90 BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



easily seen. I have elsewhere given reasons for believing that of the arms arising 

 from the IIBr axillaries those which are external are phylogenetically more 

 advanced as they are more perfected than those which are internal. This holds 

 good, of course, for all the derivatives from these axillaries. It is therefore to be 

 expected that the outer branches, being phylogenetically more advanced, will be 

 more subject to division, exactly as we see is the case in such genera as Comatella, 

 Cenometra, Ptilometra, Endoxocrinus, etc. 



In the Himerometridae and in certain zygometrids, as just explained, exactly 

 the reverse is the case; there are usually more division series internally than ex- 

 ternally- This also is to be expected, for the Himerometridse, as shown by their 

 very short brachials and curious pinnulation, are retrogressive derivatives from 

 highly multibrachiate stock, and we should therefore assume the retention of this 

 multibrachiate character longer internally than on the phylogenetically more 

 advanced outer arms. 



In the genus Comanthina (fig. 138, p. 79) the third division is normally of both 

 two and four ossicles, as in Himerometra, but the order in which they occur in 

 that genus is reversed, so that the IIIBr series are 2 externally and 4(3-|-4) inter- 

 nally. On the left posterior ray, sometimes on both posterior rays, the arrange- 

 ment commonly reverts to that of Himerometra, 4(3-(-4) externally and 2 inter- 

 nally, and may later be still further simplified by the reduction of all the series 

 of four ossicles to two. 



In the larger species of the genus Comaster, as Comaster novceguineos (figs. 

 148, 150, p. 83), the first division series gies rise to two similar branches as usual; 

 the second division series, however, bear undivided arms externally and third 

 division series internally, the third division series bear undivided arms internally 

 and fourth division series externally, and so on, this alternation sometimes continu- 

 ing for eight or nine division series. The resulting appearance is as if each second 

 division series gave rise to two stout main trunks, each of these trunks giving 

 off arms on alternate ossicles, first on one side and then on the other, and finally 

 terminating in a pair of arms themselves. The first arm is always given off 

 externally from the main trunk, so that the arrangement of the arms along the 

 trunk is exactly like the arrangement of the pinnules along one of the terminal 

 undivided arms, excepting only that the arms are given off from alternate ossicles, 

 whereas the pinnules occur on every ossicle; but as the ossicles of the division 

 series are united by nonmuscular articulations, the arrangement of the arms is 

 precisely what the arrangement of pinnules would be on an arm with every 

 alternate articulation a syzygy. 



In the smaller species of Comatella the second division series always gives 

 off internally an undivided arm and externally a third division series (fig. 154, 

 p. 83). When a fourth division series is present it is always given off from the 

 outer side of the third division series, the inner derivative of the third division 

 series remaining undivided. Tn Cotnatella nir/ra (fig. 156, p. 83), which may 

 have six or seven division series beyond the radials, the same peculiarity is found, 

 but it begins with the second division series instead of with the first, so that 

 the axillary of each second division series gives rise to two main trunks, each 



