25G BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



Tho cirri nre arranged hi more or loss crowded all<'niatiiig rows in all the 

 macro pkrcate genera except in those included in the subfamily Zenometruia! (figs. 

 109, p. 175, 110, p. 176, 111, p. 177, 208-216, p. 241, 228, p. 245, and 558, pi. 5), in 

 Atopociimis (fig. 227, p. 245), and in AtelecAnus (figs. 123, p. 192, 124, 125, p. 193, 

 218, 223, p. 243, 414, p. 319, and 573, 574, pi. S), where they are arranged in col- 

 umns. These six genera, therefore, are at once distinguishable from all other 

 comatulid genera by a glance at the centrodorsal. Moreover, they arc at once dis- 

 tinguishable among themselves; m Atelecrlnus (figs. 123, p. 192, 124, 125, p. 193, 

 218, 223, p. 243, 414, p. 319, and 573, 574, pi. 8) the cirrus sockets are bounded 

 laterally, or laterally and ventrally, by a strong horseshoe-shaped ridge, or by high 

 lateral ridges, whereas in the other genera they are mere undifferentiated pits iti the 

 general surface of the centrodorsal; there are 10 or 15 very definite columns of cirrus 

 sockets, but the surface of the centrodorsal is not marked off into radial areas. 

 This is tho case also in Lcptometra (figs. Ill, p. 177, and 219, p. 243) and ixi Adelo- 

 metra, but in the former they are entirely separated from each other, while in the 

 latter they arc closely crowded. 



In Atopocrinus (fig. 227, p. 245) the centrodorsal is very long and sharply 

 conical and is divided into 10 narrow cirriferous areas by five high serrate intci'- 

 radial and five smaller similarlj' serrate midradial ridges. Each cirrus socket 

 projects strongly over tho proximal portion of the one just below it and jiossosses 

 strong fulcral ridges which are not found in the cirrus sockets of the species of 

 ZenomctrinsE (figs. 109, p. 175, 110, p. 176, 111, p. 177, 208-216, p. 241, 228, p. 245, 

 and 558, pi. 5). 



In Balanometra, Zenometra (figs. 109, p. 175, 214-216, p. 241, and 558, pi. 5) and 

 PKatliyromdra (figs. 110, p. 17G, 208-213, p. 241, and 22S, p. 245) the centrodorsal is 

 divided into five radial areas by strongly developed ridges, furrows, or broad bare 

 areas. In Balanometra and in the Atlantic species of Zenometra (figs. 215, 216, 

 p. 241, and 558, pi. 5) there are 10 columns of cirrus sockets, two in each radial 

 area; Balanometra has tho radial areas marked off by broad furrows, and the two 

 colunms of cirrus sockets in each radial area more or less widely separated, 

 whereas in the Atlantic species of Zenometra the radial areas are delimited by 

 strong ridges, and tho two columns of cirrus sockets in each radial area are close 

 together. In the Pacific species of Zenometra (figs. 109, p. 175, and 214, p. 241) 

 and in Psathyrometra there are three or four columns of cirrus sockets in each 

 radial area, these radial areas being marked off by bare spaces not raised above 

 the general surface of the centrodorsal. In Zevx)metra triserialis the distal portion 

 of tho centrodorsal is tliickly covt^red with spines, while the three equal {■oluinns 

 of cii'cular cirrus sockets in each radial area are closel}* crowded. In Psathyro- 

 metra the dorsal pole of the centrodorsal is smooth, and the cirrus sockets are 

 arranged in tlu-ee or four columns in each radial area; they arc usually more or 

 less separated, and each cirrus socket is correspondingly soparat<!d from its neigh- 

 bors in the same column. If there arc three columns in each radial area, the 

 median column tends to bo deficient, tho outer columns converguig and meeting 

 beyond it. In one species this middle column is reduced to a single socket. If the 

 columns of cirrus sockets are crowded, the sockets become dorsoventrally elongate. 



