MUSEUM OF COMPARATIVE ZOOLOGY. 321 



four fifths of the extent of the latter, and then coats the lateral por- 

 tions of the gastro-intestinal tract, forming the anterolateral boundary 

 of the great posterior sinus. The space between the wall of the pharynx 

 and the inner layer of the sac communicates anteriorly with the ante- 

 rior sinus, posteriorly with the posterior sinus, and it is interrupted 

 at several points by the union of the pharynx and inner layer with 

 one another. It represents the system of branchial sinus." In 

 Salpa this \inion of the inner wall of the lateral atrium with the 

 branchial sac does not take place. "In side views it is not easy to 

 make out the boundaries of the lateral sacs ; but it is most important 

 to observe that, as has been already mentioned, in the middle of the 

 lateral face of the pharynx, and therefore also in the middle of the 

 lateral face of the inner wall of the sac, a series of opake rings with 

 clear centres, the rudiments of the branchial stigmata, make their 

 appearance." In this respect also Salpa differs from Pyrosoma ; no 

 branchial clefts are ever formed in connection with the lateral atria. 

 " These correspond with the points of union of the pharynx and the 

 inner wall of the sac. They are at first small, round, and very in- 

 distinct, but by degrees they elongate in a direction perpendicular to 

 the long axis of the pharynx, and their real nature becomes apparent. 

 Hence it is clear that these stigmata must eventually open into the 

 lateral sacs, as indeed they may be seen to do in such buds as that 

 represented in Fig. 30 ; and hence also it follows that the lateral sacs 

 are the rudiments of the lateral atria. 



"At first the lateral atria appear to be perfectly distinct from one 

 another, and no atrial aperture is discernible. In buds, such as that 

 represented in Fig. 29, again, they do not extend, posterioi'ly, fur- 

 ther than the sides of the alimentary canal ; but in more advanced 

 buds they are produced backward on each side until they pass beyond 

 the level of the posterior margin of the stomach, so that they now 

 constitute the entire lateral boundaries of the great posterior sinus. 

 The longitudinal section of a somewhat smaller bud shows that in 

 this condition the atria are no longer distinct, but are united together 

 below [on the neural side of] the stomach, by a comparatively nar- 

 row and short canal, which is the mid-atrium. 



"I have not traced out all the details of the process of coalescence 

 of the lateral atria ; but I suppose that each branchio-parietal portion 

 of the atrium, at first a distinct sac, is prolonged downwards and in- 



