422 THE AMERICAN NATURALIST [Vol. LVI 



brook habitat, with the penalty of loss or reduction of 

 lungs. The list is extensive, Onycliodactylus, Rliyaco- 

 triton, four species of Triturus, Salamandrina, Chio- 

 glossa, all the stock of the Plethodontidce, an assemblage 

 representing four families. The sound-transmitting ap- 

 paratus is admittedly correlated with the mode of life. 

 Therefore as a character in determining relationships it 

 must be used with extreme caution. 



The following outline classification of salamanders 

 does not counter any of the facts concerning the otic 

 apparatus, and is based on many characters. 



As regards the PletJiodontidce and the Hynohiidce, re- 

 visions of both are nearly completed, based on the exam- 

 ination of some 8,000 specimens of the first family and 

 1,000 of the second. 



The Sirenidce are the most isolated group. Scarcely 

 a character can be found to ally them with one or an- 

 other of the main stocks. The pelvis is gone, the skull 

 is that of a very specialized larva, the hyoids are those 

 of almost any larva, the tail vertebrae are very different 

 from those of any other salamander, inasmuch as there 

 is no hcTmal arch. There are flat plates on each side 

 which do not meet in the mid-ventral line. There is no 

 prearticular. 



The Proteida are only slightly less isolated. The 

 pelvis differs in having an anterior median projection and 

 no ypsiloid apparatus. The skull is larval. The bran- 

 chial arches are reduced from the primitive larval quota. 

 The prearticular is absent. 



The Amphiumid(E also have modified larval branchial 

 arches, and the pelvic girdle lacks the ypsiloid appa- 

 ratus. But Amphiuma has an adult skull which resembles 

 remotely that of the Salaniandrida. The otic apparatus 

 is that of the PletJiodontidce. There is no prearticular 

 bone. It is quite possible that this genus is descended 

 from primitive Salamandrids. 



The others have directly comparable skulls, branchial 

 arches, and pelves, and in dealing with their relationships 

 we are on much firmer ground. 



