THE EVOLUTION OF THE EXOSKELETON. 353 



may possibly have some relation to the plates that are present in the cephalothorax 

 and branchial regions in their trilobite- or merostome-like ancestors. 



5. The armor may also break up into small polygonal platelets, of uniform 

 size, that are quite separate from one another, the lines of fragmentation following 

 the "ornamental" polygonal markings visible on the outer surface of the shield 

 in Limulus, (Fig. 200, A.C.D.), in Cephalaspis, or in Ateleaspis. (Fig. 200, B.) 



6. With this process of fragmentation there is a tendency to accentuate 

 the difference between the inner layers of more highly developed bony tissue, 

 and the outer layers that still retain their primitive stratification and parallel pore 

 canals, and into which neither the bone cells nor the vascular canals have pene- 

 trated to any great extent. The isolated ridges, spines, or tubercles of the outer 

 layer, initiate the dermal denticles of the vertebrates, and represent their enamel 

 and dentinal caps; the lower layers initiate the basilar plates composed of true 

 dermal bone. (Fig. 208.) 



7. During the phylogenetic process of fragmentation the two layers may 

 undergo unequal development. In such forms as the coelolepidae (Thelodus and 

 Lanakia) apparently only the isolated epidermal denticles are retained, while 

 the underlying network of bony plates and trabeculae has largely, or wholly, 

 disappeared. In Bothriolepis and related forms, the dentinal layer is scanty, 

 or for the most part absent, while the large bony lamellae of the inner and middle 

 layers are highly developed. In the pteraspidian section all three layers are 

 present, but the underlying ones have no true bone cells, only the spindle-like 

 dilatations of the pore canals. 



8. In the antiarcha, as in the vertebrates, the placoid bones were extensively, 

 if not entirely, covered with a layer of epidermal cells. This is indicated not 

 only by their general structure, but is conclusively demonstrated by the faint 

 impressions on their outer surface left there by the nuclei. 



9. The dentinal layer of the unfragmented buckler and of the isolated den- 

 ticles, like the exoskeleton of arthropods and the enamel of vertebrate teeth, was 

 probably the product solely of underlying epidermal cells. Their outer surface 

 was probably never covered with epithelial cells, except by secondary over- 

 growths resulting from infoldings like those in the developing teeth of vertebrates. 



10. A remarkable feature of the cephalaspidas is the union of the margins of 

 the upper and lower shields by anastomosing bony trabeculae which, like those in 

 Limulus, form the solid, or cancellous cornua, and the heavy hoop-like margin 

 along the front and sides of the cephalic shield. 



In Eukeraspis (Fig. 235, D) there are peculiar chambers (marginal cells of 

 Lankester), in the bony tissue on the anterior margin of the shield. They may 

 be merely enlarged cancellffi, or possibly, enclosures formed by a deeply serrated 

 or scalloped margin like that in Thyestes, A, or like the enclosures formed by 

 marginal infoldings in the trilobites, B. 



11. In Cephalaspis (Fig. 232), there are conspicuous oval areas on the lateral 

 margins, and behind the orbits, that are formed by thickened patches of bony 



23 



