242 BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



basals or infriibasals; as tlie infrabasals of the youri<^ Aniedon and the coronal 

 j)httc3 of the urchins surround the apical system in just the way that the infra- 

 basals of Marsupites (fig. 565, pi. 7) and of Uintacrinus (fig. 572, pi. 7) surround 

 the central ])liitcs, it seems to me that we must assume that tlie central i)lates are 

 the equivalents of the entire apical system (the terminal stem plate plus the 

 columnals) of the developing Antedon. 



There is additional evidence that neither Marsupites nor Uintacrinus ever 

 possessed a stalk; this evidence is purely circumstancial, but appears to be none 

 the less good. Both Marsupites and Uintacrinus have an enormous range; now 

 we find among the jclh'fishes forms which are purely pelagic, and otlicr forms 

 which are fbced for varying periods. The extent of the range of these different 

 types is very varied, the pelagic species having the greatest, and the longest fLxed 

 the least, range. When we compare the distribution of lifarsupites and Uinta- 

 crinus with that of the recent jcllylishes we find tliat the parallel is distinctly with 

 those types which are exclusively pelagic and pass through no fixed stage, and we 

 therefore appear to be justified in assuming that Marsupites and Uintacrinus, like 

 them, were always at all stages free swimming. 



All of the numerous and diverse types of centrodorsals are ultimately derived, 

 both phylogenetically and ontogenetically, from the type characteristic of the 

 comasterids, and the segregation of the cirrus sockets into columns, with the accom- 

 panying assumption of strong interradial ridges or furrows and of a more or less 

 pronouncedly conical shape, commences after the centrodorsal has attained an 

 appreciable size. In most cases all evidence of the early stages is lost tluough 

 the erosion or resorption of the dorsal pole, but in certain small species of Psatliy- 

 rometra, as for instance in Fs. inusitata (fig. 228, p. 245), the juvenile ])ortion of the 

 centrodorsal with its alternating rows of cirrus sockets which show no trace of 

 radial segregation, but resemble those of the genus Trichometra, is retained beyond 

 the mature portion in which the cirrus sockets are in columns and the columns are 

 grouped into radial areas by the development of definite furrows. 



^Vhen the centrodorsal is of the primitive type it increases in size proportion- 

 ately with an increase in the length and stoutness of the cirri; tlius in the Comaste- 

 ridae, Zygometrida3, llimerometrida^, Stcphanometridae, ilariametridae, Colobome- 

 tridae, Tropiomctridae, Calometridae, and Pentametrocrinids, and in the genera of the 

 Antedonidse in which the primitive type of centrodorsal is retained, the species with 

 small cirri have small centrodorsals, and those with large cirri have large centro- 

 dorsals; but if the cirri are arranged in definite columns the reverse is, within 

 certain limits, tnie; species with small and short cirri have larger centrodorsals 

 than those with longer and larger cirri; tlius the species of Thalassometndse and 

 Atelecrinidae have much smaller and more sharply conical centrodorsals than those 

 of the Charitomotridsc, while the species of Zenomctrinse have, in j)roportion to 

 their size, the smallest centrodorsals of any of the Antcdonidse. 



This fact is not always easy of ajjpreciation, for as a rule species with a columnar 

 arrangement of cirrus sockets do not lose nearly so much of the dorsal pole by 

 resorption as those with the cirrus sockets arranged in alternating rows, and hence 



