411 



13 having been noted by Froriep in Torpedo. The posterior 3 or 4 

 have ventral roots corresponding to them, and the last may possess 

 a rudimentary dorsal ganglion in man and mammals. The last four in 

 the gull have rudimentary dorsal roots, and Chiarugi described 

 rudimentary roots to the posterior three in Gallus. There are 

 similarly 3 to 5 in reptiles, which have apparently no dorsal roots. 

 A dorsal ganglion has been discovered also to the last nerve of the 

 three which may survive in Selachii. Even the first spinal nerve may 

 in like manner to those in front lose its dorsal ganglion and root in 

 man and mammals. The seven which I have described above appear 

 to be identical in number and position with the seven which Sewer- 

 TZOFF discovered in Ceratodus. 



It is at once evident therefore that if these ventral roots of the 

 occipital somites are posterior to the cranial nerves and the latter 



Rudimentary dorsal roots 



-dorsal 

 ganglion 



Fig. 4. A similar reconstruction from sagittal sections of the still older embryo, '62'. 



are homologous with the dorsal roots of the spinal nerves we must 

 find room for a minimum of 13 nerves and neuromeres behind the 

 vagus, which would yield 21 rhombomeres and still more if the com- 

 plex condition of the vagus be considered. 



There is evidence to show however and it has been stated before, 

 for instance by His in 1888, that the cranial nerves are not homo- 

 logous with the spinal dorsal roots. The latter pass internal to the 

 myomeres to join the ventral roots, are not connected by ganglia with 

 the ectoderm, and are segmental in position. The cranial nerves, with 

 the exceptions which will be mentioned, pass outside the myotomes, 

 characteristically establish ganglionic connexions with the ectoderm 

 (the lateral line and epibranchial system), are independent of the 

 ventral roots, and are intersegmental in position. The two classes of 



