Editorial 
131 
^'"pq given in equation (46) and we easily find 
if < n^i and < 
cr 
1 1 f 1 ?i,>/j n,\ 1 
if > and < Tiq, 
if >»j, and > r),^ (51). 
a'-x^ can always be found, and the error of our determination of the mean from 
the j'th dichotomic line determined. But unless we are given some lengths on our 
variate line corresponding to dichotomic lines we can only choose (i) the p, q range 
to give cr, (ii) the rth dichotomic line to determine the mean. All other scale lines 
are determined by these, but the degree of accuracy of their determination depends 
on the fitting choice of /), q and r. 
(8) As illustration let us determine the probable errors of the following distri- 
bution of Intelligence in Schoolboys on the assumption that the distribution may 
be taken as normal. 
Quick Intelligent 279-.5 . j n,JN = VOOO, 000 
Intelligent 788 Ho/^= -883,00.5 
Slow Intelligent 771-5 V^^" ''^ nJN = -5.53,160 
Slow 369 J. >hlN= -230,222 
Slow Dull 139-.5 J"^" n,/N= -07.5.764 
Very Dull 41-5 ''^ n,/N= -017,371 
2389 
H,= -042,944, H.=- -142,542, i/^ = -303,847, H,= -395,391, E,= -196,487 
A, = 2-111,366, = 1-434,202, = -737,957, A, = - -133,651, =- 1-190,144 
Taking the range of Intelligent to be 100 mentaces we have : 
a {hr, — hi) = 100, or cr = 94-653 mentaces. 
Whence by (46), p. 129, putting ?■ =j^), s = q, we have 
o-„5 ^ — 2-4594 mentaces, 
and the probable error of 0-5 4= 1-6589 mentaces or is less than two mentaces, which 
is amply accurate for this type of work. 
The position of the mean is at h^a from the border of Intelligent and Slow 
Intelligent into the latter = 12-6504 mentaces from the 4th dichotomic line. 
