168 BULLETIN OF THE 



fragility, — the "rosette" of a small An(cdon being massive in com- 

 parison, — I found it impossible to preserve them intact ; but their 

 position is indicated in Fig. G 6. 



There is nothing specially remarkable about the radials and the lowest 

 ai'm-joints of Atelecrinus, but the arm-joints generally are somewhat 

 peculiar in their characters. They are rather longer than in most 

 GomatulcB, and have shallower bodies, while the muscle plates which rise 

 from about the middle of each joint arc unusually thin. There is, in 

 consequence, a series of large gaps between the muscle plates of succes- 

 sive joints, which are occupied by correspondingly large muscular bun- 

 dles (Figs. 1, 2, 7). These are not concealed from view by superficial 

 perisome as they are in ordinary Comatulce ; but the food-groove lies 

 close down upon and between the muscles, all the structui'es connected 

 with it being very much reduced and contracted together, as I have 

 sometimes found to be the case in Ant. rosacea. At the sides of the 

 groove are a few scattered " sacculi " (Fig. 2). The pinnules which are 

 borne by the twelfth and following joints are comparatively short and 

 styliform, and are composed of ten or twelve elongated joints. Their 

 ambulacra are more spotted with " sacculi " than those of the arms, and 

 are fringed with tentacles, of which I have as yet fouud no traces on the 

 arras. 



The following are the chief points of difference between Atelecrinus 

 cubensis and A. halanoides. In the former species (Fig. 7) the base of 

 the centrodorsal is much wider relatively to its height than in the latter 

 (Figs. 1,3); the cirrhus sockets are more closely packed, and the points 

 of their horseshoe-shaped rims more prominent ; while the iive processes 

 at the ventral rim which support the basals are more strongly marked 

 than in any specimen, large or small, of Atelecrinus halanoides. In cor- 

 respondence with this feature, the shape of the basals is very different 

 in the two species. In the little A. cubensis they form a kind of belt of 

 tolerably uniform height with its interradial angles somewhat produced, 

 which everywhere separates the first radials from the centrodorsal. The 

 second radials are squai'er, and the axillaries project rather more into 

 them than is the case in A. hcdanoides ; while the first brachials are 

 relatively shorter, the second longer and projecting more into the first, 

 and the three following joints also relatively longer than in the larger 

 species. 



Taking all these facts into consideration, I think it very probable that 

 we are dealing with two distinct species. Although Pourtalrs's original 

 specimen {A. cubensis) is very considerably smaller than those obtained 



