354 JOHNS HOPKINS UNIVERSITY MORPHOLOGICAL MONOGRAPHS. 



chamber remains attached to the whole ventral surface of the brain 

 (see Fig. 10, Plate LIV), forming a single large pit opening by a wide 

 mouth to the peribranchial chamber. The cells of that portion of this 

 pit which is attached to the ganglion show the structure typical for the 

 cells of the hollow disks. In this species alone is the subneural gland an 

 unpaired structure. The chain form of Salpa africana-maxima shows 

 the greatest development of the tubes and hollow disks. The latter are 

 very large (see Fig. 9, Plate LIV) and are lined by an epithelium com- 

 posed for the most part of a single layer of cubical cells whose outer 

 ends are uneven, whose protoplasm is granular and does not take up 

 haematoxylin stain, and which have an appearance as if glandular in 

 function. On the dorsal sides of the disks the epithelial cells are greatly 

 increased in number, forming an irregular mass that half fills the 

 lumen of the disk. In all species the epithelium of the dorsal wall of 

 the disk is more developed, and in some species (Salpa democratica- 

 mucronata) only this portion has any epithelial lining that can be dis- 

 tinguished in section. Probably there is a thin pavement epithelium 

 over the other portions. The ducts, one on each side as in other species, 

 are unusually large. They are lined by an epithelium of columnar cells 

 whose outer ends are very irregular. I believe we have in this species 

 a gland which is functional. The epithelium of both disks and ducts 

 has the characteristic, granular, vacuolated protoplasm, and seems to 

 be degenerating to form the secretion. The fact that the hollow disks 

 form the dorso-lateral walls of the great dorsal blood sinus, which gives 

 many branches to the dorsal lamina or gill, renders it easily credible 

 that these structures should function as a gland. If this gland be func- 

 tional in this species, it is the only case I have observed where the secre- 

 tion could be considerable. 



The nervous system of Salpa africana-maxima shows another feature 

 which is of the keenest interest. We have seen that in the chain form 

 of Salpa runcinata-fusiformis there is an antero-ventral outgrowth from 

 the ganglion. This resembles the "wart-like," antero-ventral process 

 from the ganglion of doliolum. In Salpa africana-maxima there is a 

 similar antero-ventral outgrowth from the ganglion (Plate LIV, Fig. 3), 

 and this is produced anteriorly as a rod of cells which runs forward 

 beneath the eye, soon meeting the pharynx wall. It grows smaller, but 

 still can be traced, close pressed to the basement membrane of the 

 pharynx wall, running on toward the ciliated funnel. It finally fuses 

 with the pharynx wall, but can be traced for a comparatively long 



