218 
BULLETIN OF THE UNITED STATES FISH COMMISSION. 
III. INVAGINATION. 
During the formation of the periblast the superficial layer of the blastodisc cells 
becomes differentiated from the rest to form what German authors usually call the 
“ Deckschicht,” English writers the “ epidermic stratum.” The i)rogressive flatten- , 
ing by which these cells are converted into a well-marked layer may be traced throughji’ 
Figs. 25, 26, 27, PI. xci, into Fig. 40, PL xciri. The latter section is through the stagei 
when invagination begins, and the epidermic stratum (cp. s.) is fully dilfereutiated.T 
The flattening continues through later stages until the layer is reduced to the coudi-1 
tiou of a very thin membrane, which stains deeply. I 
During the same period the blastodisc undergoes a change of shape, which is thej 
preparatory step towards invagination and the differentiation of the embryo. Duringj 
the last stages of segmentation, when the periblast wall is being formed, the under, 
surfaceof the blastodisc is either plain or slightly convex (Figs. 26 and 29,Pl.xci). Four] 
hours later (Fig. 30, PI. xci) the under surface has become decidedly concave. As thei 
hollowing out continues the concavity takes an eccentric position (Fig. 31, PL xci), and] 
til ns before any invagination occurs one part of the peripheral region of the blastoderm] 
is thicker than the rest. In a surface view of this stage the thin eccentric area appears] 
as an ill defined clear circular space, surrounded by the more opaque periphery, which* 
at one pole (p. p., Fig. 32, PI. xoi), the tail end of the future embryo, is considerably! 
thicker than elsewhere. At this pole, in Fig. 32, the invagination has already begun,] 
and hence the periphery is here separated by a sharp line from the central region.] 
But before the invagination begins this part of the edge is noticeably thicker than the] 
rest, though nowhere is the peripheral region sharply marked off from the central.* 
As to the lueans by which the center is thinned out, and the periphery, especially the' 
embryonic pole, left thicker, I can only say that in general I agree with Gotte (14). In 
the absence of any absorption of cells or extensive migration, the cause would seem' 
to lie in the direction of cell growth, which in the main determines the position ofj 
nuclear cleavage planes, and hence the direction in which new cells are pushed. i 
The change of shape which the blastoderm suffers during this i>eriod gives rise to] 
what is commonly called the subgerminal cavity, i. <?., the cavity which may be sup-f 
posed to exist between the blastoderm and the periblast layer in Figs. 30, 31, PI. xci, 
Figs. 40, 41, PI. xciii. This cavity was in almost all the embryos I examined a virtual' 
one, except at certain points where the bounding layers separate slightly, as in Figs. 
40, 41, etc. {s. (j. c.) Occasionally, however, an embryo was found in which a compara-' 
tively spacious cavity separated blastoderm and periblast, as in Fig. 47, PI. xciv. It 
will be understood that the subgerminal cavity is merely a late i)hase of the segmen-' 
tation cavity of the earlier stages, s. c., Fig. 25, etc. | 
The next preliminary stei) in the process of invagination is illustrated by Fig. 32, 
PL XCI, and by an antero-posterior section through the posterior pole of a similar blas^ 
toderm. Fig. 40, PL xciii. In the surface view the peripheral region at the embryonic 
pole is seen to be bounded internally by a sharp line, and in the section Fig. 40, the 
explanation of this line is found in the well-marked randwulst, the inner limit of ! 
which is indicated by the cells w, m'. The mode of formation of the randwulst may be ! 
inferred from the shape and arrangement of the cells and the position of the daughter , 
nuclei. Cell-growth, following the arrow a (Fig. 40) first established in the blastoderm a i 
