COLLAR CAVITIES OF THE LARVAL AMPHIOXUS. 247 
lost its apical attachment to the gut, so that the cavities 
above and below now communicate (PL 18, fig. 7). The 
remains of the septum have disappeared in the next section. 
It will be plain from this description that the collar cavity 
of the right side is continuous with the splanchnocoel. It 
has a more extensive (longitudinally) communication with 
that space than have the succeding myocoels, but essentially 
its relations are the same. We have not found anything 
comparable to the septum described by MacBride as separating 
a postero-ventral extension of the collar cavity from the 
splanchnocoel. 
Turning now to the left side of the larva, we see that the 
first myoseptum is well in advance of that of the right side. 
In PI. 18, fig. 2, the ventral horn of the collar cavity is 
already pushed down to the level of the middle of the 
notochord, and in tracing sections back the first trunk 
myocoel is found to be well established dorsally when the 
section passing through the mouth is reached — the septum 
between it and the collar myocoel having passed obliquely 
downwards and backwards to meet the upper lip. The cavity 
of the upper lip (virtual at this stage) is that of the first 
trunk somite. 
What happens in the lower lip is more difficult to make 
out. A thickened layer of ectoderm, applied to the left head 
cavity and to the antero-ventral wall of the gut, suppresses 
altogether the ventral extension of the somite (PL 18, fig. 2) ; 
just behind this thickening the somite passes down, its lumen 
often occluded, to become continuous with the mesoblast of 
the lower lip, which, in its turn, is continuous with the 
splanchnocoel (PL 18, figs. 8, 4, 5). Again, no septum 
is observable between collar cavity and splanchnocoel ; 
but, the spaces being for the greater part merely virtual, 
it is impossible to state with certainty whether there is actual 
continuity. 
The great dilatation of the collar cavities, which MacBride 
took to be the first appearance of the atrial folds, is, in our 
opinion, largely a fixation effect. It depends, no doubt, as he 
