THE METAMORPHOSIS OF ECHINODERMS, 69 
between the outer body-wall and the hydroccel, or between the 
centre of the latterand thestomach. Soon after the separation 
of the hydroccel from the anterior enteroccel a small circular 
depression appears in the outer wall of the former, as shown in 
fig. 20. As it isnot invariably present even at this stage, and 
seems to be generally absent in the next, I am inclined to regard 
it as due to shrinkage after death; but even so it marks a 
special weakness of the wall at this place, which is not un- 
important as the first indication of the position of the adult 
mouth. Inthe next stage an outgrowth of the stomach presses 
against the inner wall of the hydroccel, and a very little later 
pierces right through this organ and fuses with the body- 
wall beyond ; probably it forms a part of the permanent ceso- 
phagus, but most of the latter seems to be derived, at a much 
later stage, from the ectoderm (fig. 23). By this piercing of 
the hydroccel the water-vascular ring is formed, as Metschnikoff 
observed. This, at least, is the conclusion I draw from my 
sections; but this method of formation is unknown in other 
Echinoderms, and it is just possible that an invagination of the 
side of the hydroccel (such as occurs in Echinids) may take 
place ; it must, however, be formed and obliterated with re- 
markable rapidity to have entirely escaped notice, and I am 
not disposed to believe in its existence. After all, this un- 
usual mode of formation of the water-vascular ring is no 
more remarkable than the extreme variation in the point of 
closure of this ring exhibited by other Echinoderms (see 5, 
fig. 28). 
With the help of fig. 22, which is a dorsal view at about 
the same stage as figs. 20 and 21, the general course of the 
longitudinal dorsal mesentery can be followed ; it starts rather 
to the left of the middle line, under the posterior margin of 
the madreporic plate, and runs somewhat obliquely backwards 
on the right side of, and parallel to, the line of the terminals. 
This obliquity increases in later stages; but as at the same 
time the centre of the hydroccel pushes itself further and 
further back, the plane of this organ is always (as in fig. 22) 
parallel to the plane of the mesentery ; while the latter, as will 
