47 



or of the three, five or six in the specimens v/hich are 

 aberrant. This inference seems a likely one, from the fact 

 that in the adult medusa the tentacles which from their 

 larf^er size are presumably of the first cycle are always lo- 

 cated at the ends of the radial canals. Wliile a few excep- 

 tions are found, they fre^iuently occur in cases which have 

 apparently become abnormal by accident subsequent to meta- 

 morphosis. The inference is, then, that five-parted adult 

 medusae were f ive-ter tacled larval polyps. And this is 

 borne out by comparison of the relative numbers of each 

 kind of variation among medusae and among polyps, hargitt 

 (1901) has tabulated the number of medusae that have come 

 under his notice having three or five oi- six radial canals; 

 and he finds that about five per cent, of all medusae are 

 irregular in this regard, i.e., that they vary from the 

 normal four-parted condition. While I have not had a great 

 number of specimens of the polyps from which to compute av- 

 erages, my counts show quite a striking similarity to those 

 which are given by Dr. Hargitt for the adults. 



Among all the varieties of geometrical figures which 

 appear in the ai-rangement of parts among the various orders 

 of coelenterates , there is none, so J'ar as I can find,vihich 



