94 THE VOYAGE OF H.M.S. CHALLENGER. 



of the branchial sac, and the atrial aperture, as usual, as the dorsal side of the body, 

 then the ventral edge of the branchial sac is more extensive than the dorsal, and has 

 its anterior half fused with the mantle to form the anterior part of the body-wall, while 

 its posterior half forms the greater part of the horizontal membrane (Fig. 11, p. 93), the 

 remainder being the posterior part of the dorsal edge of the branchial sac. According 

 to this interpretation of the structure, the short endostyle described by Moseley (PI. 

 X. fig. 3, en.) would be confined to the posterior end of the ventral edge of the sac : its 

 position is shown diagrammatically by end', in the above woodcut (Fig. 11). 



The large space below the horizontal membrane in which the visceral mass is placed, 

 and which opens to the exterior Ijy the atrial aperture, I regard, then, as a peribranchial 

 cavity which lies wholly at the posterior end of the branchial sac, and which has no 

 communication with the cavity of the branchial sac except indirectly liy means of the 

 alimentary canal. 



The specimen from Station 299, off Valparaiso, in the South Pacific, which I have 

 examined, resembles Moseley's figures (PI. X. figs. 1-5) of Octacnemus bythius in 

 general shape, but differs (see PI. X. figs. 6 and 7) in having a well-marked prominence 

 upon the dorsal edge of the body in which the visceral mass is lodged,^ and upon 

 the anterior surface of which the atrial aperture opens. The breadth of the specimen 

 across the outstretched processes is now 8 cm., but was probably about 12 cm. when 

 living, judging by the amount Moseley's specimen has contracted since he measured it. 



The branchial aperture is large and is transversely elongated, while the atrial is 

 smaller and more circular. They are placed 2 cm. apart, and the atrial aperture is 

 1 cm. from the dorsal edge of the body. The rounded dorsal projection which contains 

 the viscera (PI. X. fig. 7, ad.) is roughened on its lower surface, and if the body were 

 attached to some foreign object it must have been by this part. 



The test is thin, and easily torn, over the greater part of the body; but it is thickened 

 round the margins of the posterior end so as to form slight pads at the bases of the 

 conical processes (see PI. X. fig. 8, t'.). Test cells are numerous and of various shapes 

 (PI. X. fig. 12), but are mostly of small size. No large bladder cells are present, 

 but a few rounded cells larger than their neighbours have considerable vacuoles 

 (PL X. fig. 12), and so show an approach to the formation of bladder cells. 



The mantle is thin, but has a good deal of musculature. It adheres closely to the 

 inner surface of the test. The muscle bands have the general arrangement described 

 and figured by Moseley (PI. X. figs. 1, 2, and 5). Figure 14 shows more exactly the 

 arrangement of the chief muscle bands on the conical processes. In a longitudinal 

 section of one of the processes the chief transverse muscle bands are seen cut in section 

 on the inner surface of the thickened pad of test. Besides these chief muscle bands, 

 there are numerous very fine bands formed of one, two, or three muscle fibres each, 



' Possibly the visceral mass was displaced in Moseley's specimen. 



i 



