OF THE HYDROMEDUSiE. 403 



spacious cavity, f, opening to the exterior and surrounded by a single layer of invagi- 

 nated cells, which are continuous around the edge of the orifice with the ciliated cells of 

 the surface of the body. As this invagination is very conspicuous while it is difficult 

 to trace out the structure of the more opaque endoderm, the planula bears, at first sight, 

 a very striking resemblance to an invaginate gastrula like that of the Echinoderms; but 

 more careful examination shows that the digestive cavity, g, is already present, and sur- 

 rounded by a continuous wall of endoderm cells, e, e, and that the endoderm as well as 

 the ectoderm is infolded, and that the invagination does not communicate with the diges- 

 tive cavity, and takes no part whatever in its formation. At a somewhat later stage, 

 fig. 11, the endoderm becomes drawn away from the invagination, leaving this as an ex- 

 clusively ectodermal structure. 



I have observed a similar invagination at the small end of the planula of T/urritopsis, 

 PI. 42, figs. 2 and 3, although the planula of this species is so opaque that the study of its 

 internal structure is very difficult. The fact that the invagination is present in an Antho- 

 medusa and a Leptomedusa gives a reason for believing that it occurs in other species as 

 well and that future research may show that it is not at all unusual. At first, the orifice is 

 terminal, as shown in figs. 5 and 6, and the invagination lies on the axis of the larva, but 

 one lip or edge of the opening soon grows faster than the other and thus pushes the 

 pouch on one side, fig. 11,/, which may be called ventral, since it is the surface by which 

 the planula becomes attached; but, before attachment takes place, the whole structure is 

 evaginated as shown in fig. 7, so that only a slight notch, /', remains to mark its posi- 

 tion. Lasso-cells now begin to appear at the small end of the planula, the, cilia are lost, 

 and a delicate layer of transparent cement is thrown off from the ventral surface of the 

 small end of the planula, as shown at n in fig. 8. 



This soon hardens, and, entangling foreign particles, becomes the perisarc. When first 

 attached, and for a short time after, the larva retains the shape which it had during the 

 swimming stage, but it soon elongates, as shown in fig. 9, and becomes the sessile, creep- 

 ing root, which ultimately produces a community of hydroids. For some time, its pos- 

 terior end. figs, it and 10, />, is marked by a flattened pad of ectoderm cells, separated by 

 a constriction from the ectoderm of the general surface of the body. This pad is the 

 area which was invaginated during the swimming stage. The root has no mouth nor 

 other opening to the exterii >r, and there is for some time no trace of the future hydranth ; 

 but a bud, fig. 9, soon grows out from the free end of the root and, developing a circlet 

 of tentacles and a mouth, becomes the first hydranth, fig. 10. A second bud now grows 

 out from the root on the proximal side or base of the first and this is soon followed by 

 a third and so on. As the first hydranth is formed, like all the others, by budding from 

 the root, the growth of the hydranth from the planula is rather a process of metagenesis 

 than metamorphosis and this is not the only species of which this is true. The planula 

 of Turritopsis, PI. 42, figs. 2 and 3, also becomes a root from which the hydras hud, and I 

 have observed the same thing in Hydractinia, where it has been frequently described; first 

 by Wright (64) I believe. Merejkowsky shows that the hydranth of Obelia originates in 

 the same way; that the planula becomes a star-shaped root from which the first hydranth 

 grows out as a bud, and many other cases are recorded, in some forms the planula be- 



