94 ANNALS NEW YORK ACADEMY OF SCIENCES 



geography, but it is no longer tenable. The separate portions of the 

 Gondwana Land are, however, more interesting now than ever. 



The only places for various past connections which are needed and 

 almost universally accepted are the following: 



Southern South America and perhaps the Antarctic islands. 



South America and North America by way of Central America and 

 perhaps the "West Indies. 



North America and Eurasia by the way of Greenland and the North 

 Atlantic, Alaska and Siberia. 



Southern Asia and Australia by way of the East Indies. 



Eurasia and Africa. 41 



PERMIAN REPTILES 



Does the distribution of the Permian reptiles indicate the existence of 

 a connection between Africa and South America? 



Only a few specimens of Permian reptiles have been found in South 

 America. Mesosaurus orasiliensis is the best known species. It was 

 described by McGregor (1909) from the bituminous shales of Iraty, 

 Parana, Brazil. Stereosiernum tumidum Cope is a closely related form. 

 It was found in Sao Paulo and comes from the surface of a thin layer of 

 limestone. Many fragments of it were seen by the writer near Piraci- 

 caba at a limestone quarry on the property of the Agricultural School. 

 A few well-characterized marine fossils have occasionally been found in 

 the series of beds in which Stereosternum is found. Two more species of 

 Mesosaurus are known from the Dwyka beds of South Africa. 



Mesosaurus is not a diapsid. Its unique vertebrse and ribs, as well as 

 the absence of scales, webbed feet, dorsal but no lateral temporal fenestra, 

 slender teeth, long snout, etc., separate the genus from all known reptiles. 

 Von Huene (1910) derives it from some unknown Carboniferous coty- 

 losaurian. So far, not even the antecedent type, which gave rise on one 

 hand to Mesosaurus and on the other to Stereosternum. is known. It is 



41 It must be granted that of all the evidence in favor of a continuous Gondwana Land, 

 its flora appears to be the best. But in view of the fact that when it was once formed it 

 did not appear to change, we may suggest as a future working basis that this flora offers 

 a special type of orthogenetic development which has been produced from the cosmo- 

 politan older flora by definitely directed changes in the environment during the formation 

 of the highlands where it is found. In Australia, this flora appears to have been the 

 maker of coal during the Permo-Carboniferous. It is also said to be associated with 

 marine drift and glacial deposits. Hence it appears to be a swamp flora in Australia, 

 but the presence of thick beds of coal and glacial drift in the same regions does not 

 appear to harmonize. If this is true, then the Gondwana environments of Brazil and 

 Australia are distinctly different. A continuous Permian and early mesozoic Gondwana 

 land is needed no more than a Tertiary or a recent one. We now know, however, that 

 no Tertiary Gondwana is required to explain the distribution of animals. 



