464 
BULLETIN OF THE UNITED STATES FISH COMMISSION. 
need but continued conditions, such as exist in the ovary for the hyobranchial cleft, 
to entirely replace the present mouth. To discover where the primitive mouth was is 
quite another question. We may assume that it opened into, the gill-cavity in the region 
of the first gill-slit, since here a lumen exists long before it is formed forward. In this 
connection a strand of hypoblast cells extending up from the wall of the intestine in 
the median line just in front of the chorda becomes of great interest. It is indicated 
quite early (fig. 152), or as soon as the lumen is continuous from the .gill- slit to the anus, 
and it is still striking in larvae over 3 mm. long (fig. 166), after which it gradually 
disappears. This structure is not connected with the gills nor have I been able to 
connect it with the history of any other structure. This strand of hypoblast cells may 
indeed be the vestige of the primitive gullet. I have so far not been able to trace it 
through or even into the brain. 
The thyroid gland and the hypophysis make their appearance when the larvae 
are about 1-8 mm. long. The thyroid appears as a thickening in the ventral layer of 
hypoblast just in front of the anterior end of the heart (figs. 1716 and 171c). The 
hypophysis is somewhat different from the start. The cells of the roof become columnar 
and several layers thick and a refold is formed. This condition is well demonstrated 
in fig. 166. 
Shortly afterward the infolded cells are constricted off from the rest of the 
hypoblast and lie as an independent structure at the base of the brain (fig. 177). 
Gills . — The formation of the first gill-slit has been described in connection with the 
intestine, and the hyomandibular and mandibular slits have also been considered. It 
remains now to trace the formation of the posterior slits. Soon after the formation of 
the hyobranchial a thickened mass is formed behind it; this mass extends out beyond 
the outline of the body of the embryo, and owing to its well-defined limits is con- 
spicuous in the living larva (figs. 83-91) as well as in sections. (Figs. 137, 139, 150-157, 
172 to 174.) 
This mass is largely composed of mesoblastic cells. It grows very rapidly and 
from it are derived the skeleton and the soft parts of the gill-arches. The gill-clefts 
arise as pouches extending out from the entoderm. These may be met by shallow 
ingrowths of ectoderm (fig. 150, hr. 8). The pouches are formed from the hyobranchial 
backward. They seem to be somewhat irregular at first and hollow (figs. 151 and 156), 
but later the layers of hypoblast forming them are closely appressed, all the space 
between them having disappeared. The second is forming when the larva is about 
1 mm. long (figs. 136 and 137); the third when it has reached a length of 1-8 mm. 
These measurements are after hardening; the living larvm were probably somewhat 
longer. 
The fifth slit is nearly completed at 2-5 mm. Though the slits are potentially 
completed so early, the t«svo layers of hypoblast composing them are not separated from 
each other till the larvae have attained twice this length. They are seen to be sepa- 
rating in a larva 4 mm. (fig. 137). The slits are not vertical, but extend downward and 
forward. In the early stages the thickened mass containing the rudiments of the gills 
are entirely lateral. Below there is but a single layer of mesoderm (139). After the 
potential slits have been formed the mesoderm between them grows downward and 
forward till it reaches neai'ly the median line in larvm a little over 3 mm. long. 
