504 MULLER ON THE STRUCTURE OF THE GANOIDS, 
the old formations and Polypterus and Lepisosteus; for he says 
(Poiss. Foss. ii. p. 11) “the relations of structure which connect 
the Lepidoidei, the Sauroidei, and the Pycnodontes, are more 
intimate than those which exist between these families and the 
Sclerodermi, the Gymnodontes, and the Lophobranchii.” 
The Siluroidei agree so perfectly in their anatomy with the 
Abdominal Malacopterygians, that they cannot be separated 
from them; they have merely the air canal of the swimming- 
bladder and the abdominal position of the ventral fins in common 
with the living Ganoids, characters which they also share with a 
large section of osseous fishes, which I shall call Physostomi, on 
account of their air canals, as the Cyprinoidei, Esoces, Clupee, 
Cyprinodontes, Mormyri, Characini, Salmones, Anguillares, &c. 
However, this air-passage is absent in the Sclerodermi and 
Gymnodontes, as also in many orders of osseous fishes; their 
ventral fins, when present, as in Triacanthus, are not abdominal ; 
in both points they differ from the living Ganoids, and in the 
latter character from all true Ganoids. The Ganoids can only 
be accurately defined from the characters with which we have as 
yet become acquainted, ranging among them those fishes only 
which agree with Lepisosteus and Polypterus in the scales being 
covered with enamel. If we include the bony laminz of the 
Loricarini, Lophobranchii, Ostracion, some Siluroidei, as Cal- 
lichthys, Doras, and the prickles of Diodon among the Ganoid 
structures, all accurate definition ceases. For, first, we are 
obliged to admit the naked Stluroidei and the Gymnodontes, 
merely because some of their genera are covered with laminz or 
prickles, and hence it becomes possible that families of Ganoids 
might exist, in which all the genera are unarmed, which would 
render separation and recognition impossible, as long as no 
essential characters of the Ganoids were known. Still greater 
confusion would be produced by those fishes belonging to other 
orders which have osseous carapaces, as the Peristedia, Agoni, 
and other Cataphracti with osseous laminz, the nearest in re- 
lation to which are provided with scales, as the Trigle; these 
however are not Ganoid scales. Finally, the scales of many 
Sclerodermi, as the Monacanthi and Aluteres, have but little 
resemblance to those of the Ganoids. 
Should all these animals be left among the true Ganoids, their 
characters would be so confused, that it would be utterly im- 
possible to say what would then constitute a real Ganoid, we 
