MONOGRAPH OF THE EXISTING CEINOIDS. 87 



may be narrowly produced, so that lines drawn from its tip to the outer ends of 

 the anterior sides of the axillary make considerably less than a right angle with 

 each other. Increase in the breadth of the arm bases, because of the slight curva- 

 ture of the anterior sides of the axillary, always broadens out the anterior angle, 

 which may become broadly obtuse and, through the gradual approximation of 

 the distal articular faces of the axillary and those of the first postaxillary ossicles, 

 may come to have nearly or quite straight sides- When there are several axillaries 

 in a single postradial series the anterior angle becomes progressively more and 

 more acute from the proximal outward. 



In four comatulid genera, Eudiocrinus (figs. 189-191, pp. 109-112), Thau- 

 matocrinus (part 1, figs. 113, 114, p. 181), Pcntametrocrinus (part 1, figs. 119, p. 

 185; 120, p. 187; and 121, p. 189), and Atopocrinus, the arms do not divide at all. 

 In the majority of cases there is a single division, which is invariably on the second 

 brachial ; if subsequent divisions occur they consist normally always of two or four 

 ossicles, excepting only in the genera Capillaster and Nemaster (figs. 180-182, 

 pp. 92-96), in which in the third and following divisions the first element in the 

 series of four is absent. 



In the multibrachiate species the amount of division of the series of ossicles 

 following each of the five radials is as a rule approximately the same, especially if 

 the divisions be numerous ; thus if a specimen of Comaster nova '.guinea be found to 

 have eight division series following one radial, it will be found to have eight fol- 

 lowing the other four also ; or if an example of Mariametra subcarinata be found 

 to have three division series following one of the radials, it will have three follow- 

 ing all the others. Also it is generally true that both derivatives of the first 

 division series and all the branches arising from them or their successors divide 

 exactly the same number of times, or, if the number of divisions be exceptionally 

 large, approximately the same number of times. 



Among certain species there is a tendency for each postradial series, after 

 dividing once, to divide a second time, but only on one side. When perfectly de- 

 veloped this method of division would result in the formation of 15 arms, 3 borne 

 upon each radial; but it is seldom that all of the postradial series are equally de- 

 veloped. In Comatula, NcocomateHa, Capillastcr, Ncmaster, Catoptometra, Cras- 

 pedometra, Heterometra, and especially in Ptilometra, Asterometra, Thalassometra, 

 Parametra, Gtyptometra, Chlorometra, Perissometra, and Pachylometra, there are 

 species which when adult may have anywhere from 10 to ] 5 arms, the second divi- 

 sion series being developed on only one side of from one to all five of the first 

 division series. This condition has also been noted in Antedon bifida and in Tropio- 

 metra picta. I have been unable to find any definite sequence in the place of ap- 

 pearance of these additional arms. In some of these species the number of arms 

 frequently rises above 20, in which case we usually find that the second division 

 series are developed one to each postradial series until the number of arms rises to 

 15, when the remaining second division series one by one appear, the whole series 

 of the second division being completed before the appearance of any of the divisions 

 of the third series. 



142140 21 Bull. 82 8 



