iv GENEEAL PHYSIOLOGY OF NEEVOUS SYSTEM 187 



is proved by their continuity with the axis-cylinder 

 process. Golgi has hitherto failed to discover any 

 relation between these peripheral fibrils and the 

 endocellular reticulum, which appears 'to be an 

 argument in favour of Holmgren's hypothesis, 

 although Golgi's reluctance to accept this inter- 

 pretation is easily understood. 



The pericellular network described by Golgi 

 for different cells of the cerebellum, cerebrum, and 

 spinal cord consists, in his opinion, of neuro-keratin, 

 and he believes its function to be one of insulation, 

 as he considers it entirely different to and distinct 

 from the diffuse nervous network described above. 



This fibrillary network on the surface of the 

 nerve -cells is admirably shown up by Bethe's 

 method, and probably corresponds with the peri- 

 pheral network observed by Donaggio and by Cajal 

 in 1896. 



Donaggio obtained excellent preparations of 

 vertebrate nerve-cells by his special method. As 

 seen in Fig. 122, the cells are not only penetrated 

 at the periphery by longitudinal fibrils which pre- 

 serve their individuality without anastomosing, as 

 stated by Bethe, but in addition a great number of 

 fibrils can be seen which are directed to the centre 

 of the cell, and there divide minutely to form a 

 dense network which is not stained by Bethe's and 

 Golgi's methods. The fibrillary network is con- 

 nected on the one side with the 

 fibrils that penetrate from the 

 dendrites, on the other with 

 the fibrils that form the axis- 

 cylinder. 



Donaggio's more recent pre- 

 parations (1904) show still more 

 plainly that the fibrils of which 

 the axis - cylinder is composed 

 are derived directly from the 

 endocellular fibrillary network 

 (Fig. 123). The mode of origin 

 varies according to two cellular 

 types, indicated by Donaggio. 



On tracing out the course of 

 a sensory fibre, Apathy found 

 that it breaks up within the 

 central nervous system into an 

 elementary fibrillary network (EUmentargitter\ which 



suggests 



