CLASSIFICATION OF DEVONIAN FISHES 437 
Celacanthus is so intimately connected with Undina and JZacro- 
poma, as to render the generic distinction of the three forms a matter 
of minute detail, its relations with Holoptychius, although clear and 
distinct so far as they go, are, at most, those of a member of the 
same suborder. 
But first, what are the characters of the genus Cwlacanthus? 
This question is by no means so easily to be answered as might be 
imagined, but the following facts appear to furnish a conclusive 
reply to it. 
The type species of Calacanthus, that on which the genus was 
founded by Agassiz, is the C. granulatus of the Magnesian Lime- 
stone; two figures of which are to be found in the “ Recherches,” 
while a third, representing another specimen, is given by Sir Philip 
Egerton in King’s “ Permian Fossils.” Singularly enough, neither of 
these specimens retains its head, nor are the paired fins preserved ; 
but the characters of the spinal column, of the median fins, of the 
scales, and of the tail, are so exactly those exhibited by the Undzua 
of Minster (of which sundry complete specimens enist), that the 
very close affinity of the two genera is beyond doubt. Agassiz, in 
fact, proposes to distinguish them only by their teeth ; Ca/acanthus 
having, in his opinion, conical and recurved, while Uxdina has 
flat, pavement-like and tuberculated teeth. That Miinster was 
correct in assigning such teeth to Undina I have satisfied myself 
by the examination of a well-preserved specimen of U. Kohderi in 
Lord Enniskillen’s collection; but what evidence is there that 
Celacanthus has a different dentition? Agassiz was led to believe 
that the teeth of the latter genus are conical, by the fact that the 
specimen of a fish named by him C. Minster? has such teeth. I am 
again indebted to the Earl of Enniskillen, of whose collection this 
specimen forms a part, for the opportunity of verifying the state- 
ment; but I must at the same time express my entire concurrence 
in the opinion previously expressed to me by Sir Philip Egerton, 
that the so-called “Calacanthus” Miinsteri is not a Celacanthus 
at all. 
For, as I have stated above, there can be no doubt that Ce/a- 
canthus (C. granulatus being the typical species) was, in all the 
great features of its organization, similar to Uudina; so that, con- 
trariwise, any fish which differs in essentials very widely from 
Undina can be no Celacanthus. 
But Undina has two dorsal fins, each supported by but a single 
very peculiarly shaped, interspinous bone ; it has a large caudal fin, 
whose rays are supported by interspinous bones, and which is divided 
