of the SjJotiges to the ChoanofogeUata. 367 



studied the larva? of certain Sponges, such as Oscarella (ffali- 

 sarca) lobularis, Grantia conqyressa, Leucosolenia botryoides, 

 and Halichondria sp., both in the developed state and during 

 their development, and arrived at the astonishing result that 

 these structures by no means correspond to the first develop- 

 mental stages of Metazoa, but are colonies of Choanoflagellata. 

 He infers this both from their anatomical structure and from 

 the mode of their formation. A mature " swarm-gemmule " 

 (as he calls the free-swimming sponge-larva?) of Grantia 

 conqwessa, according to Saville Kent, represents an elongate 

 vesicle, the wall of which consists of a single layer of radially 

 placed cylindrical cells. Each of these cells is said to bear at 

 its outer end a marginal collar and a central flagellum, and 

 therefore exactly to resemble the ordinary collared cells which, 

 in a single layer, line the radial tubes of that calcareous 

 sponge. In oviform swarm-gemmules of the same sponge 

 which are not yet perfectly mature the long cylindrical cells are 

 said already to possess the flagellum, but not the collar, and 

 to meet in the middle with their diminished inner extremities. 

 In a Calcisponge nearly approaching Ascandi >a pinus, Hack., 

 however, Savilie Kent once found an oviform larva, the broader 

 posterior half of which consisted of collared cells projecting 

 further from the centre. He interprets the latter as fully 

 developed individual animals, and the flagellate cells without 

 collars of the anterior part as not perfectly developed, and 

 thinks that in this way he has found the clue to the compre- 

 hension of the frequently described oviform Sycandra-huvse, 

 the anterior part of which consists of simple cylindrical 

 flagellate cells without any collar, while the hinder part is 

 composed of broad, somewhat convex, darkly granular cells 

 without any appendage whatever. Here the darkly granular 

 cells of the posterior extremity are supposed to have outrun 

 the anterior flagellate cells in their development so far that 

 they had already retracted a previously existing collar as well 

 as the flagellum, and were on the point of conversion into 

 amoeboid cells of the future sponge-syncytium. 



It is worthy of note that, with the exception of Kent, none 

 of all the naturalists who have paid attention to the develop- 

 mental history of the Sponges, and especially of the Calci- 

 spongia?, and have specially and thoroughly investigated the 

 structure of the swarm-larva?, such as O. Schmidt, Carter, 

 MetschnikofF, Barrois, Keller, and others, has detected the 

 collar on the cylindrical flagellate cells of the larva?, although, 

 according to Kent's figures, it is hardly to be overlooked ; 

 for that hyaline and more strongly refractive marginal part 

 which Barrois has represented upon isolated flagellate cells in 



