EARLY DEVELOPMENT OF GKAN'i'lA COMPRESSA. 265 



the methods used are the best known to modern cytology, I cannot say that 

 my material was, as a whole, completely satisfactory : I can only assume that 

 some oE the sponges were not transferred to the fixatives quickly enough after 

 their collection. For this reason I have been somewhat handicapped, and 

 have bad a certain measure of difficulty in working out part of my results ; 

 nevertheless,- among the large numbers of slides prepared some perfect 

 examples were procured. 



The methods which gave the best results, and which future workers may 

 use with confidence, are as follows : (1) Obampy-Kull ; (2) Hermann— or 

 Flemming — without acetic acid followed by two or three days' mordanting in 

 3 percent. K 2 Gr 2 7 ; (3) Kopsch's original method. For these techniques 

 see my late paper (4). In addition, I tried the method of Cajal and used 

 glycogen techniques. I had access to Professor E. S. Goodrich's formol- 

 Flemming material of Si/con ciliatum, and at a late stage of this work to 

 Professor Dendy's slides. I also had at my disposal the sponge collections 

 at Oxford, and at University College, London. I have therefore based my 

 remarks and figures on a very extensive collection. On my Plates the 

 collar-cells are drawn from some very successful Kopsch preparations. 



5. General Account of the Development of a Sycon, up to the Formation of the 

 Free-swimming Amphiblastula Larva. 



The ovum undergoes total and regular cleavage, resulting in the formation 

 of a hollow blastula lying beneath the collar epithelium, PI. 21. figs. 19, 20, 

 22, and 21. 



For some time the walls of the blastula are formed of cells, which are 

 subequal in size and general appearance (PL 21. fig. 21). 



Subsequently one hemisphere of the blastula, generally that which lies 

 immediately beneath the nearest gastral cavity, and touching the collar 

 epithelium, becomes modified to form a regular epithelium of deep cells, 

 with small dense nuclei like those of the choanocytes ; flagella soon appear 

 on these cells. 



The other hemisphere, generally that which lies away from the epithelium 

 of the gastral cavity, an 1 in contact with the mesoglea, becomes changed 

 very little, its cells dividing less rapid!)- and remaining clearer and larger. 

 Examined fresh, much pigmented yolk is seen to repose in the former 

 (flagellated) cells, while such granules of yolk are less evident in the latter 

 (granular) cells. According to the late Professor Minchin (9), the free- 

 swimming larva contains three cell-elements, " columnar flagellated cells at 

 the anterior pole, rounded non-flagellated cells at the posierior pole, and a 

 central mass of granular amoebocytes." In all probability the latter are 

 derived from the flagellated cells by a process of immigration inwards into 

 the reduced blastoccel. 



