
No. 425.] THE TREMATASPIDA. . 387 
The Entapophyses and the Trabecule. —1t is doubtful whether 
the endolymphatic ducts of Rohon are actual perforations of the 
shell. In sections they are seen to be deep tubular infoldings, 
probably closed at the inner end. They appear to be comparable 
with those infoldings, or entapophyses, on the dorsal shield of 
Limulus which serve for the attachment of dorso-ventral muscles. 
This interpretation is strengthened by the fact that serial sec- 
tions show the presence of another pair of bony ingrowths, just 
behind and in line with the so-called endolymphatic ducts. They 
are long deep plates that unquestionably serve for the attach- 
ment of muscles, since the frayed-out ossified tendons are still 
attached to their inner ends. These entapophyses are of great 
importance, as they indicate a similar arrangement of muscles 
» and appendages to that seen in Limulus. They probably served 
for the attachment of dorso-ventral muscles, arising from the 
dorsal surface of the cartilaginous cranium and from the various 
pairs of cephalic appendages. 
The presence of a similar set of muscles in Pteraspis and 
Cyathaspis is indicated by radiating impressions or lobes on the 
inner surface of the shell, similar to the radiating muscle 
markings on the dorsal wall of Limulus. 
Transverse sections of the whole head show that the shell is 
specially thickened in the region of the crista occipitalis, and 
that the margin of the shield is considerably strengthened by 
a network of bony trabecule uniting the edges of the dorsal 
and ventral walls. The marginal trabecule are very similar 
to those I have described in Limulus. The marginal cells of 
Eukeraspis are probably produced by a special arrangement 
of these trabecule. | 
A loose network of bony trabeculz is also developed round 
the median and lateral openings of the dorsal shield and along 
the incisions of the ventral wall where the appendages are 
attached. With the above exceptions, the inner surfaces of 
the dorsal and ventral shields are smooth. 
The minute structure of the shell of Tremataspis and other 
members of the ostracoderms can be best explained, as we have 
pointed out elsewhere, on the assumption that it is a modification 
of a three-layered dermal skeleton similar to that of Limulus. 
