THE DEVONIAN ROCKS OF CANADA. 103 
The median angulation of the dorsal body shield is usually most pronounced at or a 
little in advance of the middle of the dorsomedian plate (No. 12), and, in some specimens, 
in the posterior part of the post-dorsomedian (No. 14). The shield itself is composed of 
six large plates, as represented in the diagram. The dorsomedian (No. 12) is hexagonal 
in outline, and its length and breadth are about equal. The sutures of the two anterior 
sides of this plate seem to be invariably symmetrical, but those of its posterior sides are 
frequently not. On its posterior margin, the post-dorsomedian plate (No 14) is obtusely 
pointed in the centre and truncated somewhat obliquely at each side. 
The ventral body shield is nearly flat, but is bent upward and outward at an obtuse 
angle at the sides, immediately behind and in a line with the insertion of the pectoral 
spines. The plates of which the ventral surface is composed are represented in outline 
in the principal figure (Fig. 1) of Plate VII, and they are also numbered to correspond 
with the numbers in Pander’s original restoration of the under surface of Pterichthys. 
The principal plates of the ventral shield, to quote the words of Hugh Miller, which 
are strictly applicable to the Canadian species, “are divided by two lines of suture, 
which run, the one longitudinally down the centre of the body, the other transversely, 
also through the centre ; and they would cut one another at right angles, were there not 
a lozenge-shaped plate” (No. 16) “inserted at the point where they would otherwise 
meet.”’ At the posterior end of the ventral shield, the outer margins of the post-ventrolat- 
erals (No. 21) are obliquely and concavely emarginate, while their central portions 
together form a rather narrowly rounded lobe which projects beyond the terminal end 
of the post-dorsomedian (No. 14) of the dorsal body shield. The tail opening is not much 
narrower than the maximum breadth of the united dorsal and ventral body shields; its 
outline is transversely and narrowly elliptical, but with the ends faintly angulated in the 
middle. The breadth of the tail-opening is rather more than twice its length along the 
median line. At the anterior termination of the median suture, a small supplementary 
plate (No. 17), which corresponds to the “ os semilunare ” of Pander, is intercalated between 
the front inner margins of the two ventrolaterals (Nos. 19). This little plate, which is 
inversely triangular, and rather broader than long, with its front margin straight, is entire 
and not divided longitudinally down the middle by a suture, as one might be led to infer 
from Pander’s figures. The six plates of the ventral surface which have been described so 
far are nearly always well preserved and clearly defined ; but, in advance of these, there 
seem to be three others which are very rarely preserved at all, and whose outlines are 
very difficult to trace. Immediately in front of the central accessory plate (No. 17), there 
is a small and very thin plate (No. 18) in the median line. It appears to have been 
somewhat semicircular in outline, though its margins are imperfect in the only specimen 
in which it can be seen, and it is distinctly strengthened in the middle by a prominent 
ridge which widens and becomes more prominent anteriorly. Judging by analogies with 
the Asterolepis of Hugh Miller (but not of Pander) this may have been the hyoid plate. 
On the under surface of the front margin of the head, there are two terminal lateral 
plates (No. 15) which, no doubt, correspond to the plates which Pander calls the “lower 
maxille.” The outer portions of these plates are bounded by the inner surface of the 
anterior cranial plates, and their inner margins, which meet above, diverge obliquely and 


! The Old Red Sandstone, Edinburgh, 1861, 7th ed., p. 74. 
