¹ On the county's early history, see The History of Appanoose County, Iowa (Chicago, 1878);Google Scholar also Taylor, L. L., ed., Past and Present of Appanoose County, Iowa: A Record of Settlement, Organization, Progress and Achievement (Chicago, 1913). Appanoose County was chosen for this study because it was on the frontier in 1850, did not develop a significant urban center in the period of study, and had few foreign-born residents. We hope to carry out a series of parallel studies in the future, choosing other locations that allow us separately to eliminate each of these three criteria and examine the effects on the results obtained.Google Scholar

² Bogue, Allan G., From Prairie to Cornbelt: Farming on the Illinois and Iowa Prairies in the Nineteenth Century (Chicago, 1963), pp. 220–23.Google Scholar

³ The 1850 sample contained 631 people, of whom 325 (51.5 percent) were males; the 1860 sample contained 2,430 people, of whom 1,261 (51.9 percent) were males.Google Scholar

⁴ These statements should properly be qualified to allow for the inevitably incomplete enumeration of the censuses. For a discussion of the biases that can be introduced by underenumeration, see Galenson, David W. and Levy, Daniel S., “A Note on Biases in the Measurement of Geographic Persistence Rates”, Historical Methods, 19 (Fall 1986), pp. 171–79.CrossRefGoogle Scholar

⁵ Bogue, From Prairie to Cornbelt, p. 22. Merle Curti reports that the mean age of farm operators in Trempealeau County in 1860 was 38, while that of businessmen was 36, of artisans 32, and farm laborers 25;Google Scholar see Curti, Merle, The Making of an American Community: A Case Study of Democracy in a Frontier County (Stanford, 1959), p. 56.Google Scholar

⁶ These results are very similar to those found for Wapello County, Iowa, in 1850; see Bogue, From Prairie to Cornbelt, p. 23.Google Scholar Also see Herriott, F. I., “Whence Came the Pioneers of Iowa?” Annals of Iowa, 3rd series, 7 (04 1906), pp. 367–79; and 7 (July 1906), pp. 446–65.Google Scholar

⁷ These figures were calculated by assuming that individuals listed in a given household in the census were children of the head of that household if they had the same surname as the head and were more than 18 years younger than the head. A woman was assumed to be the wife of a male head of household if listed immediately after the head, with the same surname, and within 18 years or less of his age. In 1850, the 105 households sampled also contained seven individuals who had the same surname as the household head but were not identified as wives or children by the above definition; in 1860 the 430 households contained 30 such individuals. The 1850 sample also included 37 individuals with surnames different from the household head; in 1860 there were 159 of these individuals.Google Scholar

⁸ The ratio of children aged 0 to 9 to women aged 20 to 49 in farm households in Appanoose County from the 1860 random sample was 2.36. This is higher than the ratios reported for 1860 for farm households in all the “settlement class” categories defined by Easterlin, Richard A., Alter, George, and Condran, Gretchen A.. “Farms and Farm Families in Old and New Areas: The Northern States in 1860”, in Hareven, Tamara K. and Vinovskis, Mans A., eds., Family and Population in Nineteenth-Century America (Princeton. 1978), table 1–13. p. 36.Google Scholar

⁹ Bogue, From Prairie to Cornbelt, p. 25.Google Scholar

¹⁰ Throne, Mildred, “A Population Study of an Iowa County in 1850”, Iowa Journal of History, 57 (10 1959), p. 310; and Bogue, From Prairie to Cornbelt, p. 26. John Mack Faragher reported a persistence rate of 22.2 percent for household heads in a rural Illinois community from 1850 to 1860;Google Scholar see Faragher, John Mack, Sugar Creek: Life on the Illinois Prairie (New Haven, 1986), p. 249, fn. 14.Google Scholar

¹¹ Malin, James C., “The Turnover of Farm Population in Kansas,” Kansas Historical Quarterly, 4 (11 1935), pp. 365–67; and Curti, The Making of an American Community, p. 68. Since these earlier studies' persistence rates were calculated exclusively for farmers, it might be noted that the persistence rates for those household heads listed in the census of Appanoose as farmers were 25.3 percent for 1850 to 1860, and 36.3 percent for 1860 to 1870. It is not surprising that these are close to the rates calculated for all household heads, for farmers made up 83 percent of the total in 1850, and 71 percent in 1860.Google Scholar

¹² The difference in 1850 real wealth for those who departed and persisted during the 1850s is statistically insignificant at conventional levels. The difference in 1860 total wealth—as well as that in real wealth—between leavers and stayers during the 1860s is statistically significant at the 0.01 level. It is possible that this result observed for Appanoose, in which a positive association between wealth and persistence appeared or grew stronger after an initial period of settlement, may have been common to other frontier communities.Google Scholar For example, see Conzen, Michael P., Frontier Farming in an Urban Shadow: The Influence of Madison's Proximity on the Agricultural Development of Blooming Grove, Wisconsin (Madison, 1971), p. 128;Google Scholar and McQuillan, D. Aidan, “The Mobility of Immigrants and Americans: A Comparison of Farmers on the Kansas Frontier,” Agricultural History, 53 (07 1979), pp. 589–94.Google Scholar

¹³ For example, see Sjaastad, Larry A., “The Costs and Returns of Human Migration,” Journal of Political Economy, part 2, Supplement, 70 (10 1962), pp. 80–93;CrossRefGoogle Scholar and Mincer, Jacob, “Family Migration Decisions,” Journal of Political Economy, 86 (10 1978), pp. 749–73.CrossRefGoogle Scholar

¹⁴ For a discussion of a multivariate empirical approach to the analysis of persistence, see Kousser, J. Morgan, Gary W. Cox, and David W. Galenson, “Log-Linear Analysis of Contingency Tables: An Introduction for Historians with an application to Thernstrom on the ‘Floating Proletariat,’” Historical Methods, 15 (Fall 1982), pp. 152–69.CrossRefGoogle Scholar

¹⁵ For example, see Lewis, G. J., Human Migration: A Geographical Perspective (London, 1982), p. 83.Google Scholar

¹⁶ For findings of a positive association between wealth and persistence in nineteenth-century communities, for example, see Throne, “A Population Study of an Iowa County,” p. 321; Curti, The Making of an American Community, p. 75;Google ScholarBowers, William L., “Crawford Township, 1850–1870: A Population Study of a Pioneer Community,” Iowa Journal of History, 58 (01 1960), pp. 22–23;Google ScholarThernstrom, Stephan, Poverty and Progress: Social Mobility in a Nineteenth-Century City (Cambridge, 1964), pp. 89–90;Google ScholarConzen, Frontier Farming in an Urban Shadow, p. 128;Google ScholarKatz, Michael B., The People of Hamilton, Canada West: Family and Class in a Mid-Nineteenth- Century City (Cambridge, MA, 1975), pp. 130–31;CrossRefGoogle ScholarBarron, Hal S., Those Who Stayed Behind: Rural Society in Nineteenth-Century New Haven England (Cambridge, 1984), p. 81;Google ScholarGjerde, Jon, From Peasants to Farmers: Migration from Balestrand, Norway, to the Upper Middle West (Cambridge, 1985), pp. 161–62;Google Scholar and Winkle, Kenneth J., The Politics of Community: Migration and Politics in Antebellum Ohio (Cambridge, 1988), pp. 117–19. The following table provides some additional detail on the relationships among age, wealth, and persistence for Appanoose heads of households: As in Table 4, the rate of persistence rises with age to a peak in the forties, then declines. The ratio of the wealth of nonpersisters to that of persisters also rises with age in this sample; adding an age-wealth interaction term as an independent variable in the second version of the equation of Table 4 yielded an estimated coefficient that was negative but statistically insignificant, without affecting the magnitudes of the other coefficients.CrossRefGoogle Scholar

¹⁷ A number of other variants of these equations were estimated without changing these basic results. For example, when family size was removed from the third equation of Table 4 and replaced by two other variables—number of boys per family, and number of girls—the estimated coefficients of both were substantively and statistically insignificant.Google Scholar

¹⁸ Curti, The Making of an American Community, p. 69; also see p. 76. On older communities, see Barron, Those Who Stayed Behind, pp. 81–87;Google Scholar and Thernstrom, Stephan, The Other Bostonians: Poverty and Progress in the American Metropolis (Cambridge, 1973), pp. 39–40.CrossRefGoogle Scholar

¹⁹ In the analysis of wealth in this article, the estimates of individuals' wealth are those reported in the population census manuscripts. This source of information on wealth has been used in many investigations of the nineteenth-century economy, most extensively by Lee Soltow. Yet some historians continue to question the reliability of these data. Some research currently in progress has begun to address this issue directly. Specifically, data sets have been constructed for Utah that contain both the reported wealth from the 1870 manuscript federal census and the wealth listed on the state tax rolls for 1870 for 1,568 households. The correlation coefficient between census wealth and tax assessment wealth of households is 0.66. A regression between the two yields In (census wealth) = 1.77 + 0.787 In (tax assessment wealth), with a standard error of 0.022 on the coefficient, and an R² of 0.44. Clearly the two estimates of household wealth are highly correla ed, with tax assessment wealth averaging about 60 percent of census wealth. The obvious incentive to lower tax payments would explain why tax assessment wealth would be lower than actual wealth. Since the wealth estimates from the census are often used, as in this article, to estimate the effect of variables such as age and occupation on wealth accumulation, differences in wealth between the census and other sources may not be as important as differences in the relative influences on wealth of the various explanatory variables. Estimation of two regressions, in which wealth was specified as a function of age, age squared, duration in Utah, occupation, foreign birth, and rural residence for both census wealth and tax assessment wealth, produces very similar coefficients. More work is needed to test the accuracy of census wealth estimates, but this preliminary comparison of wealth listed in the federal census and Utah tax rolls is reassuring. A full treatment of the comparison for Utah will be presented in a forthcoming study by Clayne Pope.Google Scholar

²⁰ Price level adjustments throughout the article were done using the series presented in U.S. Bureau of the Census, Historical Statistics of the United States, Colonial Times to 1957 (Washington, DC, 1960), table El, p. 115.Google Scholar

²¹ The comparisons are to Soltow, Lee, Men and Wealth in the United States, 1850–1870 (New Haven, 1975), table 3.3, p. 76, table 3.4, p. 77, and table 3.5, p. 81. The mean real estate of Appanoose men aged 20 and older in 1850 was $352, and $944 in 1860; the mean total wealth of Appanoose adult men in 1860 was $1,347. The wealth comparisons between Appanoose County and national samples summarized in the preceding text paragraph are based on the figures for Appanoose given in this footnote. For comparison to Soltow's national estimates, these were calculated for all Appanoose adult males in the random samples who were of age 20 and above. Elsewhere in this article, except where specifically noted, wealth estimates for Appanoose County are calculated for heads of households.Google Scholar

²² Curti, The Making of an American Community, p. 78. Wealth holding patterns were similar in Utah, but with a higher degree of inequality. In Utah. the share of household heads with no real estate fell from 34 percent in 1850 to 25 percent in 1860. In 1870, however, the households with no real estate comprised 40 percent of all households. The increase was probably due to the influx of immigrants from Europe and the increase in female-headed households. The richest decile in Utah owned more of the real estate wealth–52 percent in 1850 and 55 percent in 1860—than the rich owned in Appanoose or Trempealeau County.Google Scholar

²³ The distribution of wealth in Appanoose County was much less unequal than in the United States as a whole. Thus for example in 1860 the wealthiest 10 percent of adult males in Appanoose owned 40 percent of total wealth, compared to shares for the highest decile of adult males of 68 percent of total wealth in the North, and 73 percent in the whole United States; see Soltow, Men and Wealth in the United Stares, p. 99. On p. 108 Soltow notes that rural areas generally had considerably less inequality of wealth than cities.Google Scholar

²⁴ One dollar was added to the wealth of all household heads in order to eliminate the problem posed by men without wealth in using the logarithmic transformation. It might be emphasized here that the precise slope of the relationship betweren age and wealth will vary with the treatment of these sample members with no reported wealth—it will change if they are excluded from the sample, or if values other than one dollar are assigned to them when they are included. The treatment used here is the same as that used in the studies cited below, by Jeremy Atack and Fred Bateman, Richard Steckel, and Pope, and is therefore useful for comparative purposes.Google Scholar

²⁵ Atack, Jeremy and Bateman, Fred, “Egalitarianism, Inequality, and Age: The Rural North in 1860,” this Journal, 41 (03 1981), p. 87.Google Scholar

²⁶ The figures in the text were calculated from the results reported in Atack and Bateman, “Egalitarianism, Inequality, and Age,” for all sample members (first row of table 1, p. 87). The profile reported there for farmers only is in turn less steep, with rates of growth of 5.5 percent at 20, and 3.5 percent at 30 (fourth row of table 1).Google Scholar

²⁷ Richard Steckel, “Poverty and Prosperity: A Longitudinal Study of Wealth Accumulation, 1850–1860” (unpublished manuscript, Ohio State University, 1987), table 6.Google Scholar

²⁸ The age-wealth patterns in Utah are also of interest for the comparisons made here. In 1850, shortly after settlement of Utah had begun, real estate wealth was virtually invariant with respect to age, especially between ages 30 and 60. This flat age-wealth profile was probably a result of the heavy migration costs to Utah. By 1860, however, age-wealth profiles in Utah were similar to those found by Atack and Bateman for their northern rural sample. The peak in the age-wealth profile for 1860 in Utah was about 53 years of age, with an annual rate of increase of 6.5 percent at age 20, falling to 4.6 percent at age 30, and 2.7 percent at age 40. The fact that the rates of annual increase are lower for Utah than Appanoose County suggests that the high rates of accumulation in Appanoose were not simply due to low initial wealth, since Utah households were on average less wealthy than the Appanoose households;Google Scholar see Kearl, J. R., Pope, Clayne L., and Wimmer, Larry T., “Household Wealth in a Settlement Economy: Utah, 1850–1870,” this Journal, 40 (09 1980), pp. 477–96;Google Scholar and Pope, Clayne L., “Households on the American Frontier: The Distributions of Income and Wealth in Utah, 1850–1900,” in Galenson, David W., ed., Markets in History: Economic Studies of the Past (Cambridge, 1989).Google Scholar

²⁹ The returns to early arrival have been documented for Utah in Kearl, Pope, and Wimmer, “Household Wealth in a Settlement Economy”; Kearl, J. R. and Pope, Clayne, “Choices, Rents and Luck: Economic Mobility of Nineteenth-Century Households,” in Engerman, Stanley L. and Gallman, Robert E., eds., Long-Term Factors in American Economic Growth (Chicago, 1986), pp. 215–60; and Pope, “Households on the American Frontier.” Each year's duration in Utah increased a household's wealth by 6.7 percent in 1860. Not all of the effect of duration in Utah was a rent or capital gain on land since duration also had a positive effect on income holding wealth constant.Google Scholar

³⁰ The results shown in Table 8 were obtained after eliminating two individuals from the 1860 sample, Jacob and John Coffman. Jacob Coffman's real wealth fell from $1,200 in 1850 to $0 in 1860 (he was 60 years old in 1860). In 1850, his son John, aged 22, was a member of Jacob's household, and reported no real wealth; in 1860, John was the head of his own household, and reported real wealth of $4,800, plus personal wealth of $1,500. Because both the decline in Jacob's wealth and the increase in John's were probably due in part to a transfer, it was not deemed appropriate to include them in this analysis individually. In alternative versions of the equation in which the two men were treated as a single observation, the estimated coefficient of the present-real wealth interaction remained unchanged by more than 0.01 (in sequential specifications, the age was used as the father's, then the son's).Google Scholar

³¹ It might be argued that the large coefficient of the present-in-1850 variable in Table 8 could be due not to an increase over time in the wealth of farmers who had remained in Appanoose during the 1850s, but might rather indicate that those who arrived before 1850 were wealthier at their time of arrival than was the case for men who arrived after 1850. This was probably not true, however. The 1850 mean real wealth of the 27 farmers from the 1860 sample who had been in Appanoose in 1850 was $649; the 1860 mean real wealth of the same men was $2,614. In comparison, the 1860 mean real wealth of the 279 Appanoose farmers who had not been present in 1850 was $1,207 (or approximately $1,090 in 1850 dollars). The latter group therefore had on average considerably more real wealth in 1860 than the earlier arrivals had had in 1850, and the mean ages of the two groups in the respective years were quite similar (the mean age of the early arrivals in 1850 was 36 years, while that of the later arrivals in 1860 was 38 years). The rapid increase between 1850 and 1860 in the mean real wealth of the farmers who had been present in Appanoose in 1850 points to a substantial accumulation of real wealth by these men during their time in the county.Google Scholar

³² For a study of rates of return to land ownership in Appanoose in a period that includes the decades considered here, see Swierenga, Robert P., Pioneers and Profits: Land Speculation on the Iowa Frontier (Ames, 1968), esp. pp. 192–99.Google Scholar On rising land values as the cause of increasing wealth on the frontier, see Newell, William H., “The Wealth of Testators and Its Distribution: Butler County, Ohio, 1803–65,” in Smith, James D., ed., Modeling the Distribution and Intergenerational Transmission of Wealth (Chicago, 1980), pp. 102–9. In a study of wealth accumulation in the frontier states of Texas and Arkansas in 1860, Donald Schaefer also found that with other characteristics constant, the settlers who had already been present in 1850 had greater average wealth than those who had arrived during the 1850s;Google Scholar see Schaefer, Donald F., “A Model of Migration and Wealth Accumulation: Farmers at the Antebellum Southern Frontier,” Explorations in Economic History, 24 (04 1987), pp. 130–57. He argued that the difference was the result of larger initial endowments of the earlier arrivals, the cost of migration, and relatively rapid price appreciation of land on the frontier. Considering Appanoose County, the first of these does not appear to have been the case; in fact, as discussed in the previous footnote, the typical initial endowments of earlier arrivals appear to have been lower than those of later ones. The second of Schaefer's factors appears problematic logically for Appanoose, for all settlers on the Iowa frontier had to bear the costs of migrating there, regardless of their date of arrival. Schaefer's third factor does appear relevant to Appanoose County, in view of the evidence of Tables 7 and 8, which is consistent with the presence of economic rents on real property for earlier arrivals. Another factor, not discussed by Schaefer, might also have been important: settlers might have raised their incomes substantially by moving to the frontier. Even if they had continued to save the same fraction of income as they had before migrating, their wealth would then have risen more rapidly after arriving on the frontier, and this would contribute to the observed correlation between wealth and duration of residence on the frontier. In addition, higher incomes on the frontier might have permitted the settlers to save a higher proportion of their incomes, and this would further strengthen the correlation.CrossRefGoogle Scholar

³³ Gates, Paul W., The Farmer's Age: Agriculture, 1815–1860 (New York, 1960), p. 80.Google Scholar

³⁴ Curti, The Making of an American Community, p. 163. The 30 members of the 1850 sample who persisted in Appanoose until 1860 enjoyed an annual increase in the mean value of their real estate of 16 percent, after adjusting for price level changes. The 148 members of the 1860 sample who persisted to 1870 saw their mean real estate increase by 6 percent annually, and their mean total wealth by 5 percent annually, again adjusted for price level change. The Civil War was, of course, one of many factors that could have produced the lower rates of wealth increase during the 1860s. A historian of Iowa agriculture concluded that while the war had both positive and negative economic effects on the state's farmers, on balance it probably hurt more of them than it helped;Google Scholar see Ross, Earle D., Iowa Agriculture:An Historical Survey (Iowa City, 1951), chap. 4, esp. pp. 59–60;Google Scholar also see Fite, Emerson D., “The Agricultural Development of the West During the Civil War,” Quarterly Journal of Economics, 40 (02 1906), pp. 259–78.CrossRefGoogle Scholar

³⁵ The expected number of survivors of the eight farmers in this group not found in the 1870 census in Appanoose County can be estimated from their ages by employing a model life table; see Coale, Ansley J. and Demeny, Paul, Regional Model Life Tables and Stable Populations (Princeton, 1966). If one assumes that mortality experience on the frontier can be represented by a Model West life table, level 12, then the expected number of survivors would be 5.63. If one assumes level 14, then the expected number of survivors is 5.87. It therefore appears unlikely that most of these men had died by 1870. The Civil War could of course have lowered these numbers of survivors considerably below normal levels. Yet although a substantial number of Appanoose County residents fought in the war, none of the eight men considered here was recorded as having been killed in the war in a list of the county's veterans provided in Taylor, Past and Present of Appanoose County, pp. 176–220.Google Scholar

³⁶ Bogue, From Prairie to Cornbelt, p. 51;Google Scholar also see Petersen, William J., “Population Advance to the Upper Mississippi Valley, 1830–1860”, Iowa Journal of History and Politics, 32 (10 1934), pp. 316–17;Google ScholarGates, The Farmer's Age, pp. 80–85;Google Scholar and Lebergott, Stanley, “The Demand for Land: The United States, 1820–1860,” this Journal, 45 (06 1985), pp. 195–97.Google Scholar

³⁷ Curti, The Making of an American Community, p. 163. This discussion is not intended to imply that considerable gains in wealth invariably led to migration—as shown earlier, for Appanoose the correlation between wealth and persistence was positive—but rather that it was not only the less successful who moved. On this point, see also Throne, “A Population Study of an Iowa County”, p. 321.Google Scholar

³⁸ Curti, The Making of an American Community, p. 446.Google Scholar