232 BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



207, p. 239) ; they may also be suppressed except at the interraclial angles (figs. 79, 

 p. 132, and 152, p. 221), or may be suppressed in the midradial line (fig. 196, p. 237), 

 or may even be absent altogether (figs. 162, p. 223, and 164, p. 227) ; it occasionally 

 happens that they only occur on half of the periphery of the centrodorsal (fig. 78, 

 p. 131). 



In some cases the obsolescent cirrus sockets, after losing their cirri, instead of 

 regenerating new cirri give rise to more or less elongated jointed tubercles which, if 

 the dorsal pole of the centrodorsal is spinous, may have a similarly spinous surface. 

 Occasionally these are not developed as jointed tubercles, but as attached processes 

 with their bases entirely filling the area originally occupied by the cirrus socket. 

 Both of these types must be regarded as the equivalent of an entire cirrus reduced 

 through degeneration to a single segment. 



The centrodorsal is ventrally in close apposition to the radials all around, at 

 least in the more central portion, the only break being in the interradial angles 

 where the basal rays, the small rounded or rhombic ends of which are externally 

 visible, may come between them (figs. 194, p. 237, 203, p. 239, and 208-216, 

 p. 241). 



In many species of the Comasteridse, perhaps in most of the larger forms, and 

 in many species belonging to other families, most noticeable in the Zenometrinse, 

 Thalassometridse, and Charitornetridse, deep narrow clefts extend inward between 

 the dorsal surface of the radials and the ventral surface of the centrodorsal (figs. 

 166-169, p. 229, 172, p. 231, 194, p. 237, 203, 204, p. 239, and 208-216, p. 241). 

 These clefts are most obvious in those comasterids in which the centrodorsal is 

 reduced to a stellate plate, and sunken below the dorsal surface of the radials. 

 They terminate inwardly against the inner portion of the ventral surface of the cen- 

 trodorsal, which is in close apposition with the inner portion of the dorsal surface of 

 the radial pentagon, and thus form blind cavities strictly homologous, as suggested 

 by P. H. Carpenter, with the smaller so-called interarticular pores in the stems of 

 the pentacrinites (fig. 127, p. 197; in the upper third of column). As the basal 

 rays always maintain the same relative length, they form externally five conspicuous 

 bridges separating those clefts in the interradial angles (figs. 194, p. 237, 214, 

 p. 241). 



P. H. Carpenter noticed that the ventral surface of the centrodorsal, which is 

 applied to the radials, is divided by ridges or grooves into the five trapezoidal areas 

 in which the radials are lodged, and that these are occasionally marked, toward 

 their inner borders, with more or less definite pits which receive the ends of the 

 radial axial canals (figs. 259, 260, p. 255, 262, p. 257, 280-283, p. 261, 593, pi. 15). 

 In most cornatulids every two fossae are separated by one of the five basal grooves 

 which lodge the basal star (figs. 243-249, p. 251); but if no basal star be present, 

 as in most of the macrophreate species, the radial fossae on the centrodorsal are 

 usually separated by moderately sharp ridges (figs. 280-283, p. 261). 



Internally the centrodorsal is excavated into a deep cavity for the reception 

 of the chambered organ and associated structures, and the ventral edge, especially 

 in the Macrophreata, is usually all around more or less produced inward so as to 

 result in the formation of a lip somewhat overlapping the central cavity after the 

 manner of a velum or diaphragm (figs. 66, 67, p. 93). 



