H. E. Soper 115 
p = -66, n = 30. 
r 
Calculated 
frequency »t 
Observed 
frequency 
Difference 
e 
er 
m 
■795— 
8-5 
6 
-2-5 
•74 
•735— 
15-8 
17 
+ 1-2 
•09 
■675— 
21-6 
27 
+ 5-4 
1-35 
•615— 
20-9 
22 
+ 1-1 
•06 
•555— 
15-8 
14 
-1-8 
■21 
•495— 
9-8 
10 
+ "2 
•04 
1— -495 
7-6 
4 
-3-6 
1-71 
100 4-20 
n' = 7, x 2 = 4-20, P = -650. 
It is hoped that further experiments may be shortly carried out which will 
have regai'd to the points raised and show definitely whether the distributions 
theoretically arrived at in this paper are good presentations of fact and whether 
the application of the standard types of frequency curves to the distributions of 
statistical constants in small samples is justified. 
I am indebted to Professor Pearson for drafting the lines of this investigation 
and for critical supervision. 
15—2 
