288 TVrTTT.-p.TTTJ 82, UNITED STATES NATIONAL MUSEUM. 



which might be interpreted as such recapitulation ever occurs. There is no evidence 

 in the developmental history of these genera to show that any of the ancestral 

 types from which they are derived possessed cirri much different from those which 

 they themselves possess; and, tracing the cirri backward through their ontogeny, 

 we find that, instead of becoming more different, they regularly converge toward 

 each other, which may be taken as almost certain proof that all of these forms, in 

 spite of the enormous amount of differentiation in other characters, still have 

 retained almost unchanged the primitive type of cirrus. 



A phylogenetic arrangement of the comatulids on the basis of their cirrus 

 structure is thus seen to be impossible, for the simplest type of cirrus found, with 

 little doubt the one nearest to the primitive cirrus, both of the comatulids and of 

 the pentacrinites, is characteristic of genera representing very specialized forms 

 which, judged by other characters, stand at or near the culmination of very diverse 

 lines of descent. The cirri, therefore, from a phylogenetic point of view, in show- 

 ing that no one group is especially developed beyond the others, though the lines 

 of development may be quite different, show essentially the same thing as all the 

 other available characters collectively. 



There are three lines of deviation from the primitive type of cirrus structure: 

 (1) In the direction of greater slenderness (figs. 83, p. 136, 98, p. 159, 308, p. 267, 

 376, p. 299, and 381, 382, p. 301); (2) in the direction of greater stoutness (figs. 99, 

 p. 160, 100, p. 162, 307, p. 265, and 369, p. 299); and (3) in the direction of greater 

 length (figs. 93, p. 153, 94, p. 155, 309, p. 267, 361, 362, p. 295, and 363-368, 

 p. 297). These three main lines are as a rule quite distinct, but more or less inter- 

 gradation is observable between them, especially between the two last. 



The slenderness of the cirri is one of the characteristic features of the Macro- 

 phreata and is very pronounced hi almost all of the forms, being often carried to 

 an extreme. Similar slenderness of the cirri is rare hi the Oligophreata, but is 

 found in some of the smaller or more delicate species, where as a rule it is an indi- 

 cation of the persistence or accentuation of an immature feature rather than an 

 acquired character, as in the Macrophreata. 



Slenderness is the result of the great reduction in size of each of the component 

 segments, this reduction being in the nature of a great decrease in the amount of 

 calcareous matter, as if its outer surface had been rubbed away, leaving the length 

 as it was originally. This reduction of the calcareous matter affects the central 

 portion of the segments much more than the denser ends, so that in a lateral view 

 they appear concave dorsally and ventrally, slender in the middle with prominent 

 ends (fig. 396, p. 309), or, as happily expressed by P. H. Carpenter, "dice-box 

 shaped." Owing to the fact that the length does not decrease hi proportion to the 

 decrease in thickness they become proportionately elongated, sometimes exceedingly 

 long. The slenderness is sometimes carried to such an extreme that the cirri as a 

 whole appear like very slender, almost invisible, threads, with bulky knots at inter- 

 vals marking the articulation, as hi Iridometra exguisita, Microcomatula mortenseni, 

 or Hathrometra sarsii (fig. 394, p. 309). 



Combined with slenderness resulting from a great reduction of the calcareous 

 base of the segments, there is usually a further reduction brought about by the 



