358 BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



Ill all criiioids, but especially eniphasize<l iii such species as Araclinocrinus 

 hulbosus (fig. 595, pi. 16), a most extraordinary similarity and correspondence is 

 seen between the radials and the axillaries iii the arms. An analysis of the chiasma 

 formed l)y the dorsal nerves in the axillaries shows that this is merely a redupli- 

 cation of the conditions occurring m and about the radials. 



Axillaries are always followed, on each of the derivative arms, by two ossicles 

 which are the exact counterparts of the two ossicles immediately foUowing the 

 radials. 



The first of these ossicles is invariably attached to the axillary, and no normal 

 process ever takes place which results ui separating them, though in arm redupli- 

 cation separation ordmarOy occurs between the first and second. 



Similarly, the first of the corresponding ossicles follo^ving the radial is invari- 

 ably attached to it, and never becomes separated from it, though the radial may 

 become separated fi-om the basals or from the inirabasal below it by the intercala- 

 tion of a subradial'plate, from the adjacent radials by the development of inter- 

 radials, and from the basals by the degeneration and metamorphosis of the latter. 



The first segment of the free undivided arm m the crinoids is in reality the 

 axillary from which it takes its origin. In forms which do not possess division 

 series, as those belonging to the family Pentametrocrinidae, the radial occupies 

 the place and performs, the functions of this axillary. 



We, therefore, are led to assume that in reality the radial is morphologically 

 identical with the succeedmg axillaries, an assumption which is strengthened by 

 the fact that radials are occasionally doubled — that is, to all intents and purposes 

 axillary themselves — giving rise to two similar postradial series just as do axUla- 

 ries. It was the occurrence of true axillary radials, reported from tune to time 

 in various species, which first suggested the idea, subsequently shown to be 

 abundantly justified, that the two 10-rayed genera Promachocrinus and Thauma- 

 tocrinus were derived from the corresponding 5-rayed genera Cyclometra and 

 Pentumetrocrinus by the formation of axillaries by each radial, these later becoming 

 di\'ided into two by a process of twbinmg. 



Axillaries arise through the mcomplete fusion of two originally distinct seg- 

 ments. Since radials only differ fi-om axillaries in bearing a smgle instead of a 

 double subsequent series of ossicles, we may safely infer that, like axillaries, their 

 relationsliips are ^vith the ossicles following and not with those preccdmg. 



This is shown to be the case in axillaries in species in which the arm division 

 is of the so-called extraneous type, as in Metacrinus, or in which the division 

 series are of four ossicles, as m such species as ComantJius iemietti; the axillary 

 may be joined to the preceding ossicle by synarthry (as in Antedon), by syzygy 

 (as in all the division series except the first in Comanthus hennetti), or by oblique 

 muscular articulation as in Metacrinus; and may occur on the outer of the two 

 ossicles of an interpolated division series (as in Antedon, and in all species in which 

 the division series are composed of two ossicles only), on the epizygal of the first 

 syzygial pair (as in all the division series except the first in Comanthtts hennetti, 

 and in all division series consisting of four ossicles), or fortuitously in the distal 

 part of the arm (as in Metacrinus and in all species in which extraneous division 



