214 BULLETIN 31, UNITED STATES NATIONAL MUSEUM. 



bule, wall osseous internally. Premaxillaries consolidated. Occipital 

 condyles on cylindrical jDcdestals." 



Protonopsid^: "ISTo anterior axial cranial bone. * * * Parietals 

 and prefontals prolonged, meeting and embracing frontals. Wall of 

 vestibule membranous internally. Premaxillaries separated. Occipital 

 condyles sessile." 



The following observations were made on the Ampliiumidi© : " The 

 occipital condyles and temporocervical tendon are quite as in Desmogna- 

 tlius; they have not been previously described.* In AmpMuma means 

 there is a minute non-articulated bone on the suture between the o. o. 

 frontalia and prefontalia in the situation of the lachrymal. There are 

 some approximations to Csecilia in AmphiumidfB. It does not appear 

 to have been noticed that the * * * free margin of the frontal seems 

 to foreshadow the overroofing of the orbit and temporal fossa seen in 

 C?ecilia. There is also a very large foramen or canal passing through 

 the o. maxillare from near its middle to the orbit, foreshadowing the 

 ccmalis tentaculiferus of Ciiecilia: a narrow one occurs in the same situa- 

 tion in Protonopsis. Further, the prominent horizontal anterior infe- 

 rior processes of the vertebral centra are the same in Ampliiuma and 

 CfBcilia." 



The characters assigned as above to the two families Araphiumidae 

 and Cryptobranchidfe are abundantly sufficient for retaining them as dis- 

 tinct, t The form of the occipital condyles might be excepted from this 

 estimate, and the axial bone in front of the parasphenoid proves to be 

 abnormally cut off in the specimen then examined. The Protonopsidse 

 agree with other Urodela in all of the characters given, except in the 

 exclusion of the frontals from the supraorbital border, and in the mem- 

 branous characteristic of the internal wall of the vestibule. The Am- 

 phiumicla3 differ from other Urodela in the presence of a large ethmoid 

 bone (the one referred to as ? vomer in the diagnosis above quoted), in 

 the presence of temporal ridges, and of two anteriorly directed hypapo- 

 pliyses of the precaudal vertebrfe. 



It is interesting to notice that three of the four characters just cited 

 are shared by the Cseciliidte. The presence of the ethmoid is of especial 

 importance, as it is an element constantly wanting in the Urodela. I 

 have not found it in Desmognathus, Anaides, Spelerpes, Amblystoma, 

 Srilamandra, nor Cryptobranchus, nor is it present in Necturus or in 

 Siren. It is, on the contrary, always present in C?eciliid<Ti| (see Plate 

 TX, 3). The double anterior hypapophyses are otherwise confined to the 

 same family. 



The characters of the hyoid arches also distinguish this family from 

 the Cryptobrauchidse, and they differ from those of the Pseudosauria 



* They were described by Dr. J. G. Fischer, Anatomisch. Abhandl. lib. Perenni- 

 branch. u. Derotrem., Erstes Heft, p. 61, 1864. 

 t Proceed. Anier. Philosoph. Soc, 1886, p. 442. 

 tWiedersheiai, Anatomie der Gymuophioneu, Jena, 1879. 



