390 - ARTHUR ROBINSON. 
by the increase of the trophoblast in the form of a blind sac 
towards the centre of the blastodermic vesicle is scarcely satis- 
factory, for it is evident that between the stages represented 
in figs. 8 and 10, Pl. XXIII, the increase of the epiblast and 
of the trophoblast on the distal side of the cavity (7 C) repre- 
sented in fig. 10 is sufficient alone to account for the diminu- 
tion of the distance between the proximal and distal walls of 
the yolk-sac during the same period. It appears more 
probable, therefore, that the primary cavity of trophoblast does 
not result from invagination of the trophoblast, but from the 
rapid growth of the peripheral portion of that part of it which 
would in the case of an unattached ovum gradually extend 
over the surface of the yolk-sac. Such extension, however, is 
prevented in the mouse by the close apposition of the wall of 
the yolk-sac to the uterine mucosa, and, consequently, the 
proliferating trophoblast takes the only other possible course, 
and spreads itself along the wall of the crypt in which the 
ovum lies, leaving the central portion of the cavity of the crypt 
unobliterated for a time. If this explanation is correct the in- 
vagination of the yolk-sac is not in any way due to invagina- 
tion of the trophoblast, for there is no true invagination of 
that portion of the ectoderm. 
The primary cavity of the trophoblast entirely disappears, 
and has nothing whatever to do with the secondary cavity, 
which appears at a later period. 
The secondary cavity, according to Selenka’s figures in his 
first communication (44, Taf. xi, figs. 16 and 17), is formed by 
the extension into the trophoblast of the cavity which appears 
in the epiblast at the seventh day, but in the third portion of 
his studies he figures the secondary cavity of the trophoblast 
and the cavity of the epiblast as present simultaneously and 
separately, both in the mouse and the rat (45, Taf. xvi, figs. 
57 and 63). In the rat the two cavities which are afterwards 
found undoubtedly arise independently of each other. Duval 
has so represented them (9, pl. ii, fig. 92, and pl. iii, fig. 100), 
and figures representing the rat’s ovum on Pl. XXIII are 
further confirmative of Selenka’s account. In the mouse, 
