A STATISTICAL METHOD FOR MEASURING “MARGINAL UTILITY” 183 This is our typical want equation. It applies only when Si¢r Fi S; 0." BE: pi ve., applies only as between two average families, one in Oddland and the other in Evenland, whose rations are the same, or, more precisely, whose food expenditures are exactly proportional to the food price indexes of the two countries. This implies, of course, that the want-for-one-more food unit, being dependent only on the food ration, is not dependent on the housing situation nor on any other circumstances likely to per- turb the picture. In particular, it is implied that the want-for- one-more food unit of Case 1 is not dependent on the budgets, or other circumstances of the neighbors (or else that these influences are the same in the Cases compared). Likewise, it is implied that the want-for-one-more unit of shelter is independent of other budgetary items and of the neighbors’ (or else that these influ- ences are the same in the Cases compared). (d) Equality of price indexes applicable to Cases 1 and 2, i.e. Fi=F; and R;=R;. But although this is assumed, it is not a necessary assumption. In the first place it may be pointed out, that for comparison between Cases 1 and 3 this assumption is entirely superfluous since only Fy (i.e. not Fs) and only Rj (i.e. not R,) enter into the formule.* The assumption F;—F, means that the food markets of Odd- land and Evenland compare alike at both grades of food,—the grade used by Cases 1 and 2 and the grade used by Cases 3 and 4. To make the assumption more general, the market in both Odd- land and Evenland are assumed to afford substantially the same grades A, B, C, D, ete., successively differing in cost by $1 in Evenland and by $1.3315 in Oddland. This assumption seems reasonable as between countries of the same sort of culture such as England and the United States, although, of course, it might conceivably be true that, say, the inferior grades of food in Odd- land cost 1335% as much as in Evenland, while, say, the superior grades cost 120% or 150% as much. In that case Fi would be * For the comparison later on between Cases 1 and 5, both F, and F, enter and both Re and Rs. The assumptions in question (specifically that F,=F: and R:=R;) are used in deriving formule (7) and (8); without these assumptions these formule would obviously be slightly different.