of the durso-lateral pleurochordal grooves, continuous on the one hand with the gill-slit, and on 

 the other hand extending towards the pharyngeal diverticulum (figs. 122, 121). Fig. 129 (PI. XI), 

 of the same species, illustrates the same point. Figs. 45 — 47 (PI. V), of C. gracilis, further 

 .shew that the gill-slits curve from the region of the pharyngeal diverticulum in an antero-ventral 

 direction to their external apertures. 



The gill-slits of the neuter individuals of C. sibogae are essentially similar to those of 

 the other species, but they are associated with an unusually strong development of pleurochordal 

 tissue in the more posterior („ventral") part of the pharynx. 



The probable mode of action of tbe arrangements which have just been described may 

 now be considered. I am (juite inclined to accept M.\sterman's view that an essential part of 

 the feeding mechanism of Ccp/iaiodiscns is afforded by a secretion of mucus which entangles 

 the ortranisms brought by the ciliary currents of the tentacles, and prevents them from being 

 carried out through the gill-slits. It may be presumed that a current of considerable force passes 

 through these apertures; and the need for some entangling mechanism seems probable. I am 

 disposed to think tliat the mucus required by this hypothesis is secreted by the glandular 

 patches shewn in figs. 152 — 154 (PI. XII) in the walls of the dorsal jjharyngeal diverticulum, and 

 by unicellular glands in the walls of the mouth and on the base of the operculum. M.\sterm.\n 

 ascribes the secretion to the notochord, which he terms the "subneural gland"; but it must be 

 noted that whereas the structures which I regard as mucus-glands shew a strong affinity for 

 haematoxylin, the notochord on the contrary takes ii|i staining matter to a slight e.xtent. 



If this be the case, the anterior grooves seen in the pharyngeal diverticulum (figs. 152, 

 153) may be supposed to convey the mucus towards the mouth. Here it probably passes along 

 the oral grooves to the great food-channels limited b)- the po.sterior wall of the proboscis and 

 the anterior wall of the collar. There are not wanting indications that arrangements exist which 

 might be capable of directing the mucus for a short distance along paths e.Kternal to the mouth, 

 perhaps during a temporary cessation of the ciliary current of the tentacles. Figs. 152 — 155 

 shew a perfectly definite epithelial ridge {e.r.) which is continuous with the ridge dividing the 

 two principal oral grooves on each .side, and can be traced for some distance along each side 

 of the proboscis-stalk. It is ])rol)abU: that this ridge corresponds with the division of the arms 

 of its side into two groups of three. Traced in a dorsal direction, the ridge becomes a groove 

 at exacdy the level where the arm-base meets the operculum. This transformation is effected 

 by the appearance of another epithelial ridge on the median side of the first one, a process 

 which has already been effected in fig. 152 on the right side, where e. r. is the ridge that is 

 continuous with those seen in figs. 153 — 155. The higher ridge immediately in front of it passes 

 some distance along the proboscis-stalk, dying away at the level of the proboscis-pores. 



It appears to me probable that the effect of this arrangement is as follows. The mucus 

 leaving the mouth by the outer oral grooves seen in fig. 155 probably passes along the oral 

 surface of the operculum and scjon comes into the region of the j)osteriorly directed food-grooves 

 of arms 4 — 6. Immediately on the dorsal side of fig. 152 the ridge c.r. becomes continuous 

 wMth the ventral side of the arm-base, which is traversed in an antero-posterior direction ])y the 



