24 WILLIAM V. ALLEN 



ventral cleft. These changes in the shape of the central canal 

 suggest a migration of cells, rather than a disintegration process 

 or a splitting apart of the walls as a result of pressure from the 

 secretion of cerebro-spinal fluid. The earliest Squalus series 

 showed an elliptical canal with its longest axis horizontal, this 

 changed to circular, and finally to the so-called typical embryonic 

 central canal (dorso-ventral cleft canal). For a time in the 

 medulla of Amblystoma the roof plate is much thicker than the 

 floor plate, a condition which can be attributed to the presence 

 of a number of neural crest cells in the roof plate, easily distin- 

 guishable from the other cells by their large size and spherical 

 form. After the neural crest cells had been entirely separated 

 from the roof, the roof and floor plates are found to.be about 

 equally thick. The same thinning out of the roof plate occurs 

 in Squalus in part through the giving off of neural crest cells, 

 which in this species are indistinguishable from the other cells 

 of the medulla. It is evident in Amblystoma (figs. 74 and 83, 

 R.P.) and also in Squalus that the roof plate of the medulla 

 has become no thinner than the roof plate of the spinal cord 

 through the throwing off of the neural crest cells. There is no 

 evidence that the roof plate in the medulla is rendered any 

 weaker or any more susceptible to expansion than is that of the 

 spinal cord, through the migration of gangUon cells. 



Apparently the first expansion of the roof plate in Ambly- 

 stoma (fig. 74, R.P.) and in Squalus is produced as in Petro- 

 myzon by an outward and upward migration of the roof plate 

 cells. The later, more pronounced expansion and stretching of 

 the roof plate can also be ascribed, as in Petromyzon, to internal 

 pressure due to a decided increase of cerebro-spinal fluid (see 

 fig. 72 for Squalus and 73 for Necturus). In both species the 

 dorsal and middle portions of the embryonic central canal of the 

 spinal cord are obliterated by an inward growth of the lateral 

 plates through an addition of fibers. Pressure on the cerebro- 

 spinal fluid might be increased from that source. Also a similar 

 effect would be produced through a marked proliferation of fibers 

 in the medial and ventral portions of the lateral plates of the 

 medulla, which brings about a coalescence of the ventral por- 



