A STATISTICAL METHOD FOR MEASURING “MARGINAL UTILITY’ 191 
Similarly, multiplying (11) and (4) we obtain 
SspsW ee p2/p1 (13) 
SioiW1  ¢o/ 
Before we can plot the want curve for food we need to get ¢3 
from the budget tables; and before we can do the same for rent 
we need similarly to find p;. 
Suppose we find ¢5=30% and p;=24% ; we now have all the 
data needed for calculating and plotting the two want curves 
(for food and shelter). All our data may be tabulated for refer- 
ence as follows: 
Si = $1000 per year 
¢ 1 
Fr 
*. 4 
W 
J wan 
1 ¢*] 
er 
per 
IS 
S; = $600 per year 
$2 = .50 
n2 20 
: it per “Ib.” 
1 per “sq. {i 
1 wantab 
Wa= 
S; = $1440 per year 
bs = .30 
Ps = 2D 
Vy 81.33% per “lb.” 
KR: $3 per “sq. ft.” 
Ws: = .33% wantabs 
In this table the four given magnitudes are S,, Fs, Rs, Wo, all in 
the middle column and three of them being the units of measure- 
ment assumed. 
The remaining magnitudes are all calculated from these four, 
or obtained from budget tables or from our assumed conditions. 
We could now easily plot the quantity of food and its want- 
ability from 
Sign 1000 X .40 
Ww. 75 Z $1.331 = 1.00 
these two being the “latitude and longitude” of one point (that 
for Case 1) ; and, likewise plot the analagous quantity and want- 
ability for Case 3: 
n 
Sade _ 144030 _ 394.00 
Wal «1.331 = 44. 
Such a curve would be none other than the “curve of diminishing 
utility of food” used in our text books but not hitherto reducible 
to statistics. 
The figures show that (according to our purely illustrative 
data) if the quantity (or, more strictly, index) of food consumed 
is increased from 300 to 324 the want-for-one-more unit of it 
decreases from 1.00 to .44 wantabs.