^6 KINGS LEY. [Vol. VII. 



the germ until the caudal spine appears, are scarcely notice- 

 able. 



Stage B. — The peripheral cells, on the other hand, divide 

 more rapidly, so that in five days from impregnation (Fig. 19) 

 there is a marked difference between the cells on the surface and 

 those deeper in the &g^. A more careful study shows that this 

 division of the surface cells has a peculiar character. In each 

 instance (see Fig. 20, which represents a portion of an ^^^ of 5f 

 days) the first division of the peripheral cells occurs in a plane 

 parallel to the surface of the egg. This is plainly shown in the 

 cases of the cells marked x, where the direction of the mitotic 

 spindle shows the direction of the future division — a view which 

 is confirmed by a study of the other cells. Another feature is 

 noticeable. The products of this division are unequal. There 

 is a deeper and larger cell containing a large amount of food 

 yolk and closely resembling the neighboring yolk cells ; and a 

 superficial smaller and flattened cell, richer in protoplasm and 

 containing far less yolk. In this way a blastoderm is differen- 

 tiated, but the process has in my opinion a deeper significance, 

 for by it the entoderm is separated from the rest of the egg. 

 In other words, in Limulus the two primary germ layers are 

 differentiated by multipolar delamination. 



This process is clearly allied to that multipolar delamination 

 which Morgan ('90, '91) has described as occurring in the eggs of 

 certain pycnogonids and Faussek ('91) in phalangids. While I 

 shall discuss it later, I may say here, that it probably has at 

 most a very distant relationship to the Coelenterate delamina- 

 tion, but has arisen within the Arthropod phylum. 



After the formation of the blastoderm, i.e. the separation of 

 ecto-mesoderm from entoderm, I have not been able to add 

 much to our knowledge until about eight days after impregna- 

 tion. The absence of all features which would aid in the orien- 

 tation of the ^g^ makes it necessary to cut all sections at 

 random, while the opacity renders surface views impossible. 

 In general this time is occupied by a multiplication of the 

 blastoderm cells and a consequent diminution in their size. 

 This multiplication by division proceeds at a more rapid rate at 

 one pole of the ^g^ than at the opposite, the result being that 

 soon a germinal pole may be recognized by the smaller and 

 more columnar cells, those at other portions retaining, until 

 later stages, more the appearance of pavement epithelium. 



