47^ journal of Comparative Neurology and Psychology. 



found for the small cells were nearly right but those found for the 

 large cells were less nearly right it is evident that we should expect 

 to find more cells showing smaller diameters than are actually 

 present in the nine slices examined. This leads at once to the 

 conclusion that the ordinates representing the cells of the small 

 diameters must be compounded of two heterogenous elements; 

 the true small cells plus those cells which are artificially made smaller 

 by the method of section. In the same way the ordinates for the 

 larger cells will be compounded of the measurements on the larger 

 cells minus those whichwere artificially made smaller. Thus we shall 

 find an excess of cases towards the smaller value of x and a deficit 

 towards the larger value of x. This is just what we have ob- 

 served. It was found, when the observed polygon was compared 

 with the theoretical curve that the latter considerably underesti- 

 mated the observed ordinates which correspond to the smaller 

 values of X and overestimated the observed ordinates which corre- 

 spond to the larger values of x. The facts mentioned above indi- 

 cate that the deviations of the observed polygon from the theoretical 

 curve are mainly, if not entirely, due to the method of section. 

 The greater observed excess found in the neighborhood of the 

 mean or mode is interesting, since it may be assumed that up to 

 the neighborhood of the mean or mode the cells remain nearly 

 spherical while beyond this region the increase in size is accom- 

 panied by a change in shape. 



If my argument is correct, then we should expect the greater 

 percentage deviation of the two diameters to appear more fre- 

 quently towards the larger abscissal values. Although it would 

 seem at first easy to test such hypothesis by taking the percentage 

 deviation from the averages corresponding to the different abscissal 

 values, nevertheless in practice we meet considerable difficulties. 

 Any one who is familiar with the sections of the spinal ganglia 

 prepared by the usual fixation methods, will recall that there are 

 present a number of small cells with very unequal diameters. 

 Such cells are most abundant along the periphery of the sections. 

 The general outline of such cells is either rectangular, instead of 

 curved, or the opposite boundaries are represented by nearly par- 

 allel lines. The cause of the deformation may be attributed to a 

 shrinkage of the capsule of the spinal ganglion itself. On account 

 of the presence of such deformed cells a mere comparison of the . 



