The endoderm lining the canal of the sensory club is specially differentiated. In the stalk it is more columnar than the epithelium of the stomach pockets, and is made up of cells containing a brightly staining nucleus with very little trace of cytoplasm. The cell bodies appear as if filled with a clear, non-staining fluid. Perhaps these cells give the stalk elasticity to act in connection with the thin layer of longitudinal muscle-fibres that are found just external to the supporting lamella. The epithelium of the terminal enlargement of the canal is composed of very high narrow cells, many of which show two nuclei of equal size and staining quality lying side by side. In continuation of the specialized epithelium of the perradial furrows in the floor of the stomach the inner wall of the stomach pocket shows a strip of similar densely crowded columnar cells leading from the gastric ostium downwards to the canal of the sensory club. As in the other case, the strip probably represents a specially ciliated tract, and perhaps in it we see the reason why the canal of the sensory club is almost always found to contain either spermatozoa which are shed by the male reproductive organs directly into the stomach pocket, or else floating cells of the kind to be described in the next section. The canals of the interradial tentacles arise from the peripheral gastro-vascular system much lower down than those of the sensory clubs, since these tentacles have preserved their primary positions with reference to the bell margin. Figure 16 represents a section taken at the level of the base of the pedalia which gives the connection of the tentacle canals with the gastro-vascular system. At the level below the sensory niche the four broad stomach pockets have been divided, as we have seen, into the right marginal pockets (_mp_). The figure shows that in the interradial corners the longitudinal septa (_ivl_, in the preceding figures), or lines of fusion between the two walls of the peripheral gastro-vascular space, which divide the primitively simple space into the four stomach pockets, have come to an end, leaving a connecting canal (_cc_) in each corner as all that remains of the primitive uninterrupted communication between all parts of the peripheral system. It is from these four connecting canals that the tentacle canals take their origin. From this point of origin each tentacle canal passes downwards, surrounded by the gelatine of the pedalium, into the tentacle proper. The connecting canals are of morphological importance in that they are supposed, with much reason, to represent in the Cubomedusæ the circular canal of the Hydromedusæ.