KEOTENV 65 



the external gills of the Urodela have been developed in adapta- 

 tion to their embryonic and larval, essentially aquatic, life. Con- 

 sequently the possession of such gills would be a secondary, and 

 not, strictly speaking, an atavistic feature. Normal loss of these 

 gills, exclusively pulmonary respiration, and preponderating 

 terrestrial life characterise the final adult Amphibian. These 

 cases of neoteny are therefore instances of more or less complete 

 retardation, or of the retention, of partially larval conditions. 



The whole problem is, however, by no means simple. Sala- 

 maiulra atra has become viviparous, and the whole metamor- 

 phosis takes place within the uterus ; in fact, the young have an 

 embryonic, but no larval period, if by the latter we understand 

 the free swimming and still imperfect stage. Similarly, various 

 Auura — for instance, Hylodes martinicensis — -.pass rapidly through 

 their metamorphosis, and have suppressed the stage of free 

 swimming tadpoles. On the other hand, in many newts, the 

 duration of the larval period is much prolonged, and moreover 

 is very subject to individual variation. In the Axolotl this 

 larval period is continued until and after sexual maturity is 

 reached. The extreme condition would then be represented by 

 the Perennibranchiate genera. It may seem reasonable to look 

 upon these as the youngest members of the Urodela, and the loss of 

 the maxillae in the Sirenidae and Proteidae supports this idea. 

 But it so happens that the majority of the most neotenic genera 

 are more primitive in the composition of the skull and the verte- 

 bral column than the typically terrestrial and rapidly meta- 

 morphosing genera. Witness the amphicoelous vertebrae, the 

 completeness of the pterygoids, the separate nature of the pala- 

 tine bones, and the separate splenials, as mentioned in detail in 

 the description of their skull. 



We have therefore to conclude, first, that the various Perenni- 

 branchiate genera do not form a natural group, but are a 

 heterogeneous assembly ; secondly, that they have become Perenni- 

 branchiate at a phylogenetically old stage — in fact, that they are 

 the oldest, and not the newest, members of the present Urodela. 

 At the same time, it would be erroneous to suppose that the 

 first Urodela were aquatic creatures, provided with a finny tail, 

 with small, ill-developed lungs, and with epidermal sense organs. 

 All these features are, on the contrary, directly correlated with 

 aquatic life, and are larval acquisitions, not ancestral reminis- 

 VOL. VIII 1*" 



