EMBRYOLOGY OF CRYPTOBRANCHUS 549 



The clerotremes also are not primitive urodeles. Since they 

 show a mixture of the characters of the larvae and grown-up 

 salamanders, one cannot derive them from typical larvae. Pre- 

 sumably the derotremes are descended from typical salamanders 

 which have returned completely to aquatic life. Thus the meta- 

 morphosis, which serves to adapt the larvae to land life, has lost 

 its biological meaning; it extends over a longer time, and finally 

 sexual maturity overtakes the yet imperfectly built animal. Some 

 organs complete the metamorphosis, others retain wholly or 

 in part the larval condition. So in its skull Cryptobranchus 

 (Reese '06) resembles the grown-up salamander; in its circulation 

 it retains larval characters. Versluys believes Cryptobranchus 

 to be descended from the amblystomidae. 



If we take into account only living forms, much of the data 

 thus far considered in seeking a solution of this general question 

 of the relationship of the aquatic and land-living urodeles may 

 be read either way. The key to the situation lies in the compar- 

 ison of the structures of existing urodeles with those of fossil 

 forms. We have seen that the extinct ancestor of the urodeles 

 was probably an animal whose skull was more complete than 

 that of any living urodele. In Necturus (Kingsbury '05), there 

 is a considerable reduction in the number of cranial bones, com- 

 parable to the condition in the larval Spelerpes, and contrasting 

 with the condition in the adult Spelerpes. If the stem-form of 

 the urodeles was an animal like Branchiosaurus, then so far as 

 the cranial elements are concerned the salamander, and not the 

 perennibranch, is the more primitive form. There are five 

 branchial arches in Branchiosaurus — a condition which does 

 not exclude either view of the interrelationships of the urodeles. 

 Branchiosaurus had external gills only in the larval stage; this 

 indicates that the condition found in the perennibranchs is proba- 

 bly not in the ancestral line of the caducibranchs (salamandrina). 



Von Eggeling ('11), from a comparative study of the histogen- 

 esis of the skeleton of the limbs of urodeles, favors the view that 

 the perennibranchs, derotremes and siredon are derived from 

 the caducibranchs. 



