In 1938 Broom described a reptile from the Upper Permian of South Africa as Millerina, concluding that it was a very primitive cotylosaur 'ancestral' to the mammal-like reptiles. To it he added several other genera, including one, Millerosaurus, with a pelycosaur-like temporal opening. Very well-preserved specimens of this last genus make possible a nearly complete description of the whole skeleton of these animals. They are shown by the occurrence of a typical lizard-like columella auris and tympanic cavity to be sauropsids, and are evidently far more primitive in general structure than any other members of that group. The group founded for them is shown to include, with great probability, Mesenosaurus from near the beginning of the Russian Permian reptile-containing deposits. The real resemblance of the millerosaurs to primitive captorhinids and pelycosaurs is evidence of a common ultimate derivation from anthracosaurs. The Millerosauria provide a starting point for the development of all sauropsids except perhaps the Chelonia. Thus the first appearance of 'diapsid' reptiles in the Upper Permian Cistecephalus Zone, and the immensely rapid development they show in the Lower Trias, is related to the effective disappearance of Dicynodon, and of the carnivorous gorgonopsids and Therocephalia which preyed on it, at the end of Permian time. The break is as great as that which separates the beginning of Tertiary from the end of Cretaceous times amongst land-living vertebrates.