The ossification of carpals is more likely to be complete in adult frogs than in salamanders, but some ossification of all parts named is found in several of the latter. A common ancestor of frogs and salamanders could be expected to have the following elements present and ossified in the adult: distal carpals 1-4 separate; 3 centralia; radiale, intermedium and ulnare separate. Comparison with fossils older than _Notobatrachus_ is fruitless on these points, unless we go back to forms too distant to have any special value, such as _Eryops_. This is because of inadequate preservation and because the elements are not fully ossified in many immature specimens. For the purpose of this review there is no special value in a comparison of the tarsi of frogs and salamanders, since the leaping adaptation of the former leaves very little common pattern between them. Even in _Protobatrachus_, where the legs were not yet conspicuously lengthened, the tibiale and fibulare ("astragalus" and "calcaneum" respectively) were already considerably elongated. The carpus and tarsus of _Amphibamus_ are as yet undecipherable. THE LARVA Considering the postembryonic developmental stages of modern Amphibia, there can be no doubt that a gill-bearing, four-legged larva of a salamander, in which lateral line pores and a gular fold are present, represents much more closely the type of larva found in labyrinthodonts than does the limbless, plant-nibbling tadpole of the Anura. Salamander-like larvae of labyrinthodonts are well known, especially those formerly supposed to comprise the order Branchiosauria. Many, perhaps the majority of, labyrinthodonts show some features associated with aquatic life even when full-grown, as do the lepospondyls. These features may include impressions of sensory canals on the dermal bones of the skull, persistence of visceral arches, reduction in size of appendages, and failure of tarsal and carpal elements to ossify. In fact, it appears that very few of the Paleozoic Amphibia were successful in establishing themselves as terrestrial animals even as adults. Nevertheless, in the ancestry of Anura, and that of at least the Hynobiid, Ambystomid, Salamandrid and Plethodontid salamanders, there must certainly have been a terrestrial adult, transforming from an aquatic larva. The leaping mechanism of Anura, shown in so many features of their anatomy, is perhaps to be explained as a device for sudden escape from land into the water, but it was not yet perfected in the Triassic _Protobatrachus_ or the Jurassic _Notobatrachus_.