¹ King, Willford Isbell, The Wealth and Income of the People of the United States (New York, 1917), pp. 175–77Google Scholar.

² Handlin, Oscar, Race and Nationality in American Life (Boston, 1948)Google Scholar; Novak, Michael, The Rise of the Unmeltable Ethnic (New York, 1971)Google Scholar; and Sowell, Thomas, Race and Economics (New York, 1975)Google Scholar.

³ Higgs, Robert, “Race, Skill, and Earnings: American Immigrants in 1909,” Journal of Economic History, 31 (June 1971), pp. 420–28CrossRefGoogle Scholar. The argument that Higgs accepts has been developed in much detail in Becker, Gary, The Economics of Discrimination (2nd ed.; Chicago, 1971)CrossRefGoogle Scholar.

⁴ Handlin, Race and Nationality; and Hourwich, Isaac, Immigration and Labor: The Economic Aspects of European Immigration to the United States (New York, 1912)Google Scholar.

⁵ U.S. Immigration Commission Report: Summary for Mining and Manufacturing (Washington, Government Printing Office), ch. II, p. 9Google Scholar.

⁶ Ibid. This chapter specified a two-stage method of sample selection: finding communities with many immigrants, and selecting from these communities where employment was concentrated in one or a few of the manufacturing industries in the scope of the survey (p. 10). The survey response rate was apparently high, and a time curtailment due to an order that all work be completed by July 1909 was achieved entirely by curtailing the collection of non-quantitative material (p. 10).

⁷ Handlin, Race and Nationality, pp. 104–31. Handlin's harsh criticisms concern not the statistics gathered and their reliability but misuses of the data in the Commission's summary report. Hourwich, in Immigration and Labor, also attacked the Commission's uses of its data rather than the reliability of the latter.

⁸ Immigration Commission, Abstracts of Reports, vol. 1, p. 38.

⁹ A detailed explanation of this type of discriminatory behavior is provided in Doeringer, Peter B. and Piore, Michael J., Internal Labor Markets and Manpower Analysis (Lexington, Mass.; 1971), ch. VIIGoogle Scholar.

¹⁰ See Barbara R. Bergmann, “Occupational Segregation, Wages, and Profits When Employers Discriminate by Rate and Sex,” paper prepared for the Project on the Economics of Discrimination, College Park, Maryland, 1970 (mimeographed).

¹¹ The Census of Manufactures contains data for the country as a whole, while the Immigration Commission was geographically selective. However, since the Commission's data seem quite representative of the situation in the northeast and north central regions where manufacturing was heavily concentrated, there should be little problem in combining these data sources.

¹² Reder, Melvin W., “Wage Differentials: Theory and Measurement,” in Aspects of Labor Economics (NBER, 1962), pp. 257–99Google Scholar.

¹³ Several studies using more recent data for manufacturing have uncovered a tendency for large firms to pay higher wages. The explanation generally offered for this phenomenon is the power of or the threat of unionization in these firms (see for example Stanley H. Masters, “An Interindustry Analysis of Wages and Plant Size,” Review of Economics and Statistics, 51, no. 3 [Aug. 1969], pp. 341–45). But since union membership was quite low at this time, there does not appear to be any inconsistency between our hypothesis and these other findings.

¹⁴ See for example, Brogan, Robert and Erikson, Edward, “Capital-Skill Complementarity and Labor Earnings,” Southern Economic Journal, 41 (July 1975), pp. 83–88CrossRefGoogle Scholar.

¹⁵ Higham, , Strangers in the Land: Patterns of American Nativism, 1860–1925 (New York, 1965)Google Scholar.

¹⁶ Hofstadter, The Age of Reform (New York, 1955). But see also Nugent, Walter T., The Tolerant Populists (Chicago, 1963)Google Scholar.

¹⁷ Quoted in Leinenweder, Charles, “The American Socialist Party and the ‘New Immigrants,’” Science and Society, 1968, pp. 8, 11Google Scholar.

¹⁸ All of the regression results shown using the ten-year residency variable support the finding of very little or no discrimination against Northwests using this same technique.

¹⁹ Balestra, Piero and Nerlove, Marc, “Pooling Cross-Section and Time Series Data,” Econometrica, 34 (July 1966), pp. 585–612CrossRefGoogle Scholar. Another method would be to use separate dummy variables for each industry. The BAN technique, however; is better for small samples since there are fewer parameters to estimate.

²⁰ Capital intensity and percentage of workers employed in large firms were only slightly correlated with one another so that we can analyze these effects independently.

²¹ U.S. Department of Labor, Annual Report of the Commissioner, 1890 and 1891. Washington, Government Printing OfficeGoogle Scholar.

²² The BAN procedure was not adopted here because it is primarily intended for small samples. The separate intercept method also removes the serial correlation problem, and the loss of efficiency associated with this procedure is negligible in so large a sample as we are now considering.

²³ Feller, Irwin, “The Draper Loom in New England Textiles, 1894–1914: A Study of Diffusion of An Innovation,” Journal of Economic History, 25 (Sept. 1966), pp. 320–47CrossRefGoogle Scholar, and “The Draper Loom in New England Textiles: A Reply,” Journal of Economic History, 27 (Dec. 1968), pp. 628–30Google Scholar.

²⁴ Gibb, George S., The Saco Lowell Shops (Cambridge: Harvard University Press, 1950), pp. 397–99, 410, 416–19, Appendix 10CrossRefGoogle Scholar; and Navin, Thomas, The Whitin Machine Works Since 1841 (Cambridge: Harvard University Press, 1950), pp. 217–22, 224–34, and 274–75CrossRefGoogle Scholar.

²⁵ Ibid.

²⁶ Navin and Gibb found that many southern mills paid for new machinery by common stock. The theme of capital shortage runs through southern mill histories, and Lance Davis and others have found wide regional differences in interest rates.

²⁷ McGouldrick, Paul, New England Textiles in the 19th Century: Profits and Investment (Cambridge: Harvard University Press, 1966), chapter 7Google Scholar.

²⁸ Computed by Shen, T. Y. for his Ph.D. dissertation, “A Quantitative Study of Production in the American Textile Industry, 1840–1940,” (Yale University, 1956), pp. 199–201Google Scholar.

²⁹ Census data are insufficient to pinpoint the change in immigrants (or Southeast immigrants) as shares of textile labor forces. But the 1909 Census establishes a level of immigrant males as 36 percent of all males in the industry; adjustment for the all-native-born southern branch would bring this to over a half of the New England branch male labor force. In the Immigration Commission sample, Southeast workers outnumbered Northwests by only one and a half times. Even then, however, they were slightly over a third of all workers. (1900 Census, Abstract, vol. 1, Table 5).

³⁰ Thomas Sowell, ibid., p. 160.

³¹ Friedman, Milton, A Monetary History of the United States (Princeton, 1963)Google Scholar. Arrow, Kenneth (“The Theory of Discrimination,” in Ashenfelter, O. and Rees, A., eds., Discrimination in Labor Markets [Princeton, 1974])Google Scholar concluded that capitalists could gain by discrimination as well as favored-group workers. But the gain in Arrow's model is psychic, not pecuniary, and all capitalists (or foremen) have to have the same taste for discrimination. If some are non-discriminators, they will force out the discriminators by being more efficient.

³² Higham shows a classless coalition in favor of unrestricted immigration rather than a Marxist type. In favor of unrestricted immigration were first-generation immigrants (despite exceptions noted above in the text), capitalist interests as exemplified by the National Association of Manufacturers, traditional middle-class upholders of America as an asylum for the oppressed, and idealistic members of the Left such as Eugene V. Debs. Opposed to unrestricted immigration were patrician rentiers exemplified by Henry Cabot Lodge and the Boston and New York-based Immigration Restriction League, the American Federation of Labor, much of the membership of the American Socialist Party and nativists who were often rural in origin.

³³ Quoted in Commons, John R., Races and Immigrants in America (New York, 1967), pp. 115–16Google Scholar.

³⁴ Our speculation is that shifts in the Southeast labor supply curve dominated movements along the supply curve in response to American discrimination, which would be consistent with the migration data if the supply elasticity were not too high. Using a model where the arguments were migration and wage rates in sending and receiving countries and migration flows, Jeffrey Williamson (“Migration to the New World: Long-Term Influences and Impact,” in Explorations in Economic History, (Summer 1974) did find lower labor supply elasticities in a less developed economy (Sweden) than a more developed one (England), which may be taken as consistent with our speculation. However, Williamson's parameters apply only to countries from which Northwests came and are thus not a direct test of our speculation, except for the lower supply elasticity just cited for Sweden. Moreover, his model is deficient in not including costs of migration, and he does not test whether the behavior parameters shifted over his 1870–1910 period. This may be defensible for the well-established North Atlantic routes over which Northeasts traveled, but it surely would not be defensible for the much longer and initially less developed routes taken by Southeasts. In any case, lack of reliable data on average wages paid in Russia, Austria-Hungary and Italy will preclude testing of Williamson's model for Southeast flows for some time yet.