On THE STRUCTURE AND CLASSIFICATION OF THE TrEMATASPIDAE. 
17 
the shell is bent inward all round the slit to form a short vertical fuunel, whose outer end 
opens iuto the bottom of the frontal pit and the inner end iuto the interior of the head. 
This dépréssion lias been regarded as the site of an olfactory organ (Schmidt) or para- 
physis (Rohon). Its homology with a similarly located dépréssion and slit-like opening in 
Gephalaspis is not open to doubt. 
I regard, provisionally, the three anterior opeuings as the siteof the parietal eyecomplex, 
for the followiug reasons: (1) Because of the general similarity of these opeuings to the tri- 
ocular median eye openings in Limulus and the Merostomata. (2) Because the assumption har- 
monizes with the conclusion, based on anatomical and embryological evidence, that the an¬ 
cestral Vertebrates must bave been provided with median eyes composée! of at least two 
pairs of ocelli that were of much greater functional importance than those of any recent 
Vertebrates. (3) The dumb-bell openings are not in the right place for the lateral eye or- 
bits, as the latter are usually either on the same level as the median eye, or in front of it, 
not behind it. (4) If the dumb-bell openings are the lateral eye orbits, there is no explana- 
tion for the marginal openings. 
It is not surprising that the median eyes of Tremataspis are so much larger and more 
important functionally than the lateral eyes, since this is the case in many Arachnids. 
The presence of an ocellus is not necessarily indicated either by a distinct opening, or 
by a lens, since in Limulus one pair of ocelli lies beneath a mere tubercle which in old spé¬ 
cimens may be worn off completely. A similar condition seems to prevail in Pteraspis and 
Tolypaspis. In the latter genus, the only evidence of a median eye is the presence of a 
slight excavation on the inner surface of the dorsal shield (see p. 22). Between this con¬ 
dition and that of Tremataspis where the orbits are apparently uncovered, we bave forms 
like Bothriolepis, with the orbits closed by hard plates, more suggestive of the cuticular 
covering of an Arthropod eye than the ossified sclerotic of a Vertebrate. In Gephalaspis also, 
the orbits were in some cases, asshownby the British Museum specimens, coveredby promin¬ 
ent dome-like continuations of the outer layer of the shell. 
These appearances indicate that the eyes were closed, not by mere ossification of the 
sclerotic, but by a continuation of the outer layer of the shell to form cuticular lenses or 
corneas like those in an Arthropod. Ossifications of the sclerotic are as a rule characteristic 
of much larger animais than the Ostracoderms , and ones in wliich there is a higher devel¬ 
opment of the endoskeletal tissues. Besides, I know of no case among true Vertebrates 
where the ossification of the sclerotic extends о ver the whole front of the eyeball, as it 
appears to do in Bothriolepis and Gephalaspis. 
In Limulus , there is a curious ingrowth, or infolding of the shell, over the proximal 
end of the median eye tube where the latter joins the brain. In Apus the infolding is more 
conspicuous and lies just in front of the median eye. The ingrowth closes a kind of anterior 
neuropore, or the pore through which both the median eye tube and the fore-brain vesicle 
opened to the exterior, before the final closure of the cerebral vesicle. The infokled 
Зал. Фнз.-Мат. Отд. 
3 
