No. 3-] THE GASTRULATION OF AMPHIOXUS. 571 



We have also studied the number and position of the cells 

 surrounding the blastopore at different periods, in the hope of 

 discovering some region of more rapid growth. Finally, the 

 karyokinetic phenomena of the embryo have been carefully 

 examined, for on this process Lwoff has mainly relied. It has 

 been necessary to take all these factors into consideration in 

 order to follow the changes that take place during gastrulation. 



Kowalevski and Hatschek have shown that the blastula is 

 composed of larger and smaller cells, and several more recent 

 authors have observed the same fact. MacBride makes the 

 surprising statement that the cells of the blastula wall are all 

 of equal size. He has probably confused cross-sections of the 

 blastula with longitudinal ones, otherwise it is difficult to see 

 how such a mistake could have been made. 



The yolk granules almost completely fill the cells around the 

 vegetative pole and gradually diminish in number toward the 

 animal pole. 



The gastrulation begins by a flattening of ihe vegetative part 

 of the Q^g (PI. XXXIII, Fig. i). The lower hemisphere then 

 turns in and slowly obliterates the segmentation cavity (PI. 

 XXXIII, Fig. 2). The invagination is not entirely radially 

 symmetrical, for the inturned cells bend over more toward one 

 side of the embryo than toward the other (PI. XXXIII, Fig. 2). 

 At this time the segmentation cavity almost disappears at one 

 side, which becomes subsequently the dorsal side of the embryo, 

 but a portion of the segmentation cavity is left around the 

 remaining part of the inturned cell plate, and, in general, 

 is largest exactly opposite the dorsal side, and in this way a 

 dorsal and a ventral side of the gastrula are distinguishable at 

 an early stage of gastrulation. The outline of the embryo may 

 also be used, as Hatschek demonstrated, to determine points of 

 orientation, as shown in PL XXXIII, Fig. 2. The outline of 

 the dorsal side is less rounded th?n that of the ventral side. 

 This seems to be constant ; but, if the sections do not pass 

 dorso-ventrally, this difference cannot be made out. 



A further point of orientation is found in the appearance of 

 the inturned cells of the dorsal side. These cells are somewhat 

 smaller than the other invaginated cells, and they resemble 



