458 CLASSIFICATION OF DEVONIAN FISHES 
6. As Roemer has pointed out, Paleoniscus has orbital plates very 
like those of Acanthodes. 
7. The production of the pectoral arch into long backwardly di- 
rected processes in Diplacanthus and Chetracauthus is the very reverse 
of an Elasmobranch character, seeing that the like only obtains, so far 
as I know, in some Siluroids. 
8. Acanthodes is provided with two very long filaments, beset with 
short lateral branches, which proceed from the region of the mouth, 
and such oral tentacles are to be found only in Ganoids and Siluroids. 
Under these circumstances the safest course probably is to regard 
the Acanthodide as a distinct suborder of Ganoids. 
The genera Cephalaspis, Pteraspis, Auchenaspis, and Menaspis 
certainly form.a family by themselves, to which the title of CEPHA- 
LASPID.E may be conveniently applied ; but the position of this family 
is not readily determinable. No one can overlook the curious points 
of resemblance between the Siluroids, Calléchthys and Loricaria, on the 
one hand, and Cephalaspis, on the other, while in other respects, they 
may be still better understood by the help of the Chondrostean Ganoids. 
Compare, for example, Scapirhynchus with Cephalaspis, or the great 
snout of Pteraspis with that of Spatularza. I am inclined to place the 
Cephalaspids provisionally among the Chondrostei, where they will 
form a very distinct family. 
The affinities of two genera remain for discussion, the one being 
the well-known Chezrolep?is of Agassiz, the other, the new genus 
Tristichopterus, described by Sir Philip Egerton in the course of the 
following Decade. 
Chetrolepis contains fishes with moderate-sized heads and markedly 
heterocercal tails ; with a single dorsal fin, a single anal, pectorals, and 
ventrals. The median fins are situated forwardly, the dorsal being 
over the posterior part of the anal; and the ventral fins are so for- 
ward as to be almost close to the pectorals. None of these fins are 
lobate. The body is covered with minute rhomboidal scales, which 
do not overlap one another, so that the skin has quite the aspect of 
shagreen. Nevertheless, according to Pander, the structure of these 
bony scales is not so like that found in the Squalidz as that of the 
scales of Diplacanthus. 
The head is usually crushed, and its component elements dis- 
placed, but according to Professor Pander, whose account is largely 
borne out by the specimens I have examined, the middle of the roof 
of the cranium, from the posterior edge of the occiput to the anterior 
