On THE STRUCTURE AND CLASSIFICATION OF THE TrEMATASPIDAE. 
29 
Shell. The median eye was large, complex, and important functionally. It consisted of two 
pairs of ocelli, one pair completely united in the median line and the other nearly so. The 
lateral eyes are reduced in size and in functional importance. The nasal pit is unpaired and 
lies behind the median eye. The mouth was small, circular, and situated near the centre of 
a group of oral plates. No upper and lower jaws were present. 
The liead of the Peltocephalata may be regarded as a modification of the céphalothorax 
of an arthropodan ancestor, consisting of three principal groups of segments, namely: the 
pre-oesophageal, including ail the parts derived from the cephalic lobes; the true thoracic 
segments : and the highly modified vagus segments, formed by the forward migration of 
from two to four abdominal segments and their complété union with the thorax. The brain, 
like that of Vertebrates, probably consisted of three groups of neuromeres derived from 
these three sources. 
The Peltocephalata hâve their nearest relatives among the known Invertebrates in the 
Trilobites and the Merostomata, having retained the general shape of the body, structure 
of the head, and mode of life characteristic of these Arthropods. 
Sufficient data are as yet unavailable for a permanent arrangement of the Peltocepha¬ 
lata into orders and families, but some modifications of the old arrangement may be made 
to advantage. The old subdivisions into Osteostraci and Heterostraci, proposed by Lanke- 
ster, should be abandoned as they do not mark natural divisions. 
The discovery of heavily armored, oar-like appendages in Cyathaspis and Tremataspis 
and their probable presence in Pteraspis and Tolypaspis unité these généra more closely 
with one another, and with the Asterolepidae than ever before. It is therefore inadvisable 
to isolate the Pteraspidian section, merely on the absence of multipolar bone cells unless the 
Pterichthydae are united with the remaining families under the heading Osteostraci. But 
such an arrangement would not sufficiently emphasize the resemblance between the oar-like 
appendages of Cyathaspis , Tremataspis, and Pterichthys, and the différence between these 
appendages and those of Cephalaspis. 
Moreover the Pteraspidae approach the Asterolepidae more closely than do the Trema- 
taspidae , in the division of the cephalic buckler into separate plates, and in its séparation 
into a true cephalic portion bearing the median eyes, and a thoracic one to which the oar- 
like appendages are attached. On the other hand, Tolypaspis, which must be placed close 
to Pteraspis and Cyathaspis on accout of the minute structure of the shield, shows no trace 
of a subdivision of its dorsal shield into separate plates. 
Tt seems to me, therefore, that we must recognize four subdivisions of the Peltocepha¬ 
lata of about equal value viz : The Pteraspidae , Tremataspidae , Asterolepidae and Cephalas- 
pidae. The Cephalaspidae are separated from the other families on account of the very pe- 
culiar shape of the head and the position and character of the appendages, although on the 
other hand, as shown by the connecting form Thyestes , a close relationship between Cepha- 
