378 Journal of Comparative Neurology. 



Their neurites correspond closely to the central processes of the 

 spinal ganglion cells. This correspondence is complete in those 

 rare cases in which there is a single neurite dividing into rostral 

 and caudal (ascending and descending) branches. Their den- 

 drites also may be compared with the peripheral processes of 

 the spinal ganglion cells, being distributed to the integument. 

 The comparison is striking in the case of spinal ganglion cells 

 such as those shown in Figs. 4 and 7, which send their periph- 

 eral processes up to the dorsal surface. The existence of cells 

 within the spinal cord which correspond in function and in the 

 disposition of their processes to the spinal ganglion cellls may 

 be explained by the hypothesis that they are derived ontogenet- 

 ically from the neural crest, from which the spinal ganglia arise, 

 but have failed to migrate with those ganglia. The distribu- 

 tion of the dendrites of these belated cells suggests that they 

 may be compared most closely with the cells whose peripheral 

 processes go into the dorsal rami of the spinal nerves. 



Comparative anatomical investigations seem to show that 

 in the phylogeny of vertebrates the spinal ganglion cells have 

 been derived from the spinal cord. In Amphioxus (Retzius 

 '90) the sensory roots of the spinal nerves, which are without 

 ganglia, arise from cells within the spinal cord. Retzius de- 

 scribes three varieties of such cells : transverseiy disposed bi- 

 polar cells, one of whose processes divides into a peripheral 

 fiber and a fiber of the longitudinal tract, while the other pro- 

 cess goes to the longitudinal tract of the opposite side ; multi- 

 polar cells ; and longitudinally disposed bipolar cells sending 

 one process into the sensory root and the other into the longi- 

 tudinal tract of the same side. Retzius points out that Am- 

 phioxus is the only vertebrate whose dorsal roots have no gan- 

 glia but arise from cells within the cord. KoUiker ('96, p. 158) 

 suggests that the cells which give rise to the sensory nerves in 

 Amphioxus are homologous with the neural crest of the em- 

 bryos of higher vertebrates, and that this crest is therefore to 

 be considered as derived from the cord rather than as a direct 

 derivative of the ectoderm. 



In Petromyzon Freud ('78) found longitudinally disposed 



