1887] Metschnikoff on Germ-Layers. 421 
scribed by F. E. Schulze (33) ), that we may suppose the differ- 
entiation of an amceboid form of individual to have been the first 
step in the historic development of the endoderm. 
At any rate, I believe the peculiarities of Protospongia can be 
more easily harmonized with my view (called by some writers 
‘the Parenchymella theory) than with any of the above-discussed 
theories of other investigators. But how does the Parenchymella 
theory agree with the facts of embryology in general, and of the 
Medusz, as given in the preceding part of the book, in particular? 
In discussing this question we must, in the first place, recall the 
a priori conclusion to which I came regarding the multiplication 
of the hypothetical Metazoo-Flagellata. Reasoning from the 
fact that the first three segmentation planes (sagittal, frontal, and 
equatorial) in so many and various groups of animals follow the 
three dimensions of space, and consequently represent transverse 
and longitudinal division, I concluded that the ancestors of the 
Metazoa also possessed these two kinds of division. Gradually, 
however, the direction of division became more fixed, so that 
while one form divided exclusively or predominantly in a longi- 
tudinal plane another related form divided transversely. That 
such a condition of affairs as we have here sketched is not im- ' 
possible is shown by the life-history of the several species of 
Salpingoeca already referred to. We must therefore suppose 
that in our colonies of Metazoo-Flagellata certain of the super- 
ficial cells became amceboid and migrated into the centre of the 
colony, as occurs to-day in Protospongia, and that certain other 
cells divided transversely into two segments, one of which re- 
tained its position at the surface, while the other came to lie 
within the central space. Figure 3 illustrates these processes in 
a diagrammatic fashion. This double method of forming the 
endoderm, by the immigration of some cells and the cutting off 
of the central segments of others, is actually employed in those 
species that have a mixed delamination. For instance, Polyxenia 
leucostyla (Fig. 4). In the next place, transverse division became 
predominant in some forms (Fig. 5) and longitudinal division in 
others, in which latter case the en rm was formed by the 
immigration of superficial cells (Fig. 6). In this manner mixed 
delamination split up into primary delamination, on the ‘one 
hand, and multipolar immigration, on the other. Secondary 
delamination is to be regarded as a mere modification of mixed 
