EARLY DEVELOPMENT OF CRIBRELLA OCULATA. 401 



wall of the pre-oral coelom in the ' neck region,' but in this case the position of these 

 parts does not remain constant. In fig. 52 the tip of the ventral horn is opposite the 

 future radius 1, but in stage E it bends upwards, pushing the ventral wall of the 

 pre-oral coelom upwards and backwards. Hence in figs. 83 and 84 it is seen to have 

 reached the base of radius 5, and in stage F it forms an aboral radial process lying over 

 radius 5 (figs. 96, 97, 98). In doing this it has pushed the ventral wall of the pre-oral 

 coelom so far upwards, and later backwards, towards the dorsal wall, that this pre-oral 

 coelom, instead of, as in stage D (figs. 56 and 57), extending dorso-ventrally from inter- 

 radius 4/5 to radius 1 (with an aperture of about 160°), is now reduced to a narrow 

 canal lying in inter-radius 4/5, from which it runs forwards to the main bulk of the 

 pre-oral coelom in the pre-oral lobe. (See A, B and C, diagram 2.) 



Thus the inter-radius of the stone-canal is 4/5 of the larva, and "the plane of adult 

 bilateral symmetry is aloug inter-radius 4/5 — radius 2 of the larva. This plane passes, 

 as we have seen, at right angles to the sagittal plane of the larva, and it is at an angle 

 of 72° to the larval coronal plane, which naturally passes along radius 3 and inter- 

 radius 1/5. 



The dorsal horn of the hypogastric ccelom forms the 'aboral radial process over 

 hydrocoelic radius 4, the posterior portion over radius 3, and the ventral horn, as is 

 clearly indicated in figs. 86 and 87, forms the processes complementary to 2 and 1. 

 The last is the latest, as it cannot be formed by the ventral horn till it reaches radius 5 

 as detailed above. From this it is clear that the ventral horn provides three aboral 

 processes and the dorsal one ; hence if we number the aboral processes clockwise, 

 similarly to the hydrocoelic, we find that I. lies over 5, II. over 1, III. over 2, IV. over 

 3, and V. over 1. Thus is produced the appearance of the torsion actually observed 

 in so many asterids by previous workers. Had the dorsal horn grown forwards 

 equally with the ventral horn, the stone-canal would have been formed in inter-radius 

 1/5, the dorsal horn would have supplied the two dorsal aboral elements, and there 

 would have been no want of conformity in the numbering. It should be noted that 

 there is, at least in Cribrella, no true torsion at all ; the hypogastric ring does not twist 

 as a whole ; its dorsal termination is fixed throughout, and its ventral part merely 

 moves round by elongation till it reaches radius 5. In other asterids at present 

 described the left posterior ccelom lies loose as it is constricted off from the anterior 

 coelom, and is not attached to the posterior end of the mesenteron ; it is therefore free 

 to rotate, and does so till the same end is attained as in Cribrella. I am inclined to 

 think that Cribrella is primitive in this respect, and that therefore the torsion in 

 asterids of oral and aboral elements upon each other through 72° is primarily 

 due to these two facts — (1) the true madreporic plane lies through inter-radius 4/5 on 

 the dorsal side of the larva, compelling the hypogastric ccelom, in assuming axial 

 symmetry, to grow only ventrally ; (2) in species with free hypogastric ccelom this 

 part grows symmetrically in dorsal and ventral horns, and then has to twist through 

 72° in order to bring the same elements in opposition as in Cribrella, in which there 



