The brain of Acipenser. 91 
1. Body of Cerebellum. 
The granular layer. — The lateral portion of the body 
gives the appearance in haematoxylin preparations of a coarsely re- 
ticulated mass of medullated fibres among which are an enormous 
number of very small cells closely packed together. These cells 
are more numerous here than elsewhere, except a restricted part 
of the lateral lobe, so that this is one of the most deeply staining 
parts of the brain (Phots. 14, 15). These granule cells are the 
most minute cells in the brain of Acipenser, measuring 8—12 by 
8—16 u. They have rounded or ovoid bodies and are usually pro- 
vided with two dendrites arising from the opposite poles of the cell. 
The dendrites are very short and have only a few hook-shaped or 
claw-like branches. In this respect they resemble very closely the 
granule cells of the Selachian brain (ScHAPER, 798). The neurite 
arises from some part of one of the dendrites. In the caudal part 
of the boby the course of the neurites is very different from that 
of the neurites in the cephalic part. In the caudal part the neu- 
rites run dorsally and turn mesad to enter the “fimbria” in which 
they cross to the other side. Here they turn forward along the 
dorso-lateral surface of the body to reach the dorsal border of the 
lateral lobes, around which they continue as a distinct and compact 
bundle to form the greater part of the cerebellar crest. There is 
a gradual transition between the caudal and cephalic parts of the 
body with regard to the course of the granule cell neurites. 
In the cephalic part the neurites run at first ventro-mesad 
(Phot. 14), enter the median or molecular portion, turn dorsad 
and at the dorsal surface of the cerebellum run laterad in the 
molecular layer of the lateral lobe of the same side. The neurites 
of the granule cells do not bifurcate in the regular manner which 
is typical in the higher Vertebrates and which has been described 
by Scuarer for Teleosts (93) and Sharks (98). In Acipenser 
when the neurites reach the molecular layer they bend toward the 
lateral lobes (or in those lobes turn caudad) without dividing. In 
the body of the cerebellum, however, these neurites frequently give 
off two or more collaterals in the granular layer. I have never been 
able to trace these collaterals into the molecular layer. 
Besides the granule cells there are present a small number of 
larger cells which are most numerous in the near vicinity of the 
tractus tecto-cerebellares I and II (see page 128) and in the ventral 
