Such a hypothesis as the foregoing satisfactorily explains the extraordinary mosaic shell of the leather-back, and is perhaps an acceptable explanation of the rather strange fact that the horny shields of turtles do not correspond with the bones below them, as might be expected. Unfortunately this hypothesis lacks sufficient proof. About the only evidence that is offered in its support is the existence of a row of bones along the middle line in the Cretaceous _Toxochelys_, and notably in _Archelon_, both aquatic forms. It is urged that these bones, the epineurals of Wieland, are really the remains of an external layer that persisted in these turtles. However, they might have been new ossifications, such as we know did occur in not a few of the land tortoises later over the tail and limbs. Aside from _Proganochelys_ in the vast interval of time from the Triassic to the Eocene no other evidence of such an external dermal layer has been discovered. The chief argument against such divergent ancestry of the turtles in two chief lines of descent is the fact that in its other structure _Dermochelys_ shows great resemblance to other sea-turtles of the Cretaceous times—so much resemblance that it seems impossible that the ancestors of the leather-back should have paralleled them in almost everything except the shell. On the other hand, those who disagree with this view believe that the modern leather-back is the descendant of such Cretaceous marine turtles as _Protostega_ or _Archelon_, some of which had lost nearly all of the costal plates and had the neurals and marginals reduced. It is urged that some of these early marine forms, after they had practically lost the ordinary bones of the carapace, for some reason or other found a bony shell again necessary for their welfare. Possibly they had become littoral in habit; possibly they again became subject to new and dangerous enemies in their unprotected condition, notwithstanding their great size; perhaps the zeuglodons were among their enemies. Now, as we have seen, an animal never takes a back track and recovers a thing it has once lost. It was impossible for the ancestors of the leather-back again to acquire an orthodox shell, and they forthwith proceeded to acquire quite another kind that would serve the same purpose. Possibly the truth lies between the variant views, in the theory recently expressed by Versluys: “The shell of tortoises and turtles is formed by a combination of two layers of dermal ossifications, a thecal layer and a more superficial epithecal layer, the latter generally represented by the marginals only. The leather-back is a member of the Cryptodira, and is allied to the other marine turtles. The problem of the origin of the aberrant shell of the leather-back seems to find its solution in the hypothesis that it is a secondary proliferation of the marginals and such other epithecal elements as were present in its thecophorous ancestors.”