OF ERRORS OF JUDGMENT AND ON THE PERSONAL EQUATION. 
Table XIII. 
2(39 
= T729 
= -.5503 ± -0210 
o'x'sj'i = T793 
= -5002 + -0226 
•2» 31 — 
= -1852 
P = -4478 + -0241 
r 3- 12 — 
1 
We have now the complete data requisite for analysing the experiments on the 
bisection of straight lines. We place all the correlation coefficients together in 
Table XIV. for comparison of the two methods of deducing results. 
Table XIV.—Correlation in the Judgments as to Midpoint of Lines. 
Errors measured in terms 
of length of line. 
Errors measured 
absolutely. 
>'■’3 
’>31 
>'l-2 
■3627 + -0262 
•1139 ± -0298 
•2053 ± -0289 
•3596 + •0263 
•1242 + -0297 
•2223 ± -0287 
in, -13 
p-2, 31 
P3, 12 
•5615 + -0207 
•4980 ± -0227 
•4379 ± -0244 
•5503 + -0210 
•5002 + -0226 
•4478 ± -0241 
We conclude at once that:— 
(i.) Within the limits of the probalJe errors of the ol)servations tlie correlations of 
the errors in judgment, whether measured alrsolutely or in terms of the length of the 
line bisected, are sensibly the same ; and this is true not only for the correlation in 
absolute (r) hut also for the correlation (p) in relative judgments. 
(ii.) Thus while we have shown that tlie erj’or in lusecting a line is not proportional 
to the length of the line, and indeed not at all or only slightly correlated with it, 
yet the observed correlation of judgments cannot arise solely from the use of a I'atio 
or index. For this correlation still exists, if 'we deal Avith the absolute errors. It is 
thus not a purely spurious correlation. 
(iii.) The correlation varies considerably from one pair of ol)servers to a second. We 
thus are forced to conclude that it is not a result of a common varying external 
cause, but must in part or wholly be due to a common element in the personalities of 
tAvo experimenters, Avhich is affected in the sajue Avay, and differently from some other 
common element in the personalities of another pair of observers. 
Thus the bisection experiments entirely coiifirm the conclusions Ave haA^e formed as 
a result of the bright-line experiments. In both cases there is a real personal 
correlation of judgment, only in the tAAm series it is differently masked by or com- 
