EASTMAN: STRUCTURE AND RELATIONS OF MYLOSTOMA. 11 



plates, arranged as in Mylostoma, these afterwards fusing into one. 

 Neoceratodus therefore reproduces ephemerally a stage which remains 

 permanent in the Devonian genus, and is probably to be regarded as an 

 inheritance of primitive conditions. 



Sufficient arguments have now been given, we think, to support the 

 claim that the dentition of Mylostoma and Dinichthys is constructed 

 distinctly upon the Dipnoan type. We learn from recent forms that 

 the external ensheathing bones of the mandible may become greatly 

 reduced. In the Arthrodiran jaw this process has merely been carried 

 a little further than in Protopterus. The presence of a supernumerary 

 pair of dental plates in Mylostoma is satisfactorily accounted for by 

 the ontogeny of Neoceratodus, which reveals the primitive nature of 

 Mylostomid type of dentition, and suggests for it a common origin with 

 the Ceratodont. Indeed, the evidence derived from this latter source 

 shows that Mylostoma has departed less widely from primitive Cerato- 

 donts with respect to its dentition than have true Dipterines ; for 

 amongst the latter no vomerine teeth occur, nor is any form known 

 which has retained more than a single pair of palato-pterygoid dental 

 plates, whereas their ancestors may reasonably be presumed to have 

 had two. This point is in harmony with other facts making for the 

 conclusion that modern lung-fishes are not directly descended from 

 Dipterines. 



Cranial characters indicating Dipnoan affinities of Arthrodires. — 

 From the interpretation of jaw-parts just given, it is obvious that no 

 secondary upper jaw occurs amongst Arthrodires. Certainly the sub- 

 orbital, which is simply a cheek-plate, has nothing whatever in common 

 with a maxillary arch, nor is there the slightest reason for believing 

 that it supported, or was otherwise associated with, or even in contact 

 with the " shear-tooth " in Dinichthys. Aside from the suborbital and 

 dental elements already accounted for, there are absolutely no plates 

 left which can by the greatest stretch of imagination be homologized 

 with the maxillae and premaxillae of ordinary fishes. We are there- 

 fore prohibited from classing Arthrodires amongst Teleostomes, and at 

 the same time must recognize their agreement with Dipnoans in one of 

 their most distinctive characteristics. 



Another important fact must also be considered. It is well under- 

 stood that the cranial roofing-bones of modern lung-fishes are not readily 

 to be homologized with those of Ctenodipterines, or indeed of any other 

 group with the exception of Arthrodires. Now, how are we to explain 

 this remarkable coincidence except upon the hypothesis of common 



