1887] | Metschnikoff on Germ-Layers. 423 
the first to differentiate, once accomplished, the neighboring parts 
of the blastoderm are also involved in the process, and the invag- 
inated sac gradually becomes larger. The next stage in the pro- 
cess of abbreviation is not to be observed in the Meduse. We 
may conceive it to consist in the still. earlier differentiation of 
the endoderm cells, as a result of which all the cells destined to 
invaginate are already marked out in the blastula as peculiar 
elements. The flattened blastospheres found in Lumbricus and 
the Ascidians, for instance, must be looked on as having been 
formed by some such abbreviation. In these blastospheres the 
ectoderm and endoderm areas are about equal. It needs no 
- explanation to see that this early differentiation of the endoderm, 
carried far enough, will lead to the amphiblastula, will then make 
itself perceptible in the segmentation, and will finally be ex- 
_ pressed in the structure of the egg itself. 
It has been generally acknowledged since Kowalevsky’s work 
on Euaxes that an amphigastrula (epibolic gastrula) may arise 
from an archigastrula by precocious differentiation. But it must 
also be admitted that a similar form (to the amphigastrula) may 
be derived from a mixed delamination by means of unequal seg- 
mentation. This latter view is supported by Polyxenia leucostyla. 
(The segmentation of this medusa is variable and in some eggs 
decidedly unequal, in which case the segmenting eggs strongly 
resemble epibolic gastrulz.) It is thus evident that amphigas- 
- trula may come by two different ways from two different starting- 
points, and this embryonic form is hence polyphylitic. 
