100 THE MEDUSAE. 



boid ectoderm-cells, although I have seen no indication of the occurrence of 

 such a process. The ridges, as in Cunina, are proliferating regions, from 

 which the true buds or planulae are given off, but, in Pegantha, budding 

 instead of being restricted to them, may take place from the undifferentiated 

 endoderm in other regions of the canal (PI. 27, fig. 5). The buds are solid, 

 ciliated, morula-like structures, in which the cell boundaries are easily dis- 

 tinguishable (PI. 27, figs. 6, 7). A considerable number were found lying free 

 in the gastric cavity, but since no further advanced stages were observed, it 

 is probable that they now pass out through the mouth of the parent, to con- 

 tinue their development either independently or in some other host. 



Trachomedusae. 



Most authors now recognize in this order six families: — Petasidae, Pectyl- 

 lidae, Geryonidae, Halicreasidae, Aglauridae, and Trachynemidae. Of these 

 the first four are well marked, and probably represent natural phylogenetic 

 groupings, but the separation between the last two seems to be based on 

 unsatisfactory grounds. Thus, as Maas (: 06'', p. 494) has already pointed 

 out, the presence or absence of a gelatinous peduncle, considered by Haeckel 

 ('79) the most important character separating Aglauridae from Trachyne- 

 midae is of much less taxonomic importance than has usually been sup- 

 posed, — a conclusion strongly supported by the fact that the new genus 

 Tetrorchis in the present collection, very closely related to Aglaura in so far 

 as the structure of the gonads is concerned, lacks a peduncle. 



The conformation of the gonads, whether extending along the radial 

 canals or hanging free in the bell cavity, seems to me of but little more 

 systematic significance, since this criterion, if rigidly applied, must lead to 

 the removal of Agliscera from the Aglauridae, although its closest natural 

 allies are undoubtedly the genera Aglaura and Aglantha. The form of the 

 bell is certainly not a family characteristic, for either a high or a low bell may 

 be associated with a peduncle, or with either linear or pendent gonads ; nor 

 can we use the otocysts, whether free or enclosed, as a family character, as 

 Browne (: 04) has done, without confusing other structural characters which 

 apparently are fully as important. There is one character which seems 

 to me to be more significant than has usually been supposed ; that is, whether 

 the tentacles are of one or of two kinds, but even this can hardly be made 

 the basis of family distinction, since to do so would result in separating Rho- 

 palonema from genera to which it is universally believed to be closely related. 



