THE STRUCTURE OF THE SPINAL GANGLIA 163 
cated form is shown in fig. 6. Here the axon breaks up into a 
number of fibers which unite with each other to form a plexus 
out of which a single axon is again formed. Sections through 
much more complicated networks formed by splitting axons can 
often be seen. 
4. In another variety, closely related to those just described, 
the axon arises from the cell by two or more roots, each of which 
has the appearance of an axon and forms a conical expansion at 
the point of origin from the cell. Each of these roots may branch 
repeatedly. These branches then reunite with each other form- 
ing a more or less complicated network, out of which a single axon 
arises (fig. 7). 
Groups 3 and 4 are closely related in that the cells of the latter 
differ from those of the former only in the fact that the splitting 
involves the initial portion of the axon. In both cases it is rare 
to find a-myelin sheath on the fine fibers forming the plexus. 
These two groups correspond to Dogiel’s types v and vi taken 
collectively, but the basis of separation into the two groups is 
different. It is quite bewildering to read Dogiel’s description of 
types v and vi with their subvarieties and try to determine 
what the basis of classification of his seven subvarieties into these 
two major groups might have been. For this reason it has 
seemed best to adopt as a basis of classification the more obvious 
and apparently more fundamental difference in the hehe of the 
axon by a single or by a number of roots. 
According to Dogiel the axons of both types, after exhibiting 
their peculiar plexiform arrangements course as single axons for 
some distance and finally divide in the manner of a 7 or Y into 
central and peripheral fibers, a point which could not be verified 
in the relatively thin sections with which I worked. It was, how- 
ever, chiefly the origin of the axon from several roots, the split- 
ting of the axon or itsroots, and the formation of plexuses in its 
course which most needed confirmation; the final division of the 
axon as described by Dogiel agrees so well with our former knowl- 
edge of the ganglion that it may safely be accepted. The cells 
included in groups 3 and 4 are fully as numerous as those in 
group 2; and if one might be permitted to make a very rough 
THE JOURNAL OF COMPARATIVE NEUROLOGY, VOL. 22, NO. 2 
