472 journal of Comparative Neurology and Psychology. 



the following conditions, within the limit of random sampling, 

 must be fulfilled : 



]//?i = o; .92- 3 = o; ^ ^ ^^ V — = o; and J, modal divergence =0. 



5/?2-6/?i-9 



By examining the analytical constants given in Table I, it is 

 seen that all those constants for both the cell-bodies and nuclei 

 are considerably greater than zero even when their respective 



210 

 200 



12 i6 20 24 28 32 36 40 44 48 micra. 



Fig. I. Frequency polygon and fitted curve for variation in the diameters of the cell-bodies. 



probable errors are considered. This at once leads to the con- 

 clusion that in both cases the frequency distribution can not be 

 represented by the normal curve. Furthermore a considerable 

 deviation of those constants from zero, i. e.^ skewness, as well as 

 modal divergence, indicates that they can never be represented by 

 any other symmetrical curve since the deviation in excess and 



