OF THE HYDROMEDUS^E. 381 



the epithelium of the sub-umbrella, although such a statement was not, twenty years 

 ago, intrinsically improbable. 



My own observations show that Haeckel really observed the origin of the chymiferous 

 tubes, although he failed to discover that they are formed by the differentiation of the 

 walls of the digestive cavity, instead (if those of the sub-umbrella. 



At the stage shown in fig. 9, the oral layer of endoderm has been pushed in, by the 

 formation of the sub-umbrella, until it is nearly in contact with the aboral wall, and 

 the digestive cavity is thus reduced to a thin dome which is concentric with the sub- 

 umbrella and extends to the bell margin. At four points on the four inter-radii 

 and near the bell margin, the two layers of endoderm have come into contact with each 

 other and fused to form four shield-shaped areas of adhesion, fig. 9, i. The stomach 

 is thus divided, by the four areas of adhesion, into first, a spacious axial chamber or 

 stomach proper, which reaches more than half-way down the bell; second, four short, 

 wide, radial canals, I; and third, four short arcs of the circular tube, m, which unite the 

 distal ends of the radial tubes with each other. 



In older medusae I have traced the gradual extension of the four areas of adhesion, 

 figs. 10 and 11, i, i, until four narrow, sharply defined radial canals, I, and a circular 

 canal, m, are produced, and I think there can be no doubt that, in a younger medusa 

 than the one shown in fig. 9, the areas of adhesion would be still smaller, and that, in a 

 still younger medusa they would be entirely absent, while the stomach would extend to 

 the bell margin as a continuous cavity without interruption. 



While I have not found a larva in this condition, a reference to Fritz Midler's 

 (55) and Haeckel's papers (30) will show that both these authors have seen and cor- 

 rectly figured this stage of development. The larva shown in Haeckel's PI. 4, fig. 35, 

 is like our fig. 9, except that the areas of adhesion have not yet appeared, and the four 

 quadrate, interradial areas, of which Haeckel speaks on p. 136, are, beyond doubt, the 

 areas of adhesion. 



Although both Fritz Muller and Haeckel were led astray in their interpretations, I 

 believe that their figures correctly represent the larvae, but this is not true of the figures 

 which have been given by other authors. 



Drawings which are touched up at home, from sketches made at the seashore, are 

 very apt to become conventionalized, and I cannot help believing that the sharply de- 

 fined radial canals which are shown by Leuckart (47) and Fewkes (6$, pi. 7, fig. 2) 

 in young Geryonids at about the same age as our fig. 9, were introduced into the 

 drawings upon theoretical grounds, rather than from observation. This is certainly the 

 case with Gegenbaur's figure (25) for he represents the canals as interradial. 



It is interesting to note that the endoderm cells do not completely disappear in the 

 areas of adhesion, even in the adult. The Hertwigs giv-e (69, PI. 4) two sections, 

 figs. 2 and 9, through the bell margin of Carmarina hastata, and in each section they 

 show a double layer of endoderm cells, r', t', in contact with each other, running up 

 toward the axis of the bell from the oral side of the circular tube, with the epithelium of 

 which they are continuous, although the cavity of the tube does not extend between 

 them. 



