228 BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



lids, -when we make due allowance for the cramped eonditions under which in the 

 latter cirrus formation occurs. 



The alternation and the irregular crowding of the cirrus rows in the comatulids 

 is the result of mechanical restraint cons('(|uent on the comparativeh" very slow 

 growth of the centrodorsal. This crowding and accompanyuig ii-regularity in 

 position is most marked in those comatulids in which the centrodorsal is the least 

 specialized, the species with the more highly specialized and longer types of centro- 

 dorsal having, as would be expected, more nearly attained a balance between the 

 CiUcareous and the uncalcified constituents of the stem. 



A few comatulids, belonging to the genus ChJorometra, have the cirri in five 

 radial colunms, exactly as the cii-ri occur on the pentacrinite stem (fig. 207, p. 239) ; 

 many have them confined to the radial portions of the centrodorsal, in which 

 the)' may occur in two, three or four columns, or more or less irregularly (figs. 192, 

 194, 196, 198, p. 237, 200, 203, 204, p. 239, 208-216, p. 241, and 227, 228, p. 245). 

 All of these species have long and liighly specialized centrodorsals. One species 

 always {Comatula purpurea), and several often, have from 5 to 10 cirri confined 

 to the interradial angles of the calyx (figs. 79, p. 132, 160, p. 223, 163, p. 225, and 

 182, p. 233) ; but in these cases these are alwaj^ the latest cirri to be formed, and 

 have persisted after the repression and resorption of all the others, wliich were 

 radial in position. 



W. B. Carpenter observed that the precise epoch of growth at wliich the 

 separation of the comatulids from the larval stem occurs varies greatl)^; thus, for 

 example, the young of the species of Hathrometra retain the stem until 20 or 30 

 cirri have appeared on the centrodorsal, which now conceals the basals, and the 

 pinnules are developed upon all the lower bracliials; whereas in Anfedon and in 

 certain other genera the stem is discarded when there are only 10 well-gi'own cirri 

 on the centrodorsal, the basals are still visible, and the lowest portions of the arms 

 are devoid of pinnules. The absolute size which is reached by the mature larve 

 before dropping off its stem also varies considerably, even mthin a single species. 



At the end of the pentacrinoid stage, when the centrodorsal of AntedonsepaT&tes 

 off from the lower part of the larval stem, "a minute 5-rayed perforation remains 

 at its dorsal pole, wliich corresponds to the central canal in the stem" of the stalked 

 species that gives passage to the neurovascular axis. This is very soon closed up 

 by calcareous deposit. In a number of fossil forms it has been noticed that this 

 opening is a characteristic feature, in some species extending "into a large stellate 

 impression wliich occupies a considerable space on the lower surface of the centro- 

 dorsal, and in the fossil condition is more or less obliterated." 



P. II. Carpenter believed that in these fossil species tliis opening in the centro- 

 dorsal at the dorsal pole is a larval character preserved in adult life; but I am 

 firmly of the opinion that it is a purely secondary feature, produced after death 

 by the erosion of the doi-sal pole, which in many of the recent species is in life 

 very tliin and composed of a rather loose calcareous deposit. A small amount of 

 erosion here would suffice to open the central cavity of the centrodorsal to the 

 exterior, without producing niucli, if any, change in the remaining more dense 

 portions of that centrodorsal, or in the radials. The large stellate central opening 



