DEVONIC FISHES OF THE NEW YORK FORMATIONS 1 45 



from all species of Coccosteus and Dinichthys, and points to ancestral rela- 

 tionships with Titanichthys, in recognition of which we have given it the 

 generic name. Indeed, no other course is open to us than to attach signifi- 

 cance to a character which is common to the two genera under comparison 

 (Titanichthys and Protitanichthys), but occurs not elsewhere amongst 

 members of the same family. We refer to the large size and transverse 

 elongation of the pineal plate, which, according to our view, indicate an 

 initial modification along the line of descent culminating in Titanichthys. 

 As yet our acquaintance with intermediate forms connecting the archetype 

 and extreme limits of specialization is very inadequate. But when we take 

 into consideration the long geological interval separating the two genera in 

 point of time, it is scarcely risking too much to claim this small composite 

 form as a possible forerunner of the Upper Devonic Titanichthys. 



The head shield, itself imperfectly preserved, is unaccompanied by any 

 other bones of the skeleton. Although, as a rule, little dependence is to 

 be placed upon theoretical association of parts, the question may properly 

 be raised whether there are any other Coccostean remains occurring in the 

 same horizon which correspond in general proportions and superficial orna- 

 ment to the head shield under discussion. Attention rests immediately 

 upon two forms: the lower dental plate described as Liognathus 

 spatulatus, and the dorsomedlan plate known as Coccosteus Occi- 

 dent a 1 i s [ante, p. 1 1 5]. The possibility is not remote that all three of 

 these detached parts which have received separate names may actually 

 belong to a single species ; in default of proof, however, we have no other 

 procedure than to maintain each provisionally as an independent species, or 

 even genus. At the same time it may not be unworthy of comment that 

 the peculiar spiniform prolongation observed in Coccosteus occiden- 

 tal is reappears in an abbreviated form in the triangular termination of 

 the corresponding plate in Titanichthys. Such a coincidence certainly sug- 

 gests the idea of an inherited characteristic, and confirms us in the belief 

 that prototypes of the most highly specialized Coccosteans are to be sought 

 as early as the Ulsterian stage in this country. 



