STRUCTURE OP THE LARVA OF SPONGTLLA LAOUSTRTS. 455 



(10) That there is always a residue of eells with vesicular 

 nuclei which retain their blastomeric characters, and which are 

 therefore capable of giving rise to the whole sponge. Some of 

 these or perhaps all of them become wandering cells, and ulti- 

 mately give rise to the gemmule which is capable of producing 

 both the dermal and the gastral layers of the sponge. 



(11) That the collar-cells multiply by karyokinetic division, 

 and that owing to the multiplication of the collar-cells in the 

 flagellated chambers the latter become separated into two 

 groups, and so produce two daughter chambers. 



IV. Theoretical. 



I do not intend to embark on a complete discussion of the 

 position of the sponges in the animal kingdom. However, I 

 have a few considerations to bring forward in favour of what I 

 consider to be their true relation to the Protozoa on the one 

 hand, and to the Metazoa on the other. My chief reason for 

 not wishing to debate this question at length is the publica- 

 tion, only a year and a half ago, of a very able article by 

 Mr. Minchin. I shall take all he has said for granted, and 

 refer the reader to his article (12). However, I must quote a 

 few expressions from the concluding paragraph of his paper. 

 After giving an account of the views held by difi'erent ob- 

 servers, and the arguments for and against such views, he says, 

 " We have two theories to choose between ; either to regard 

 sponges as descended from choano-flagellate ancestors indepen- 

 dently of the Metazoa, or to regard them as true Metazoa 

 composed of the two primary layers, ectoderm and endoderm, 

 which have become reversed in position in the adults.'^ He 

 then states that the choice "will depend on which of these two 

 assumptions is the most difficult. If sponges are Metazoa, the 

 collar-cell occurring in no other Metazoa must have been inde- 

 pendently acquired. If sponges are descended from choano- 

 flagellates, then the sexual reproduction, the segmentation of 

 the ovum, and the formation of two germ layers must be 

 processes analogous and not homologous with the similar 



