98 



BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



rounded or denticulate flange-like processes, not coming into direct contact with 

 those of the neighboring ossicles (part 1, figs. 41, 42, 78, 85, 87, pp. 77, 131, 139, 

 143). When, however, the division series and arm bases are more or less turned 

 upward they come into contact with those adjacent on either side, with the result 

 that the possibility of lateral growth is greatly limited, and the ossicles become 

 sharply flattened or " wall-sided " laterally but, in compensation, deep dorsoven- 

 trally (part 1, figs. 43, 44, p. 77). Ordinarily, this lateral flattening is confined 

 to the elements proximal to the first brachial syzygy, ossicles which for phylo- 



genetic reasons are broader than those 

 following, but in extreme cases it may 

 be traced almost as far as the twentieth 

 brachial of the free undivided arm (fig. 

 208, p. 146). 



In a few instances species with broad 

 and patent central structures have de- 

 veloped an excessive broadening of the 

 division series and first two brachials, 

 and in these cases the ossicles which have 

 become broad have come into lateral con- 

 tact and have developed the sharply flat- 

 tened sides. A good example of this is 

 seen in Comanthus solaster. 



Carpenter believed that the lateral 

 flattening of the division series and lower 

 brachials of the free arms, or "wall-sided- 

 ness" as he called it, was, together with 

 the commonly correlated presence of 

 strongly developed side and covering 

 plates, an attribute of deep water species, 

 and that it was a systematic character of 

 the highest significance; in fact, his 

 ^'Basicurva" "Spinifera" and "GranuUferaP specific groups are based chiefly upon 

 it, though the two last invariably possess in addition "plated" ambulacra. 



Carpenter was to a considerable extent justified in this conclusion at the time 

 he wrote, but subsequent investigation has shown that this character, however 

 suggestive it may be, is far too uncertain to be relied upon, appearing in widely 

 divergent groups and being suddenly suppressed in certain species belonging to 

 others in which it is usually seen. 



In the Comasteridse (figs. 120, 133, 137. 138, 148, 150, 152, 154, 156, 158, 169, 

 171, 180-188, pp. 79-108) the division series are almost invariably widely free 

 laterally and strongly roimded dorsally and laterally, the only exceptions being in 

 a few species of Co7nanthus (compare fig. 171, p. 86) in which the elements of the 

 division series are especially broad. 



Fig. 183. — Lateral view of specimen of Comatilia 

 ihidometrifobmis fhom "albatross" station 



2670. 



