114 JOURNAL OF COMPARATIVE NEUROLOGY. 
the tip is reached, where it dissolves into a brush or a series of 
hooks spreading over a relatively considerable area. There are 
no gemmules. 
The axone is exceedingly slender. It may arise directly 
from the cell-body, but it usually takes its origin from one of 
the dendrites, either near the base or at some distance from the 
cell; Fig. 15 illustrates the two modes of formation. The 
course of the axone is invariably peripheral, pushing through 
the intervening thickness of the granular layer into the molec- 
ular layer. At the boundary between the two layers, the course 
changes abruptly to a horizontal one for a short distance, and 
hence the entire course of an axone can rarely be traced in one 
section. Inthe molecular layer, the axone divides into two 
branches which, with the original stem, form a T-shaped figure 
(Fig. 16). The two branches pursue a course parallel with the 
surface of the cerebellum and the sides of the fold in which 
they run. It is thus seen that they pass through the dendritic 
tops of the neurones of PURKINJE, comparable to telephone 
wires passing through the tops of the trees along a highway. 
b. Golgi Neurones.—A few neurones of the granular layer 
have an altogether different character from the ones just de- 
scribed. These lie in the upper levels of the layer. Such a 
neurone is shown in Fig. 17. The cell-body is always a little 
larger than that of the typical granule neurone, and its form 
is more rounded. A few club-like dendrites radiate from it for 
a short distance, branching but sparsely. The size of a dendrite 
is increased at intervals by slight swellings. 
The axone passes downward into the deeper levels of the 
granular layer, instead of upward. It gives off collateral 
branches soon after its origin, and it breaks up into a number of 
fine terminal twigs before any great distance has been traversed. 
This neurone is homologous with the variety described by 
Gotei (94) from the human cerebellum, and by him made a 
representative of his second type of nerve-cell. The branching 
of the axone is far less profuse, however, than in the mammal. 
We thus see that the granular layer of the cerebellum of 
Mustelus is marked by the presence of the same varieties of 
