Tlir iiu>tli(Ml consisted in counting,- all the iinclroli in 

 eri'i-i/ seel ion of a complvtc si-ricn tlinuiijli the f^iinjjjlion, 

 and all the fibers in a cross section of the root between 

 the ganglion and the cord, generally close to tlu; gang- 

 lion. The tissue is hardened in osmic acid. A nuc- 

 leolus may be pushed to one side or dragged out of a 

 cell by the edge of the knife, but is never cut in two. 

 So we run no risk on that score of counting a cell 

 twice ; and double nucleoli are so rare that they may 

 be left out of the account.' My rule throughout was to 

 count everything that could be construed to be a fiber 

 in the root, and nothing but what was most certainly 

 a cell in the ganglion, thus throwing all the doubtful 

 cases upon the same side. 



With the above figures approximately correct, either 

 apolar cells in the ganglia must be very numerous, 

 or the relations of fibers to cells must be more complex 

 than formerh' supposed. A most careful teasing of 

 spinal ganglia of a frog, using a fine jet of water 

 instead of needles, I think demonstrates the following 

 points : 



1. Apolar cells do not occur in the spinal ganglia of 

 the frog in any considerable numbers, none having 

 been observed. 



2. Typical bipolar cells do occur. Three have been 

 noted up to date. 



:3. The axis cylinder of the cell process is often seen 

 to divide and enter the cell as a spiral and straight 

 fiber. 



4. At the angles of the " T " the axis cylinder of the 

 cell process maj' be observed to divide and pass 



' In counting the last ganglion, T. kept careful account of all 

 double nucleoli. In the 4456 cells 38 were found. In all cases they 

 were found in the small cells, lying among the fibers or close to the 

 nerve fiber axis of tliu ganglion. 



