This study was made possible by support from the National Science Foundation. We also acknowledge the aid of our able assistant, Richard H. Rowland, in the preparation of this paper, as well as the use of the computer services at Western Data Processing Center, University of California at Los Angeles, where we were assisted by George Diehr.

1 For example, differences in unemployment rates between areas were used in a study of labor mobility; see Makower, H., Marschak, J., and Robinson, H. W., “Studies in Mobility of Labour: A Tentative Statistical Measure,” Oxford Economic Papers, I (1938), 83–123 CrossRefGoogle Scholar; “Studies in Mobility of Labour: Analysis for Great Britain,” Pt. I, ibid., II (1939), 70-97; and Pt. II, ibid., IV (1940), 39-62. This approach has also been incorporated in a migration model by Somermeijer, W. H. (“Een Analyse van de Binnenlandse Migratie in Nederland tot 1947 en van 1948-1957,” Statistische en Econometrische Onderzoekingen (1961), pp. 115–74Google Scholar; English summary, p. 144).

2 In this paper the term “guberniia” is also used to refer to an oblast.

3 Tsentral'nyi statisticheskii komitet, Pervaia vseobshchaia perepis’ naseleniia rossiiskoi imperii, 1807 g. (St. Petersburg, 1899-1905)Google Scholar, Vypuski 1-89; and TSK, , Naselennye mesta rossiiskoi imperii v $00 i bolee zhitelei (St. Petersburg, 1905 Google Scholar; “Pervaia vseobshchaia perepis’ naseleniia rossiiskoi imperii, 1897 g.“). Data in the 1897 census are based on the de facto population.

4 The urban population in the 1897 census comprises the uezd and guberniia administrative centers, including their contiguous built-up areas, as well as a significant number of legal cities (zashtatnyi and bezuezdnyi). A number of centers that were urban in terms of function were not legal cities; for example, thirty-five centers with populations between 15, 000 and 41, 000 were not included in the urban population. These centers had a total population of 694, 674. Also very small uezd centers, which were actually little more than agricultural villages, were considered urban. We have determined the urban population living in all cities of 15, 000 or more according to the contemporary major economic regions of the USSR. This was necessary in order to compare population changes from 1897 to 1959 in conjunction with a study of internal migration. The larger study required the establishment of comparable territorial units within the USSR. The results were presented in our article, “Regional Population Changes in Russia and the USSR Since 1851,” Slavic Review, XXV, No. 4 (Dec. 1966), 663-68; and the methods are explained in detail in our monograph entitled Population Changes in Russia and the USSR: A Set of Comparable Territorial Units (San Diego: San Diego State College Press, 1966).

5 Pereselencheskoe upravlenie, Itogi pereselencheskogo dvizheniia za vremia s 1896 po XQOQ (St. Petersburg, 1910); and Itogi pereselencheskogo dvizheniia za vremia s 1010 po 1914 (St. Petersburg, 1916).

6 In the census tabulation of place-of-birth data by uezd, cities are considered as separate units. Consequently, intra-guberniia urban migration must include migrants into a city from the uezd in which the city is located, and intra-guberniia rural migration includes migrants who moved out across the boundary of any city within the uezd into a rural area.

7 Taeuber, Conrad and Taeuber, Irene B., The Changing Population of the United States (New York, 1958), p. 95.Google Scholar

8 Because this study is concerned with internal migration, data on migration to foreign countries were not considered. From 1860 to 1889 there was a net outflow of 1, 100, 000 people, and from 1890 to 1915, 3, 300, 000. The emigrants came predominantly from western Russia, and the largest single group was Jewish. The effect of our disregarding this emigration is to exaggerate slightly the percentage of out-migrants and in-migrants in a number of western guberniias. See Walter F. Willcox, ed., International Migrations (New York, 1931), II, 523.Google Scholar

9 These were St. Petersburg, Moscow, Petrokov (in Polish, Piotrköw; containing the city of Lodz [Łödź]), and Warsaw guberniias.

10 No breakdown into urban or rural out-migration was given in the 1897 census.

11 For a detailed treatment of this subject in English, see Pavlovsky, George, Agricultural Russia on the Eve of the Revolution (London, 1930), pp. 61–146 Google Scholar; and Robinson, Geroid T., Rural Russia Under the Old Regime (New York, 1949).Google Scholar

12 Pavlovsky, pp. 76-79.

13 For detailed discussion of rural migration to the eastern area, see George J. Demko, “The Russian Colonization of Kazakhstan, 1896-1916” (unpubl. Ph.D. diss.. Department of Geography, Pennsylvania State University, 1964); Skliarov, L. F., Pereselenie i zemleustroistvo v Sibiri v gody stolypinskoi agrarnoi reformy (Leningrad, 1962)Google Scholar; Treadgold, Donald W., The Great Siberian Migration (Princeton, 1957)CrossRefGoogle Scholar; and Iamzin, I. L., Pereselencheskoe dvizhenie v Rossii s momenta osvobozhdeniia krest'ian (Kiev, 1912).Google Scholar

14 Treadgold, pp. 67-68.

15 Skliarov, pp. 61-68.

16 Alexander Gerschenkron, “The Rate of Industrial Growth in Russia,” Journal of Economic History, VII (1947), 144-57.

17 Urban wages appear higher than agricultural wages. The average annual wage of workers in cotton manufacturing, one of the lower paid industries in Russia in 1897, was 180 rubles per year (see Liashchenko, Peter I., History of the National Economy of Russia to the 1917 Revolution [New York, 1949], p. 545 Google Scholar). The median annual wage for agricultural workers in forty-three of the European guberniias from 1882 to 1891 was 57 rubles (see Chuprov, A. I. and Posnikov, A. S., Vliianie urozhaev 1 khlebnykh tsen na nekotorye storony russkogo narodnogo khoziaistva, I [St. Petersburg, 1897], 176 Google Scholar). The agricultural wage refers primarily to the landless peasants and those with small allotments, who probably predominated in the rural to urban migration.

18 Unfortunately, the noncensus data indicating the standard of living for the agricultural sector (e.g., average annual peasant allotment) are available only for the fifty European guberniias. The eastern areas of Russia usually had higher proportions of rural migrants, and these were the areas with relatively more land. However, this part of the analysis is limited to European Russia.

19 The median proportion of the total in-migrants provided by these top five guberniias was 57 percent.

20 When comparing differences in these variables from census data for 1897, weare not examining the differences at the time the migrants moved. The data indicate merely where people were living in 1897 and where they were born, not the year they moved. A person may have been born in guberniia A and then have moved to guberniia B in 1890, where he was enumerated in 1897. At the time he left guberniia A, the socioeconomic characteristics were undoubtedly different from those in 1897. This is also true for guberniia B. However, if guberniia B had an increase in the proportion literate between 1890 and 1897, guberniia A probably also experienced an increase. Therefore the differences might remain fairly constant. Another possible distortion is that migrants themselves might influence the value of the variables. This bias, however, works in both directions and probably was not great, because in-migrants as a proportion of the total guberniia population amounted to less than 10 percent in about three-fourths of the guberniias. Still another problem is that differences alone do not take into consideration other factors, such as whether an area was newly developing. If so, although it might not have been highly industrialized, its potential might have attracted many migrants. Such limitations are found in any single statistical measure. What the top-five method does show, however, is the net movement as of a certain date and the difference in the socioeconomic characteristics as of that date.

21 The three groups are defined below for the total and rural population. Total population Percent of in-migrants Number of guberniias Low 0-4.99 33 Medium 5-9.99 29 High 10 and over 27 Rural population Percent of in-migrants Number of guberniias Low 0-2.99 28 Medium 3-5.99 33 High 6 and over 28 An urban group is not given because it would be devoid of any significance. In the total category there is no problem—people have left the guberniia of their birth and are now living in another. For the rural category, it can be assumed that very few people move from urban to rural areas. Therefore, this grouping generally shows the movement of people from the rural area of one guberniia to the rural area of another. For the urban category, people enumerated in an urban area may have come from the urban area in another guberniia as well as from its rural area, and probably the rural out-migrants dominated. The urban category would have value for comparison only if we could compare migrants from the urban area of one guberniia with the urban area of another, but for this wc lack data.

22 As noted earlier, we are relating only economic differences by guberniia. Obviously other factors, such as cultural differences, distance, and availability of transportation, also influence migration and will vary by guberniia. The effect of economic factors alone on migration cannot be isolated, and, as a consequence, the observable relationship between the two, the proportion of migrants and income variables, will be weakened. However, if it can be assumed that each of these noneconomic factors in each of the three groups of guberniias has a somewhat random distribution, the effect on migration of each of these factors will tend to be offset by one another within the same group.

23 Because of the multinational character of the Russian Empire, cultural differences significantly influenced the movement of people. The most mobile group was the Eastern Slavs. However, even the Eastern Slavs migrated primarily to areas where they predominated, that is, the southern steppe, Siberia, and the North Caucasus. Of course, these were also areas where there was a surplus of agricultural land. In Stepnoi Krai, however, where there was surplus land, Slavs gradually displaced the native Kazakhs. In areas not densely settled by Eastern Slavs, the influence of culture was more pronounced. Central Asia, the Transcaucasus, and the Polish guberniias had little unused land. Of the million out-migrants from the Polish guberniias, 68 percent went to other Polish guberniias and 13 percent went to adjacent guberniias in European Russia where there was a significant Polish minority. Also, intra-guberniia migration was significantly higher than the empire average in the Polish guberniias. Out-migration in the Transcaucasus was much less than in the Polish guberniias, but 66 percent of the 163, 754 out-migrants went to other guberniias in the Transcaucasus. In Central Asia, excluding Stepnoi Krai, out-migration and intra-guberniia migration were very low, perhaps reflecting the influence of the tribal social organization. Of the 28, 608 out-migrants, about half went to guberniias within Central Asia.

24 For example, if two variables such as in-migration as a proportion of the total population and percentage of the labor force in manufacturing are to be compared, the values of each variable for each of the 89 guberniias would be ranked from the highest to the lowest, and the statistical agreement of these two ranks would be measured by Spearman's Rank Correlation Coefficient formula, and the significance of the relationship could be tested. A correlation with a plus sign would denote a positive relationship, and a minus, a negative one. It should be kept in mind that the correlation results merely support or do not support hypotheses, and thus do not definitively prove or disprove the existence of a causeand- effect relationship.

25 Rank correlation is used rather than more complex techniques such as standard correlation or regression, because the top-five method is based on a ranking of differences in the values for the variables and the use of the median difference. Thus, throughout we have used an ordinal measure. Such a measure is adequate for our purposes. Given the limitations of the data, it appears that other techniques such as multiple correlation, while not detracting from the analysis, would not provide any additional insights. In addition, we merely want to determine the direction of movement rather than the value for any coefficients associated with the variables in any complex relationship. Such a precise model is not warranted by the data.

26 Unless otherwise indicated, “rural” and “urban” hereafter refer to the definition of the 1897 census.

27 To determine whether there were significant regional variations in the statistical relationships, data were grouped by guberniia into the following categories: chernozem, nonchernozem, Central Asia, Transcaucasus, Moslem areas, and European Russia. Selected variables were correlated according to these groupings; the correlations were generally not significant or were lower than those obtained from all eighty-nine guberniias. It can be inferred that within these groupings variables were not as closely related as between them.

28 This is so for the following reasons: first, the top-five method is based on only about 50 percent of the in-migrants; and, second, median differences for the three groups used in the top-five method might increase with the proportion of in-migrants, and yet the corresponding rank correlation based on all guberniias might not be statistically significant.

29 Iamzin, I. L. and Voshchinin, V. P., Uchenie o kolonizatsii i pereseleniiakh (Moscow, 1926), p. 39 Google Scholar.

30 Iamzin, I. L., Pereselencheskoe dvizhenie v Rossii s momenta osvobozhdeniia krest'ian (Kiev, 1912), p. 181.Google Scholar

31 William Leasure, J. and Lewis, Robert A., “Internal Migration in the USSR: 1897-1926,” Demography, IV, No. 2 (1967), 479–96.CrossRefGoogle Scholar