GENEOLOGICAL. 385 



the result was two quite separate and independent 

 twins of half the normal size. Each of the isolated 

 cells segments like a normal ovum, and gives origin, 

 through blastula and gastrula stages, to a half-sized 

 metameric larva. 



If the shaking has separated the two first segmen- 

 tation cells incompletely, double embryos like Si- 

 amese twins result, and also form short-lived 

 (twenty-four hours) segmented larva}. 



Similar experiments with the four-celled stages 

 succeeded, though development never continued long 

 after the first appearance of metamerism. Com- 

 plete isolation of the four cells resulted in four 

 dwarf blastulse, gastrulaj, and even larvaa. Separa- 

 tion into two parts of cells resulted in two half-sized 

 embryos. Incomplete separation resulted in one of 

 three types (a) double embryos, (&) triple em- 

 bryos one twice the size of the other two and (c) 

 quadruple embryos, each a quarter size. 



The eager observer proceeded to shake up the 

 eight-celled stages, but in no case did he succeed in 

 rearing a gastrula from an isolated unit of the eight- 

 celled stages. Flat plates, curved plates, even one- 

 eighth size blastula3 were formed, but none seemed 

 capable of full development. 



Thus, a unit from the four cell stage may form an 

 embryo, but a unit from the eight cell stage does not. 

 For various reasons it seems likely that this is due 

 to qualitative limitations, not merely to the fact 

 that the units of the eight cell stage are smaller. For 

 although the separated cells of the eight cell stage 

 have considerable vitality, and swim about actively, 

 the difference between macromeres and micromeres 

 has by this time been established; in fact the eells 

 have begun to be specialised, and have no longer the 



