20 SIRENOID AND CROSSOPTERYGIAN GANOIDS. 



compare it with the Cestracionts. At a later time he more plausibly referred Ceratodus 

 to the Chimaridce} From their occurrence together in the Rhaetic beds of the Severn, 

 Agassiz hinted that the spines named Nemacanthts might belong to the same fish. 



Perhaps no other palaeontologist could have better interpreted the remains in 

 question. It appears that such criticism as was offered was either unimportant or mis- 

 taken. 2 Putting aside for further consideration the questions of the zoological position of 

 Ceratodus and the number of species, we may here remark that Agassiz was right as to 

 most of his conclusions, distinctly wrong only as to the anterior and posterior ends of 

 the teeth. His conjecture respecting Nemacanthus has proved unfounded. 



We are no longer obliged to divine the structure of Ceratodus by means of obscure 

 fossil remains. In 1870 the genus was ascertained by Mr. Gerard Krefft to be still 

 extant in the rivers of Queensland. Specimens transmitted to Europe have been 

 carefully described by Dr. Gimther, 3 and, though difficulties yet remain, the general 

 structure of the fish is fairly well known. 



The disposition of the teeth in this genus is now ascertained to be as follows : — An 

 osseous palato-pterygoid arch supports two palatal dental plates, whose horns or denticles 

 project outwards, the larger and more prominent being anterior. In advance of these 

 are two cutting-teeth imbedded in the " vomerine " (mesethmoid) cartilage. The man- 

 dible supports a similar pair of dental plates. These are attached by their inner edges 

 to the splenial bone, which invests the inner side of the jaw, while on the outer side they 

 overlie the cartilaginous centre of the ramus, and to some extent the angular bone, with 

 which they have a fibrous connection. In biting, the ridges of the dental plates inter- 

 lock, though the opposed surfaces are not so accurately adjusted as altogether to prevent 

 a sliding movement ; the vomerine teeth are unopposed. All the dental plates are anky- 

 losed to the supporting bones. 



In the recent species the palatal tooth bears six distinct horns, and the inner margin 

 is prolonged backwards as an imperfectly developed seventh. The first mandibular horn 

 is received between the first and second palatal horns, the last mandibular between the 

 sixth and the (imperfect) seventh of the palatal. This rule would enable us to dis- 

 tinguish readily between the fossil palatal and mandibular teeth, were the number of 

 horns known in all the species. But we find nearly all the fossils to possess either four- 

 or five-horned teeth, and we cannot at once tell how far the number is related to specific 

 difference. Two difficulties occur at once, each of which complicates the other. 



In order to trace gradations and discover gaps in the numerous fossil examples, we 

 require to know that we are comparing similarly placed teeth. On the other hand, till 

 the gradations and gaps have been so far investigated that species are satisfactorily 

 defined, we cannot well test our rules for distinguishing upper and lower teeth. In 



1 'Poissons Fossiles,' vol. i (introduction), pp. xxxvii, xxxix, xlix. 



2 See, for example, Plieninger (' Palaontologie Wiirttembergs,' p. 88). 



3 « Phil. Trans.,' 161, pi. ii (1872). See also Prof. Huxley in ' Proc. Zool. Soc.,' 1876, p. 24. 



