298 BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



serrations may be all small and subcqual; but usually the projection is slightly 

 V-shaped, with a comparatively large tubercle at the apex of the V flanked on 

 either side by from two to five or six other smaller tubercles; perhaps the com- 

 monest arrangement is a large median tubercle with two or three smaller ones 

 (forming the sides of the V) on either side. Distally the median tubercle gradually 

 increases in size, the lateral tubercles at the same time gradually diminishing until 

 in the outer portion of the cirrus the median tubercle only remains, forming a prom- 

 inent dorsal spine. The resolution of the broad rounded finely serrate transverse 

 ridge into a dorsal spine follows the same lines as described for the dorsal processes 

 of the Thalassomelridse. 



In the subfamily Comactiniuase a curious dimorphism of the cirri is found, 

 exactly com])arable to a similar state of affairs in the antedonid genera Antedon and 

 Compsometra. The most ])erfected type of cirrus in Comactinia and in Comatula (figs. 

 76, p. 129, and 327, p. 281) has from 10 to 15 segments, of which the more proximal 

 (not including the basal) are elongated, centrally constricted, and broadly oval in 

 cross section, and the distal are short, broader than long or squarish, not constricted 

 centrally, but much flattened laterally, so that in lateral view the cirri appear to 

 increase considerably in diameter distally. The more primitive type of cirrus 

 possesses the same number of segments in the same species, but the segments are 

 subequal, becoming only slightly, if at all, shorter distally than they are in the 

 earlier part of the cirri, and the cirri appear in lateral view of equal diameter through- 

 out, as the distal portion is only very sUghtly flattened (fig. 328, p. 281). 



These two very distinct types of cirri are correlated with the proportionate 

 amount of basal swelling in the arms and the shortening of the segments in the 

 earlier pinnules. In specimens or species in which the arms do not ex})and outward 

 from the first brachial (figs. 78, p. 131, 80, p. 133, and 108, p. 174), the cirri will be found 

 always to be of the second type; but if the arms gradually expand up to about the 

 twelfth or fourteenth braclual, slowly tapering from that point onward (figs. 76, p. 129, 

 and 1 07, p. 1 73) , then the cirri will be found to be, possibly with one or two exceptions, 

 of the first tyi^e. Among the Comactiniinse, and to a lesser extent among the 

 AntedoniniB, the earlier pinnules of specimens or of species with swollen arm bases 

 and the first type of cirnis are composed of proportionately shorter and broader 

 segments than those with arms which taper evenly from the base to the tij) and 

 with the second type of cirrus. 



In Comatula pectinata or in C. purpurea, where the arms of the anterior ray 

 may be evenly tapering but the arms of the other rays swollen, there is frequently 

 a mixture of these two cirrus types, the proportion of the second to the first being 

 about the same as the proportion of slender to stout arms. 



Both of these cirrus t_y]ies occur frequently in the same specimen in Comatula 

 pectinata and in C. purpurea; both also occur, but, so far as I have seen, never in 

 the same specimen, in Comactinia echinoptera. In Comatula rotalaria, C. etlieridgd 

 and C. micraster only the second type is found; but all three of these species lose 

 their cirri before acquiring the swollen arms so characteristic of the adults. 

 Strangely enough, though the swelling of the arms is carried to an extreme in 

 Comatula Solaris and in Comatvlella brachiolata, the cirri of these two species are 



