8 THE VOYAGE OF H.M.S. CHALLENGER. 



more or less obliterated. But there can, I think, be no question that in Antedon 

 ■perforata, Antedon rugosa, Antedon striata, and other species from the English Chalk, 

 together with some foreign species like Antedon tourtiie, Antedon semiglobosa, and 

 Antedon retzii, the inferior surface of the centro-dorsal was marked during life by a 

 large stellate opening which was considerably more than would be necessary for the 

 simple downward passage of the neuro-vascular axis of the stem. It seems to me very 

 probable, as I have explained elsewhere, 1 that the peripheral parts of this opening, which 

 are radially situated, may have given passage to tubular extensions of the body-cavity 

 into the stem, such as existed in Barycrinus, Cuprcssocrinus, and in other Palaeocrinoids. 

 An indirect confirmation of this view is afforded by the characters of the stem in the 

 Bourgueticrinidge, which resembles that of the young Comatula in all essential points. 

 The stem-joints of this family contain a set of five radial spaces which communicate with 

 one another from joint to joint, and probably also through the top of the stem with the 

 body-cavity within the calyx. The presence of these same radial spaces in the stems of 

 fossil Comatula? would account for the perforation of the lower surface of the centro- 

 dorsal, which would have effected the communication between the portions of the body- 

 cavity derived from the right peritoneal sac, that lie in the stem and in the calyx 

 respectively. In the ordinary species of Antedon the calycular portion of the ccelom is 

 much broken up by the rosette, and by the calcareous network which rests above it and 

 occupies the central funnel of the radial pentagon (PI. IV. fig. 36) ; but, as I have shown 

 elsewhere, 3 there are five median grooves on the ventral surface of the radials which 

 extend outwards in a similar position over the skeleton of the rays and arms, and 

 lodge the lowest portions of their cceliac canals. They are more distinct in some species 

 than in others, but are well shown in Antedon carinata (PL III. figs, id, 3a), Antedon 

 disciformis (PI. IV. fig. 26), and in Actinometra lineata (PL V. figs. 2a, 2c). When 

 these grooves pass from the ventral to the inner faces of the radials and descend into the 

 central funnel, they become closed into canals by the union of their edges with those of 

 the spout-like radial processes of the rosette. These canals, which I have called the axial 

 radial canals, are therefore the proximal ends of the five cceliac canals of the arms and their 

 extensions into the pinnules. As a general rule they become closed up by calcareous 

 tissue, and so do not reach the dorsal surface of the radial pentagon, which presents no 

 real openings except the central one occupied by the rosette (PL I. fig. 8c). The five 

 radial and five interradial processes of this structure are separated by passages which lodge 

 the paired branches of the five primary cords proceeding from the nervous envelope of 

 the chambered organ. These ten openings are well seen in PL I. figs. 6c, 8c ; PL III. 

 figs. 4c, 56; and also in PL V. figs, lc, 2c, 2e, bd, 5e, but in Antedon quinquecostata 

 and Antedon disciformis there are five additional openings on the lower surface of the 



1 On some New Cretaceous Coniatulre, Quart. Journ. Geol. Soc, 1880, vol. xxxvi. pp. 556, 557. 

 8 Trans. Linn. Soc. Land. (Zool.), 1879, ser. 2, vol. ii. pp. 77, 78. 



