er Devonian and Carboniferous periods. As there
are, however, no calculations more doubtful or more to be suspected than
those on which the size and bulk of the extinct animals are determined
from some surviving fragment of their remains,--plate or bone,--I must
attempt laying before the scientific reader at least a portion of the
data on which I found.

[Illustration]

This figure represents not inadequately one of the most characteristic
plates of the Asterolepis. A very considerable fragment of what seems to
be the same plate has been figured by Agassiz from a cast of one of the
huge specimens of Professor Asmus ("Old Red," Table 32, Fig. 13); but
as no evidence regarding its true place had turned up at the time it was
supposed by the naturalist to form part of the opercular covering of the
animal. It belonged, however, to a different portion of the head. In
almost all the fish that appear at our tables the space which occurs
within the arched sweep of the lower jaws is mainly occupied by a
complicated osseous mechanism, known to anatomists as the hyoid bone and
branchiostegous rays; and which serves both to support the branchial
arches and the branchiostegous membrane. Now, in the fish of the Old Red
Sandstone, if we except some of the Acanthodians, we find no trace of
this piece of mechanism: the arched space is covered over with dermal
plates of bone, as a window is filled up with panes. Three plates,
resembling very considerably the three divisions of a pointed Gothic
window, furnished with a single central mullion, divided atop into two
branches, occupied the space in the genera Osteolepis and Diplopterus;
and two plates resembling the divisions of a pointed Gothic window,
whose single central mullion does _not_ branch atop, filled it up in the
genera Holoptychius and Glyptolepis. In the genus Asterolepis this
arch-shaped space was occupied, as I have said, by a single plate,--that
represented in the wood-cut; and the nail-shaped bone rose on its
internal surface along the centre,--the nail-head resting immediately
beneath the centre of the arch, and the nail-point bordering on the
isthmus below, at which the two shoulder-bones terminated. Now, in all
the specimens which I have yet examined, the form and proportions of
this plate are such that it can be very nearly inscribed in a
semi-circle, of which the length of the nail is the radius. A nail five
inches in length must have belonged to a plate ten inches in