NATURAL CLASSIFICATTOX OF THE TRILOBITES 113 



ment of the pygidium for dividing the trilobites into grou[)s 

 larger than families, and it seems evident from the present 

 state of knowledge that it is impossible to make this charac- 

 ter of more than family or even generic value. Many of the 

 genera which must naturally be included in the Archiaspides 

 have pygidia that cannot be said to be rudimentar}^ obsolete, 

 or wanting in function. Even those genera having pygidia 

 with few segments, as Mesotiacis, Holmia, Paradoxides, Sele- 

 nojjeltis, Dicranurus, Bronteus, Harpes^ etc., show in many 

 other more important characters that they are highly differ- 

 entiated and specialized forms, and that this feature is one 

 expression of such development. The futility of the scheme 

 is at once evident when a comparison is made between allied 

 genera which present marked differences in the size and 

 segmentation of the pygidium; as Phaeops and Dahnanites, 

 Ceraurus and Pncrinurus, Galymmene and Homnlonotus, Harpes 

 and Trinudeus, Mesonaeis and Zacanthoides, Paradoxides and 

 DiJcelocephalus. 



The last classification to be noticed is that of E. J. Chap- 

 man, ^^ in 1889, in which four sub-orders or primary groups are 

 proposed, differing considerably from any previous arrange- 

 ment, and based upon arbitrary features of general structure 

 and configvu-ation, especially the form of the glabella, whether 

 wide, conical, or enlarged. Twenty-seven families are rec- 

 ognized. In this scheme Tnnueleus, Ampyx^ and JEglina 

 form one section ; Paradoxides and Acidaspis, together with 

 phaeops and Uncrinurus, another; all under one sub-order. 

 Omitting the Agnostidae, there are here considered in a single 

 sub-order the most cliaracteristic representatives of nearly all 

 the types of triloljite structure. Proetus, Cyphaspis, and 

 Arefhusina fall into three sections, under two sub-orders, 

 although these genera, on account of their great similarity 

 in essential points, are placed in a single family by most 

 authors. A further analysis of this classification in its 

 broader lines would be unprofitable. It is sufficient to state 

 that the facts obtained from the study of the ontogeny of 



8 



