DETWEEN I947 and I957 the price of )manufactured goods, other than foods, rose 33 per cent. The price of constituent crude materials rose 20 per cent. (Their spot market prices fell io per cent.) Production workers' payrolls per unit of output rose I7.5 per cent. The growing gap between these costs and selling prices was partly filled by certain fringe benefits of production workers, by depreciation accruals, and by taxes. A further substantial portion of the margin over direct costs is attributable to the work performed by non-production workers in manufacturing whose number rose 55 per cent while the number of production workers increased only 2 per cent. Payments to non-production workers were a major factor in causing the compensation of all workers in manufacturing to increase almost twice as fast as payrolls for production workers alone. The rise in prices over the decade has been blamed on a demand-pull or a cost-push, or both. Rising costs of the articles currently purchased push against operating margins and presently against prices. In this sense, union demands for higher wages or suppliers' demands for a higher price of materials constitute a direct rise in costs that tends to push prices up. But the figures just cited indicate that costs have risen in other ways also. Manufacturers have added hundreds of thousands of workers engaged in marketing, advertising, administration, research. They have bought vast quantities of new machines for which depreciation must be charged as a current cost. Had they not changed their input in these and other ways, production labor per unit of output would have risen more than it did. Indeed, if each decision to alter. the character and quantities of input had been a perfect textbook example of economic behavior, the total cost-push would presumably have been minimized by the shifts in inputs. But insofar as this has not actually been the case -insofar as the shifts have been inefficient with respect to minimizing costs or maximizing contemporary consumer satisfaction the rapid rise in the cost of non-production labor and equipment could bear directly on the inflationary process. In the closing section of this paper I shall argue that there are strong reasons to suspect that this has actually been the case. If so, the matter has not been accorded the attention it deserves. For the figures indicate a potential quantitative importance of substantial scope. It is this potential quantitative importancethe extent of the shifts in cost structure-which is examined in the following pages. First we review recent trends in the unit cost of production labor and next in prices of crude materials. The data are examined over the postwar decade ending with the business peak in I957.1 The third section shows how trends in these major direct costs are associated with shifts of very considerable proportions in the rest of the cost structure of manufacturing. The final section speculates on the probable bearing of these shifts on the inflationary process, and points to questions that require answers.