VIVIPAROUS FISHES OF THE PACIFIC COAST. 
457 
of the hind gut, which, however extends even further in later stages, the anus forming 
some distance behind the original place of the vesicle. The lower one remains for some 
time as a space in the yolk and the upper one is pushed upward and disappears through 
a proliferation of the cells of its wall when about fifteen pro tovertebrae have been formed. 
I will describe the successive phases of this structure as they could be made out by 
my material. The conditions described by Henneguy for the trout approach nearer 
those of the middle vesicle of Cymatogaster than any others made known so far. 
As stated above, Kupffer’s vesicle makes its appearance early ; when three proto- 
vertebrae have been formed it is a well-defined structure. Before this stage it has not 
been seen. At this time it is a broad, short, low space between the yolk and the 
hypoblast, a short distance behind the end of the notochord. The hypoblast cells form- 
ing its roof are high columnar. The greater part of the floor is composed of the yolk, 
but in a few places cells are also found on the floor of the vesicle, but these are not 
regularly arranged. (Longitudinal sections 67, 68, 70; cross-section, fig. 73.) The 
columnar arrangement of the hypoblast cells is continued forward for some distance 
in front of the vesicle. The antero posterior extent of the vesicle is much less than its 
lateral extent. The outer angles of the vesicle are pushed slightly upward and out- 
ward, giving the roof an angular appearance and greatly resembling the conditions of 
the gills in slightly older larvae. For these reasons I at first considered this stage of 
Kupffer’s vesicle as the first indication of the lumen of the alimentary canal in the 
region of the gills. 
In a larva with ten protovertebrae it has become a large shallow subcircular 
cavity in the entoderm and is still floored by the yolk. From its roof a small dome- 
shaped cavity extends still deeper into the indifferent or hypoblastic mass of cells 
forming this part-of the larva. The contour of the yolk is not affected by the vesicle 
(fig. 75). The cells in the tail of this larva are somewhat disarranged, so that the 
relations of the vesicle with the surrounding structure can not be definitely made 
out. It is, however, only in this region that the entoderm and yolk are not in close 
contact. The lumen of the intestine has nowhere appeared as yet. 
The larva at this time forms an almost complete circle around the yolk. As the 
larva is straightened out and the lumen of the intestine is formed, and even before any 
lumen appears (fig. 76), Kupfifer’s vesicle enlarges rapidly. The arrangement of the 
entoderm cells in front of the vesicle shows that it is from the first an enlarged por- 
tion of the alimentary tract, which remains without a floor long after a floor is formed 
in all other parts. At this time the vesicles differ greatly in shape in different speci- 
mens, but all the larvae examined show the following structure: 
First. A low cavity in the entoderm with considerable lateral extent. Its dorsal 
wall of columnar hypoblast is a direct continuation of the roof of the alimentary tract. 
Second. A dome-shaped cavity usually in the anterior half of the vesicle extending 
dorsad (figs. 76, 77, 144). This dome-shaped portion, which was evident in the earlier 
stage, forms, as will be seen later, the lower half of the neurenteric canal. 
Up to this stage the vesicle is formed exclusively at the expense of the entoderm 
which surrounds it above and on the sides, and frequently forms a floor at the margin 
of the vesicle. The contour of the yolk has not been affected. Over the center of the 
dome the cells are usually somewhat irregularly placed as compared with those of its 
sides, but a neurenteric canal I have not been able to find in this stage. 
