282 BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



beyond the middle of the cirri the superior tension of the larger dorsal ligament bun- 

 dle prevents the production of the distal dorsal edge of the cirrals at the same rate 

 as the distal ventral edge is produced. As the potential growth of the cirrals is the 

 same all around the edges (both distal and proximal) the excess growth of the 

 dorsal part of the distal edge over what is possible owing to the restrictions conse- 

 quent on the curvature is accommodated by an eversion of the distal edge and its 

 production into a prominent dentate frill, which later is usually specialized and 

 developed into a more or less sharp dorsal spine. In case the curvature of the 

 cirri is not very sharp, the excess of stereom deposited on the dorsal side of the 

 cirrals may be evenly distributed, taking the form of a swelling of the dorsal side, 

 as in Comanihus bennetti, or (secondarily) of a longitudinal carination (fig. 369, p. 299) ; 

 such swelling or carination often occurs hi combination with more or less pronounced 

 spines. 



The opposing spine approaches nearer the terminal claw than to the dorsal 

 spines in structure. As there is practically no motion possible between it and the 

 terminal claw, its origin could not have been quite the same as that of the dorsal 

 spines. It is probably the result of excess growth localized on the distal dorsal 

 border of the penultimate segment for purely mechanical reasons, its subsequent 

 molding into a sharp spine resembling in all essentials the terminal claw being due 

 to the same causes that operated in the case of that element. 



The dorsal spines or dorsal processes proximal to the opposing spine form a 

 finely graduated series from the most primitive or rudimentary toward the base of 

 the cirri to the most highly perfected on the antepenultimate segment (figs. 365- 

 367, p. 297). In many cases the change is slow and uniform, and there is a pro- 

 gressive specialization segment by segment to the end. This is especially to be 

 noted in spiny cirri xvhich are short or of moderate length; in long cirri the spines 

 commonly become perfected at some distance from the tip, and no further change 

 is visible from that point onward. 



. This gradual development of the dorsal processes is correlated with (indeed, as 

 previously shown, probably dependent upon) a similar gradual increase in the 

 amount of dorsoventral motion possible between adjoining segments. Very con- 

 siderable dorsoventral motion is allowed between the two to four or five basal seg- 

 ments; the next following are very closely united, and there is a very slow gain 

 in the scope of possible motion until the tip of the cirrus is reached; in very long 

 cirri the maximum is attained at some distance from the end and is continued to 

 the tip. There is practically no motion possible except in the planes including 

 the longitudinal (dorsoventral) axis of the body, as the fulcral ridges of the joint 

 faces all run straight across these from side to side; the basal segments collectively 

 allow of flexion through about 180, so that the cirri may at this point be bent 

 directly downward or directly upward so as to extend vertically (parallel to the 

 longitudinal axis) between the arms; no motion is possible between the smooth 

 proximal segments, and the scope of the motion permitted by the more distal seg- 

 ments is much more limited than that allowed between the basal; the outer part of 

 the cirri (beyond the rigid middle portion) can not be raised further than to bring 

 all of the segments into a straight line, and often a broad spiral is the extreme in 



