NARCOMEDUSAE. 429 



grown downward from the bases of the tentacles, which retain their primitive position. 

 Haeckel's assumption that the tentacles have secondarily migrated upward is not borne out 

 by the embryological evidence and appears to me to be untrue. I believe the bell in Narco- 

 medusa? to be merely a newly-formed, collar-like expansion of the side walls of the actinula, 

 and not comparable with the bell of Anthomedusae. While this book has been in press Goette, 

 1907 (Zeit. fur wissen. Zool., Bd. 87, p. 289), has presented an able exposition of the view I here 



advocate. 



We should then regard the Narcomedusae as medusiform actinulae, and not as being directly 

 comparable with the Anthomedusae and Leptomedusae, which arise by budding from hydroids 

 and in which the tentacles grow out from the bell-margin after the bell has been formed. The 

 entodermal sensory-clubs of Narcomedusae can not be considered to be homologous with the 

 ectodermal lithocysts of the Leptomedusae, but are evidently modified tentacles. 



The species of Cunina and Cunoctantha exhibit very interesting parasitic stages in their 

 development, for in their asexual generation they infest the gastrovascular spaces or gelat- 

 inous substance of other medusae such as Geryonta, Tumtopsis, etc. 



Stschelkanowzeff, 1905-1906, studied the life-history of the common Cunina probosadea 

 of the Mediterranean, and the development in this case is probably typical for the more complex 

 life-histories, such as are to be witnessed in the genus Cunina. In C. probosadea the eggs of 

 the free-swimming medusa develop in the ectoderm, but when mature they migrate into the 

 entoderm and are fertilized in the stomach-wall of the mother-medusa, where they remain 

 attached and develop rudimentary tentacles, sensory-clubs, bell and velum, but practically no 

 gelatinous substance. The mouth breaks through, and finally the sexual products (male or 

 female) develop in the ectoderm of the subumbrella of the young medusa, which still remains 

 in the stomach-cavity of its mother. The daughter medusa then degenerates into a mere sac 

 filled with genital products, which escape into the water through the mouth of the free-swim- 

 ming mother-medusa. These eggs then develop into free-floating planula larvae which attach 

 themselves to the stomach or lips of the medusa Geryonta, where they grow and finally develop 

 numerous asexually produced medusa-buds which are set free to form the large free-swimming 

 sexual generation. There are thus three generations, one large, sexual, free-swimming; one 

 internal, degenerate, sexual; and one free-swimming actinula-form, which reproduces asexually. 



External medusa-buds are produced by budding upon the gonads of Cunoctantha fowleri 

 Browne, and internal stolons are budded from the entodermal layer of Cunina peregrina 

 and Pegantha laevis Bigelow. 



Being developed from transported, free-floating or parasitic larva;, and being independent 

 of shore conditions for their development, the Narcomedusae are of world-wide distribution, 

 being in a peculiar sense creatures of the open ocean. They are common in the Mediterranean, 

 but are quite rare elsewhere, and are all but unknown from polar regions. Most of the species 

 are of extreme tenuity and die so rapidly in aquaria and are so difficult to study alive, that but 

 little is known of them even to-day. 



Haeckel (1879, p. 301) divides the Narcomedusx into two suborders, Porpylotae with 

 "otoporpae" at the bases of the sensory-clubs, and Cordylotae, without these small sensory 

 tracts. It appears inadvisable to separate suborders upon so slight a distinction and, indeed, 

 in this case it leads to the wide separation of forms which are actually closely related. For 

 example, Haeckel defines his four families of the Narcomedusae as follows: 



poRPYLOT.*. 



1. CunanthiDjE: With perradial stomach-pouches and with a marginal ring-canal system. 



2. Peganthid.1: Without radial stomach-pouches and with a marginal ring-canal system. 



CORDYLOT.S. 



3. ^Eginidje: With interradial stomach-pouches and with a marginal ring-canal system. 



4. Solmaridje: Without ring-canal system or "festoon canal." 



Later researches by H. V. Wilson, Maas, H. B. Bigelow, VanhofFen etc., have demon- 

 strated that some of the Cunanthidae lack a marginal ring-canal system, while others possess it, 

 and it is also very variable in ^Eginidae, being sometimes absent and sometimes present. It 

 therefore appears to be an unsuitable character upon which to attempt the separation of 

 families. 



