SALPA IN RELATION TO EVOLUTION OF LIFE. 209 



The craspedota of modern times fall into two divergent groups, 

 which must have separated from their common stem very early. 

 In the one group are the trachomedusse, the anthomedusfe and the 

 leptomedusae ; agreeing in the possession of circular and radial 

 chymiferous tubes, diiferentiated out of the simple larval digestive 

 cavity by the formation of areas of adhesion between its oral and 

 aboral walls ; and in the other we have the narcomedusoe, in which 

 chymiferous tubes are absent and the digestive cavity is formed out 

 of that of the hydra-like larva in a much more simple and direct 

 way, as Wilson's account (The structure of Cunocantha in the adult 

 and larval stages, by H. V. Wilson, Studies from the Biol. Lab., 

 Johns Hopkins Univ.) which has recently been corroborated by 

 the researches of Maas (Ueber Ban und Entwicklung der Cuninen- 

 knospen. Zoologischen Jahrbiichern, V, 272) shows. 



While the lines of descent represented by these two types are 

 quite distinct we cannot doubt their origin in a common ancestor 

 from which both the hydra-like larva, and the bell, sub-umbrella 

 velum, and marginal tentacles, sense organs and nervous system of 

 all the craspedota have been inherited, neither can we doubt that 

 this primitive veiled medusa was still more primitively derived 

 from a hydra-like form. 



Now the fact that the sessil funnel-shaped hydra is restricted to 

 a part of the members of one of these primary subdivisions of the 

 craspedota, while the solitary star-shaped floating hydra-like larva 

 is found in both of them ; in the larva of cunina, as well as in that 

 of the geryonids, and in the aetiuula of tubularia, shows that the 

 ancestral form which the hydra-larva represents was of the latter 

 type, and that the more familiar hydranth is a secondary modifica- 

 tion of the more simple and ancient form. 



In a more extensive discussion of this subject (The Life History 

 of the Hydromedusse, Mem. Boston Soc. Nat. Hist., 1886) I have 

 referred to the history of parasitism and its effects, in the narco- 

 medusse, in order to show how easily this primitive larval type 

 may become converted into a polymorphic cormus with alterna- 

 tion of generations. Nothing could have been further from my 

 mind than a belief that the cormi of ordinary hydroids have 

 been phylogenetically derived from the parasitic cormus of cunina 

 larvae. 



6 



