TROCHOPHOIIE OF HYDKOIDKS UNriXATUS (kUI'OMATI's). 567 



exactly equal ; the two nuclei of the cells are opposite oiie 

 another^ and show no. tendency to rotate as Coiiklin (7) has 

 described in Crepidula. The subsequent divisions follow 

 in lapid succession. With the third cleavage four slightly 

 smaller upper cells ai-e separated by a dexiotropic division 

 from the lower macromeres. In the fourth cleavage tlie 

 inicromeres of the second group are of the s^me size, and 

 ai-e very slightly smaller than the macromeres. 



Invagination produces a typical gastrula. Gastrulatiou 

 usually commences about seven or eight hours after fertilisa- 

 tion, and consists of a sinking in of the ventral ectoblastic 

 plate, all the eutomeres of which are alike during the earlv 

 stage of the process. Gastrulation is of the modified embolic 

 type, with considerable preparatory flattening of the ventral 

 plate. The cells about to sink in elongate, and their nuclei 

 take up a position at their inner swollen ends. AVhile this 

 flattening is taking place the apical portion of the gastrula 

 is rounding out, the apical tuft of cilia commences to 

 appear, and the endoderm cells sink in till they come in con- 

 tact Avith the inner wall of the ectoderm, in the region of the 

 " rosette cells." At first there is a complete obliteration of 

 the segmentation - cavity, the endoderm folding up close 

 against the ectoderm; but in the immediate filling out of 

 the gastrula, which takes place almost simultaneously, the 

 ectoderm is again drawn away and the segmentation-cavity 

 reappears (Text-fig. 18). At this stage a number ot viscid 

 protoplasmic threads are seen connecting the two layers, and 

 one blastomere with another. They have been observed in 

 Podarke by Treadwell l'43), in Serpula by Soulier (38), and 

 in Thalassema by Torrey (41) ; I have already drawn atten- 

 tion (35) to them in Eupomatus, and pointed out that they 

 are probably similar to the filose strands first described by 

 Andi'ews (2), and considered by him as cell connections. 

 Prof. Loeb has suggested to me, however, that they are 

 rather more in the nature of the fine cytoplasmic strands so 

 frequently seen in membrane formation during fertilisation 

 than definite cell communications. 



