452 CLASSIFICATION OF DEVONIAN FISHES 
seven are closely articulated together, while the other two, small and 
comparatively insignificant, (and not represented in the dorsal view, 
fig. 19) were placed loosely at the sides of the posterior end of the 
great median plate of the seven. This plate a corresponds in width, 
anteriorly, with the cranial bone S. O.; it widens a little behind the 
middle of its length, and then rapidly tapers to a point. From the 
middle of its under surface it sends down a strong bony crest, deeper 
behind than in front, while its lateral edges overlap and unite, by a 
squamous suture, with the plates S.s. and 4, 
S.s. is a four-sided plate, articulated with Pa. Ep. in the manner 
before mentioned, while behind it overlaps the triangular plate 4, and 
below is overlapped by the plate ¢. The latter is so constantly thrown 
out of its place in specimens where the connexion between a, 6 and S.s 
is perfectly retained, that I suspect it rather overlapped than was 
suturally united with S.s. 
The ventral shield appears to me to have had no direct connexion 
with the dorsal. JI have examined a Jarge number of specimens with 
reference to this point, but I have never discovered the least evidence 
of a sutural union between any two elements of the two shields, though 
the respective constituents of each shield are constantly met with in all 
stages of union and disunion. Of the elements of the ventral shield, 
two are median and symmetrical, four lateral and in pairs. The two 
latter, upon each side, are broad at their remote ends and narrower 
at their adjacent ends, whose outer edges are, besides, somewhat 
bent up. Of the median plates, the posterior is rhomboidal and 
articulates with all the others ; the anterior has the form of an elon- 
gated isosceles triangle, whose base, directed anteriorly, is rounded off 
and forms the middle of the anterior margin of the ventral shield. 
The stout, doubly curved, clavicle-like bones Mn., found, in com- 
plete specimens, on the under side of the head, have one edge beset 
with minute denticles for a short distance; and there are two other 
flat, elongated, bones, devoid of sculpture upon their outer surfaces, 
which lie between them and the anterior edge of the ventral shield. 
Beside the parts now described, the only other bones known to 
belong to Coccosteus are the neural and subcaudal arches, the fin-rays 
and their supports, and the curved ossicles which lie just behind the 
body armour, and were perhaps connected with ventral fins ; but I 
enter into no particular description of these, as they are not essential 
to my present purpose. 
For some years past I had suspected that the modern Siluroids 
presented more analogies to the seemingly aberrant Devonian fishes 
than any other members of the class Pisces, and from the examination 
