422 Metschnikoff on Germ-Layers. [May 
delamination, from which it is distinguished by the late appear- 
ance of any difference between the ectoderm and endoderm cells. 
It is necessary to assume that multipolar immigration is a 
more primitive form than unipolar immigration (Fig. 7), since the 
contrary supposition leads us into great difficulties, as we have 
already seen. The transition from a multipolar to a unipolar 
immigration (where the seat of migration is always the hinder 
end of the larva) is, on the other hand, an easy conception, ` 
especially as the latter is prone to occur in blastula larve that 
are very active, while the former is observed in motionless or 
sluggish embryos. It is as well to recall here the great pre- 
' dominance in the metagenetic Medusz of cell-immigration over 
„transverse division. 
If there be no difficulty in deriving unipolar from multipolar 
immigration, there is likewise none in reducing invagination to 
the former. We learned in the third chapter that Laodice cruci- 
' ata is distinguished from other metagenetic Craspedota by the 
fact that the posterior pole of the blastula is occupied by a con- 
tinuous area of transparent cells, These cells, however, do not 
immigrate all at once, but one after another as in other Meduse 
(Fig. 8). There is then formed a parenchymatous endoderm, 
which gradually acquires a cavity, the endoderm becoming 
i . The stage with the area of transparent cells is 
strikingly like the blastula stage of Mausithoé marginata, Atlas, 
_ Plate X. (invaginate gastrula), in which the cells at the posterior 
_ pole likewise differ from the other cells of the body. Let us 
_ suppose the development of Laodice to be still further abbre- 
viated. The endoderm cells, still at the surface but already dif- 
ae ' d, will no longer immigrate one at a time, but will inə 
--vaginate in a body, and thus in a more direct way establish a 
> ee (Fig. 9). The invagination of those cells that are 
