274 BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



With the increasing differentiation in size of the ventral and dorsal ligament 

 masses in the cin-i comes also a progressive differentiation of the fosste which con- 

 tain them, and these come to resemble those on tlie earlier pimmlo segments. 



It is probable that the pinnules and the cirri represent the original type of 

 crinoi(hd appendage, and that these appendages were an-anged in five pairs, the 

 two components of each pair being, so to speak, back to back; but both the j)in- 

 nules and the cirri have become enormously reduphcated, while in addition the 

 former have come to lie along either side of long body processes of subsequent 

 development. 



When the origin of the cirri and of the cirrals is understood it becomes at once 

 evident why no brandling ever occurs in the cirri, such as frequently occurs in the 

 distal jjortion of tlie arms and at the bases of the pinnules. The cirri are true 

 uniserial outgrowths, both phylogenetically and ontogenetically, like the legs of 

 arthropods; and, like the legs of arthropods, tjjey may bifurcate at the base, though 

 tliis never happens except within the central cavity or within the substance of the 

 centrodorsal. 



At first tlie lines of division between tlie cirrus segments are, when tlie cirri 

 are viewed laterally, perfectly straight and at right angles to the longitudinal axis 

 of the cirri (figs. 553, 558, pi. 5) ; at this time also the cirri are straight and almost 

 or quite uniforml}' jointed processes. Correlatively with the gradual change of the 

 cirrus segments toward the adult type the portion of the line of division ventral to 

 the transverse articular ridge gradually leans distally, while the portion dorsal to the 

 transverse articular ridge, to a lesser degree, leans proximally. 



The amount of departure from a straight line exliibited by the lines of division 

 between the cirrus segments is in general proportionate to the motion to be accom- 

 modated. Tlius in cirri with long, proximal and short distal segments the lines of 

 division separating the former are almost straight and perpendicular to the longi- 

 tudinal axis of the cirri, while those separating the latter are obtuse angles (figs. 

 322, p. 277, 327-329, p. 281, and 339, p. 285). In the case of enormously enlarged 

 cirri, such as those of the species oi Asterometra (figs. 94, p. 155, and 362, p. 295), 

 however, the short outer segments, being physiologically too remote from the source 

 of nutrition, always remain in a comparatively undeveloped state, and the lines of 

 division between them are straight or nearly so. 



The obliquity of the course of the lines of division between the cirrals is the 

 result purely of mechanical considerations. If the central canal runs through 

 the middle of the segments, so that the ligaments on either side of it are in a state 

 of equilibrium (fig. 5876, pi. 13), the lines of division are straight and at right 

 angles to the longitudinal axis of the cirri; but if the central canal is ventral to 

 the center of the cirrus segments, so that the dorsal ligament bundles are larger 

 than the ventral (fig. 587a, pi. 13), a constant contraction operates, not only 

 within the ligament bundles themselves but also within their continuation in 

 the interior of the segments, which is proportionate to the difference in size 

 between the two ligament bundles; and tliis results in giving to the cirrus a curve 

 dorsalward proportionate to the difference in size and strength between the ventral 

 and dorsal ligament bundles, and in pulling distally the whole mass of the segments 



