A STATISTICAL METHOD FOR MEASURING ‘MARGINAL UTILITY” 189 
of family budgets beyond the lower incomes, those of working- 
men. 
And just as, through Case 2 in Evenland taken as a yardstick, 
we are enabled to compare Cases 1 and 3, both in Oddland, so 
Case 3 could be used as a yardstick to enable us to compare 2 and 
4—Dboth in Evenland—and then go on to 6, 8, etc. In this way 
we could construct a series of points on a corresponding curve for 
Evenland. 
Comparison Between Two Countries Possible 
Moreover, not only can we thus compare wantabilities between 
different families in one and the same country under the same set 
of prices and general conditions and subject only to differences in 
income, but we can also make comparison between the two coun- 
tries, involving different prices as well as different incomes. 
All the foregoing calculations are supposedly worked out by 
using the two sub-groups specified, food and rent. But the same 
method applies with any other two sub-groups—food and cloth- 
ing, for instance, or clothing and rent, as long as the three speci- 
fied assumptions apply. 
Moreover, the same method may be applied to two different 
times instead of two different places, using, say, 1927 instead of 
Oddland and 1900 instead of Evenland. 
Wantability Curve for Any Commodity Group 
Thus far the only curves of want constructed relate to total 
income, giving quantitatively the “law of diminishing utility” by 
which the subjective value of a dollar diminishes as the number 
of dollars in one’s income increases. But by similar methods we 
may construct wantability curves for the sub-groups, food, rent, 
clothing, ete. 
Let us take the food group, for instance. The money expendi- 
tures for food in Cases 1 and 3 were S; ¢1 and S3 ¢3; while the 
physical quantities—what we first called “pounds,” but what, 
more exactly, may be described as an index of food consumption— 
are Bh and bh . The corresponding marginal wants,—i.e., for 
food per “pound,”—we found to be W, F; and Ws Fs. These last 
four expressions relating to food, the first pair being “physical” 
quantities (or indexes thereof) and the second pair being their 
AN