122 J. B. JOHNSTON, 
a much deeper stain in haematoxylin sections than the epithelium 
fo the saccus (compare Phots. 18 and 20). There is an indistinct 
basement membrane. Some of the tubes open into common crescent- 
shaped space between the caudal and cephalic parts. This is in the 
brain of a fish four feet (1!/, meters) in length. In the young fish 
the central cavity is much larger (Phot. 22). 
The caudal portion, in the sagittal sections which I am de- 
scribing, is sharply characterized by the presence of large and con- 
spicuous bundles of nerve fibres taking a red stain. In fact, these 
nerve bundles appear to constitute nearly one-half the bulk of this 
caudal portion of the hypophysis. The nerve bundles come from the 
thin ventral wall of the corpus mammillare. A part of these fibres 
also continue back along the ventral wall of the corpus mammillare 
to enter the saccus. These nerve bundles are separated from the 
epithelium of the hypophysis by a distinct membrane which has the 
appearance of the basement membrane in the saccus. The nerve 
fibres are stained red, the epithelium blue, and the nerve fibres 
never pierce the membrane to approach the hypophysial cells. Indeed, 
the epithelium is often widely separated from the membrane sur- 
rounding the nerve fibres by connective tissue. Within the bundles 
of nerve fibres there occasionally appear groups of epithelial cells 
stained blue and in some sases these enclose small cavities. 
Among the epithelial cells there are conspicuous everywhere slender 
cells stained intensely blue which pierce the nerve bundles and reach 
the surrounding membrane. In a few cases I have been able to 
demonstrate ciliated cells similar to these of the saccus lining the 
small cavities. In these sagittal sections, which are of the brain of 
a fish 11/, meters in length, there is no connection visible between 
these small cavities and the cavity of the corpus mammillare. But 
in sagittal sections by the GoLGI method I have found the nerve 
bundles surrounding narrow tubes which are directly connected with 
the corpus mammillare (Phot. 58). Finally in frontal haematoxylin 
sections of the brain of a young fish such as were used for GOLGI 
sections, I have found numerous tubular extensions of the ventral 
wall of the corpus mammillare into this part of the hypophysis 
(Phots. 19—21). These are lined with the characteristic ciliated 
epithelium which is distinct from the epithelium of the hypophysis. 
The two kinds of epithelium are readily distinguished in haemato- 
xylin sections by the depth of stain (Phot. 20). These evaginations 
from the corpus mammillare are also supplied with a thick layer of 
