xli 



Some of these characters are diagnostic, that is, they distinguish the 

 forms from all others ; others may be shared with isolated forms of widely 

 separated groups; but the agreement of the " Plectognaths" among them- 

 selves in the many common characters justifies their association together, 

 and the characters that are peculiar to them sanction their isolation as a 

 group. 



Three well-defined groups exhibit the principal modifications under which 

 the fishes possessing these common characters are developed. They are 

 principally distinguished by the development of the scapular arch (the hy- 

 percoracoid is atrophied in the Gymnodonts), the degree of union of the jaws 

 and the dentition, and by the squamation. But while the external differ- 

 ences between these forms are doubtless very considerable, they all share 

 the common characters above enumerated and other less salient ones, and in 

 view of this much nearer connection, in contrast with other forms, seem 

 most decidedly deserving of retention together, in contrast with other fishes, 

 whatever rank may be conferred on the group. Their differences sink 

 into comparative insignificance, when compared with their common charac- 

 ters, and seem not entitled to more than subordinal value, while the group 

 of which they are constituents may be most aptly considered an order, as 

 has been done by almost all ichthyologists. The Scleroderms have fur- 

 nished the chief basis for dissent as to the homogeneous character of the 

 order, and have been deemed more related to ordinary Acanthopterygian 

 types than to the other admitted Plectognaths And it is quite true that 

 they (and especially the Triacanthids) are much more similar to the ordi- 

 nary fishes than are the typical Plectognaths. This, however, is quite 

 explicable by the supposition that they are the most generalized, and repre- 

 sent the immediate line of descent, while the others are more specialized. 

 That the likeness, however, is superficial and illusive, is evident from the 

 disagreement from the types they must resemble in form, in anatomical 

 characters, and their agreement therein with the other Plectognaths, as 

 already indicated. Prof. Cope has considered the relations of the order 

 (through the Triacanthidse, on the one hand, and the Chaetodontidae and 

 Acronuridae on the other) to be most intimate with the Teleocephals at 

 the point indicated, and M. Dareste has contended that the Balididse are 

 especially related to the Acanthuridse. As there seems to be no proof of 

 any nearer relations elsewhere, the hint furnished by the agreements induc- 

 ing such belief may be followed in the arrangement and sequence of the 

 order as well as of the families constituting it. 



PEDICTTLATI. 



The only order adopted remaining for consideration is that of Pedicu- 

 lati. The natural character of the association of forms combined therein 

 is obvious, and has never been questioned, and the comparatively slight 



