1 Bendix, R., Nation Building and Citizenship (Berkeley: University of California Press, 1977), 90.Google Scholar

2 Three important landmarks in the vast literature concerning the role of the franchise in the development of citizen-state relations are Mill, J. S., Considerations on Representative Government (New York: Henry Holt, 1873);Google ScholarMarshall, T. H., Citizenship and Social Class (Cambridge: Cambridge University Press, 1950):Google Scholar and Rokkan, S., Citizens, Elections and Parties: Approaches to the Comparative Study of Processes of Development (New York: McKay, D., 1970).Google Scholar

3 Ibid.

3 Ibid., 78.

5 Adam Przeworski has pointed out (in conversation) that the “lumpiness” of the enfranchisement process can be attributed, in part, to the fact that suffrage extensions involve the political incorporation of social groups. For a discussion of the importance of exogenous threats to stability, see Przeworski's “Institutionalization of Voting Patterns, or Is Mobilization the Source of Decay?” American Political Science Review 69 (1975), esp. 58.

6 A discussion of the piece-wise character of suffrage extension can be found in Converse, P., “Of Time and Partisan Stability,” Comparative Political Studies 2 (1969), 139-71;Google Scholar and Trevelyan, G. M., History of England, Vols. 1,2 and 3 (New York: Doubleday, 1953).Google Scholar

7 Rokkan, , Citizens, Elections and Parties, 31.Google Scholar

8 Bendix, , Nation Building and Citizenship, 168-69. Emphasis added.Google Scholar

9 For a discussion of the Scandinavian cases see Deny, T. K., A History of Modern Norway 1814–1972 (Oxford: Clarendon Press, 1973)Google Scholar, 51; Verney, D. V., Parliamentary Reform in Sweden 1866–1921 (Oxford: Clarendon Press, 1957);Google Scholar and Bendix, , Nation Building and Citizenship, 144, n. 56.Google Scholar

10 See Griffin, E. G., “The Adoption of Universal Suffrage in Japan” (unpublished doctoral dissertation, Columbia University, 1965), 172-73;Google Scholar and Butler, D. E., The Electoral System in Britain Since 1918 (Oxford: Clarendon Press, 1963), 35 and 135ff.Google Scholar

11 Much of this section is devoted to a formalization of extant discussions of the role of participation in augmenting legitimacy. Exemplary of this tradition is E. E. Schattschneider's argument in his book, The Semi-sovereign People: A Realist's View of Democracy in America (New York: Holt, Rinehart and Winston, 1960), 112: “A greatly expanded base of popular participation is the essential condition for public support of government. This is the modern problem of democratic government. The price for support is participation” (emphasis added). Other examples of the use of this economic metaphor may be found in Seymour, C. and Friary, D., How the World Votes, Vol. 2 (Springfield, Mass.: Nichols, C. A., 1918), 56;Google Scholar and Board, J., The Government and Politics of Sweden (Boston: Houghton Mifflin, 1974), 31.Google Scholar

12 For a survey of the literature on the relation between legitimacy and enfranchisement see Freeman, J. R., “Franchisement and Political Stability: A Historical-Formal Analysis“ (unpublished doctoral dissertation, University of Minnesota, 1978).Google Scholar

13 The reason for formulating the supply condition in this way will become apparent momentarily. Each supply line in Figure 1 is of the form E₁ = S₁L₁+η₁ where η₁, is the E-intercept at time t₁, that is, the point where S₁ (theoretically) crosses the E axis. (In actuality, of course, L is never zero.) The equation can be rearranged to read L₁=E₁/s₁–η₁/s₁ from which it follows that dL₁/dE₁=lS₁ at time t₁.

14 See Seymour, and Friary, , How the World Votes, Vol. 2, 61, 81;Google ScholarBendix, , Nation Building and Citizenship, 88;Google ScholarLijphart, A., The Politics of Accommodation (Berkeley: University of California Press, 1968).Google Scholar

15 An instance of this is the discussion of Thrane's agitation in Norway, in Derry, A History of Modern Norway 1814-1972, 41.

16 Verney, , Parliamentary Reform in Sweden 1866–1921, 64.Google Scholar

17 A detailed discussion of Gladstone's views on the suffrage is provided in Lambert, J., “Parliamentary Franchises Past and Present,” Nineteenth Century 26 (1889), 954.Google Scholar For a description of elite attitudes on this issue see Butler, , The Electoral System in Britain Since 1918.Google Scholar

18 Ostrogorski, M., Democracy and the Organization of Political Parties, Vol. 1 (London: Macmillan, 1902), 99.Google Scholar

19 For a discussion of this problem see Klingman, D., “Temporal and Spatial Diffusion in the Comparative Analysis of Social Change,” American Political Science Review 74 (1980), 123-37.CrossRefGoogle Scholar

20 Cortés, F., Przeworski, A., and Sprague, J., Systems Analysis for Social Scientists (New York: John Wiley, 1974), 261.Google Scholar See also Cortés, F. and Przeworski, A., “Sistemas partidistas, movilizacion electoral y la estabilidad de sociedadas capitalistas,” Revista Latinoamericana de Ciencia Politico 2 (1971), 220-41;Google Scholar and Przeworski, A., “Institutionalization of Voting Patterns.“Google Scholar

21 The series of events which precipitated the five major British and three major Swedish suffrage reforms are discussed at length in Freeman, “Franchisement and Political Stability: A Historical-Formal Analysis.”

22 For example, suppose that at t₁, elites demand L₁*(t) in Figure la and consider the magnitudes of reforms which would occur as a result of a crisis at some later time. If the next major crisis occurred at t₄, it would result in an increase in the enfranchisement level to E*₄ whereas if the crisis did not occur until t₆, then a larger suffrage reform equivalent to E*₆ would obtain. Clearly, then, we have (from the diagram) the inequality E*₆–E*₁ >E*₁–E*₁. Thus the comparatively larger size of the 1906 Finnish reform can be understood in terms of the same underlying political process which accounts for the smaller reforms experienced in such countries as Britain. Legitimacy crises did not occur as regularly in the Finnish case as in most other West European countries.

23 Say we set the demanded legitimacy at L₁ in Figure la, that is, we extend the legitimacy demand line upward so that it intersects all nine legitimacy supply lines. We then see that the upward shift in the legitimacy supply lines means a small change in E(t) is needed to maintain L₁ even though elites do not demand greater legitimacy (L(t) is fixed at L₂). Of course, the magnitude of these changes often will be quite small. Hence the marginal alterations in E are not likely to arrest the buildup of legitimacy discrepancies. A more thorough diagrammatic analysis of this phenomenon is carried out in Freeman, J. R. and Snidal, D., “Franchisement and Legitimacy: A Diagrammatic and Empirical Analysis“ (unpublished manuscript, July 1979).Google Scholar

24 Of course, demographic or economic forces (for example, deflation) also can work in the reverse direction, retracting the franchise. For a discussion of this possibility see Lambert, “Parliamentary Franchises,” 948-49; and Pole, J. R., “Suffrage and Representation in Maryland from 1776 to 1810: A Statistical Note and Some Reflections,”Google Scholar in Silbey, J. H. and Sveeney, S. T. (eds.), Voters, Parties and Elections (Lexington, Mass.: Xerox College Publishing, 1972), 61–71.Google Scholar

25 See Cortes and Przeworski, “Sistemas Partidistas,” 225.

26 Lijphart, The Politics of Accommodation, 107.

27 Seymour, C., Electoral Reform in England and Wales (Hamden, Conn.: Archon Books, 1970), 486Google Scholar, emphasis added. See also Ostrogorski, Democracy, 45.Google Scholar

28 Since E(t) is the proportion of the adult population eligible to vote at time t, the required constraint is 0≤E(t)≤l for all t.

29 A few comments should be made about the substantive interpretation of the interactive term in this equation, [1 – E(t)]. Consider the factors which play a role in determining the relative rate of growth of L*_p which can be expressed as (dL*_p/dt)/L*_p. On the one hand, those factors contributing to the increasing need for state legitimacy can be assumed to be exogenous to the process of enfranchisement and are embodied in the parameter g which captures the overall growth in required legitimacy. On the other hand, the factors which inhibit elites from relying on enfranchisement as a means of gaining legitimacy are more closely tied to their perceptions of the costs of extending the franchise. First, elites were concerned that alterations in voting rights would destabilize their base of electoral support. The importance of this concern varied directly with the relative rate of growth in the franchise at any point in time (that is, (dE_t/dt)/E_t). Second, elites believed that successive franchise extensions incorporated progressively less qualified and less desirable elements of the population into the electorate (groups of people who were less likely to support the prevailing order at the polls). This concern about the “allegiance” of newly enfranchised voters can be represented by the ratio of enfranchised to disenfranchised adults at time t, E/(l –E), so that as this quotient increases so does elite concern about the political reliability of new groups of voters. Combining these dual concerns regarding the relative size and allegiance of newly enfranchised groups of voters into the overall cost of legitimacy through enfranchisement, we can reformulate the demand for increasing suffrage legitimacy net of these same costs as (1') (dL*_p/dt)/L*_p=g – [E/(l – E)][(dE/dt)/ E] Solving for LJ in (1') gives (1) which together with (2) and (3) implies (5). Hence this substantive interpretation of elite preference preserves the upper bound corresponding to universal suffrage; (1') introduces the growth constraint in a logically equivalent fashion.

30 This discontinuous movement of L(t) (even as L*(t) moves continuously) was represented in Figure la. However, the present formulation differs from that depicted in Figure la since L*(t) does not continue to grow without bound but rather the rightward shift of the L*(t) lines slows as E(t) approaches unity.

31 Just as equation 1 was shown to embody the causes of growth in L*_p (footnote 29), equation 2 can be shown to capture the factors which are responsible for the growth in L_s. To see this, consider the expression for the relative rate of growth in legitimacy supply: (2') (dL_s/dt)/Ls=-r+[(dE/dt)/E] The first term on the right side of (2'), –r, represents the steadily declining rate of legitimacy supply due to the diffusion and development of democratic expectations. And the second term embodies the increase in legitimacy supply that is an outgrowth of the relative magnitude of suffrage reform. Thus, it is possible to interpret our representation of the shift in legitimacy supply schedules in terms of factors which determine citizen's willingness to support the state in return for voting rights.

32 The representations in equations 1 and 2 assume constant rates of increase in both the legitimacy level which elites prefer and also in the decline in the slopes of the legitimacy-supply lines. Since we have no a priori reason to believe these are the best assumptions, the empirical accuracy of some alternative formulations of L*_p and L_s were studied in conjunction with a variant of the present model (see J. R. Freeman and D. Snidal, “Franchisement and Legitimacy”). These alternative representations allowed for both a time variant rate of growth in legitimacy preference and a time variant rate of decline in the slope of the supply lines. Two of these representations were found to yield implausible results while statistical criteria permitted us to determine that the model used in the text was the best of the two remaining candidates. This result held up under the various refinements of the model discussed below. Although the model based on constant rates of growth and time invariant decline has shortcomings (for example, it is not possible to identify the r and g paramters individually), we can be relatively certain that (4) is a reasonable repre sentation of the processes of enfranchisement in our sample of West European countries.

33 Technically speaking, (5) is not identical to the standard logit formulation because here P(t) is defined only for positive time, t>0. This particular transformation is employed so as to allow linear estimation of the coefficients. Our use of least squares regression and various statistical tests requires assumptions of normally distributed and independent errors on P(t). Given the transformation from P(t) back to E*(t), this is equivalent to assuming that deviations from the equilibrium path of enfranchisement will be greatest at intermediate values of time (for example, 1850-1920).

34 See Marshall, , Citizenship and Social Class; and Butler, , The Electoral System in Britain Since 1918, xxiii.Google Scholar

35 See Rokkan, , Citizens, Elections and Parties, 84–85.Google Scholar

36 For example, the 1849 Dutch reorganization was omitted because no major revision of the 1815 Dutch Constitution was enacted; elections simply were made direct rather than indirect. See Lijphart, , The Politics of Accommodation, 77.Google Scholar

37 For a detailed discussion of these and other measurement issues, see Freeman, J. R. and Snidal, D., “Franchisement and Legitimacy.”Google Scholar

38 Rokkan's discussion suggests that universal suffrage is the situation in which men and women are enfranchised on an equal basis. Our diagrammatic argument defines universal suffrage as the circumstance where the entire adult population possesses the right to vote (E=E*=1). If we analyzed only those extensions included in Rokkan's typology, then, we would limit ourselves to an examination of major suffrage reforms and ignore the numerous revisions of the suffrage enacted between 1920 and 1970. However, as we noted at the end of the previous section, it is possible that our model is not applicable in this period when franchise extensions alone would not eliminate legitimacy discrepancies. For this reason, we investigate the validity of the model both when these late reforms are and are not included in the sample. The fact that the parameter estimates are essentially the same in each case suggests that the model accounts for late reforms as well as for earlier ones.

39 The French Revolution, in particular, represents a major turning point in the development of West European democracy. For instance, Bendix argues that “the French Revolution brought about a fundamental change in the conception of representation: the basic unit was no longer the household, the property, or the corporation, but the individual citizen...” (Nation Building and Citizenship, 114).

40 It is well known that a diffusion curve like that in equation 4 will always fit a single country's history quite well (for example, see Feller, W., An Introduction to Probability Theory and Its Applications [2nd ed.; New York: John Wiley, 1966]Google Scholar). But there is no reason why the combined sample of nine different suffrage reform histories should conform to the theoretical expectation as well as they do. Thus the striking empirical result is that the model is able to explain the histories of enfranchisement in all nine countries simultaneously.

41 There is evidence that the Italian war experience constituted a very strong and peculiar interruption of the democratization process. See P. Converse, “Of Time and Partisan Stability.”

42 Residuals were scrutinized by visual inspection of plots, by a quasi-jacknifing method of checking the stability of coefficients when the model is reestimated excluding each data point in turn and by the use of the “hat matrix” to investigate the influence of particular data points. Excepting the Italian reform of 1945, no data point was found by any of these methods to markedly affect the estimates reported here. The robustness of the model estimates (with respect to outliers) is evidenced by the fact that when any one of the other forty points is excluded, none of the coefficients change by even as much as two-thirds of its associated standard error. For a discussion of these methods see F. Mosteller and J. Tukey, “Data Analysis Including Statistics,” in Lindzey, G. and Aranson, E. (eds.), The Handbook of Social Psychology, Vol. 2 (Menlo Park, Calif.: Addison-Wesley, 1968);Google Scholar and Hoaglin, D. and Welsch, R., “The Hat Matrix in Regression and ANOVA,” The American Statistician 32 (1978), 17–22.Google Scholar

43 The constant in the regression equation now represents the initial conditions in the “protracted absolutist” countries and the coefficients for the dummies represent deviations above or below this level.

44 In interpreting these coefficients, the reader should remember that the estimates are –ln(C) not C. Accordingly, the more negative the estimated constant, the larger is the estimate of C in equation 5. Again, it is important to keep in mind that the estimate of the constant for each group of countries is the sum of the overall regression constant and the dummy for that particular “inherited style of rule.” For example, in the case of city oligarchic and absolutist heritage countries in column three, –ln(C)= –5.96 +(–1.19)= –7.15.

45 Seymour, C., Electoral Reform in England and Wales.Google Scholar

46 Some minor heteroskedasticity was detected in the residuals against time but was judged not to pose a problem. The results were also checked for autocorrelation (between present error and the error at the last major reform for the same country) but none was detected.

47 A second cluster of reforms can be seen in the aftermath of the Second World War. Major reforms occurred in Sweden, Finland, Norway, Austria and the Netherlands between 1944 and 1946 and in Belgium in 1949. However, no shift was detected in the equilibrium legitimacy path at this point suggesting that the turmoil of the war caused states to adjust the franchise but caused no change in the underlying demand and supply conditions of legitimacy. The only other clustering of reforms in one period occurred with five countries experiencing major reforms between 1881 and 1887. However, this grouping appears to have no better explanation than coincidence since there is evidence neither of a shift in the underlying legitimacy conditions nor of any particular cause for reforms to have occurred at this time. Finally, the “hat matrix” (see footnote 42 above) suggested that the pre-1848 data points might be affecting the estimates, but dropping these points from the sample did not affect the results.

48 Morgan, D., Suffragists and Liberals: The Politics of Women’s Suffrage in England (Totowa, NJ: Rowman and Littlefield, 1975), 149-50.Google Scholar

49 This excludes Finland which had only one case in the small sample and two in the large sample.

50 The definition and measurement of suffrage reforms in absolutist heritage countries (Italy, Germany, and Austria) is especially problematic. Major electoral inequalities existed in Prussia and Austria. The Austrian Curiae and Prussian three-class system were qualitatively different from such inequalities as the Belgian plural voting. The processes of unification in Italy and Germany make it extremely difficult to obtain accurate adult population figures for those countries. For example, to construct such measures for the German Confederation it would be necessary to determine the adult population of 22 German states (or, alternatively, the adult populations of Baden Württemburg, Bavaria, and parts of Hesse-Darmstadt). Because both types of difficulties plague the German case and because there is evidence that the democratization of Germany differed somewhat from that of Italy and Austria, we decided to treat Germany as a separate case here. Most of the important French franchise extensions (and retractions) occurred before 1850, and the measurement of those suffrage reforms also is problematic. Estimates of French enfranchisement levels and adult populations are unreliable, but the most important reason for treating the French case separately is the fact that that country experienced several major enfranchisement reversals. An in-depth analysis of enfranchisement in France is very much needed.

51 See Anderson, E. N., Social and Political Conflict in Prussia 1858-1864 (Lincoln: University of Nebraska Studies, no. 12, 1954), chap. 8.Google Scholar

52 The three-class system provided for voting according to the proportion of taxes paid in each electoral district; the system was highly inequitable (see Anderson, E. N. and Anderson, P. R., Political Institutions and Social Change in Continental Europe in the Nineteenth Century [Berkeley: University of California Press, 1967\, 316).Google Scholar This three-class system had been employed in Baden and in the ussian Rhineland prior to 1848. Indeed, one of the problems in analyzing the German case is that each state has a different history of democratization dating back to Stein and Hardenburg’s liberalization of the municipal franchise in 1808.

53 See Anderson, and Anderson, , Political Institutions and Social Change, 330-31;Google Scholar and Seymour, and Friary, , How the World Votes, Vol. 1, chap. 17.Google Scholar

54 Thus David Thomson emphasizes “the accumulative progress in the assertion of constitutional rights and civil liberties.” He goes on to point out that “even the ‘personal government’ of Napoleon III was forced to retain universal male suffrage and an elected assembly, although the democratic functioning of these institutions was foiled by governmental pressure and management of elections…” (Democracy in France: The Third and Fourth Republics [3rd ed.; London: Oxford University Press, 1960], 17).Google Scholar

55 Of course, there is some question as to whether the franchise had any meaning under Napoleon Bonaparte’s rule. Be that as it may, the suffrage retraction of 1815 clearly represented a severe case of disenfranchisement, especially when compared to the optimal enfranchisement curve for the continuous representation and protracted absolutist countries.

56 See Trevelyan, , History of England, Vol. 3, 173.Google Scholar

57 After the Opposition made great gains in the elections of 1830, Metternich and Tsar Nicholas advised Charles to make certain concessions to his opponents. Instead, Charles responded by restricting the press, dissolving the newly elected Chamber, and altering the electoral system so as to change the number of deputies and reduce the number of eligible voters from about 90,000 to 25,000. Seymour, and Friary, , How the World Votes, Vol. 1, 341-42.Google Scholar

58 Castlereagh’s advice had to do with what he called “the principle of exclusion.” In Castlereagh’s opinion, “Tyrants [might] poison or murder an obnoxious character, but the surest and only means a constitutional sovereign has to restrain such a character is to employ him. Office soon strips him of his most dangerous adherents–he becomes unpopular, can be laid aside at pleasure, and sinks to his true lead.“ See the letter from Castlereagh to Sir Charles Stuart of May 8,1815 reprinted in Webster, C., The Foreign Policy of Castlereagh (London: G. Bell and Sons, 1963), 545-48.Google Scholar

59 Charles, ’ actions are explained by Seymour and Friary (How the World Votes, Vol. 1, 342)Google Scholar in the following way: “There is no clearer instance in modern history of the ancient saw, that whom the gods would destroy they first make mad.”

60 Thomson, , Democracy in France, 31.Google Scholar

61 In particular, it is important to consider elites’ desires to maintain control over governmental institutions and how that objective takes priority over the goal of legitimation. See Freeman, J. R., “The Logic of Franchisement: A Decision Theoretic Analysis,” Comparative Political Studies 13 (1980), 61–95;Google Scholar and Freeman, “Franchisement and Political Stability,” chaps. 3, 8.

62 In his article, “Finland,” in Rokkan, S. and Meyriat, J. (eds.), International Guide to Electoral Statistics (Paris: Mouton, 1969)Google Scholar, O. Rantala reports that prior to the 1906 reform, about 126,000 Finns possessed the right to vote. The discrepancy between E* and E therefore was about .75–.08=.67. Perhaps the existence of this large discrepancy from E* explains why the general strike of 1905 made such an impression on Finnish elites. See E. Allardt and P. Pesonen, “Structural and Non-Structural Cleavages in Finnish Politics,” in Lipset, S. and Rokkan, S. (eds.), Party Systems and Voter Alignments (New York: Free Press, 1967).Google Scholar

63 Przeworski, “Institutionalization of Voting Patterns”; see also Griffin, “Universal Suffrage in Japan.”

64 Converse, “Partisan Stability”; and Przeworski, “Institutionalization of Voting Patterns.”