224 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY. 



observations confirm those of Miss Piatt. According to Sedgwick ('94) 

 the third nerve is formed directly from the neural crest as are the dorsal 

 cranial nerves [?], but arises as a diffei-entiation of the reticulum formed 

 by the breaking up of the neural crest, and first makes its appearance as 

 a projection of nuclei from the mesoceplalic ganglion. His observations 

 thus do not essentially differ from those of Miss Piatt, their conclusions 

 diflFering chiefly by reason of diff'erence in theoretical views as to the 

 mode of nerve development. My own evidence differs quite fundamen- 

 tally from that given by pi-evious investigators, since I find that the 

 nerve develops after the manner described for spinal ventral nerves in 

 Selachii and other Vertebrates, as an axis-cylinder process from " neuro- 

 blast " cells in the ventral horn of the midbrain. At the earliest stage 

 iu which I have been able to detect the oculomotorius the extent of its 

 development and its relationships are such as are shown in Figures F 

 to H, Avhich represent sagittal sections of a Squalus embryo with 

 52 somites (approximately 8 mm. long). At this stage the thalameu- 

 cephalon is just becoming differentiated from the primary forebrain 

 (eucephalomere I). The identification of the fibrillar process as the 

 oculomotorius is made easy by a comparison of its point of attachment, of 

 the direction of its long axis, and of its histological appearance with those 

 of an embryo with 54 somites, where the oculomotorius is alread}' con- 

 nected with the mesocephalic ganglion. Under higher powers of the 

 microscope the nerve appears as a deeply staining, highly refractive 

 process, clearly distinguishable by these characteristics from the granu- 

 lar and faintly staining processes of the mesenchyma cells at the base 

 of the midbrain. Owing to a shrinkage, which however appears in very 

 few of the specimens killed by the fixing agent used (vom Rath's fluid) 

 and always most markedly in the region ventral to the midbrain, the 

 mesenchyma cells and the roots of attachment of the nerV^ have broken 

 away from the base of the brain. Since, however, similar deeply stain- 

 ing processes are seen to extend from cells in the ventral horn of the 

 medullary tube towards the points where the roots may be supposed to 

 have once united with the wall of the brain, the inference seems war- 

 ranted that the nerve is made up of the processes of these cells. The 

 latter show the characteristics described by His ('89) for the neuro- 

 blasts of the spinal cord, viz. a highly chromatic nucleus surrounded by 

 a thin, very deeply staining protoplasmic ring, which is pi-olonged into 

 the axis-cylinder process. The precipitation of osmium serves to render 

 the processes quite opaque and easily traceable among the remaining, as 

 yet undiflFerentiated, cells of the medullary wall, and to make it possible 



