OH. 6
CURRENCY STANDARDS
85
Now, whilst the conception of a price index asbeing the price of a composite commodity (Edge-worth’s “ power of money to purchase advan-tages ”) has been becoming increasingly prevalent(all the old-fashioned index-numbers were almostentirely unweighted, whilst the best new ones, asfor example the U.S. Bureau of Labour Index, arecarefully and elaborately weighted), yet this otherconception has not been rooted out and still maintainsa demi-sway, a traditional influence over the statisticalworld. The conclusion of the British AssociationCommittee of 1888 (in spite of their recommendinga weighted index-number for practical purposes as“ commanding more confidence ”), that “ the scientificevidence is in favour of the kind of index-numberused by Prof. Jevons, provided there is a large numberof articles ”, has never been expressly repudiated bythe economic world.
Nevertheless I venture to maintain that such ideas,which I have endeavoured to expound above asfairly and as plausibly as I can, are root-and-brancherroneous. The “ errors of observation ”, the “faultyshots aimed at a single bull’s-eye” conception ofthe index-number of prices, Edgeworth’s “ objectivemean variation of general prices ”, is the result of aconfusion of thought. There is no bull’s-eye. Thereis no moving but unique centre, to be called the generalprice-level or the objective mean variation of general
Indices, chap. 1 ; see also Bowley, “Notes on Index Numbers” (EconomicJournal, June 1928, pp. 217-220). But the method seems to me to be mis-conceived, except for negative purposes, i.e. to show that there is no regular-ity in the shape of the dispersion, unless it were to be applied to a greatnumber of cases. To apply it in a few cases, as its exponents have done,proves nothing, except as a corroboration of something we have reason toexpect a priori. If it were to be shown that the curve of dispersion is ofthe same type in a great number of different contexts, then one would takenotice ; but the investigations, so far as they have gone, show nothing of thekind. It is worth mentioning, however, that M. Olivier and Prof. Bowleyboth conclude that, as a matter of curve-fitting, the geometric curve fitsbetter, in the cases which they have examined, than the arithmetic.