Representatives of several genera of the modern Caudata have been dissected in order to make a direct comparison of the fossil alimentary canal with that of the recent forms. The alimentary tract of Desmognathus fuscus Raf. from the vicinity of Ithaca, New York, resembles in a marked degree that of the fossil form. The nearest approach to the condition there represented is found, however, in an immature branchiate individual, some 47 mm. in length, of Diemyctylits torosus Esch., from a fresh-water pond on Orcas Island in Puget Sound. The presence of this species on the island is very suggestive. It is of extreme interest, too, that the condition represented in the alimentary tract of the fossil branchiosaurs should resemble so closely that of an immature rather than a mature form.
(j) The vertebral column is clearly and readily separable into cervical, dorsal, sacral, and caudal regions. The neck is always short, with from 5 to 10 vertebræ, cervical ribs often present. The dorsal region is not long, but varies from 20 to 30 in the constituent vertebræ. There is a single sacral vertebra not always to be readily distinguished from those of the dorsal and anterior caudal series. The tail may be very short or extremely long, with neural and hæmal spines elongate and flattened into an oar-like appendage. The distal caudals are in some species cartilaginous, apparently always so in the Branchiosauria.
(k) The atlas and axis are unknown among the American specimens, but we are able to infer from the structure of the other vertebræ what this must have been; and our inferences are partly confirmed by the conditions existing in the European forms ([187]). The atlas, apparently, consisted of a pair of neurocentral plates which are partly ossified, partly embedded in cartilage, judging from the edges of the plates which have been preserved. The centrum seems not to have been present in the atlas, or if present it was only very slightly developed and quite free from the neural pieces and largely embedded in cartilage. A fairly accurate picture of the condition of the atlas and axis may be seen on examining a cow, pig, or chick embryo ([378]) in the early stages of vertebral development, which has been cleared by the Schultze method (Amer. Journ. Anat., VII, No. 4, 1908).
(l) The dorsal vertebræ, as well as those of the other series, present a primitive character ([fig. 8]) in the persistence of the notochord ([540]). Among the Branchiosauria the notochord was not at all or but slightly constricted intravertebrally, but among the Microsauria it was constricted so far that the notochordal remnants in each centrum resemble an hour-glass.
The structure of the vertebræ among American forms agrees fully with that outlined by Credner, Fritsch, and others for the European species. The details of structure are so fully given by Zittel ([642, pp. 346-353]) and by Schwarz ([540], [541]) that it will not be necessary to state more here as to their structure, since there is nothing new to add concerning the American species.
The temnospondylous vertebra of the same nature and type as exhibited by the Permian forms has its representatives ([94], [478]) in the Coal Measures. Spondylerpeton spinatum Moodie ([478]) ([plate 4, figs. 1, 2]) and Eryops sp. ([plate 18, fig. 2]) indicate the embolomerous and rachitomous types of vertebral structure. The occurrence of these widely different types of vertebral structure indicates a long history for the group prior to the Coal Measures. This history is further indicated by footprints in the Mississippian and Devonian of this continent.
Fig. 8.—Vertebræ and ribs of Amphibia from the Coal Measures of Linton, Ohio. Originals in Geol. Inst. Berlin. (All after Schwarz.)