile ancestors, they had biconcave vertebrae. The brain was relatively poor. We are still dealing with a type intermediate in some respects between the reptile and the modern bird. The gannets, cormorants, and pelicans are believed to descend from some branch of this group.