436 CLASSIFICATION OF DEVONIAN FISHES 
In order to make this clear, however, I must enter at some length 
into a historical and anatomical criticism of the Ccelacanths as a 
family of fishes. 
In establishing this family (“ Recherches,” vol. ii. p. 168), Professor 
Agassiz dwells particularly upon the hollow fin rays of the typical 
genus ; the absence of joints in some part of the length of most of 
those fin rays; the presence of interspinous bones in the caudal fin; 
the continuation of the vertebral column between the two lobes of 
that fin, and the prolongation of the caudal extremity beyond it 
as a filamentary appendage. With Cewlacanthus, Undina, Macro- 
poma, Hoplopygus, Uronemus, Holoptychius, Glyptosteus, Glyptolepis, 
Psamimolepis, Phyllolepis, Ctenolepis, and Gyrosteus, are associated ; and 
it is a curious circumstance that while Holoptychius takes its place 
among the Ccelacanths, without any special demonstration of its right 
to that position, Professor Agassiz hesitates touching J/acropoma, and, 
while admitting it into the family on account of the striking analogy 
of its general physiognomy,, and of the form, arrangement, and struc- 
ture of its fins, adds: “I must admit that side by side with these 
“resemblances, the two types exhibit profound differences,”. . . 
“which will perhaps, in the long run, necessitate another arrangement.” 
The idea that Celacanthus inclined more to Holoptychius than 
to Macropoma, appears to have found still more favour with Pro- 
fessor Agassiz at the time of the publication of his great work on 
the Fishes of the Old Red Sandstone; and the consequences of this 
inclination were the more important from the fact, that Agassiz 
held that the teeth, properly distinguished by Professor Owen 
under the name of Rfzzodus, belonged to Holoptychius. For 
Glyptolepis and Platygnathus were undoubtedly closely allied to 
Floloptychius, while Dendrodus, Lamneodus, and  Cricodus had 
much in common with Rfizodus ; hence, as these dendrodont teeth 
were conceived by Agassiz to belong to the fish whose bony plates 
and scales had received the names of Asterolepis, Bothriolepis, &c., 
it was natural that he should include all these genera under the 
common title of “ Ccelacanths ;” while J/acropoma and Undina were 
regarded with doubt, and, in fact, almost excluded from the group 
(‘“ Vieux Gres Rouge,” p. 64). 
Here, however, I cannot but believe, that the founder of fossil 
ichthyology has, for once, gone off upon a wrong scent. For later 
investigations have made it, to say the least, extremely improbable 
that Asterolepis (Ag. & Miller) has anything to do with Crecodus 
or with Holoptychius, whatever may be the relation of the two 
latter genera; and I shall now endeavour to prove that, while 
