There is no sacrum, that is, there were no united vertebrae posteriorly for the attachment and support of the pelvis, as no such support was needed. In only one other group of aquatic reptiles was the sacrum lost, though it has wholly disappeared in the cetaceans and sirenians among mammals. The chevron bones of the tail, usually bony arches on the under side of the tail for the protection of the blood-vessels, in crawling reptiles, were very imperfectly developed in the later forms, though normal in shape in the early ones. The ribs are numerous, long, and slender, very much resembling those of the fish-eating dolphins. They usually had, however, two attachments to the body of the vertebra and none to the arch, differing in this respect from all other animals. Of the shoulder bones, the scapula or shoulder-blade, as usual among water animals, is short and broad. In the place of a sternum the coracoids joined each other broadly in the middle, just as they did in the oldest known land reptiles. And there were clavicles and an interclavicle. Below the abdomen behind were numerous slender bones called ventral ribs. The pelvis is very weak, and was suspended below the spinal column in the fleshy walls of the abdomen. The hind legs were so small that little support was necessary for them, and, because they were not used either for the support of the body or for propulsion, they did not require a firm union with the skeleton. Doubtless had the ichthyosaurs continued to the present time, they would have lost entirely the hind legs, as have the cetaceans. [Illustration: FIG. 55.—Pectoral girdle of _Baptanodon (Ophthalmosaurus)_, an American Upper Jurassic ichthyosaur. (After Gilmore.)] It is in the limbs that most extraordinary differences from all other animals are seen. So great are these differences that it has been a puzzle to naturalists to understand how they could have arisen. In no other animals above the fishes, that is, in no other reptiles, in no amphibians, birds, or mammals, are there ever more than five fingers or five toes, the number with which air-breathing animals began. Fingers and toes may be lost and often are lost in all groups of life, until a single one in each limb may remain, as in the domestic horse. An increase of fingers and toes, however, seems to be an impossibility in evolution, and doubtless of real fingers and toes it is an equal impossibility. All naturalists are now agreed that a specialized character can never revert to a generalized condition, or rather to a generalized structure, that an organ once functionally lost can never be regained by descendants. A character once lost is lost forever; horses of the future can never have more than one finger or one toe in each limb. [Illustration: FIG. 56.—Front paddle of _Ophthalmosaurus_ (after Andrews): _h_, humerus; _r_, radius; _u_, ulna; _p_, pisiform; _re_, radiale; _int_, intermedium; _ue_, ulnare.]