220 
BULLETIN OF THE UNITED STATES FISH COMMISSION. 
face of the randwulst ia as marked a manner as is shown in Fig. 45. Round the rest ?| 
of the blastoderm edge the marginal epidermic cells are often larger than the others, i! 
but there is nothing striking in their behavior, and after the first stages of invagina- 
tion the peculiar character of the cells at the embryonic pole is lost (Figs. 43 and 44, ^ 
ri, xciiT). 
The germ ring, the origin of which has now been described, continues its cen 
tripetal growth for a short while around the entire edge of the blastoderm, but espe- 
cially in the region of the embryonic pole. In a surface view, Fig. .34, PL v, an hour 
later than Fig. 33, the inner outline of the germ ring {(j. r.) is seen to have encroached 
upon the central clear space (the thin roof of the subgermiual cavity), and in a section,] 
Fig. 43, PI. xom, tlie short, thick tongue of cells shown in Fig. 41 is seen to have] 
h. (two layers of cells). Still an! 
grown out into a thin sheet of uniform thickness, jjr 
hour later this part of the germ ring has approached considerably nearer the center,] 
ii 
Fig. 44, PI. xciii, preserving its characters of a simple two-layered sheet, which in 
deed it does not lose until the formation of the notochord begins. The growth of the 
germ ring I'ound the rest of the blastoderm edge may be gathered from a comparison' 
of Figs. 46 and 48, PI. xciv, both longitudinal sections througb the anterior pole of 
the blastoderm (Figs. 48 and 44 are parts of same section). The ingrowth {v. mes.) is 
seen to be very slight compared with that at the embryonic pole, and moreover the 
cells of the under layer are not arranged in two strata. i 
The positive means by which the ingrowth is effected is undoubtedly cell division.^ 
(See the nuclear spindles in Figs. 46 and 48.) But as Agassiz and Whitman {1. c.) havei 
pointed out, the width of the ring is probably also increased in a passive manner byi; 
the spreading of the blastoderm round the yolk. I 
There is one very sti'iking and theoretically important feature which comes out in* * 
the comparison of the early and later stages of the germ ring. If Fig. 41, 1^1. xciii,Tl 
be compared with Fig. 44, PI. xciii, it will be seen that the germ ring has not oniyl; 
grown in a centripetal direction, but that it has also been splitting off in a centrifugall ^ 
direction from the randwulst. This point is brought out even better in a comparison 1 ' 
between Figs. 46 and 48, PI. xciv. In other words, accompanying the ingrowth or||' 
invagination, there has been a backward (centrifugal) delamination, slight, indeed, but! ' 
still a fact. Gotte, the discoverer of invagination in the Teleosts, is the only writer* 
who speaks of the centrifugal delamination in the randwulst, though, as I shall point ' | 
out later, Gbtte’s randwulst is really a stage of the germ ring in which the actual ,| 
ingrowth has already begun. 
The significance of the centrifugal delamination lies in the suggestion it gives of the i | 
way in which invagination may be converted into delamination. It is, to be sure, very 
doubtful whether there is any vertebrate in which the primitive hypoblast is really 
delaminated from the upper layer. But in the very similar case of the origin of the | 
mesoblast from the primitive hypoblast, there is no doubt that in certain animals 
delamination has superseded the folding process (see section on mesoderm), and any 
occurrence which points to the way in which the one method may be converted into the 
other, seems worth recording. 
In the cut, Fig. 3, the concentric lines of cell growth by which the randwulst is 
established, meet in what may be called the apical line a, which marks the inner edge 
of the randwulst. From this zone the ingrowth iirogresses ceutripetally, and a cen- 
