centrum of the proatlas is the so-called intercentrum of the atlas, necessitating the view that the axial intercentrum is merely an accessory or provisional bone developed below the odontoid to fill out what would otherwise be an unoccupied space!
Positive evidence of the proatlas has been discovered in several genera of the Cotylosauria, but no complete specimen has yet been discovered; it is doubtless present throughout the order. It is present in many if not all forms of the Theromorpha and Therapsida. In Ophiacodon (Fig. 78) and Dimetrodon (Fig. 79) of the former group, it is a small bone on each side, articulating in front by a facet on the exoccipital, behind with an anterior zygapophysis on the arch of the atlas, both surfaces looking more or less downward. These articular surfaces appear to be present in all known genera. In the Crocodilia, occurring as far back as Jurassic times, it is a single bone in the adult, roof-shaped, arising from paired cartilages. In Iguanodon (Fig. 80 l), of the predentate dinosaurs, as also in several genera of the Sauropoda, and the Triassic Plateosaurus of the Theropoda, it is paired, as in the modern Sphenodon (Fig. 80 d), also articulating with the atlas. A roof-shaped, unpaired proatlas has been described in Rhamphorhynchus, a Jurassic pterosaur. It has also been reported in the chameleon lizards and the mammals Erinaceus and Macacus. As an abnormal element it was also found by Baur in a trionychoid turtle (Platypeltis spinifer, Fig. 32), partially fused with the occiput, and articulating with the arch of the atlas in the primitive way, from which he concluded that the real body of this vertebra had become permanently fused with the basioccipital. Probably it will be eventually discovered in many other extinct reptiles.
Atlas (Figs. 78, 79, 80). There is no vertebra in the known amphibians which can be homologized with the atlas of reptiles. By some the so-called atlas of the amphibians is thought to be represented by the proatlas; or it may have entirely disappeared. In the earliest reptiles (Fig. 79), the atlas is temnospondylous in structure, that is, composed of a paired arch resting in part upon a large, wedge-shaped intercentrum, in part upon a single large, embolomerous, notochordal pleurocentrum, all of them loosely connected with the axis, the arch of the atlas or neurocentrum articulating in the usual way by zygapophyses.
In its highest development, in the mammals, the arch and inter-
