CLASSIFICATION OF DEVONIAN FISHES. 



37 



At any i-ate, I think the 'prima facie case in favour of the 

 Teleostean nature of Coccosteus is so strong, that it can no longer 

 be justifiable to rank it among the Ganoids, " sans phrase," but 

 that even those who will not allow it to be Teleostean must attach 

 to it the warning adjunct of incertw sedis. 



No one doubts that wherever Coccosteus goes, Pterichthys must 

 follow, and though the structure of the last-named fish is, in some 

 respects, more difiicult of interpretation than that of the former, in 

 others it is strikingly Siluroid. For example, I know of no piscine 

 structure that is even remotely comparable to the proximal joint 

 of the pectoral limb of Pterichthys, except the corresponding arti- 

 culation of the pectoral spine and fin of the Siluroids. And again 

 the example of Ostracion shows that the box-like cincture of the 

 body of Pterichthys is by no means foreign to the Teleostean group, 

 though it cannot be paralleled by fishes of any other order. Whether 

 the other "Placodermi" of Pander, such as Asterolepis (Ag. and 

 Miller) really belong to the same group as Coccosteus and Pterich- 

 thys, or not, is a question which can perhaps be hardly settled at 

 present ; although, provisionally, I am much inclined to associate 

 them together. In principle, the cranial structure of Asterolepis, 

 is very similar to that of Coccosteus. 



Having disposed of the undoubted Elasmobranchs, of the Crosso- 

 pterygian Ganoids, and of the "Placodermi" of the Devonian epoch, 

 several important and rather difficult groups remain for discussion. 

 These are the Acanthodidse, the genera Cephalccspis and Pteraspis, 

 and the genus Cheirolepis. 



The ACANTHODID^ have hitherto been ranked among the G'anoids, 

 but the following considerations have often led me strongly to 

 suspect that they might be Elasmobranchs : — 



1. Their dorsal spines are similar in form and mode of implantation 

 to those of the Elasmobranchii, except perhaps that the surface of 

 the implanted portion is less different from the rest than in the 

 latter order., 



2. Their dermal ossicles are more like shagreen than scales. 



3. As Roemer has pointed out, their lateral line runs between 

 two rows of these ossicles, and is not formed by separate canals or 

 grooves in successive scales as in most Ganoids and Teleosteans. 



4. They seem to have had no distinctly ossified cranial bones. 



5. They have no opercular apparatus, but as Sir Philip Egerton 

 long ago pointed out to me, their branchial arches are naked. 



6. The sternal part of their pectoral arch seems to have had 

 no bony connexion with the head. 



