EMBRYOLOGY OF THE SEA BASS. 
227 
the lowest cells of the streak which have assumed a columnar shape, and have thereby 
ditt'ereutiated themselves from the rest. The line of demarcation between them and 
the neurenteric streak is always more evident in transverse than in longitudinal 
sections. 
Posteriorly they, together with the mesoderm and ectoderm, fade away into the 
caudal mass (Fig. 59, behind n). Anteriorly they are continuous with the flat ento- 
derm. These columnar cells subsequently form the roof and walls of Kupffer’s vesicle. 
Brook (3) has tried to show that the entoderm lamella, instead of delaminating 
from the invaginate germ ring, is derived from the periblast. Kupfter has always 
maintained that the periblast is the source of the entoderm (24, 25, 20); but it seems 
impossible that this should be the case in any Teleost, when so many observers have 
failed to find that the periblast takes any share in forming the i)ermanent layers. 
Notochord . — By the time the entoderm layer is completely established the notochord 
has assumed its ultimate shape of a somewhat cylindrical rod. (Fig. 61, PI. xcv, 
n. c.) Anteriorly it thins away (Fig. 62, PI. xcv) and in the neck region disappears. 
The two layers of polygonal cells which primitively constituted the chord (Figs. 55 
and 56, PI. XCV) have in Fig. 59, PI. xcv, begun to assume the well-known shape char- 
acteristic of chorda cells. They are already much interlocked, and a few hours later 
Fig. 65) each cell, in a longitudinal section, extends the whole width of the chorda. 
In the early stages of the chorda the number of cells which together compose a cross 
section (Fig. 61) is much greater than in later stages. For, as the cells become flat- 
tened antero-posteriorly, they spread out in the transverse plane, and consequently two 
or three come to compose a cross section (Fig. 77, PI. xcvii, 39 hours). When this 
stage is reached it is next to impossible to determine the cell outlines. 
The next stage in the histological differentiation of the chorda is brought about 
by vacuolatiou. The beginning of vacuolation is shown in Fig. 110, PI. ci (trans- 
verse section through an embryo of 53 hours). The vacuolation continues until, before 
the time of hatching, the protoplasm is reduced to a thin peripheral layer (Fig. 127, PI. 
cm), in which the nuclei are situated, and a few strands which cross the central cavity. 
By this time a well-defined sheath of high staining power is found round the chorda. 
After hatching the sheath becomes more conspicuous and the protoplasmic layer even 
thinner than before. (Compare Fig. 144, PI. cv, part of a transverse section through 
a larva 3 days after hatching. The notochordal sheath is indicated by a heavy line, 
n, c. s.) 
The vacuolation which probably puts an end to cell multiplication, does not 
extend into the posterior portion of the chorda. The chorda cells on the contrary 
retain in this region their embryonic, protoplasmic character. Compare Figs. Ill and 
114, PL CI, transverse sections through the tail and head regions, respectively, of an 
embryo 59 hours old, and Fig. 119, PI. cii, through the tail of a 65-hour embryo. At the 
tip of the tail there is found throughout embryonic life and for 3 days after hatching 
(possibly for a much longer time) a mass of undifferentiated cells (caudal mass) in 
which the chorda ends. At the time of hatching this mass of cells has become very 
small. The iiosterior growth of the chorda undoubtedly depends upon the preseuee 
of these undift'eren tinted cells, and possibly upon the embryonic character of its own 
cells in this region, though I have not observed nuclear figures. 
If a comparison be made between Fig. 53, PI. xciv, and Fig. 83, PI. xcvm, it will 
be evident that a considerable forward extension of the chorda takes place after the 
