VIVIPAROUS FISHES OF THE PACIFIC COAST. 
453 
Some of the cells of the floor of the intestine may be derived from the roof directly. 
If so, such cells are derived from the region just below the chorda (fig. 127) and dem- 
onstrate the bilateral condition of the enter on. The same bilateral condition becomes 
still more evident later, when frequently two lumina make their appearance, especially 
in the anterior region (fig. 122). Kupffer’s vesicle has now become much larger 
(figs. 75, 110, 113), but since it is such a striking feature of the stages succeeding the 
one under consideration it will be dealt with in a separate chapter. 
In a slightly older larva (fig. 76), which differs from the one just described in being 
straightened out, the hypoblast presents in a sagittal section the same appearance 
seen in the preceding stage. But a layer of horizontal cells is now present between 
the yolk and the main mass of hypoblast (figs. 117, 125, 126, 127). In other words, 
the ingrowth of hypoblast has now reached the median line, and the floor of the enteron 
has been completed. Transverse sections demonstrate the completion of the floor in 
the whole of the postcephalic region exclusive of Kupffer’s vesicle, which is now a very 
large cavity (fig. 76). 
Soon after the completion of the floor as far as just described, the lumen appears 
both anteriorly and posteriorly (fig. 77). It is formed by the separation of the two 
layers of cells already formed. The separation usually begins laterally and grows 
towards the middle line. This is especially true of the cephalic portion (figs. 121, 
122, 127, 130). In the front part of the body the hypoblast is about as high as wide 
(figs. 23, 130, 131), while in the posterior region it is many times as wide as high, 
encircling the whole of this portion of the yolk, and the lumen formed here is a 
A-shaped slit with very long arms, and occasionally an upward extension at the 
median line (figs. 124, 133, 134). The posterior lumen is, therefore, from the first 
potentially very wide, for as soon as the lower layer is separated from the upper to 
its full extent the intestine is relatively very large in this region. 
As stated above, the lumen is formed in front and behind simultaneously. The 
gullet or middle region remains a solid rod (figs. 77, 131, 132). This appearance led me 
at first to suppose that the posterior intestine is precociously developed; but, as we 
have seen, the floor of the entire intestine is formed continuously from before backward, 
and the solid mass is the result of the retardation of the formation of the lumen 
between the two layers of the hypoblast, after such a separation has taken place in 
front and behind. I am not aware that a similar condition has been found in other 
teleosts more than once before. A similar condition is, however, described by 
Balfour (Elasmobranch Fishes, p. 217) for elasmobranchs. In elasmobranchs the 
oesophagus has a well-developed lumen like the remainder of the alimentary tract, 
but “its lumen becomes smaller and finally vanishes, and the original tube is replaced 
by a solid rod of uniform and somewhat polygonal cells.” Although a lumen does not 
at first appear in Cymatogaster , its equivalent does, if we bear in mind that the 
floor is formed after the exact method used in elasmobranchs to form a lumen, i. e., 
by the ingrowth below of the marginal cells of the hypoblast. What morphological 
significance this retrogressive development of the part of the intestine in elasmo- 
branchs and in teleosts has I am unable to suspect, unless indeed Balfour’s conjecture 
(II, 61) may be correct. He has found a solid oesophagus in the salmon (sp.) long 
after hatching. “ It appears not impossible that this feature in the (esophagus may 
be connected with the fact that in the ancestors of the present types the oesophagus 
was perforated by gill-slits; and that in the process of embryonic abbreviation the 
