98 A, A. W. HUBRECHT. 
by which nature, starting from simple methods of vascularisa- 
tion of the external embryonic membrane, has tended to 
create a very wide range of adaptation to all the different 
possibilities of nutrition which are offered to the embryo. 
The uncommon diversity which we observe in the endless 
varieties in the arrangement of the foetal membranes of the 
mammals would be as good as inexplicable if we held on to 
the derivation of the monodelphian mammals out of yolk- 
laden ancestors with ornithodelphian characters. And since, 
after Hill’s (97) investigations, we must assume that the 
didelphian mammals are not descended from Ornithodelphia 
but from monodelphian, placental ancestors, they no longer 
form an imaginary transition between Ornithodelphia and 
Monodelphia. 
Thus a more direct phylogeny of the latter should form 
the object of diligent research, towards which the present 
paper is a first attempt. 
Cuapter 1V.—THe PART PLAYED BY THE ’ROPHOBLAST IN THE 
NvtritioNn AND THE ATTACHMENT OF THE Empryo. 
In Chapter II we have discussed the trophoblast as a larval 
layer which is of great importance in monodelphian and 
didelphian mammals, but which, in the Sauropsida, has 
diminished both in importance and in distinctness parallel to 
the development of oviparity, and of which we may presume 
that even among lower vertebrates rudiments are yet re- 
tained. 
Suggestions have been thrown out on p. 18 that the 
original significance (protective, locomotor, or otherwise) of 
the ancestral larval layer may have gradually become con- 
verted into an adhesive and a nutritive one. For this I want 
in this chapter to adduce all the evidence available. 
