¹ Niles' Weekly Register occasionally commented on the importance of skilled engineers. See Vol. XXIX (Oct. 22, 1825), pp. 101–02.Google Scholar That acute observer of ante bellum America, Michel Chevalier, was also aware of the importance of skilled engineers. See the quotation in Hill, Forest G., Roads, Rails and Waterways (Norman: University of Oklahoma Press, 1957), p. 4,Google Scholar which is also evidence of Hill's concern with the subject. For a similar view by another modern scholar see Calhoun, Daniel H., The American Civil Engineer (Cambridge: The Technology Press, 1960), p. viiiGoogle Scholar.

² The majority of primary printed materials (largely annual reports of antebellum canals and railroads) that were useful are located at Harvard's Baker Library, the Bureau of Railway Economics Library, or the Library of Congress. For a considerably, more complete discussion of the sources from which the basic data comprising Table 1 were gleaned, see the author's doctoral dissertation, “Rates of Return on Investment in Technical Education in the Ante-Bellum American Economy” (Unpublished, University of Texas at Austin, 1969).

³ Several additional comments on the organization of Table 1 are in order. The table includes a few interpolations and extrapolations, but these amounted to less than five percent of the total number of observations. No such “guesses” were made for more than five consecutive years, and none were carried over peaks or troughs in the business cycle.

Since antebellum civil engineers occasionally received some of their income in kind (e.g., a free pass along the railroad for which they worked), such payments were added to the engineers' cash salary whenever they could reasonably be valued in money.

Finally, it should be noted that the figures in column 3 are almost entirely yearly extrapolations of daily wage rates received by junior engineers, based on a 313-day work year. Should the reader feel that 313 days is an overstatement of the actual work year, he may, of course, divide the figures in column 3 by 313 and multiply by his chosen employment period.

⁴ The formula is . 100.

⁵ Found in Calhoun, The American Civil Engineer, p. 167.

⁶ Ibid., p. 169.

⁷ Aldrich, “Rates of Return …,” ch. iv. Our salary data for the 1830's are drawn from projects located in the following states: Alabama, Delaware, Florida, Georgia, Illinois, Indiana, Louisiana, Maryland, Massachusetts, Michigan, New York, North Carolina, Ohio, Pennsylvania, Rhode Island, South Carolina, and Virginia.

⁸ Stanley Lebergott has used the considerably smaller decline in coefficients of variation of farm-labor wage rates from 1818 through 1830 as “one empirical indication of the rapidity with which an integrated national labor market tended to develop.” Manpower in Economic Growth (New York: McGraw-Hill Book Co., 1964), pp. 134–35Google Scholar.

⁹ Fishlow, Albert, American Railroads and the Transformation of the Ante-Bellum Economy (Cambridge: Harvard University Press, 1965)Google Scholar, Table 54, p. 399, and Cranmer, Jerome, “Canal Investment 1815–1860,” in Trends in the American Economy in the Nineteenth Century, National Bureau of Economic Research, Studies in Income and Wealth, XXIV (Princeton: Princeton University Press, 1960)Google Scholar, Table 3, 555–56.

¹⁰ Fishlow, American Railroads …, p. 399, and Cranmer, “Canal Investment 1815–1860,” pp. 555–56.

¹¹ Fishlow, American Railroads …, p. 399.

¹² The best single discussion of the technical problems associated with the emergence of the railroad may be found in Calhoun, The American Civil Engineer, ch. iii.

¹³ By administrative problems we mean those difficulties associated with the organization of railroad construction, such as communication, control and delegation of authority within a large engineer organization, plus the maintenance of sufficient labor and supplies to avoid bottlenecks. For example, see Gates, Paul W., The Illinois Central Railroad and its Colonization Work (Cambridge: Harvard University Press, 1934), pp. 94–98CrossRefGoogle Scholar, for the Illinois Central's difficulties in maintaining an adequate labor force.

¹⁴ A. D. Chandler has advanced a thesis similar to ours—that the railroads' expansion in size presented them with unique administrative problems in their day-today operations (“The Railroads: Pioneers in Modern Corporate Management,” Business History Review, XXXIX (Spring 1965), 16–40)Google Scholar. It is perhaps no coincidence that the major innovators in corporate management were former engineers (Chandler, p. 40).

¹⁵ The Illinois Central and the Mobile and Ohio Railroads (two of the largest projects of the 1850's, with construction costs of $24 million and $17 million respectively) each offered its chief engineer the immense salary of $10,000 per annum. See Minutes of Directors, entry for February 5, 1859, Illinois Central Railroad Archives, and Ninth Annual Report of the President and Directors of the Mobile and Ohio Railroad (Mobile, 1858), p. 15.

¹⁶ Admittedly, both average cost per mile and total construction cost are proxy variables and may well reflect influences other than those we have suggested—such as corruption.

¹⁷ If scale economies associated with construction were significant, the statement in the text would not hold. However, correlation between project size and average cost per mile is low. See fn. 21 below.

¹⁸ Because of insufficient evidence on the time shape of construction costs, none of the variables are deflated. However, since our concern is with the change in the relative explanatory power of the independent variables (and with their relative variation) between the two periods, the fact that all variables may exhibit time trends is of no concern.

¹⁹ United States Bureau of the Census, Preliminary Report on the Eighth Census (Washington: Government Printing Office, 1862), pp. 214–29Google Scholar.

²⁰ Equations of the form LogX₁ = Log A + B₂LogX₂ + B₃Log X₃ were also fitted to the data, but yielded R²'s no higher than the linear form and so the results are not presented.

²¹ The sample sizes are, respectively, 22, and 35. Figures in parentheses are standard errors. The F ratios are, respectively, 6.99 and 8.23; hence both regressions are significant at the one percent level. The Durbin-Watson statistics are 2.3 and 2.7, leading us to reject the hypothesis of any serial correlation at the five percent level.

²² It seems unlikely that the extreme variation in the significance of the independent variables between the two decades is attributable to multicollinearity, as the squared coefficient of partial correlation between the two independent variables is .30 for the 1830's and only .18 for the decade of the 1850's.

²³ Lebergott, Manpower in Economic Growth, p. 134.