724 EDWARD T. BROWNE. 



As the shape is constant for each species in all stages of development, it is easy to identify 

 the early and intermediate stages, which are liable to be described as distinct species and 

 perhaps placed in different genera. 



The revision of the genera belonging to the Petasidae has not involved any alterations in 

 generic names. The structure of the sense organs has been taken for the character of the 

 two subfamilies, Petnachnidae and Olindiadae, instead of the absence or presence of centripetal 

 canals, which was used as the subfamily character by Haeckel. The genera Aglauropsis and 

 Gossea have been placed among the Olindiadae, and so have the following additional genera, 

 Gonimiemus, Gonionemoides, and Vallentinia. The structure of the tentacles has been used as 

 the principal character of the genera belonging to the Olindiadae. 



A new genus, Amphogona, has been instituted for Pantachogon apsteini of Vanhoffeu. This 

 medusa is bisexual, having male and female reproductive organs on alternating radial canals. 

 I believe this to be the first recorded case of hermaphroditism amongst the Craspedote 

 medusae. The new genus has been placed with the Aglauridae, but it also possesses characters 

 of the Trachynemidae and looks like a connecting link between the two families. 



Order ANTHOMEDUSAE. 



Family Williadae, Forbes, 1848. 



Character of the family. Anthomedusae with 4, 6 or more radial canals, each having 

 one or more lateral branches (except in the earliest stage) running to the margin of the 

 imibrella. Stomach with 4, 6 or more lobes, upon which the gonads are situated. Mouth 

 with four or more lips, or with a folded margin. Tentacles simple, evenly distributed (not 

 arranged in groups) round the margin of the umbrella. 



The Williadae were classified by L. Agassiz (1862) as a family belonging to the Tubulariae 

 (Gymnoblastea-Anthomedusae), but Haeckel (1879) removed the family to the Leptomedu.sae. 

 He considei-ed the gonads to be upon the radial canals and not upon the stomach. In 1893 

 I succeeded in connecting Willia stellata, Forbes, with the Gymnoblastic hydroid Lar sahel- 

 larum, Gosse, and cut a series of sections which showed that the gonads were upon the stomach 

 and its lobes. 



Haeckel placed the Williadae as a subfamily of the Cannotidae, and used the mode of 

 branching of the canal system as the basis for classification. He introduced two new genera, 

 namely, Dyscannota and Willetta, on the supposition that A. Agassiz had wrongly identified 

 his specimens. It appears to me that the number of main radial canals is of greater im- 

 portance than the mode of branching of the canal system and should be used for the generic 

 character. The mode of branching, which carries with it the number of tentacles, would be 

 of more value as a specific character. 



The development of the radial canal system shows that the earliest free-swimming stage 

 has unbranched canals. A young Proboscidactyla has four radial canals, without branches, and 

 four tentacles. A young Willia has six unbranched radial canals and six tentacles. The 

 branching of the canals appears later and passes through a definite series of stages. 



A single abnormal specimen in the Maldive Collection, belonging to the Williadae, led 

 me to investigate the literature relating to all the species and I have now attempted to 

 make a revision of the family. 



