Historically, index numbers of prices have been constructed for the primary purpose of measuring changes in the value of money, and this remains today a major objective. But this is not the only purpose which such measurements now serve. Index numbers of prices are widely followed for the light they throw on the processes of production, exchange and distribution, and on the dynamic changes, secular or cyclical in character, which affect the operations of an economy. This purpose is, of course, related to the first, but it does not follow that what is an adequate index number for the first purpose will be satisfactory for the second. Current index numbers of prices do not, in fact, yield the information we should like to have concerning economic changes. The tremendously important movements of the last eighteen months, for example, are far from clear to us. One reason for this is found in the deficiencies of the price record. Three main requirements of price index numbers intended for use in the study of economic changes may be set down. They should be based upon comprehensive compilations of price series, representative of transactions in all important commodity markets. They should be so constructed that changes occurring in different parts of the price system might be followed. They should be comparable with index numbers measuring changes in non-price elements of the economy. Each of these points requires brief elaboration. The character of the individual series which should be used in constructing index numbers of prices has been ably discussed in Professor Copeland's paper. His emphasis is on the intensive aspects of the problem before us-on the improvement of the quality of the individual bricks to be used in our building. I should endorse most heartily all that he has said, urging only that the price-compiling authorities give us the price series in unadulterated form, as well as the derived composite prices. For such pure series, relating, over a period of time, to precisely the same commodity, to the same market, and to transactions in which buyer and seller are clearly defined, are essential raw material for all work in the field of prices. But more bricks are needed, as well as better bricks. If an economic intelligence service is to function properly we sorely need more information concerning the numerous price transactions through which