Much of our understanding of structural change with industrialization is based on empirical studies that describe patterns in the relationship between consumption, intermediate and capital goods [1; 2, for instance]. Thus, classification of an industry’s output, prices, imports or exports by use is of primary importance. Even if the data available describe individual products, there is some ambiguity in use classification—safety razor blades (metal products) are clearly con¬sumption goods but sewing machines (nonelectric machinery) are consumption goods if owned by housewives and capital goods if owned by tailors. Far more serious problems, however, arise when large industry sectors, like whole four digit industries, must be classified by use. If there is no basis for dividing any industry’s output (or prices or imports or exports) among its alternative uses, then the entire industry must be classified as a single lump in one of the three use categories on the basis of some judgment— even if heroic—or else such aggrega¬tion by use must be abandoned. But ad hoc classifications that lump entire four-digit industries under one or another heading are not appropriate if there exists a reasonable basis for dividing each industry’s output among uses. Then, instead of assigning all metal products to capital goods (as did Chenery) it is possible to assign some portion to consumption goods (representing razor blades, et al), some to intermediate goods (rivets) and some to capital goods (spades).