242 BULLETIN 100, UNITED STATES NATIONAL MUSEUM. 



of R. alexandri. In R. syinbolicus the plates are keeled, the edge of 

 the keel being very thin, uneven, and rough but not spiny. In R. 

 mimicus there is no keel. 



As there are 2 names in the field for the same group, I have exam- 

 ined the species with the idea of instituting two subgenera. About 

 the onlj^ characters I find are as follows: Rosaster (including alex- 

 andri^ nanirms, and perhaps con-finis) is small and the gonads are 

 rather closer to the margin, although this may be due to the fewer 

 abactinal plates, especially of the interraclial regions. The adambu- 

 lacral spines are generally few — 4 to 7 or 8 or, leaving out conflnis, 

 4 or 5. In Nereidaster (including symhoUcus, hipunctus, and mi- 

 micus) the size is larger. The gonads are spaced from the margin 

 in a line parallel to the median radial (but this is strongly indicated 

 in R. alexandri), and the furrow spines are numerous — 8 to 15. In 

 the present state of our knowledge I do not feel justified in advocat- 

 ing this distinction. I fancy the slight difference in the gonads will 

 break down when more species are examined, and the difference in 

 the number of furrow spines is bridged over by R. confinis, which 

 is intermediate between the extremes. Even R. symholiciis may have 

 as few as 8, 



It should be mentioned in this connection that R. alexandri differs 

 from the other species, with the possible exception of R. confinis, in 

 having more spiniform granules on the abactinal plates, these being 

 thorny and 1.5 to 2 times as long as wide, blunt, or pointed. The 

 furrow and subambulacral spines are unusually long. One of the 

 latter is generally enlarged over the other 1 or 2. These charac- 

 ters may be of considerable importance, but they seem too slight for 

 a generic division. In Mediuster tenellus Fisher a somewhat anal- 

 ogous elongation of the granules into short spinelets has taken place. 



Nymphaster protentus approaches this group rather closely, but 

 it is useless to speculate on external resemblances. The arrangement 

 of the gonads and the internal structure of the abactinal skeleton is 

 unknown. 



Nymphaster fyorifer possibly belongs to Bfediaster. The internal 

 anatomy is not known. 



The following synopsis will contrast the differences between 

 Nymphaster, Rosaster, Mediaster, and allies: 



c\ With internal radiating ossicles connecting the strongly tabulate plates of 

 the papuhir areas; gonads three to many arranged in series parallel to 

 the radius, or following the contour of margin. 

 ^\ With calcified interhracliial septa, supplementary internal actinal inter- 

 mediate plates, unusually large, and fewer abactinal plates, and large 



triangular papular areas Ncctria. 



6*. With membranous interbrachial septa, no internal actinal intermediate 

 plates, and with moderate or small abactinal plates and papular areas. 



