= 0 1 
cl 
A- = 0 J 
N 
h = -i; 
d 
A- = 0 1 
N 
/( = 0 
W 
A- = -1 
In 
h = -1 
)d 
k = -1 
hence h = -06969) d. _ 
286 Siqyplementary Tables for High Tetrachoric Correlations 
We have djN = 6/216 = -02778, 
= 1 (1 _ a,) = -47222, ift = 1 (1 _ = 48148, 
or h = -06969, k = -04644. 
From the Tables for r = — -95 
, hence h = -06969) d 
F = -038,334, 
•033,024 ' 
= -033,024 
= -020,318 
J 
From the last column on right we deduce : 
/i = -06969, A; = -04644, J, = -031,756. 
For the given values of /; and A- we have accordingly : 
r = 1-00, r = - -95, 
djN = 0, djN = -031,756. 
Thus for djN = -02778, we have r = - -9562. 
Accordingly for the table returned to its original form, we conclude that the 
association between jiidgraents of sex made by two independent and competent 
observers from the femur is measured by a tetrachoric correlation of + -956. 
Illustration II. The following data are for the French long bones in the male, 
maximum lengths: 
Femur: Mean = 452-28 mm. Humerus: Mean = 330-10 mm. 
Standard deviation = 23-72 mm. Standard deviation = 15-38 mm. 
Correlation of femur and humerus = -8421. 
(a) Find the percentage of cases in which a humerus of under 300 mm. will 
be combined in the same individual with a femur of over 480 mm. 
Here /i = 27-72/23-72 = 1-16863 ; A- = - 30-10/15-38 = - 1-95709. 
Since h is positive, k negative and r positive, we must replace our system by 
h = 1-16863, k = 1-95709, and r = - -8421. 
Ouv tables for r = — -80 and — -85 show that for the given values of /; and 
A' the recjuired frequency would probably be less than 1 in 50,000,000. We may 
conclude therefore that no such individuals would occur in the total French male 
population. This is a result whose order would hardly be appreciated without 
an examination of the present tables. 
