If the displacement of labor cannot be seen in the figures for the absolute size of the working population occupied in the making of goods, it can be seen in the measures of its relative size. If we convert the tabulation to the form of percentages of total nonagricultural employment for each census year, the trend emerges with some clarity (see percentage column of table, below).
In view of the untrustworthy nature of nineteenth-century statistics, it would perhaps be wrong to draw from them any other conclusion than that the percentage of those gainfully occupied who were to be found in these goodsproducing industries fluctuated in a fairly narrow range, between 45 and 50 percent of nonagricultural employment. And this situation, strikingly enough, continued until 1920; thereafter, the percentage moved consistently downward to the 33 percent figure of the 1970 census. The balance between the growth of production on the one side and the growth of productivity on the other held for a century and was, it would appear, finally broken in the decade of the 1920s, when employment in these manufacturing, extractive, and construction industries began for the first time to fall off as a proportion of all nonfarm employment.
But the more striking tendency is the marked change in occupational composition within these industries. As has already been pointed out, the separation of conceptualization from execution—the removal of all possible work from the shop floor, the point of execution, to the officeand the further necessity of maintaining a shadow replica of the entire process of production in paper form, brings into being large technical and office staffs. Statistics from all the principal capitalist countries indicate that there has been a rapid rise, starting before the turn of the century, in the proportion of those not employed directly in production. Characteristically, there were in manufacturing industries around the start of this century somewhere between five and ten nonproduction employees for every hundred employed in production, and by the post-World War II period this had risen to more than twenty per hundred. The figures given for United States manufacturing industries are as follows:3 \*
\* This table is constructed in three parts because there is no single continuous series covering the entire 150 years since the first occupational census. Nor is it possible to splice the three series together, since they were each constructed on somnewhat different principles. The first two sections of the table (by, respectively, P. K. Whelpton in 1926 and Alba Edwards in 1943) are attempts at reconstruction of census data; the final portion is from Bureau of Labor Statistics figures as gathered in their monthly payroll surveys. Despite the lack of a continuous series constructed on a single set of principles for the entire period, and despite the unreliability of early occupational statistics, the trends are clear, both as to numbers and percentages.²
Labor and Monopoly Capital
Effects of Management and Technology
165
\* The United States Census data for the ratio of nonproduction to production workers in manufacturing shows “a secular trend upward, beginning in 1899, with some variation in slope but unmistakable in direction. This means that if one is looking for causal forces, the place to look is the whole twentieth century." This is the conclusion of George E. Delehanty, who has done one of the most thorough investigations of this subject. Delehanty points out that the series maintained by the Bureau ofLabor Statistics on a different basis gives a different picture, showing a constancy in the ratio up to 1952, and then an upward trend. After reviewing the evidence for both series, Delehanty is forced to conclude that it is impossible to choose between them on the basis of