30 FOSSIL TURTLES OF NORTH AMERICA. 



elongated reptiles, in the skulls of which there were many chelonian characteristics 

 and whose bodies were protected by an armor of elongated plates overlying the ribs. 

 The stegocephahan genus Archegosaurus, of the Permian, possest a ventral armor 

 of elongated plates. We may confidently expect that in the Permian there will yet 

 be found an animal possessing such a combination of these characters, together 

 with other features, that we can recognize in it the ancestor of the order of turtles. 



As to the relationships of the turtles to the orders of reptiles, other than the 

 Cotylosauria, only a few words can be said. The opinion is generally maintained 

 that their relationships are closer to the Plesiosauria than to any other order. Baur 

 (Jour. Morphology, I, 1887, p. 98) has enumerated a number of resemblances 

 between the two groups. Among other things, he states that the pelvic arch of the 

 Nothosaurida?, among the most primitive of the Plesiosauria, is only comparable 

 with that of the Testudinata. It may be remarked that the pubes in both the turtles 

 and the plesiosaurs are greatly expanded forward and laterally. The ischia are not 

 so much alike, those of the plesiosaurs being much more extended backward than 

 those of the turtles. Baur says that in the humerus of the lower Plesiosauria there 

 is an ectepicondylar foramen; but, according to Fiirb ringer, there is only an 

 entepicondylar foramen. Baur does not compare the skulls of the two orders. The 

 strongest argument in favor of the relationships of the two orders is found in sup- 

 posed resemblances in the shoulder-girdles. Probably the best presentation of this 

 is to be found in Fiirbringer's elaborate paper in the Jenaische Zeitschrift, xxxiv, 

 1900, page 326. In that treatise the author takes the ground that the process 

 directed interiorly and inwardly from the scapula in the turtles is a procoracoid, a 

 view adopted by the present writer. Fiirbringer also holds that the procoracoid 

 was present in the Plesiosauria. This opinion has been accepted by some authors 

 and disputed by others. To the present writer the tract of bone in question appears 

 to be simply an extension forward and mesially of the scapula, whether that 

 extension is regarded as an acromion or a "ventral ramus" of the scapula. The 

 tract advanct gradually from the scapula to the midline, and there is no evidence 

 that it ever was a distinct bone. The fact that the two clavicles do not lie between 

 the ends of these processes does not appear to be decisive, for in numerous reptiles 

 the clavicles extend laterally on the scapula?. Opposed to the idea that there are any 

 special relationships between the turtles and the plesiosaurs are the facts that the 

 latter always possess large supratemporal fenestra?, the former never; and that the 

 most primitive plesiosaur, Larwsaurus, has the digital formula belonging to the 

 Diapsida and not that of the Synapsida. 



At best, the relationship between the turtles and the plesiosaurs is not close; 

 and the most that we can say is that possibly the cotylosaurian stocks of the two 

 orders were a little closer to each other than to stocks that gave rise to the other orders. 



THE GEOGRAPHICAL DISTRIBUTION OF LIVING TURTLES. 



At the present day turtles exist in all the larger divisions of the earth, wherever 

 there is sufficient heat, at least a moderate amount of moisture, and a supply of food. 

 They occupy too all the warmer seas. Fig. 9 is an outline map that includes all the 

 lands and seas in which these reptiles are found. The areas occupied by the land and 

 fresh-water forms are indicated by parallel rulings, except that certain small islands 

 are made black; but no attempt has been made to show the distribution of the marine 

 species. Their realm may be taken as comprising all the seas between the ruled areas. 

 This map and those succeeding it have been compiled with considerable care; but 

 it is not possible in all cases to determine exactly the limits of the groups. 



