164 Trans. Acad. Sci. of St. Louis. 
The mean weight is obtained for each grade by adding the 
figures in the grade column, beginning at the bottom, until 
the sum cannot be increased by the next number without 
exceeding 50 per cent of the total number in the grade. 
This point is marked in the columns of the above table 
by an underscore. Somewhere in the first number above 
this line lies the mean of all the observations in the col- 
umn. Its position may be determined by interpolation. 
An example will make this clear. Let us calculate the 
mean weight of boys of nine years in Grade II.* Fifty 
per cent of the 1,195 boys in this age and grade is 597.5. 
Adding up the column we reach 556 at 56 pounds. The 
next number in the column would make 621, which is more 
than 50 per cent of 1,195. The mean is therefore greater 
than 57 pounds and less than 58 pounds. The difference 
between 597.5 and 556 is 41.5, which is 64 per cent of 65. 
Thus the mean weight is 57.64 pounds. 
The mean weight in pounds for each grade is printed i in 
DOUBLE FACED type in Table No. 2 ( bows): and Table No. 3 
(girls), along with the number of observations (lower case 
type) from which each weight was calculated. The mean 
weight at each age irrespective of grade is also given. In 
Table No. 4, the weights in the two preceding tables are ex- 
pressed in kilograms, the figures for boys being in DOUBLE 
FACED type. 
* This method is that used by Galton. See Natural Inheritance, London, 
1889, Chapter IV 
