48 BULLETIN 82, UNITED STATES NATIONAL MUSEUM VOLUME 1 



or slightly curved or more or less abruptly bent in the middle. Sometimes this rod 

 becomes stout and roughened at both ends, or irregularly broadened and perforated 

 at one or both ends, or, rarely, converted into an irregular cribriform plate or an irregular 

 branching spicule. In some cases there are two rods, one longer and one shorter, 

 smooth or roughened, plain or with variously modified ends, rarely converted into an 

 irregular plate. Occasionally, especially in the species of Antedon, ambulacral deposits 

 are absent. (For the details of these deposits see vol. 1, part 2, pp. 264-265, and 

 beyond.) 



Some of the species of Antedon may also have spicules in the tentacles. 



Relationships of the genera. Since in this subfamily the characters differentiating 

 the various genera he almost wholly in the oral pinnules, and since in the comatulids 

 as a whole the transformation of the oral pinnules from food collecting to tactile or 

 defense organs represents specialization, the species in which this is at a minimum may, 

 other things being equal, be considered as the most generalized. 



The minimum of specialization in the oral pinnules occurs in Antedon mediterranea 

 and in A. adriatica in which PI is the only outer pinnule to be modified, and, generally 

 speaking, differs but little from P 2 and the following pinnules except in length. The 

 assumption that these species are more or less primitive in character is borne out by the 

 very generalized nature of the cirri which, though with numerous segments, are quite 

 featureless and resemble, in a broad way, the first cirri of pentacrinoids, as well as by 

 the similarly generalized nature of the IBr scries which resemble those of pentacrinoids 

 more than do those of any other species. 



In the other species of Antedon, excepting A. petasus and nuttingi, P 2 though of 

 the same size as P 3 , does not bear a gonad and may lack the ambulacral groove; the 

 cirri, strongly recurved distally and with short distal segments, are of a more specialized 

 kind than those of mediterranea and adriatica, and the same is true of the very short 

 IBr series. The other Atlantic species appear in every way to be comparable to the 

 Pacific species formerly included in Compsometra in which, however, the cirri may be of 

 the short-segmented type (as in A. incommoda), or greatly attenuated and elongate, 

 or more or less generalized, while the segments of the lower and middle pinnules have 

 strongly everted and produced distal ends. 



In Antedon petasus P 3 is similar to PI and intermediate in length between that 

 pinnule and P 3 ; in its other features this species closely resembles the second group of 

 species of Antedon. A. petasus is so very like the species of Mastigometra in every 

 way as to render its separation from them a matter of some difficulty, and to suggest 

 that perhaps Mastigometra had best be considered as a synonym of Antedon. 



In Annametra Pj and P 2 are alike, both similarly modified, with 18 to 35 segments, 

 and are more different from P 3 than is true of the preceding examples, while the cirri re- 

 semble those of all the preceding except Antedon mediterranea and A. adriatica. 



In Euantedon PI and P 2 are of the same character, with relatively few segments, 

 not more than 23, and differ from P 3 more strongly even than in Annametra; the cirri 

 also are modified in a rather different direction. 



In Argyrometra PI and P 2 become still more reduced, with less than 15 segments, 

 and the cirri, broadly speaking, are like those of Euantedon. 



Of the genera in which three pairs of pinnules are modified as oral pinnules, Irido- 

 metra has them all long, but similar. Andrometra, with P 2 enlarged and assuming the 

 functions of PI as seen in Antedon, appears to be more generalized than the other 



