454 
BULLETIN OF THE UNITED STATES FISH COMMISSION. 
stage with the perforated oesophagus became replaced by a stage with a cord of 
indifferent cells (the oesophagus being in the embryo quite functionless) out of which 
the nonperforated oesophagus was directly formed.” 
The extent of the gill-pouches in the stage figured in 76 is seen in figs. 115 and 116. 
A large part of the gill-pouches lies in front ot the auditory thickenings. A little 
later the two layers composing the rudiments are beginning to separate and touch the 
ectoderm (fig. 128). An opening does not exist. The hypoblast grows out and up 
till it reaches the ectoderm, when its distal cells separate. Later the ectoderm also 
gives way and the first gill-slit, the hyobranchial, is formed in front and below the 
auditory capsule. I have never found the spiracular opening which some authors 
(Hoffmann) claim to be the first opened. The hypoblast of the spiracle extends out to . 
the epiblast, but a canal is never formed in Cymatogaster and the ectoderm never 
parts. The remaining gill- slits are not formed till much later. The hyobranchial is 
functional as soon as opened, the ovarian fluid entering the intestine through it. Its 
early formation is another of the precocious features of Cymatogaster due to its vivi- 
parity. The fact that the hyobranchial is formed by an upper outward growth makes 
it resemble the formation of the spiracular slit in the fish described by Hoffmann. 
In the stage with 12 protovertebrse (fig. 78) the intestinal lunpen is continuous 
from a little ways behind the origin of the chorda to Kupffer’s vesicle, part of which 
now forms a dilation of the intestine at the posterior end. An anus has not been 
formed. As will be seen in the chapter on Kupffer’s vesicle, this structure has 
separated into three parts, only the middle one of which remains connected with the 
intestine. By an ingrowth of cells similar to that found in the formation of the floor 
of the anterior part of the intestine, a floor has been formed to Kupflfer’s vesicle and 
the depression in the yolk separated from the upper parts. A new roof has also been 
formed for this part of the intestine, separating a dorsal, dome-shaped upper part 
from the median portion which remains permanently as part of the intestine (fig. 38). 
The walls of the intestine at this 12-proto vertebrae stage are everywhere two or 
more cells thick. The lumen does not grow forward to form the mouth until much 
later, and the cephalic portion with the development of the gills will be described in 
another chapter. I shall here continue the description of the postcephalic portion, to 
which growth is mostly restricted at this time. 
An outline of a larva about 0-9 mm. long, with 15 protovertebrse, is represented in 
fig. 5. Spermatozoa are now found in large numbers in the intestine, to the walls 
of which they are seen to cling by their heads. Their tails at this time are fetill very 
mobile. Large numbers are seen in cross-sections. They have gained entrance 
through the hyomandibular gill-slit. The anus is probably formed at this time, though 
I was unable to determine this in living specimens. In a stage with only one addi- 
tional protovertebra (20) it is certainly open (figs. 86 r 136). The thickness of the 
walls has been reduced to a single layer of cells, except in the region of the gullet, 
where two layers are frequently found. The cells are highest behind the yolk, and 
here the nuclei of the cells are nearer the bases of the cells than in other regions 
(figs. 136, 141, 142). Over the yolk the cells are lower and have the nuclei nearer 
their center. In the gullet, where the walls are largely two cells deep, the inner layer 
of cells has become ciliated (not shown in figures). This ciliated region remains an 
important structure, as will be seen in later stages. In life the cilia are so active that 
they can not be seen, nor are they very evident in sections. When particles of food 
