1907.] MEDUS.E FROM AFRICAN LAKES. 653 



Limnocnida too, shows a similar type of structure excepting that 

 the gonads are situated on the manubriun in a zone in which bud- 

 formation normally occurs, and would consequently belong to the 

 Anthomedusffi or to a division of medusfe with manubrial gonads 

 which, according to the System of Haeckel, are alDSolutely distinct 

 from the Olindiada?, Liinnocodiuin, or any other medusae Avith 

 radial gonads. We have therefore an indication that the time has 

 arrived for a reconsideiution of the value of the characters upon 

 which this System is founded. May they not tend to an artificial 

 rather than to a natural classification of medusfe ? 



On the whole we incline to the opinion that the hard and fast 

 division of medusae into those with radial and those with manubrial 

 gonads must be abandoned in the case of these fresh- water forms. 

 We cannot imagine that the detailed resemblances which exist 

 between Liinnocodium, Limnocnida, and the Oliudiadte are the 

 result of convergent evolution. A simpler hypothesis is that these 

 forms ai-e descended from a common ancestor, but that the place 

 of development of the germ-cells has changed. 



Arguing from known facts about the migratory proclivities 

 of germ-cells in Hydroids in general, and from the history of the 

 germ-cells of Ohelia in paiticular, in which Leptomedusan the 

 germ-cells, although maturing in pouches of the radial canals, 

 originate in the wall of the manubrium itself, there is good 

 gi-ound for the view tha,t Limnocnida in respect of its manubrial 

 gonads preserves an early stage in the phylogenetic history of 

 Limnocoditcm. Just as Odontornithes are none the less birds 

 because they have teeth, so Limnocnida would be a Tracho- 

 medusan in spite of its manubrial gonads. 



One other taxonomic problem has still to be dealt with. In 

 Haeckel's system the Olindiadse are regarded as a subfamily of 

 the Trachomedusan Petasid^e, which have blind centripetal canals 

 between the radial canals, and are thereby distinguished from the 

 Petachnidfe. Goto, however, does not consider the presence of 

 such canals to be of any systematic moment, since they may be 

 present or absent in closely related genera. It is to be remembered 

 that such centripetal canals, as well as the marginal i"ing of 

 nematocysts, are well marked Trachomedusan features. 



The Olindiadte have usually been regarded as Trachomedusae, 

 until four years ago when Seitaro Goto made an examination of 

 young stages of the sense-organs of Olindioides formosa and of 

 Gonionema depressum. He found that the first rudiment con- 

 sisted of a small segregation of ectodermal cells hardly distin- 

 guishable from the rest, closely applied to the endoderm of the 

 circular canal at the point where the two kinds of cells meet 

 (pi. ii. fig. 15), and he goes on to add that "there cannot be any 

 reasonable doubt that the rudiment consists exclusively of ectoderm 

 cells, since tlie boundary line between the two cell-layers is always 

 distinguishable with a good objective." 



In consequence Goto removed the Olindiadse from the Tracho- 

 medus?e and ranged them with the Leptomedusae, considering 



