NARCOMEDUSAE 4Y 



would then be narrowed until represented merely by the eight (?) ridges or 

 pillars seen in our smaller specimen. The fact that in the larger specimen 

 there are eighteen or nineteen such gastric ridges, instead of seven or eight, 

 is difficult to explain. 



The affinities of this peculiar genus must remain problematical until more 

 satisfactory material with tentacles intact can be studied. In the structure 

 of the canals and of the bell margin the present specimens closely resemble 

 Callinema ornata Verrill, especially the fragmentary specimen recently re- 

 corded by Browne (: 08) under that name from the " Scotia " collection. The 

 structure of the gonads (apparently not preserved in Browne's individual) puts 

 identity, or even close relationship, with that species out of the question. 



The most remarkable feature of Poralia is, of course, that it is probably 

 octoradial in its younger stages, a conclusion of which the "Valdivia" speci- 

 men leaves little doubt, although the latter had only seven gonads. As 

 Vanhoffen (: 02", p. 41) has pointed out, octoradial Medusae had previously 

 been known among discophores only as abnormalities of Aurelia aurUa. Since 

 however they are extremly rare even in this form, this second capture of 

 Poralia practically eliminates the possibility that this genus can be abnormal. 

 The structure of the gonads gives us but little assistance in this question, 

 as the development of the genital folds directly from the aboral wall of 

 the stomach, without the formation of gastro-genital pockets or well-marked 

 subgenital pits, appears to be the primitive condition among the Discophora, 

 occurring among Cyanidae and the more primitive Pelagidae and Ulmaridae. 



Craspedotae. 



NarcomedusaE Haeckel, 1879. 



The collection contains twelve species of Narcomedusae, representing 

 ten genera, on which to test the classifications of the order recently pro- 

 posed by Maas (: 04% : 04"=) and Vanhoffen (: 07). These two plans rest 

 on radically different bases; for while Maas has discarded the presence 

 or absence of a peripheral canal system as sound ground even for generic 

 diagnosis, Vanhoffen believes that this character, in conjunction with the mode 

 of development, whether direct or indirect, is sufficiently important to war- 

 rant the division of the Narcomedusae into two suborders, Diocheteumena 

 having, and Adiocheteumena lacking, canals. There are, however, strong 

 objections to assigning such importance to the presence or absence of the 



