ScHAPER, Structure of Selachian Cerebellum. w 



in the axis-cylinder of a granule cell, which I have figured in 

 figure 9 (Plate II). We see here the nervous process divide fork- 

 like while still in the gramdar layer. I recollect having once be- 

 fore happened upon a similar condition in the cerebellum of a 

 mammal. We have here, doubtless, a developmental anomaly 

 viz. : an abnormal division of the axis-cylinder. 



I have above called attention to the fact that most of the nerve 

 processes of the small granule cells appear to run in a more 

 or less direct course to the molecular layer ; frequently, how- 

 ever, one meets with cells also where the axis-cylinder first pro- 

 ceeds horizontally for a considerable stretch and then wends up- 

 ward in a wide curve (Plate II, fig. 1 1 below). I have unfortu- 

 nately not been able to follow such an axis-cylinder into the 

 molecular layer. I conjecture however that these axis-cylinders 

 are devoted to those parts of the cerebellar plate where a gran- 

 ular layer is lacking. We also find here in the molecular layer 

 numerous axis cylinders of granule cells ascending and are 

 thereby forced to the conclusion that these portions of the cere- 

 ' bellum are supplied by the granule cells of otJier regions. 



It remains to be mentioned that the neuraxons of the 

 granule cells on their entrance into the molecular layer appear 

 to gain somewhat in caliber and are more thickly studded 

 with varicosities than in the granule layer (Plate II, fig. ii). 

 Shortly before their termination in the molecular layer they usu- 

 ally turn upwards abruptly and thus send their terminal arbori- 

 zations to the upper portions of this layer (Plate II, fig. 12). 



The ^' large granule cells" (Plate II, figs. 13, 14, Plate III, 

 fig. 15) are, as in the other vertebrates, cells of the ''Golgi-type," 

 i. e. those whose axis-cylinders break up in their terminal arbori- 

 zations soon after their origin. The very volwninotis cell-body is 

 usually round and in size often exceeds that of the Purkinje cell. 

 The extremely coarse protoplasmic processes are small in number 

 and exhibit no great tendency to branching. The nervous process 

 arises in the majority of cases directly from the cell body, only 

 seldom from a dendrite (Plate II, fig. 13). Its terminal arborization 

 is less complex than in the higher vertebrates and is distributed 

 in an entirely irregular manner among the small granule cells. 



