The discussion of P, the price-level, in the statistics of Kemmerer and Fisher need not be extended. P, for the equation of exchange, and for the quantity theory, is a _weighted_ average, each price that goes into it being weighted by the number of exchanges involving the commodity of which it is the price. The weighting of P should correspond to the elements in T, the volume of trade, and should vary from year to year, as the elements in T change.[449] Now Kemmerer's P is weighted as follows: wages, 3, security prices, 8, wholesale prices, 89.[450] If our conclusions with reference to the composition of the volume of trade, as developed in the chapter on "Volume of Money and Volume of Trade," are valid, this weighting gives us a P which has no relevance to the equation of exchange. The wholesale items should have a weight of not more than one-sixth of the total for 1909. Certain commodities, as wheat and cotton, in which there is heavy speculation, should be given great weight, and securities should have, probably, the greatest weight of all. If "trade" is to be extended to cover transactions in bills of exchange and loan transactions (as it is by Kemmerer),[451] then P should contain these things, weighted more than all else put together, particularly if call loans are included. The weights should be radically altered from year to year. We should then get a P which would fit the "equation of exchange"--though what else it would be good for is hard to say! The same criticism applies to Fisher's P. It is dominated by wholesale prices.[452] It therefore has no relevance to an equation of exchange in which only one-sixth at the very most of the items are wholesale items. Neither Fisher nor Kemmerer alter their weights in P at all, to correspond to yearly alterations in the composition of T. As _indicia_ of changes in the _absolute value_ of money, Kemmerer's and Fisher's index numbers, or other index numbers of numerous wholesale prices, with a substantial weighting of wages, are probably better than an index dominated by stocks. Stocks fluctuate more widely than wholesale prices and wages, their values are more affected by variations in business confidence, and by variations in the rate of interest. For measuring _the value of money_, the index numbers here criticised are very good. But for the purpose for which they are chosen, namely, to fill the equation of exchange, and to measure variations in a _price-level_ of the sort the quantity theory and the equation of exchange are concerned with, they are simply irrelevant. If it were really true that such an index number varied with the quantity of money, then the quantity theory would be effectively disproved!