XVI ECHINODERMATA 467 



tube. In this case this water-tube becomes completely divided into 

 anterior and posterior portions, which we may term right anterior 

 and right posterior coeloms respectively, just as the left is 

 normally. This formation of a second hydrocoele was first described 

 by us in Asterina (1896). 



Somewhat later, near the mid-dorsal line but to the right of it, a 

 small sac is formed. According to Field (1894) it just appears as a 

 solid bud of cells in the blastocoele, but in Asterina gibbosa (Fig. 365) 

 it is certainly budded off from the posterior wall of the right side of 

 the anterior coelom, with which it remains connected for some time 

 by a solid string of cells. The bud is, from the first, nearly but not 

 quite solid ; it would be correctly described as a very thick-walled 

 evagination of the right anterior coelom. It is soon cut off from the 

 anterior coelom, and then its cavity rapidly enlarges and it becomes 

 thin-walled. 



In Asterias Goto (189*7) has seen this cavity connected by a string 

 of cells with the left anterior coelom, but it is practically certain 

 that a re-examination of this point will show that Asterias and 

 Asterina agree in essentials. Gote's observations are exceedingly in- 

 complete, and it is iby no means clear that the " scattered string of 

 cells," which he saw connecting the sac with the left anterior coelom, 

 represents the original connection of the sac with the coelom. 



This sac may be termed the madreporic vesicle. According to 

 Gemmill it executes slow pulsations. It may be compared to the 

 pericardial vesicle of Balanoglossus (see p. 575). We formerly 

 (1896) regarded the madreporic vesicle as the vestigial right hydro- 

 coele, but the observations of Gemmill, who has seen this vesicle and 

 a well-developed right hydrocoele present in the same larva, render 

 this view untenable. 



METAMOKPHOSIS OF ASTERIAS 



With the formation of the madreporic vesicle, and of the rudiment 

 of the water vascular system, the Bipinnaria has reached the summit 

 of its development as a free-swimming organism. It now begins to 

 prepare to take up a fixed life, and with this change in habits the 

 metamorphosis may be said to begin. From the anterior end of the 

 larva, between the prae-oral and post-oral bands, there grow out three 

 clubbed arms which are not ciliated in the larva of A. vulgaris, but 

 in the larva of A. rubens and of ^. glacialis the median ventral " arm " 

 of the prae-oral ciliated band is continued on to them. These arms 

 contain diverticula of the anterior coelom. One is median and dorsal 

 (br.med, Fig. 360), and the other two are situated symmetrically to 

 the right and left of it {br.lat, Fig. 362). These processes are termed 

 the brachiolar arms, and the larva is now termed a Brachiolaria. 

 It still swims, but it occasionally attaches itself to the side of the 

 vessel in which it is contained by the brachiolar arms, which appar- 

 ently act as suckers. Goto states that the cells forming the walls of 



