248 M. MIYAJIMA : 



however, reasonable to suppose that the centrally placed smaller 

 cells which are continuous with the endoderm of the branchlet 

 belong to that layer. The cells which presumably belong to the 

 ectoderm and form the main part of this globule seem to be 

 mostly distended. In this globule the nematocysts (Fig. 15, 7i.) 

 are found in a large number ; hence the central globule may be 

 regarded as the battery. 



The lateral globules (Fig. 14, !.(/.) are mostly spherical and 

 consist of compactly packed cells rich in cytoplasm. The meso- 

 derm prolonged from the branchlet distinctly separates the 

 ectoderm from the central cell-mass. After examining many 

 sections I was able to find a few globules which enable us to 

 see that the clusters are true gonophores. In such globules 

 (Fig. 16), one is able to see that the ectoderm cells at the tip 

 are grouped into a mass forming the " bell-nucleus " which 

 pushes the endoderm in as a cup. This part of the endoderm 

 is arranged into a regular layer one cell deep and is easily dis- 

 tinguishable from the remaining part. Owing to the section 

 (Fig. 16) having been cut slightly obliquely, the cavity in the 

 endoderm seems irregular and very limited. In leality, there 

 is a wide cavity occupying the whole interior of the globule, 

 which communicates with that of the branchlet. I could not 

 detect gonophores developed any further than this in our speci- 

 men. January is probably not the season in which the ripen- 

 ing of the sexual products takes place. 



The terminal branch thus bears two sorts of globules, the 

 one being a nematocyst- battery and the other a true sexual organ. 

 Hence the dendritic gonosome of this animal is a peculiar organ 

 wliich bears on a common stalk the sexual and defensive elements. 



