I am grateful to Jeremy Atack, Michael Bernstein, Michael Edelstein, Carol Heim, William Parker, Gavin Wright, as well as Claudia Goldin and an anonymous referee for their comments. Earlier versions of the paper were presented to the Columbia University Seminar in Economic History and the 26th Annual Cliometrics Conference. Financial assistance for the research on transportation improvements was provided by the ENO Foundation.Google Scholar

¹ See Rubin, Julius, “Urban Growth and Regional Development,” in Gilchrist, David T., ed., The Growth of the Seaport Cities 1790–1825 (Charlottesville, 1967), pp. 14–15;Google ScholarSchlotterbeck, John T., “The ‘Social Economy’ of an Upper South Community: Orange and Greene Counties, Virginia, 1815–1860,” in Burton, Orville Vernon and McMath, Robert C., eds., Class, Conflict and Consensus Antebellum Southern Community Studies (Westport, 1982), pp. 3–28;Google ScholarHahn, Steven, The Roots of Southern Populism: Yeoman Farmers and the Transformation of the Georgia Upcountry (New York, 1983), part 1;Google Scholar and Barron, Hal S., Those Who Stayed Behind: Rural Society in Nineteenth-Century New England (New York, 1984).Google Scholar

² The following two paragraphs summarize the argument in my unpublished paper “Slavery, Plantation Settlement and Regional Development in the Antebellum Cotton South,” presented at the 45th annual meeting of the Economic History Association in Sept. 1985. In South Carolina, Georgia, and Alabama the Upcountry includes the counties in the Upper Piedmont and Appalachian Valley regions, and the Cotton Belt comprises the Lower Piedmont, Central Coastal Plain, and Tennessee Valley regions.Google Scholar

³ This argument is advanced by Hahn, Roots of Southern Populism, pp. 48–49;Google ScholarAllman, John M. III, “Yeoman Regions in the Antebellum Deep South: Settlement and Economy in Northern Alabama, 1815–1860” (Ph.D. dissertation, University of Maryland, 1979), chap. 6;Google Scholar and Ford, Lacy K., “Social Origins of a New South Carolina” (Ph.D. dissertation, University of South Carolina, 1983), chap.5.Google Scholar

⁴ Early frosts and delays in shipments lowered expected cotton yields and farm-gate prices and magnified their fluctuations. Consequently, specialization in cotton was less profitable in the Upcountry than in the Cotton Belt and an even greater risk for small producers. For a summary and critique of this argument, see Weiman, David F., “The Economic Emancipation of the Non-Slaveholding Class: Upcountry Farmers in the Georgia Cotton Economy,” this Journal, 45 (03 1985), pp. 72–77.Google Scholar

⁵ Genovese, Eugene D., “Yeomen Farmers in a Slaveholders' Democracy,” Agricultural History, 49 (04 1975), pp. 331–42;Google Scholarand Hahn, Roots of Southern Populism, esp. chap. 2.Google Scholar

⁶ Hilliard, Sam Bowers, Hog Meat and Hoecake: Food Supply in the Old South, 1840–1860 (Carbondale, 1972), pp. 126–28.Google Scholar

⁷ Smith, Alfred G. Jr, Economic Readjustment of an Old Cotton State South Carolina 1820–1860 (Columbia, 1958), p. 74;Google Scholar and Hilliard, Hog Meat and Hoecake, pp. 198–200.Google Scholar

⁸ Wright, Gavin, Political Economy of the Cotton South (New York, 1978), pp. 68–74.Google Scholar

⁹ This point is emphasized by Wright, Gavin, Old South, New South: Revolutions in the Southern Economy since the Civil War (New York, 1986), pp. 39–43.Google Scholar

¹⁰ To test the representativeness of the sample counties, a discriminant function explaining the division between Cotton Belt and Upcountry counties was estimated using the published census data for counties in the cotton economy of 1860 Georgia. The estimated discriminant function correctly classified both counties in the Upcountry region at a 1 percent confidence level.Google Scholar

¹¹ Weiman, “Slavery, Western Settlement and Regional Development,” pp. 38–39.Google Scholar

¹² For example, one-third of pioneering households owned slaves, and 13 percent of the slaveholders owned at least 15 slaves; Manuscript Census, Floyd County, Georgia, 1840. The early history of Rome and the Rome Railroad Company is told by Battey, George M. Jr, A History of Rome and Floyd County (Atlanta, 1922), pp. 33–34, 107.Google Scholar

¹³ American Railroad Journal, 2 (09 26, 1846), p. 620.Google Scholar

¹⁴ White, George, Statistics of the State of Georgia (Savannah, 1849), p. 249;Google ScholarAmerican Railroad Journal, 4 (12 23, 1848), p. 818.Google Scholar

¹⁵ U.S. Patent Office, Annual Report (Washington, D.C., 1852), p. 329;Google Scholarsee also the circulars for the years 1848 (p. 502), 1849 (pp. 202–3), 1852 (pp. 73–74), and 1855 (p. 238); and Southern Cultivator, 8 (March 1850), p. 41 and (June 1850), p. 89.Google Scholar

¹⁶ Ibid.

¹⁷ U.S. Census Office, Census, Eighth, 1860, Population of the United States in 1860 (Washington, D.C., 1864), vol. 1, p. 74.Google Scholar

¹⁸ Georgia, vol. 12, Dun, R. G. & Company Collection, Baker Library, Harvard University Graduate School of Business Administration, pp. 151–54.Google Scholar

¹⁹ U.S. Patent Office, Annual Report (1856), pp. 191, 238.Google Scholar

²⁰ Martin, Thomas H., Atlanta and Its Builders: A Comprehensive History of the Gate City of the South (Atlanta, 1902), vol. 1, pp. 20–21.Google Scholar

²¹ Railroad, George and Banking Company, “Annual Report,” reprinted in American Railroad Journal, 2 (10 10, 1846), pp. 645–46.Google Scholar

²² Georgia, vol. 9, Dun, R. G. & Company Collection, pp. 127–36.Google Scholar

²³ The samples contain 370 households from DeKaIb County, or every third household, and 404 households from Floyd County, or every fourth household. A more thorough description of the sampling procedure and tests of the representativeness of the sample populations are presented in Weiman, David F., “Petty Commodity Production in the Cotton South: Upcountry Farmers in the Georgia Cotton Economy, 1840 to 1880” (Ph.D. dissertation, Stanford University, 1983), appendix A.Google Scholar

²⁴ Gallman, Robert E., “Self-Sufficiency in the Cotton Economy of the Antebellum South,” Agricultural History, 44 (01 1970), p. 7.Google Scholar

²⁵ Wright, Political Economy of the Cotton South, p. 58.Google Scholar

²⁶ Seed requirements, expressed as a percentage of the crop set aside for planting, are 5 percent for corn, 12 percent for wheat, 7 percent for oats, 11 percent for rye, 9 percent for peas and beans, and 10 percent for potatoes. The feed requirements for livestock are 35 bushels of corn for horses and oxen, 30 bushels for mules, 5 bushels for dairy cows, and 0.25 bushel for sheep. The estimated seed requirements for all crops, except potatoes, were derived from the average per acre seed requirements and yields in the region reported in the U.S. Patent Office, Annual Report for the years 1848 (p. 501), 1849 (p. 144), 1850 (pp. 194–95, 231–32, 393–94, 400–1, 460–61), 1851 (pp. 319, 323, 331), 1852 (pp. 82, 89–90), and 1855 (p. 191).Google Scholar These estimates are similar in magnitude to those used in previous studies. Because this source provided no information on seed requirements for root crops and feed requirements for livestock, I used the estimates reported by Battalio, Raymond and Kagel, Jonathan, “The Structure of Antebellum Southern Agriculture: South Carolina, A Case Study,” Agricultural History, 44 (01 1970), p. 28, fn. 7;Google Scholar and Ransom, Roger and Sutch, Richard, One Kind of Freedom: The Economic Consequences of Emancipation (New York, 1977), p. 250. In general, I have followed assumptions made by Ransom and Sutch on feeding practices and levels of feed to compare the estimates of food supplies and self-sufficiency in foodstuffs on tenant and small owner-operated farms in 1860 and 1880.Google Scholar

²⁷ A survey of the circular reports published in the U.S. Patent Office, Annual Report, from agents in the Piedmont and Valley regions of Alabama, Georgia, and South Carolina between 1848 and 1855 attests to the widespread use of this practice. As an Alabama farmer admitted, farmers in the region have “…no real system of raising stock…the cattle live half the time on Uncle Sam's pasture” (1851, p. 331). See also the published circulars in the U.S. Patent Office, Annual Report for the years 1850 (pp. 233, 401), 1851 (p. 320), 1852 (p. 83), and 1855 (pp. 20, 50, 61).Google ScholarHutchinson, William K. and Williamson, Samuel H., “The Self-Sufficiency of the Antebellum South,” this Journal, 31 (09 1971), pp. 595–600, present indirect, quantitative evidence of the importance of informal feeds in the antebellum South, as well as a summary of contemporary feeding practices drawn from published sources. I am grateful to William Hutchinson for these references.Google Scholar

²⁸ The net output of feed crops excludes corn fodder and so underestimates the supply of livestock feed by approximately 11 percent. Farmers picked an average of 1,200 to 1,800 pounds of fodder for every 100 bushels of corn harvested, and fodder has half the nutritional value of corn. See U.S. Patent Office, Annual Report, 1848 (p. 501–2), and 1849 (p. 145); Hilliard, Hog Meat and Hoecake, p. 138;Google Scholar and Baily, L. H., ed., Cyclopedia of American Agriculture. Vol. 3: Animals (New York, 1908), pp. 111–12.Google ScholarCorn fodder is not included in the supply of informal feeds that are used to derive the weight of swine from grazing below; see Hutchinson and Williamson, “The Self-Sufficiency,” pp. 600–1.Google Scholar

²⁹ The lower estimate of 20 bushels per adult in assumed by Ransom and Sutch, One Kind of Freedom, pp. 251–52. The intermediate estimate of 26.4 bushels is based on the standard rations of adult slaves, 16.25 bushels of corn meal and 182 pounds of meat, reported by Battalio and Kagel, “The Structure of Antebellum Southern Agriculture,” pp. 29–30. These rations are converted into bushels of corn under the assumption that swine attained a sustainable weight of 90 pounds on forage. This assumption is consistent with the estimates made by Hutchinson and Williamson, “The Self-Sufficiency,” pp. 600–2. The maximum estimate is based on the rations of adult male slaves, 19.5 bushels of meal and 208 pounds of meat, reported by Gallman, “Self-Sufficiency,” pp. 18–19. They were converted into bushels of corn using the above assumption.Google Scholar

³⁰ Tenants were assumed to pay a share rent equal to one-third of their corn and wheat crops. See Reid, Joseph D. Jr, “Antebellum Southern Rental Contracts,” Explorations in Economic History, 13 (01 1976), pp. 69–83;CrossRefGoogle ScholarHahn, Roots of Southern Populism, pp. 64–66;Google ScholarBode, Frederick A. and Ginter, Donald E., Farm Tenancy and the Census in Antebellum Georgia (Athens, 1986), chap. 5. The criterion used to identify “certain tenants” is thoroughly explained and analyzed by Bode and Ginter.Google Scholar

³¹ Tyron, Rolla, Household Manufactures in the United States, 1640–1860 (Chicago, 1917), pp. 370–71, 374–75, identified the Upcountry counties of the South Atlantic states as one of the exceptional regions where the levels of household manufacturing per capita in 1860 exceeded $2.Google Scholar

³² See Hahn, Steven, “Hunting, Fishing, Foraging: Common Rights and Class Relations in the Postbellum South,” Radical History Review, 26 (10 1982), pp. 38–43;Google Scholar and King, J. Crawford Jr, “The Closing of the Southern Range: An Exploratory Essay,” Journal of Southern History, 48 (02 1982), pp. 53–70.CrossRefGoogle Scholar

³³ For evidence on the sale of corn, wheat, fodder, as well as cotton to local storeowners, see the Ledgers of Thomas Morris, Georgia Department of Archives and History. See also Atherton, Lewis E., The Southern Country Store, 1800–1860 (New York, 1968), chap. 5;Google Scholar and Hahn, Roots of Southern Populism, pp. 32–33.Google Scholar

³⁴ Wright, Political Economy of the Cotton South, p. 56.Google Scholar

³⁵ See Atack, Jeremy and Bateman, Fred, “Self-Sufficiency and the Origins of the Marketable Surplus in the Rural North, 1860,” Agricultural History, 58 (07 1984), esp. pp. 306–12;Google Scholar and McInnis, Marvin, “Marketable Surpluses in Ontario Farming 1860,” Social Science History, 8 (Fall 1984), pp. 395–424. Households with estimated food deficits were assumed to deplete stocks of food or lower their standard of living rather than purchase foodstuffs. To determine the marketable surpluses of wheat, the maximum adult consumption of wheat was assumed to be 4 bushels. This, of course, assumes that households had adequate supplies of food and feed crops to satisfy their remaining consumption requirements. Otherwise, the consumption of wheat was increased.CrossRefGoogle Scholar

³⁶ Wholesale prices of farm products in 1860 are taken from Ransom and Sutch, One Kind of Freedom, p. 203. For references on the sale of food crops by yeomen farmers to store owners, see fn. 33.Google Scholar

³⁷ The supply of field labor is defined as free, male household members between the ages of 15 and 65 and all slaves between the ages of 10 and 65. The greater participation rates of slave women and children in field work is assumed by Wright, Political Economy of the Cotton South, pp. 82–83, and is indirectly measured by the withdrawal of their field labor following emancipation; see Ransom and Sutch, One Kind of Freedom, pp. 44–47. Robert Fogel and Stanley Engerman, “Instructions for Computing Output and Input Data from the Parker-Gailman Sample for 1860 to be used in Production Function Estimates” (unpublished manuscript, n.d.) make the alternative assumption that there is no difference in participation rates of slave and free women. I am grateful to Gerald Friedman for providing me with a copy of the instructions used by Fogel and Engerman to calculate labor inputs. The different assumptions of field labor force participation rates for women and children do not greatly affect the results reported here.Google Scholar

³⁸ The likelihood ratio test statistic for this hypothesis is distributed asymptotically as chisquared with 2 degrees of freedom and equals 1.6. It is not significant at a 5 percent confidence level.Google Scholar

³⁹ The estimated standard error of the slope of the improved acreage variable in the range, 50 to 199 improved acres, is 0.032. The Wald test statistic of the null hypothesis, which is asymptotically distributed as chi-squared with one degree of freedom, equals 9.35, and rejects the hypothesis at a 1 percent confidence level.Google Scholar

⁴⁰ A likelihood ratio test that the sum of these coefficients equals zero cannot reject the hypothesis at a 5 percent confidence level.Google Scholar

⁴¹ The estimate is equal to the coefficient of the dummy variable for tenants divided by the coefficient of the improved acreage variable.Google Scholar

⁴² The potential volume of long-distance trade in food crops equals the estimated supplies of marketable surpluses of corn and wheat, measured in corn-equivalent bushels, less the consumption requirements of the nonagricultural population, assumed to 26.4 bushels per adult consumer. The lower-bound estimate is based on the net supply of food crops for the total farm population, which equals the supplies of marketable surpluses of corn and wheat less the food deficits of those households that were not self-sufficient in foodstuffs. The upper-bound estimate is based on the total volume of marketable surpluses only.Google Scholar

⁴³ Farms with at least 100 improved acres in this section comprised only 20 percent of all farms, but accounted for over one-half of the total cotton output in 1860. The southern section of the county includes the militia districts of Evans, Diamond, Pantherville, Lythonia, and Barnes. I am grateful to Cindy Platto, archivist at the DeKalb Historical Society, for providing me with the information about the location of these districts. See also Smith, George Gillman, The Story of Georgia and the Georgia People (2nd edn., Baltimore, 1968), p. 363. Specialization in cotton did not follow the route of the Georgia Railroad, which went through the Lythonia, but also the Stone Mountain and Decatur districts.Google Scholar

⁴⁴ Danhof, Clarence H., Change in Agriculture: The Northern United States, 1820–1870 (Cambridge, Mass. 1969), p. 18.Google Scholar

⁴⁵ Western, and Railroad, Atlantic, Report of the Superintendent and Treasurer (Atlanta, 1858–1860). In 1860, 25.4 thousand bushels of corn and meal, 90.5 thousand bushels of wheat, 8.1 thousand bushels of other crops, 12.6 thousand barrels and 69.3 thousand sacks of flour, as well as large quantities of livestock and barreled pork were shipped to Atlanta from the Rome and East Tennessee and Georgia railroads.Google Scholar

⁴⁶ The interpretation of the regression coefficients is put forward by Atack, Jeremy, “Farm and Farm-Making Costs Revisited,” Agricultural History, 56 (10 1982), p. 673.Google Scholar

⁴⁷ The increment to the real wealth of the household equals the value of structures plus the value added to the improved acreage. The estimate is based on the acreage holding of improved acreage for medium-sized farmers. In contrast, a household with the same amount of land in Floyd County would earn $1,275 from farm-making.Google Scholar

⁴⁸ Weiman, “The Economic Emancipation,” pp. 82–93,Google Scholar documents the timing of this transformation and the formation of a marketing and transportation system in the region, and identifies the private sources of capital for the region's development. For evidence on state aid for railroad construction, see Summers, Mark W., Railroads, Reconstruction, and the Gospel of Prosperity Aid under the Radical Republicans, 1865–1877 (Princeton, 1984).Google Scholar

⁴⁹ See Wright, Old South, New South, pp. 115–23;Google ScholarHahn, Steven “The ‘Unmaking’ of the Southern Yeomanry: The Transformation of the Georgia Upcountry, 1860–1890” in Hahn, Steven and Prude, Jonathan, eds., The Countryside in the Age of Capitalist Transformation: Essays in the Social History of Rural America (Chapel Hill, 1985), pp. 179–204;Google Scholar and Weiman, “Petty Production,” pp. 409–42.Google Scholar