The authors are affiliated with the University of Illinois, Indiana University, and the University of Kansas, respectively. They wish to thank Hugh Aitken, Stanley Engerman, Stefano Fenoaltea, Herman Freudenberger Louis C. Hunter, Larry Neal, Richard Sylla, Tom Ulen, Paul Uselding, and the anonymous referee of this Journal for their comments on earlier versions of this paper. Funding was provided by the National Science Foundation to Fred Bateman under grant SOC-75-20034, to Tom Weiss under grant SOC-75-18917, and by the Illinois Investors in a Business Education, the College of Commerce, and the Research Board of the University of Illinois to Jeremy Atack.

¹ Ashton, Thomas S., The Industrial Revolution, 1760–1830 (Oxford, 1948), p. 58Google Scholar; Musson, A. E., “Industrial Motive Power in the United Kingdom, 1800–70,” Economic History Review, 2nd Ser., 29 (08 1976), 415–39Google Scholar.

² Robert Thurston, for example, devoted 229 pages of his history of the steam engine to transportation applications and only 40 pages to the stationary engine. See Thurston, Robert H., A History ofthe Growth of the Steam Engine (Ithaca, 1939)Google Scholar.

³ The lone American engine, built by Joseph Brown, powered the Hope Furnace in Rhode Island. The accounts, however, leave in doubt whether this engine was the prime mover at the furnace. More likely it pumped water to drive a waterwheel that in turn operated the bellows. See Pur-sell, Carroll W. Jr, Early Stationary Steam Engines in America (Washington, D.C., 1969), p. 9Google Scholar. Lord, John, Capital and Steam Power 1750–1800 (London, 1923), p. 151Google Scholar, gives statistics on steam power usage in Britain.

⁴ See U.S. Census Office, Twelfth Census of the United States, 1900: Manufactures (Washington, D.C., 1902), vol. 7, pp. cccxv–cccxxxviiiGoogle Scholar.

⁵ Only one study, by Louis C. Hunter, embraces as wide a time span as the one we present. A copy of this book was received as this paper was being typed. A quick reading suggests many points of agreement between Hunter and us despite our differences in approach. See Hunter, Louis C., A History of Industrial Power in the United States, 1780–1930: Waterpower in the Century of the Steam gine, vol. 1 (Charlottesville, VA, 1980), especially pp. 481–535Google Scholar. Von Tunzelman in his chapter on the United States primarily reexamines the evidence presented by Peter Temin which is based on the Woodbury Report (U.S. Congress, Report on the Steam Engine in the United Stales, House Doc. 21, 25th Cong., 3d Sess., 1838). See Tunzelman, G. Nicholas von, Steam Power and British Industrialization to 1860 (Oxford, 1978)Google Scholar, and Temin, Peter, “Steam and Waterpower in the Early Nineteenth Century,” this Journal, 26 (06 1966), 187–205Google Scholar. The Woodbury Report and the published census data from 1870 form the basis of Fenichel's study of the growth of manufacturing power usage. See Fenichel, Allen H., “The Growth and Diffusion of Power in Manufacturing, 1839–1919,” in Output, Employment, and Productivity in the United States after 1800, NBER Studies in Income and Wealth, vol. 30 (Princeton, 1966), pp. 443–78Google Scholar.

⁶ See, for example, DuBoff, Richard B., “The Introduction of Electric Power in American Manufacturing,” Economic History Review, 2nd Ser., 20 (12 1967), 509–18Google Scholar.

⁷ With the exception of the Woodbury Report made in 1838, no statistics on steam power were published until the Ninth Census (see U.S. Census Bureau. Statistics of Wealth and Industry of the United States, vol. 3 (Washington, D.C., 1872), pp. 392–93)Google Scholar. Water power statistics are only available beginning with the Ninth Census.

⁸ Computed from the manuscripts of the Third Census. See National Archives, Records of the 1820 Census of Manufactures, Microcopy 279 (Washington, D.C., 1964)Google Scholar. See also, U.S. Census Bureau, Statistics of Wealth and Industry, pp. 392–93Google Scholar.

⁹ U.S. Census Office, Twelfth Census, vol. 7, pp. cccxv–cccxxxviii, especially p. cccxxxvGoogle Scholar.

¹⁰ Based upon manuscript census data for 1860, we estimate that in 1860 there were approximately 46,250 waterwheels in operation in the United States producing about 806,500 h.p., whereas the al most 25,600 steam engines generated about 621,000 h.p. By 1870, the Ninth Census estimated steam horsepower at 1,215,711 and hydraulic horsepower at 1,130,431. See U.S. Census Bureau, Statistics of Wealth and Industry, pp. 392–93Google Scholar.

¹¹ These published statistics have been collected and analyzed by Fenichel, “The Growth and Diffusion of Power.”

¹² The 1820 Census, for example, which included a question on the kind of machinery being used, made no returns for Louisiana, and the returns for almost every other state omitted at least one county. (See American State Papers, Series 3, vol. 4, Finance, Doc. 675, pp. 298–99, for a list of the omissions). The McLane Report (U.S. Congress, Documents Relative to the Manufactures in the United States, House Exec. Doc. 308, 22d Cong., 1st Sess., 1833) omits the southern states and those west of the Ohio, and the Woodbury Report has no, or only partial, returns for Alabama, Arkansas, Illinois, Indiana, Iowa, Kentucky, Mississippi, Missouri, Tennessee, Vermont, and Wisconsin.

¹³ Temin mentions the 1820 Census as noting that there were “about a dozen” steam engines then in use. This estimate is too low and appears to be based upon the published census summary. See American State Papers, Series 3, vol. 4, Finance, Docs. 662 and 675 (Washington, D.C., 1858), andGoogle Scholar, Temin, “Steam and Waterpower,” 189Google Scholar.

¹⁴ See McLane Report, vol. 2, pp. 235–650, especially Doc. 14. Also see the Woodbury Report, pp. 191–95.

¹⁵ See Hutton, F. R., “First Stationary Steam Engines in America,” Transactions of the American Society of Mechanical Engineers, 15 (1894), 982-97Google Scholar; also “The Oldest Steam Engine in the United States,” Engineering News, 30 (11 9, 1893), 370Google Scholar.

¹⁶ “Manufacturing Establishments at Cincinnati,” American Railroad Journal, 3 (12 27, 1834), 806–07Google Scholar.

¹⁷ See, for example, Griliches, Zvi, “Hybrid Corn: An Exploration in the Economics of Technological Change,” Econometrka, 25 (10 1957), 501–22Google Scholar; Mansfield, Edwin, “Technical Change and the Rate of Imitation,” Econometrka, 29 (10 1961), 741–66Google Scholar; and , Mansfield, “Intra-firm Rates of Diffusion of an Innovation,” Review of Economics and Statistics, 45 (11 1963), 348–59Google Scholar.

¹⁸ See , Griliches, “Hybrid Corn,” 504Google Scholar. The fit of the data points to the logistic curve was close, as shown by the R2 statistics in Table 2.

¹⁹ The estimate for the Middle Atlantic region used data for 1820, 1833, and 1850–90, whereas the estimates for the South and Midwest used 1820 and 1850–90 data. Those for New England and the Mountain and Pacific regions were based on 1833 and 1850–90 data.

²⁰ See , Griliches, “Hybrid Corn,” 505–06Google Scholar.

²¹ See Redlich, Fritz, “The Philadelphia Water Works in Relation to the Industrial Revolution in the United States,” Pennsylvania Magazine of History and Biography, 69 (07 1945), 243–56Google Scholar. The publication of the Young Steam Engineer's Guide was a part of the reason why the courts were unwilling to enforce Evans's patent protection. See Evans, Oliver, Patent Right Oppression Exposed; or Knavery Detected (Philadelphia, 1813)Google Scholar.

²² The Woodbury Report data indicate that each city had 44 engine makers.

²³ See Haites, Erik, Mak, James, and Walton, Gary M., Western River Transportation (Baltimore, 1975), Table B-lGoogle Scholar, and Hunter, Louis C., Steamboats on the Western Rivers (Cambridge, MA, 1949), p. 123Google Scholar.

²⁴ , Haites, , Mak, and , Walton, Western River Transportation, pp. 130–31Google Scholar. Not all these steamboats, however, used new engines. See , Hunter, Steamboats, p. 112Google Scholar.

²⁵ , Griliches, “Hybrid Corn,” 505Google Scholar.

²⁶ For example, the distance of major population centers in each region from the Pittsburgh-Philadelphia area, which was the center of the nation's steam engine production, correlates quite highly with the dates of innovation from Table 2 and in the anticipated direction (r =.834). Similarly, these dates show a strong inverse correlation (r = -.801) with 1840 coal production. In this latter instance, however, the direction of causation is not clear.

²⁷ Given the short average life-span of these early steam engines, this estimate is remarkably high. See the Woodbury Report, pp. 156–67.

²⁸ Woodbury Report, pp. 41–7, 330–31.

²⁹ These early estimates were made by the Army in their evaluations of suitable sites for a national armory on the western rivers. See, for example, U.S. Congress, Message from the President of the United States Transmitting a Report of the Commissioner … to Establish a National Armory on Western Waters, House Doc. 55, 18th Cong., 2d Sess. (Washington, D.C., 1825)Google Scholar.

³⁰ These contemporary estimates were mostly made for a narrow range of New England, Midwest, or Middle Atlantic sites (see Table 3). There is no evidence on water power costs for the 1870s. Moreover, estimates for other decades frequently are repetitions of earlier ones.

³¹ James, for example, acted as consultant in the establishment of the steam-powered Cannelton Mills in Indiana in 1848. See Wilson, Harold S., “The Indiana Cotton Mills: An Experiment in North-South Cooperation,” Indiana History Bulletin, 42 (05 1965), 75–83Google Scholar.

³² See , Justitia, Strictures on Montgomery on the Cotton Manufacturers of Great Britain and America (Newburyport, 1841)Google Scholar. The pamphlet Practical Hints on the Comparative Cost and Productiveness of the Culture of Cotton and the Cost and Productiveness of Its Manufacture, by James, Charles T. (Providence, 1849)Google Scholar, was addressed to Smith. Further articles in DeBow's Review for 1848, 1849, 1850, an d 1853 all bear the stamp of James's argument and Hamilton Smith's approval. See DeBow's Review, 5 (532–35), 7 (128–34), 8 (7–19, 550–55) and DeBow, James D. B., The Industrial Resources of the Southern and Western States (New Orleans, 1853)Google Scholar. The pro-water group found their journal outlet in Hunt's Merchants' Magazine, and James also published responses to articles appearing there. See James, Charles T., Letters on the Culture and Manufacture of Cotton Addressed to Freeman Hunt, Esq. (New York, 1850)Google Scholar.

³³ The Pacific Mills, capitalized at $1.2 million, was the largest manufacturing venture of the time. The water power owned by the Essex Company amounted to 11,000 gross h.p. in 1880 and sold for $14.08 per horsepower per year. See U.S. Department of the Interior, Census Office, Reports on the Water-Power of the United States at the Tenth Census, vol. 16 (Washington, D.C., 1885), pp. 25–30Google Scholar.

³⁴ In fairness it should also be mentioned that although Manning was superintendent of the Amos-keag Manufacturing Company, which at the time used steam power, Amoskeag also owned the Merrimack water rights at Manchester, N.H. It will be noted from Table 3 that his estimates of steam and water power costs are quite similar although he ultimately recommends in favor of steam.

³⁵ We assume that steam engines required the services of both a fireman and an engineer. Hence the “2.0” in the equation for C(S).

³⁶ The properties of this specific simulation model are described in Atack, Jeremy, “Fact in Fiction? The Relative Costs of Steam and Water Power: A Simulation Approach,” Explorations in Economic History, 16 (10 1979), 409-37Google Scholar. Computer simulation as a method is described in Naylor, Thomas, et al., Computer Simulation Techniques (New York, 1966)Google Scholar. Also see Schaefer, Donald and Weiss, Thomas, “The Use of Simulation Techniques in Historical Analysis: Railroads versus Canals,” this Journal, 31 (12 1971), 854–84Google Scholar, for another example of the method in historical analysis.

³⁷ As high pressure engines exhausted steam under high pressure, this steam could be used for heating with only a very slight increase in fuel consumption (which was due to the back pressure generated by piping the steam into heating lines). Incremental fuel consumption for heating rose with falling steam pressure, for example, in compound engines. See Emery, “Cost of Steam,” Schedule A.

³⁸ Available evidence suggests that water-powered plants had to shut down for one or two months a year because of ice, floods, or drought which made it dangerous or impossible to operate a waterwheel. Turbines were less affected by ice and floods, but could still be damaged by chunks of ice.

³⁹ See the sources to Table 3 for complete references.

⁴⁰ See Eavenson, Howard N., The First Century and a Quarter of American Coal Industry (Pittsburgh, 1942), especially pp. 385–86Google Scholar. For the 1830s, Eavenson gives two independent price quotes for Pittsburgh, both of 4 cents a bushel.

⁴¹ See Table 3 for complete references. McElroy was employed as a consultant engineer on water power projects; he gave no detailed breakdown of how he arrived at the steam power cost estimate.

⁴² This in particular should be kept:in mind when studying Table 4.

⁴³ The original grantees at Lawrence paid a lump sum of $10,000 and an annual payment of $300 (being interest at 6 percent on the balance remaining on the $15,000 valuation of the water rights). Current cost of using water power at Lawrence to these original grantees was therefore only $3.52 per theoretical horsepower or $10.56 when all costs (historic sunk costs and variable costs) are included. The Essex Company sold new rights at $1,200 per mill-power or $14.08 per theoretical horsepower. For the original grantees, no plausible levels of steam power cost elements would have induced a switch to steam.

⁴⁴ Estimates of breast wheel efficiency are given in the Cyclopaedia of Arts and Sciences (London, 1861), pp. 765–70Google Scholar; by calculations made using data in U.S. Congress, Message… to Establish a National Armory, Doc. 120, p. 49 for the Springfield Armory, and by experiments reported in the Franklin Institute: Morris, Ellwood, “Remarks on Reaction Water Wheels Used in the United States and on the Turbine of M. Fourneynon,” Journal of the Franklin Institute, 34 (10/11 1842), 217–27, 289–304Google Scholar. Estimates of turbine efficiency are given by Emerson, James, “Water-Wheel Tests at Lowell and Other Places,” Journal of the Franklin Institute, 93 (03 1872), 174–80Google Scholar. Also see Francis, James B., “Experiments on the Humphrey Turbine Water-wheel at the Tremont and Suffolk Mills in Lowell, Mass.,” Transactions of the American Society of Civil Engineers, 13 (1884), 295–302Google Scholar, and Manning, Main, Webber, and McElroy (see Table 3). See also the Appendix to this article.

⁴⁵ The relationship is denned by:

where 62.5 is the weight in pounds of one cubic foot of water. See Atack, “Fact in Fiction?”, 427–30, for a sensitivity analysis. Comprehensive estimates on water-right costs are available only for 1880.

⁴⁶ See, for example, the references listed for the 1830s and early 1840s in Table 3. Also see Unwin, W. Caw-thorne, “The Cost of Steam Power,” Cassier's Magazine, 5 (02 1894), 352–54Google Scholar, and see , Atack, “Fact in Fiction?” 428Google Scholar, for the effect on cost estimates.

⁴⁷ Calculated from the Bateman-Weiss random samples from the 1850 manuscript Census of Manufactures. Funding for the collection of these samples was provided by the National Science Foundation.

⁴⁸ U.S. Census Office, Twelfth Census, vol. 7, p. cccxxxvGoogle Scholar.

⁴⁹ All sample statistics were calculated from the cost probability distributions derived from the simulation model. Thus the mean cost, E(X), was calculated using the formula: E(X) = ∑_iX_i P(X_i) where P(X_i) is the probability of cost level X_i.

⁵⁰ The mean of relative mean costs of steam to water power between 1830 and 1899 were 1.38, 1.22, 0.91, 0.68 and 0.91 for New England, Middle Atlantic, Southern, Midwestern, and Mountain and Pacific states, respectively. They correlate with the product of the ceiling level and rates of adoption (that is, rates of adoption, adjusted to take into account that differing fractions of plants in each region will adopt steam) at -0.86, indicating that the cheaper steam power was relative to water, the faster steam power was adopted. This procedure was suggested by , Griliches, “Hybrid Corn,” 517Google Scholar.

⁵¹ A cord of hardwood contained about four fifths the Btu energy of a ton of coal, but weighed considerably more and was more bulky, making transportation much more expensive.

⁵² We estimate the steam horsepower in American factories in 1850 at approximately 181,000 h.p. Assuming a 12-hour work day, 309 days a year, and a range of coal consumption rates of 4.1–7.3 lbs./ h.p./hr. with a mean of 5.5 lbs./h.p./hr., coal consumption by steam engines would be between 1.4 million tons and 2.5 million tons a year, compared with a U.S. coal production of 8.4 million tons. See , Eavenson, American Coal Industry, p. 434Google Scholar.

⁵³ The 1880 census ( U.S. Department of the Interior, Report on the Manufactures of the United States, vol. 2 (Washington, D.C., 1883), p. 501Google Scholar) estimated steam horsepower at 2,185,000 h.p. If they consumed coal at a rate of 2–3 lb./h.p./hr., then for the standard work year of 309 12-hour days, total consumption would be 8.1–12.2 million tons compared with Eavenson's production estimate of 79.4 million tons. See , Eavenson, American Coal Industry, p. 434Google Scholar.

⁵⁴ A variety of examples is given in U.S. Department of the Interior, Water Power, vols. 16Google Scholar and 17. Other contracts, for example, those for water rights along the Brandywine are also similar. See Accession 169, Brandywine Creek, MSS in Eleutherian Mills Historical Library. See also Papers of the Brandywine Mill Seat 6, Longwood MSS, Eleutherian Mills Historical Library.

⁵⁵ U.S. Department of Interior, Water Power, vol. 16, p. 26Google Scholar.

⁵⁶ Ibid.

⁵⁷ Ibid., pp. 27–28.

⁵⁸ Estimated from graphs 4 and 5 in , Atack, “Fact in Fiction?”, 431 and 434Google Scholar.

⁵⁹ Coal and Coal Trade Journal, vol. 1 (New York, 1889-), published weeklyGoogle Scholar.

⁶⁰ For original grantees, coal would have had to be available for about $3.00/ton in order to induce abandonment of the water rights. See , Atack, “Fact in Fiction?”, 433Google Scholar.

⁶¹ U.S. Department of the Interior, Water Power, vol. 16, p. 27Google Scholar.

⁶² See graphs 4 and 5 in , Atack, “Fact in Fiction?”, 431 and 434Google Scholar.

⁶³ U.S. Department of the Interior, Water Power, vol. 16, pp. xxxiii–xxxvGoogle Scholar. See also Swain, George F., “Statistics of Water Power Employed in Manufacturing in the United States,” Journal of the American Statistical Association, 1 (03 1888), 44Google Scholar.

⁶⁴ U.S. Department of the Interior, Water Power, vol. 16, pp. xxx–xxxiiGoogle Scholar. These data refer only to what the Census described as “large” developed water powers, those of about 2000 h.p. or more. The Census was careful, however, to point out that the list was incomplete, especially for the Mississippi and its tributaries. For the South, the Census also gave the following caveat: “on account of the comparative absence of abrupt falls on the larger streams in this region, many of the powers so enumerated would admit of development only at a cost which would in many cases be perhaps almost prohibitory.” Ibid., p. xxx.

⁶⁵ See U.S. Census Office, Twelfth Census, vol. 7, p. cccxxxvGoogle Scholar.

⁶⁶ See U.S. Department of the Interior, Water Power, vol. 17, pp. 41–49Google Scholar for a full description of the water power at Appleton and its development. The low prices of water rights at Appleton did not reflect the full costs of developing them since the canals were built by the Corps of Engineers.

⁶⁷ Most important of these advantages were freedom of location and the multiplication of prime movers within a plant. See DuBoff, “Electric Power.”

⁶⁸ See, for example, Emery, Charles E., “The Relations of Electricity to Steam and Water-Power,” Journal of the Franklin Institute, 142 (09 1896), 165–86Google Scholar.

⁶⁹ From the results given in Table 2 it can be seen that the date when adoptions had reached 90 percent of the ceiling level may be calculated by [(2.197 - a)/b] + 1820. This yields estimates of August 1906, February 1880, January 1901, December 1879, and February 1881 for the New England, Middle Atlantic, Southern, Midwestern, and Mountain and Pacific states, respectively.

⁷⁰ U.S. Department of the Interior, Census Office, Report on Manufacturing Industries of the United States at the Eleventh Census: 1890, vol. 6 (Washington, D.C., 1895), pp. 758–59Google Scholar.

⁷¹ U.S. Census Office, Twelfth Census, vol. 7, p. cccxxxvGoogle Scholar.

⁷² In 1890, 62 percent of New England's population was classified as urban (in towns of 2,500 or more) compared with 58 percent in the Middle Atlantic, 37 percent in the West, 33 percent in the Midwest, and only 13 percent in the South. See U.S. Bureau of the Census, Historical Statistics of the United States, Colonial Times to 1970 (Washington, D.C., 1975)Google Scholar, series A202–3.

⁷³ See Table 4 for some evidence on this.

⁷⁴ U.S. Department of the Interior, Water Power, vol. 16, p. 789Google Scholar.

⁷⁵ Ibid., p. 847.

⁷⁶ That is, there was no more water available, so that power gains had to come through raising the efficiency of the prime mover.

⁷⁷ In the 1830s and 1840s the typical plant used less than 20 h.p. but by the 1890s it used more than triple that amount. Indeed, by 1895 the average power of steam engines in use in the Massachusetts cotton textile industry was 316 h.p. whereas the average waterwheel power was 180 h.p. See Census of the Commonwealth of Massachusetts: 1895, vol. 5 (Boston, 1898), p. 579.Google Scholar In 1875 these figures were 156 hp. and 113 hp., respectively ( Census of Massachusetts: 1875, vol. 2 (Boston, 1877), pp. 337–38)Google Scholar. See also U.S. Department of the Interior, Water Power, vol. 16, p. 27 which, for example, at Lawrence shows the grantees of the Essex Company also using steam engines. Pacific Mills in particular had grants of 36 mill-powers.(3,069 hp.) but also had 2,400 hp. of steam power in 1882Google Scholar.

⁷⁸ See the sensitivity tests in , Atack, “Fact in Fiction?”, 427–30Google Scholar.

⁷⁹ See, for example, Davis, Lance E., “The Investment Market 1870–1914: The Evolution of a National Market,” this Journal, 25 (09 1965), 355–99, especially 360–63, 394–99Google Scholar.

⁸⁰ Homer, Sidney, A History of Interest Rates (New Brunswick, NJ, 1963), pp. 323–24Google Scholar.

⁸¹ Homer, , Interest Rates, p. 324Google Scholar. A similar estimate is used for the appraisal of Royce Areas in Illinois and Wisconsin: Walter R. Kuehnle and Company, Appraisal of Royce Areas 147 and 148 in Illinois and Wisconsin - 1829 (Chicago, 1958)Google Scholar, prepared for the U.S. Department of Justice for Case Nos. 13-L, 15-K, 18–1, 29-J, 40-J and 217 before the Indian Claims Commission. Although most midwestern states had usury laws limiting interest well below these levels, Bogue has described how these restrictions were circumvented in the mortgage market. See Allan Bogue, From Prairie to Corn Belt: Farming on the Illinois and Iowa Prairies in the Nineteenth Century (Chicago, 1963), especially ch. 7.

⁸² Macaulay, Frederick R., Some Theoretical Problems Suggested by The Movement of Interest Rates, Bond Yields and Stock Prices in the United States Since 1856 (New York, 1938), pp. A246 and A248Google Scholar. Also, , Homer, Interest Rates, p. 318Google Scholar.

⁸³ Bigelow, Erastus B., The Tariff Question (Boston, 1862), pp. 204–5Google Scholar, and Martin, Joseph G., Seventy-three Years History of the Boston Stock Market (Boston, 1871), pp. 37–40Google Scholar. These and other sources for short-term rates are discussed in , Macaulay, Movement of Interest Rates, pp. A341–51Google Scholar.

⁸⁴ Hunt, F., ed., Merchant's Magazine and Commercial Review (New York, monthly, 1839)Google Scholar contains irregular mention in the “Monthly Commercial Chronicle” from 1841 onward of rates on domestic bills in New York and statistics on prices and yields on stocks, both state and private, sold in New York.

⁸⁵ Macaulay, Movement of Interest Rates, Table 10 and, Homer, Interest Rates, Table 44 and p. 327.

⁸⁶ , Homer, Interest Rates, pp. 323–6Google Scholar.

⁸⁷ Financial Review, (New York, 1860–1921)Google Scholar, 1874 onward (annual). The National Union List of Serials lists no holdings earlier than 1873.

⁸⁸ , Davis, “The Investment Market,” especially 360–63, 394–99Google Scholar; , Homer, Interest Rates, pp. 323–26Google Scholar; Emery, “Costs of Steam Power,” Schedule A; Manning, “Comparative Cost,” 504; Main, “Cost of Steam,” 114; , Webber, “Water-Power,” 51Google Scholar; and, , McElroy, “Water Power,” 68Google Scholar.

⁸⁹ Lebergott, Stanley, Manpower in Economic Growth: The American Record Since 1800 (New York, 1964), Tables 6–2 and A-25Google Scholar.

⁹⁰ See appendixes to House Doc. 55, 18th Cong., 2d Sess.; House Exec. Doc. 120, 25th Cong., 2d Sess.; and, House Doc. 133, 27th Cong., 3d Sess.

⁹¹ U.S. Bureau of the Census, Historical Statistics, series D718, D734, and D847–48.

⁹² U.S. Department of the Interior, Census Office, Report on the Statistics of Wages in the Manufacturing Industries With Supplementary Reports on Average Retail Prices of Necessities of Life and on Trade Societies, and Strikes and Lockouts, also known as the Weeks Report (Washington, D.C., 1886), vol. 10, especially pp. 503–16, 544–63Google Scholar. U.S. Congress, Report on Wholesale Prices, On Wages and on Transportation, Senate Report of Committee 1394 (also known as the Aldrich Report), 52nd Cong., 2d Sess. (1893), pp. 293–1560.

⁹³ Emery, “Costs of Steam Power,” Schedule A; , Manning, “Comparative Costs,” 503Google Scholar; , Pray, “Steam Horsepower,” 14–19Google Scholar; , Main, “Cost of Steam,” 113Google Scholar; and , McElroy, “Water Power,” 68Google Scholar. , Justitia, Strictures on Montgomery, pp. 1–75Google Scholar. See also Montgomery, Cotton Manufacture of the United States.

⁹⁵ See House Doc. 55, 18th Cong., 2d Sess.; House Exec. Doc. 120, 25th Cong., 2d Sess.; and House Doc. 133, 27th Cong., 3d Sess. The Corps estimated hourly coal consumption at about 6.35 pounds per horsepower in 1843 compared with 10.18 pounds per horsepower for the 1820s.

⁹⁶ DeBow's Review, 5 (01/06 1848), 532–35Google Scholar. Also DeBow's Review, 7 (08 1849), 128–34Google Scholar; and DeBow, James D. B., The Industrial Resources, pp. 128–30Google Scholar.

⁹⁷ Emery, “Costs of Steam Power,” Schedule A; and Unwin, “Cost of Steam Power,” 353–54Google Scholar.

⁹⁸ , Manning, “Comparative Costs,” 503Google Scholar; , Pray, “Steam Horsepower,” 14–19Google Scholar; , Main, “Cost of Steam,” 113Google Scholar; and , McElroy, “Water Power,” 68Google Scholar.

⁹⁹ The upper-bound time trend on fuel consumption per horsepower-hour closely followed the equation:

¹⁰⁰ Wickstead, Thomas, “On the Effective Power of the High Pressure Expansive Condensing Engines in Use at Some of the Cornish Mines,” Transactions of the Institution of Civil Engineers, 2 (1838), 67Google Scholar.

¹⁰¹ , Eavenson, American Coal Industry, especially pp. 385–405Google Scholar, but see also the references listed in the index by states.

¹⁰² See footnote 90.

¹⁰³ “Coal Trade Review,” column in Coal, vols. 1 and 2 (published weekly). Also Coal and Coal Trade Journal, vol. 1 (published weekly).

¹⁰⁴ Justitia, Strictures on Montgomery; , Clowes, “Steam and Water,” 446–47Google Scholar; DeBow's Review, 5, 534; DeBow's Review, 7, 131; , DeBow, Industrial Resources, p. 129Google Scholar; , Manning, “Comparative Costs,” 503Google Scholar; , Main, “Cost of Steam,” 113Google Scholar; and , McElroy, “Water Power,” 68Google Scholar.

¹⁰⁵ U.S. Bureau of the Census, Historical Statistics, series E129.

¹⁰⁶ , DeBow, Industrial Resources, p. 129Google Scholar, and , Manning, “Comparative Costs,” 502Google Scholar.

¹⁰⁷ , Emery, “Costs of Steam Power,” Schedule A. These findings are corroborated by data in Kent, William, ed., The Mechanical Engineer's Pocket Book (New York, 1916), p. 1011Google Scholar, quoting from the Buckeye Engine Company Catalogue that puts the minimum cost per horsepower at $12.50 for a Buckeye Engine rated at 350 h.p. Costs rose quite sharply for small engines ($20/h.p. for 50 h.p. engines/and more gradually for larger engines ($15/h.p. for an 800 h.p. engine).

¹⁰⁸ U.S. Department of the Interior, Census Office, Reports on the Water-Power of the United States at the Tenth Census (Washington, D.C., 1885)Google Scholar, vol. 16 (New England, Middle Atlantic and South), and vol. 17 (Midwest).

¹⁰⁹ , Swain, “Statistics of Water Power,” 1–44Google Scholar.

¹¹⁰ Cyclopaedia of Arts and Sciences vol. 8, (London, 1861), pp. 765–70. See also House Exec. Doc. 120, 25th Cong., 2d Sess., 49, and House Doc. 133, 27th Cong., 3d Sess., 18Google Scholar.

¹¹¹ Sweeny, Frank R. I., “The Burden Waterwheel,” Transactions of the American Society of Civil Engineers, 79 (1915), 708–26, especially 715–20Google Scholar.

¹¹² , Morris, “Remarks on Reaction Water Wheels,” 218–19, 300–04Google Scholar.

¹¹³ Francis, James B., Lowell Hydraulic Experiments, 3rd ed. (New York, 1871)Google Scholar. See also Mills, Hiram, “Experiments with a 35-inch Bodine Turbine,” Journal of the Franklin Institute, 89 (06 1870), 377–79Google Scholar; Emerson, “Water Wheel Tests at Lowell”; Francis, James B., “Report on a Test-Trial of a Swain Turbine Water Wheel,” Journal of the Franklin Institute, 99 (04 1875), 249–531Google Scholar; Francis, “Experiments on the Humphrey Turbine Water-Wheel”; and Thurston, Robert H., “The Systematic Testing of Turbine Water Wheels in the United States,” Transactions of the American Society of Mechanical Engineers, 8 (1887), 359–420Google Scholar.

¹¹⁴ , Manning, “Comparative Cost,” 501Google Scholar; Main, , “Cost of Steam,” 118Google Scholar; and, , Webber, “Water-Power,” 44–54Google Scholar.

¹¹⁵ U.S. Bureau of the Census, Historical Statistics, series D847, yielded the following regression equation:

U.S. Department of the Interior, Census Office, Report on the Statistics of Wages, pp. xxviii-xxxiii, gives some statistics by state and industry on hours worked per day as far back as 1835.

U.S. Department of the Interior, Census Office, Report on the Statistics of Wages, pp. xxviii–xxxiiiGoogle Scholar, gives some statistics by state and industry on hours worked per day as far back as 1835.

¹¹⁶ See, for example, House Exec. Doc. 120, 25th Cong., 2d Sess., 56–7; House Doc. 133, 27th Cong., 3d Sess., 23–4; Emery, “Cost of Steam Power,” Schedule A; , Manning, “Comparative Costs,” 502, 504Google Scholar; and, , Main, “Costs of Steam,” 114Google Scholar.

¹¹⁷ The earliest insurance estimates are given in Manning, “Comparative Costs,” 504.

¹¹⁸ See for example, Emery, “Cost of Steam Power,” Schedule A or , Manning, “Comparative Costs,” 502Google Scholar.

¹¹⁹ U.S. Bureau of the Census, Historical Statistics, series D846.

¹²⁰ See for example, , Main, “Costs of Steam,” 115–16Google Scholar.