In Urodela it is invested by a dentary, splenial and articular bone. A splenial is absent in Anura, most of which have a pair of ossified lower labials ( = Mento-Meckelians)at the symphysis. The extent of the gape of the mouth depends on the inclination of the quadrate which is forwards and downwards in Proteus and the lower Urodela, backwards and outwards in Anura. The hyoid of Urodela consists of a ceratohyal and generally a hypohyal element: it is united to the quadrate and stapes by ligament. In Anura it generally resembles that of the Frog (p. 81), and is continuous with the floor of the tympanum. The Urodela have four branchial arches, the two posterior rudimentary and in adult Salamanders lost. There is in them a basi-hyo-branchial as well as a basi-branchial copula. The larval Anura have four branchial arches represented by four rudimentary cerato-branchial elements, but there are also four large extra-branchial cartilages united at their dorsal extremities and lying subcutaneously.
1 Cope states that the Permian Trimerorhachis has but one condyle (American Naturalist, xviii. 1884).
The skull in Gymnophiona is remarkable for the large size of its bones, and for the very complete case they make. Epicrium has a ring of orbital bones like the Stegocephali. The mandible articulates with the quadrate by a concave surface as in Fish. The skull of the extinct Stegocephali is similarly complete. It has remarkably large epiotics and paired supra-occipitals (not homologous probably with the supra-occipital generally so-called). There are prae- and post-frontals, post-orbitals, and a supra-temporal which lies externally to the squamosal. A jugal and quadrato-jugal are nearly always present. The dorsal surface of the skull is often polished and sculptured as in many Ganoid fish, and in this respect it is approached by a few living Anura, e.g. Pelobales1. Remains of two branchial arches have been discovered, and there is reason to believe that they persisted in some adults, e.g. Branchiosaurus, whilst they are lost in others, e.g. Melanerpeton.
The vertebral column contains a large number of vertebrae in Ichthy-oidea and Gymnophiona (250 or more); a more restricted number in Salamanders; and in living Anura only eight praesacral vertebrae and one sacral with a coccygeal style. It is divisible into a cervical, thoracic, sacral and caudal series in Urodela. A single more or less ring-like cervical vertebra devoid of ribs is found in all Amphibia. In most Urodela it carries a well-marked 'odontoid' process, traces of which are sometimes seen in Anura, and which points to the loss, partial or complete, of an anterior vertebra. The remaining vertebrae carry in Urodela double transverse processes, the upper, feebly marked in Gymnophiona, an outgrowth of the arch, the lower of the centrum. They become rudimentary in the tail. These processes are single in Anura. Articulating processes are well developed. The neural arches are in living Amphibia anchylosed to the centrum. The neural spines are almost obsolete in Anura, but in Urodela they are prominent, sometimes forked terminally and articulating one with another; or, in Salamanders, flattened sideways, lamelliform and comblike, a condition which obtains in some Stegocephali (Nectridea), where the inferior arches of the tail are similarly conformed.
These latter arches are present also in Urodela, and are homologous with the neural arches and similarly developed. The centra of the vertebrae are formed by ossification of the cellular sheath of the notochord. They constrict the notochord intra-vertebrally, while it is also constricted inter-vertebrally by a growth of cartilage. If the inter-vertebral cartilage is little developed the vertebrae are biconcave, as in Ichthyoidea or extinct Stegocephali, and some Salamanders. But in the majority of the latter the cartilage is large in amount and the inter-vertebral constriction great; and in the highest forms, e.g. Triton, is segmented,forming opisthocoelous vertebrae, whilst theintra-vertebral chorda is replaced by marrow. The inter-vertebral cartilage grows rapidly in Anura, and the chorda persists for a long time intra-vertebrally; the centra are as a rule procoelous, but in a few, e.g. Pipa, Discoglossus', opisthocoelous, or even variable, as in Pelobates. In these exceptional instances the centra are formed for the most part above the notochord, not round it, i.e. are epichordal as opposed to perichordal. The vertebral column terminates in many Urodela with an imperfectly segmented cartilaginous rod.
In some Stegocephali a vertebral centrum, devoid of neural arch and transverse processes, is intercalated between every two vertebrae formed on the ordinary type. In others the centra are stated to consist of a ventral piece (inter- or hypo-centrum) and two lateral pieces (pleuro-centrum or centrum proper). These two forms have been distinguished as embolomere and rhachitome. They occur, however, apparently in the vertebral column of one and the same animal, the former in the tail, the latter in the thorax, and therefore are not of classificatory value. And it is not certain that the hypo- and pleuro-centra belong to the same vertebra (Fritsch). The neural arches of some Stegocephali are complex (Branchio-saurtis). Ribs are present except on the cervical vertebrae and posterior caudal. In Anura they are very small, united to the transverse processes, and traces of a suture are rarely observable, e. g. Discoglossus. They are bifurcated proximally in Urodela, a process articulating with both upper and lower transverse processes, and are developed by the union of a short dorsal and longer ventral rod. They do not surround the body, and in the genus Spelerpes, among Salamandrina, are absent on two posterior thoracic vertebrae, hence truly lumbar.
The extreme shortness of the ribs in living Amphibia is perhaps due to atrophy. They are long in some Stegocephali, but there is no evidence of a ventral union.
1 The significance of the channels, sometimes termed 'mucous-canals,' on the face in certain genera of Stegocephali is not known.