636 MR. H. G. NE^A'TH OX THE 



anfiinglich, d. h, so lange es noch mit dem Urdarme zusammen- 

 hangt, dorsalwiirts von diesem um erst nach seiner Abschniirung 

 an dessen linke Seite zu rlicken, sondei'n befindet sich von vorn- 

 herein an dieser Seite." 



By the middle of the third day the formation of thestomodfeum 

 has begun and the segmentation of the coelom, in C. normani, is 

 complete, though in some lai-vte the connection between hydrocoel 

 and posterior coelom is not yet lost. The hydrocoel still points 

 forward in the axis (roughly speaking) of the animal, and from it 

 project on either side the rudiments of the radial canals and of 

 the oral tentacles (PI. II. fig. 9). The posterior part of the anterior 

 vesicle is now bent, at an obtvise angle to the hydrocoel, towards 

 the dorsal body- wall, and in many larvae has acquired an opening 

 to the exterior — the madreporic pore. There is only a very 

 shallow invagination of ectoderm in the formation of the opening, 

 and it will probably be correct to regard the whole of this postero- 

 dorsal limb of the water- vascular system, which is later converted 

 en bloc into the stone-canal, as being homologous with the 

 anterior coelom of other Echinoderm larvte (cf. Bm-y, 2). At this 

 time — as also, more markedly, in the next stage — there is no con- 

 tinuous clear lumen in the hydrocoel, the stone-canal, or the 

 posterior coelom, and it is doubtful, I think, whether the madre- 

 poric pore ever functions. Even after the tentacles are well 

 established, and can be protruded and retracted, their lumen is 

 obliterated in some places by the vacuolated inner ends of their 

 cells. 



The relation of the tentacle-rudiments to the undifferentiated 

 residue of the hydrocoel is precisely similar to that of the rudi- 

 mentary radial canals, with which they alternate. There is no 

 association among any of the ten outgrowths to indicate their 

 future arrangement with reference to the water-vascular ring. 

 This is a stage obviously somewhat later than the most advanced 

 three-days-old larva of C. planci described by Ludwig, in which 

 the hydrocoel had the shape of an irregular horse-shoe, slight oiit- 

 pushings of which were identified as the rudiments of the radial 

 canals. In C. normani there is a complete suppression of the 

 typical curved hydrocoel crescent, owing to the large size and close 

 crowding together of its lobes, and to the thickness of their walls. 

 On this account the interpretation of this stage is peculiarly 

 difficult in the absence of intermediate stages between it and the 

 one next following. My identification of the hydrocoel lobes in 

 fig. 9 must therefore be considered as being provisional only, and 

 it must be pointed out that it is based upon the determination 

 of the point of closure of the ring-canal on the following day — 

 which I have only made out in C. saccicola. 



The Pentacula of C. saxicola. 



The third-day (65 hours) larva of C. saxicola shows a consider- 

 able advance upon the last stage described. The stomod^um 



