No. 2.] THE DEVELOPMENT OF THE NEWT. 327 



changes in the Hving o.^^. Despite the considerable difficulty 

 of removing the membrane I have succeeded in a number of 

 cases in performing this operation, and in tracing the exter- 

 nal changes in the Q^g thus freed. The development of such 

 eggs is normal, and the time of appearance of the different 

 external features is the same as in eggs on which the mem- 

 brane is allowed to remain. The first stages of the formation 

 of the blastopore are best observed with the aid of a plane 

 mirror placed underneath the flat-bottomed watch-glass in 

 which the Q%g rests. The reflected image of the lower pole 

 may be easily studied with the dissecting microscope and the 

 successive changes accurately determined. For the study of 

 the blastopore when it shifts to the side, I place the Q.gg in a 

 thin-walled glass chamber in which it can be examined with 

 the microscope-tube in a horizontal position. 



Segmentation cavity. — The segmentation cavity is formed 

 at the division of the ^gg into eight cells, and enlarges pro- 

 gressively up to the time of beginning invagination (Figs. 46- 

 48). The roof of the segmentation cavity is shown by median 

 sections to consist of a single layer of cells, while the floor is 

 composed of a mass of large yolk-cells with ill-defined bound- 

 aries. The cells of the roof are somewhat columnar and, up 

 to the time when invagination begins, divide for the most 

 part vertically. Scott and Osborn ('79) state that in Triton 

 tceniatiis the roof of the segmentation cavity is only one cell 

 thick, in which respect it agrees with that of the Qgg of 

 Petromyzon and differs from that of Rana. v. Bambeke ('80) 

 and Hertwig ('82) have not been able to confirm this state- 

 ment. They maintain that there is no essential divergence in 

 this respect between the frog and the newt. However this 

 may be with regard to the particular species considered by 

 them, I am confident that in Diemyctyhis the roof of the seg- 

 mentation cavity up to a comparatively late period is only one 

 cell thick. In other words, the horizontal division of the roof 

 cells does not occur so early in the newt as it does in the frog. 

 This is clearly seen if we compare Fig. 48, Plate XVIII with 

 Gotte's figures of Bonibijiator (Entwickelungsgeschichte, Figs. 

 2d> and 29, PI. II), or with other figures of Anuran eggs at this 



