I am indebted to Stanley Engerman, Ronald Lee, Gary Saxonhouse, Jonathan Pincus, Heywood Fleisig, Ralph Schlomowitz, David Denslow, and Mark Aldrich for suggestions, though they bear no responsibility for the assertions contained in the paper. Michael Olin supplied valuable research assistance. My research was partially supported by a Rackham Grant from the University of Michigan.

¹ See Conrad, Alfred H. and Meyer, John R., et al. , “Slavery as an Obstacle to Economic Growth in the United States: A Panel Discussion,” Journal Of Economic History, XXVII (December 1967), 528–29Google Scholar; Aitken, Hugh (ed.), Did Slavery Pay? (Boston: Houghton Mifflin Company, 1971); pp. 273–74, 284–85, 288.Google Scholar

² Aldrich, Mark, “Flexible Exchange Rates, Northern Expansion, and the Market for Southern Cotton: 1866–1879,” Journal Of Economic History, XXXIII (June 1973), 399–416.Google Scholar

³ Meyer, John R., “Comment on Papers by Engerman, Goldin, and Kahan,” Journal Of Economic History, XXXIII (March 1973), 104.Google Scholar This is not a literal replication of Meyer's statement, which was: “Was the realized cotton price in the South in 1866 lower than it would have been without the Civil War?” Since the New York price of cotton was 43¢/lb. in 1866—virtually the highest peacetime price observed in history—this question cannot be taken both literally and seriously. What must be intended is the question whether, had U.S. cotton production been at its 1865–66 level without a Civil War, the price would have been higher than it actually was.

⁴ Indian exports to Britain are from Ellison, Thomas, The Cotton Trade of Great Britain (New York: Augustus M. Kelley, 1968)Google Scholar, Table l.( Brazilian exports are from Stein, Stanley, the Brazilian Cotton Manufacture (Cambridge: Harvard University Press, 1957), p. 198.Google Scholar Egyptian exports are from Todd, John A., The World's Cotton Crops (London: A&C Black Ltd., 1915), p. 421.Google Scholar Todd identifies the Egyptian table as “The Egyptian Cotton Crop,” but it is clear that the figures are exports. Cf. Owen, E. R. J., Cotton and the Egyptian Economy, 1820–1914 (Oxford: Clarendon Press, 1969), pp. 34, 73.Google Scholar Foreign cotton prices are from Ellison, The Cotton Trade (India and and Brazil), and Todd, Cotton Crops (Egypt).

⁵ The significance of this result should not be overstated. For India and Brazil, the figures are UK prices of Indian and Brazilian cotton varieties—which may not reflect local prices very well. For India, the statistical “superiority” of US prices is in any case quite marginal.

⁶ Such an effect is most commonly suggested for India, where diversion from the China market was also a factor. See Silver, Arthur W., Manchester Men and Indian Cotton 1847–1872 (Manchester, The University Press, 1966), especially pp. 30–32.Google Scholar Silver does indicate, however, that delays in information and shipment may have caused this diversion to exhibit a one-year lag.

⁷ The cotton crop is from Hammond, The Cotton Industry, Appendix. British GNP figures are from Phyllis Deane, “New Estimates of Gross National Product for the United Kingdom, 1830–1914,” Review of Income and Wealth, Series 14 (June 1968). U.S. GNP figures are from Berry, Thomas Senior, Estimated Annual Variations in Gross National Product, 1789 to 1909 (Richmond: The Bostwick Press, 1968), pp. 42, 32.Google Scholar

⁸ For India, this result tends to confirm recent research suggesting that the expansion of exports during the Civil War. did not come at the expense of home consumption or exports to China, but rather involved an expansion of acreage devoted to cotton. See Hametty, Peter, “Cotton Exports and Indian Agriculture, 1861–70,” Economic History Review, XXIV (August 1971), 415–428.Google Scholar

⁹ See Wright, Gavin, “An Econometric Study of Cotton Production and Trade, 1830–1860,” The Review of Economics and Statistics, LIII (May 1971), 111–120Google Scholar; Temin, Peter, “The Causes of Cotton-Price Fluctuations in the 1830's,” The Review of Economics and Statistics, XLIX (November 1967), 463–470.Google Scholar

¹⁰ See Silver, Manchester Men and Indian Cotton, chapters II-V; Owen, Cotton and the Egyptian Economy, chapters I-III; Hametty, Peter, Imperialism and Free Trade (Manchester: Manchester University Press, 1972), chapters 2–4.Google Scholar

¹¹ Silver, Manchester Men, pp. 242, 290, Bhatia, B. M., Famines in India (Bombay: Asia Publishing House, 1963).Google Scholar Ironically, Bhatia attributes the increased incidence of famine in large measure to the opening of Indian agriculture to the pressures of world demand during the 1860's.

¹² Logan, Frenise A., “India's Loss of the British Cotton Market After 1865,” Journal of Southern History, XXXI (February 1965), 42–44Google Scholar; Silver, Manchester Men, Appendix A; Harnetty, “Cotton Exports and Indian Agriculture,” 428; Harnetty, Imperialism and Free Trade, 82–100.

¹³ Sandberg, Lars G., “Movements in the Quality of British Cotton Textile Exports, 1815–1913,” Journal Of Economic History, XXVIII (March 1968), 8, 10–11.Google Scholar Sandberg seems to view the series somewhat differently, referring to a “virtually continuous” fall to 1890. But after the increase from the extraordinarily low levels of 1861–66, no downward trend is evident in his figures. It is only in 1889–90 that the index falls below the 1859–60 level to any great extent, and it quickly returns to its former level and starts upward in 1891.

¹⁴ Indian cotton did find a somewhat better welcome in parts of the continent, where textile industries at an earlier stage of development could make use of a lower-quality variety. Hence the use of British import figures overstates the degree of collapse in Indian supply. Nonetheless, the time series of total Indian cotton exports, which do not become available until the late antebellum period, is qualitatively similar to the narrower series. Indeed actual total exports are below predicted exports to Britain in every year but one after 1870. See Statistical Abstract Relating to British India (London: George Eyre and William Spottiswoode), No. 7, pp. 28–29 (1873)Google Scholar, No. 10, pp. 50–51 (1876), No; 20, pp. 126–127 (1886).

¹⁵ My understanding of the Brazilian case has benefited from access to the unpublished research of David Denslow.

¹⁶ Owen, Cotton and the Egyptian Economy, p. 189.

¹⁷ Aldrich, “Flexible Exchange Rates,” 401–405. For the underlying conceptual outline, see Kindahl, James K., “Economic Factors in Specie Resumption,” Journal of Political Economy, LIX (February 1961).Google Scholar

¹⁸ Henry Schultz estimated the elasticity of domestic demand for 1875–1895 between − .436 and − .510, both strongly significant. See The Theory and Measurement of Demand (Chicago: The University of Chicago Press, 1938), pp. 304–305.Google Scholar Two-stage least square estimates of American demand after the-war using American GNP figures as the shift term, show elasticities ranging from − .40 to − .88. These are really not far from the Aldrich estimates, but he may have underestimated the coefficient to some degree by measuring cotton prices relative to cloth prices (thereby removing by assumption a derived-demand component of the curve for raw cotton).

¹⁹ Leff, Nathaniel H., “Development and Regional Inequality: Origins of the Brazilian Case,” Quarterly Journal of Economics, LXXXVI (May 1972), 243–262.Google Scholar

²⁰ See Ray, Parimal, Indian's Foreign Trade Since 1870 (London: George Routledge & Sons, Ltd., 1934)Google Scholar, esp. ch. I for discussion of structural-change in India's foreign trade. India was on a silver standard until 1893, which meant that the rupee fluctuated in value relative to the pound and other gold-standard currencies (Ray, ch. VI).

²¹ Leff does present evidence that variations in the exchange rate explains only 27 percent of the variations in the milreis price of cotton during 1822–73, but 67 percent during 1874–1913 (p. 256). This evidence is difficult to interpret in light of the fact that Brazilian currency was depreciating markedly during the entire nineteenth century. Hence Leff's case is substantially less well-established than Aldrich's, since it is not even shown that the exchange rate was “high” relative to any standard; to the extent that exchange-rate variations were causea by monetary disturbances, the evidence has no relevance at all. What the evidence seems to reflect primarily is the instability of the Sterling cotton price during 1822–1873 and its relative stability during 1874–1913. Again, I am indebted to David Denslow for discussion of the point.

²² The Theory and Management of Demand, Chapter VIII.

²³ It is reasonable to suppose that measurement errors will be concentrated in the quantity term. By minimizing the sum of squares strictly in a vertical direction, ignoring the horizontal error component, one will obtain a demand curve which appears to be flatter than it really is.

²⁴ All equations are strictly log-linear, estimated over the period 1866–95. Total supply is defined as the cotton crop in the current crop year plus stocks of raw cotton retained from the preceding year. From Watkins, J. L., Production and Price of Cotton for One Hundred Years, U.S. Department of Agriculture Division of Statistics, Miscellaneous Series, Bulletin No. 9 (Washington, 1895).Google Scholar

²⁵ Watkins, , The Cotton Crop of 1899–1900, U.S. Department of Agriculture Division of Statistics. Miscellaneous Series, Bulletin No. 19 (Washington, 1901).Google Scholar

²⁶ Wright, “An Econometric Study of Cotton Production and Trade, 1830–1860,” pp. 117–119.

²⁷ Despite rising production, there is no upward trend at all in British cotton goods consumption during this period. In fact, consumption per head of cotton goods did not exceed the 1860 level until after World War I. See Tyson, R. E., “The Cotton Industry,” in Aldcroft, Derek H. (ed.), The Development of British Industry and Foreign Competition, 1875–1914 (London: George Allen & Unwin Ltd., 1968), p. 103.Google Scholar

²⁸ The change in American market share from about ¾ to about ½ (1866–75) explains just about the full amount of the change: the antebellum figures imply that η_T is about .7 and E_S is about .25: − 2(.7) − (1)(.25) = 1.65. This explanation suggests that the elasticity should have come back down over time, and Schultz' evidence indicates that this is indeed the case.

²⁹ Engerman, Stanley, “The Economic Impact of the Civil War,” in Fogel, Robert and Engerman, Stanley (eds.), The Reinterpretation of American Economic History (New York: Harper and Row, 1971), p. 372.Google Scholar

³⁰ See Robson, R., The Cotton Industry in Britain (London: Macmillan & Co. Ltd., 1957), p. 6.Google Scholar

³¹ The technique is used by Fogel and Engerman for the antebellum years. See “The Economics of Slavery,” in The Reinterpretation of American Economic History.

³² Aldrich shows a demand curve for American exports of cotton which appears to show a rate of shift of five percent per year for the period 1866–79. This is misleading, however, because the regression includes another variable BR, the real value of British cotton textile exports (which in fact declined over the period). Taking the mean value of BR into account, Aldrich's regression has effectively a much lower constant term than mine, explaining the high growth rate over this short period.

³³ Figures for Britain and other countries from The Fontana Economic History of Europe, vol. 4, Statistical Appendix, 1700–1914, Industry Table 11.Google Scholar

³⁴ Tyson, “The Cotton Industry,” esp. pp. 116–118,125–127.

³⁵ This point is made by Fischbaum, Marvin and Rubin, Julius, “Slavery and the Economic Development of the South,” Explorations in Economic History, VI (Fall 1968).Google Scholar

³⁶ Richard Easterlin, “Regional Income Trends, 1840–1950,” in Fogel and Engerman, The Reinterpretation, p. 40.

³⁷ Some evidence for this effect may be found in the effect of cotton prices on land sales in the antebellum years (Wright, “An Econometric Study,” p. 116), but this relationship is hardly the whole story. Steven DeCanio does find substantially higher elasticities for the postwar period than does Aldrich, though of course they are far less than infinite for particular states. See “Cotton ‘Overproduction’ in Late Nineteenth-Century Southern Agriculture,” Journal Of Economic History, XXXIII (September 1973).Google Scholar

³⁸ Ransom, Roger L. and Sutch, Richard, “Debt Peonage in the Cotton South After the Civil War,” The Journal Of Economic History, XXXII (September 1972).Google Scholar

³⁹ As of 1880, the ratio of gross income in cotton to its 1860 level was 45 percent higher than a similar ratio for corn, 20 percent higher than the same ratio for total gross income. The annual figures are collected and discussed in Lerner, Eugene M., “Southern Output and Agricultural Income, 1860–1880,” Agricultural History, XXXIII (July 1959).Google Scholar