238 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY. 
structure of the trochlearis to be worthy of trust, or even of serious 
consideration. 
Before closing my account of the development of the cranial nerves 
and their chief branches in Squalus, I wish to call attention to a phenom- 
enon seen in still later stages of development, already noted by me in a 
former paper (’97, p. 455). It appears to me a matter of considerable 
morphological importance that the ganglion of the dorsal nerve of van 
Wijhe’s eighth somite (fourth post-otic) —the ventral root of which 
forms at this stage the first of the five hypoglossus roots — unites in 
late stages of development with the ganglion cells near the root of the 
vagus. Kupffer ('90) was the first to make evident the morphological 
importance of the clearly marked distinction between dorsal and lateral 
(epibranchial) ganglia in embryos of Cyclostomata. While in the em- 
bryos of Selachii there is not such a clearly marked distinction, there 
nevertheless exist at the roots of the vagus groupings of ganglion cells, 
or at least of neural-crest cells (quite distinct from the lateral, epi- 
branchial ganglia of this nerve, the ganglion nodosum), which in my 
opinion are to be regarded as homologous with the dorsal ganglion of 
the vagus of Ammocotes.! The evidence of the union of dorsal seg- 
mental ganglia in the vagus is as follows. During development the 
continuous neural crest, in the occipital and trunk regions of Squalus 
becomes differentiated into clearly marked ganglia, lying opposite the 
myotomes and connected by a cellular “dorsal commissure ” (Balfour, 
’81), as far forward in the embryo as van Wijhe’s seventh somite. Oppo- 
site the sixth and seventh somites no distinct ganglia appear; but instead 
a wide sheet of cells, lying in close juxtaposition to the extended roots of 
the nerve, is observable. While in early stages the ganglion of the eighth 
somite is separated by a considerable interval from the roots of the vagus, 
in later stages it approaches these, and in embryos of 30 mm. is seen to 
be in union with them as a well marked ganglionic appendage. In later 
stages, its fusion appears complete. The ganglion cells do not degen- 
erate, but send axis-cylinder processes both centripetally and centrifu- 
gally, the latter forming the posterior of the roots of the vagus nerve. 
The ganglion of the second hypoglossus root (ninth somite) does not, 
however, so fuse with the vagus, but is seen in embryos of 50 mm. as a 
group of cells without nerve relations, so far as I am able to determine, 
enclosed in the cartilage of the cranium. It apparently disappears in 
1 These are probably the homologues of the intracranial ganglia of Ganoids 
(see Allis, ’97, p. 747). 
