I020 The American Naturalist. [November, 
surface. This keel subsequently becomes separated from the superficial 
layers of the ectoderm, and afterwards a central canal is developed in 
it. In the ganoids, which approach the elasmobranchs in structures 
there is, as shown by Selensky ^ a medullary groove of peculiar 
form, which suggests a transition from the solid keel to the open 
groove ; again in amphibia there is evidence that the delamination is 
still preserved to a slight extent in that group. These considerations 
lead me to the hypothesis that the nervous system of vertebrates was 
primitively a solid axial thickening of the ectoderm, and within the 
class of ganoids became modified into a groove, perhaps simply by 
more precocious development of the central canal ; the groove type 
has been kept in elasmobranchs, amphibians and amniota. Balfour 
(Comp. Embryol., II., 303) thus defends the opposite view : "It seem, 
almost certain that the formation of the central nervous system from a 
solid keel-like thickening of the epidermis is a derived and secondary 
mode ; and that the folding of the medullary plate into a canal is 
primitive. Apart from its greater frequency, the latter mode of 
formation of the central nervous system is shown to be the primitive 
type by the fact that it offers a simple explanation of the presence of 
the central canal of the nervous system ; while the existence of such a 
canal cannot easily be explained on the assumption that the central 
nervous system was originally developed as a keel-like thickening of 
the epiblast." 
It is not possible at present to decide positively between the two 
views, but the view which I am inclined to adopt is further justified 
by the development of the central nervous system in annelids, which 
is formed by the coalescence of a pair of linear cords : these cords 
arise each side of a cilated longitudinal furrow, first as a single row of 
ectodermal cells, subsequently as several rows ; while still united to the 
external ectoderm they extend towards one another below the ciliated 
cells of the furrow, and unite in a single nervous band. In leeches 
and arthropods the development is very similar. In all these cases 
the bands split off from the ectoderm. It appears, then, that in the 
nearest ' invertebrate allies of the vertebrates the nervous system devel- 
opes as a thickening along the inner surface of the ectoderm, and 
delaminates from that layer. It seems to me very natural to suppose, 
therefore, that the strikingly similar process in the lowest vertebrates 
is the primitive one, and that the canalization of the medullary plate 
was evolved within the vertebrate series. 
e exception of Amphio 
