¹ According to Schattschneider, E. E., ‘The nature of the nominating procedure determines the nature of the party; he who can make the nomination is owner of the party.’ Party Government (New York: Holt, Rinehart and Winston, 1942), p. 64.Google Scholar

² Epstein, Leon, Political Parties in Western Democracies (New York: Frederick A. Praeger, 1967). P. 220.Google Scholar For an account of candidate selection in England see Ranney, Austin, Pathways to Parliament (Madison: University of Wisconsin Press, 1965)Google Scholar, and Jackson, Robert J., Rebels and Whips (New York: St Martin's Press, 1968).Google Scholar

³ Epstein, , Political Parties in Western Democracies, p. 225.Google Scholar Our knowledge of candidate selection beyond the Anglo-American context is rather spotty; but the bits of information available certainly confirm Epstein's assertion. For Austria see Shell, Kurt L., The Transformation of Austrian Socialism (New York: State University of New York Press, 1962), pp. 115–18.Google Scholar For Germany see Kitzinger, U. W., German Electoral Politics (Oxford: Clarendon Press, 1960), pp. 57–79Google Scholar; and Loewenberg, Gerhard, Parliament in the German Political System (Ithaca: Cornell University Press, 1967), pp. 67–72.Google Scholar For Italy see Chasseriaud, J. P., Le Parti Dimocrate Chretien en Italie (Paris: Fondation Nationale de Science Politique, 1967), pp. 214–22.Google Scholar For Norway see Valen, Henry and Katz, Daniel, Political Parties in Norway (Oslo: Universitetsforlaget, 1964), pp. 21 and 29.Google Scholar

⁴ See Michels, Robert, Political Parties, trans. Eden, and Paul, Cedar (Glencoe: Free Press, 1949).Google Scholar

⁵ For excellent discussions of the impact of the primary on the American political parties see Key, V. O. Jr., Politics, Parties and Pressure Groups, 4th edn. (New York: Thomas Y. Crowell Company, 1964)Google Scholar, Chaps. 14 and 16 and American State Politics: An Introduction (New York: Alfred A. Knopf, 1956).Google Scholar A number of studies have demonstrated how under certain conditions party leaders manage to retain control over nominations despite the direct primary. See, for example, Sorauf, Frank J., Party and Representation: Legislative Politics in Pennsylvania (New York: Atherton Press, 1963), pp. 52–6Google Scholar; Seligman, Lester G., ‘Political Recruitment and Party Structure: A Case Study’, American Political Science Review, LV (1961), 77–86CrossRefGoogle Scholar, and Snowiss, Leo M., ‘Congressional Recruitment and Representation’, American Political Science Review, LX (1966), 627–39.CrossRefGoogle Scholar

⁶ Each of these parties is fragmented along linguistic lines; their French names are Parti Social Chrétien, Parti Socialiste Belge and Partide la Libérté et du Progrès. The Flemish equivalents are as follows: Christelijke Volkspartij, Belgische Socialistische Partij, and Partij voor Vrijheid en Vooruitgang. The CSP and BSP were established immediately after World War II. In 1961, the PLP replaced the Liberal Party, the oldest Belgian party which dated back to the 1840s. Until the early 1960s, these three traditional parties consistently won more than 90 per cent of the popular vote in the elections for the Chamber; in the 1954,1958 and 1961 elections, they won a combined vote of 92.7, 96.0 and 90.5, respectively. In the last three elections, however, several minor parties, representing the interests of Belgium's two major linguistic communities, have cut into the electoral monopoly previously enjoyed by the three largest parties. Thus, the total vote for these parties has dropped to 84.1 in 1965, 8o.6 in 1968 and 73.7 percent in 1971.

⁷ There are a number of excellent studies describing the operation of the Belgian primaries. Among the most complete studies are Debuyst, Frédéric, La Fonction Parlementaire en Belgique: Mécanismes d'Accès et Images (Brussels: Centre de Recherche et d'Information Socio Politiques, 1967), pp. 165–280Google Scholar and DeWachter, Wilfried, De Wetgevende Verkiezingen Als Proces Van Machtsverwerving In Het Belgische Politiek Bestel (Antwerp: Standaard, 1967), pp. 110–66.Google Scholar One of the few accounts of candidate selection in Belgium available in English is Wachter, Wilfried De, ‘The General Elections as a Process of Power Achievement in the Belgian Political System’, Res Publica, XI (1967), pp. 369–412.Google Scholar

⁸ It should be noted that there are other isolated cases of primaries among European mass parties. Most notable among these is the case of the Swedish Socialist Party which until 1949 held primaries in isolated cases. See Andren, Nils, Modern Swedish Government (Stockholm: Almqvist & Wiksell, 1961), p. 33.Google Scholar

⁹ By leaders, I mean the members of particular national and constituency committees. For the CSP these committees include the Flemish and Walloon National Committees and for each constituency the Committees of the Arrondissement. For the BSP these committees include the national Bureau and Central Committee and the constituency Bureaus and Central Committees. For the PLP these committees include the Political Committee and the Permanent Bureau at the national level and equivalent committees in each constituency. For a detailed discussion of the exact composition of these committees see Centre de Recherche et d'lnformation Socio-Politiques, Série Documentaire, No. 6.

¹⁰ The information for this article was gathered through sixty personal interviews with constituency association leaders of the three major parties as well as a perusal of the detailed press reports of the selection proceedings. The interviews, which were conducted between May and November 1967, consisted primarily of open-ended questions designed to collect information on how candidates were selected in each of the constituencies.

¹¹ That the CSP held more primaries than its opponents may be related to the fact that, through 1965, the chances that an association organized a primary varied directly with the number of deputies it expected to elect. Thus for the 1961 and 1965 elections, 81 per cent of all associations expecting to elect three or more deputies held primaries. But in the same years only 33 per cent of those associations that hoped to elect one deputy organized primaries and only 5 per cent of those which did not expect to elect any deputies bothered to poll their members. Hence, primaries were organized quite frequently in the CSP for many of its associations elected three or more deputies. Primaries were organized less regularly in the BSP for somewhat fewer or its associations elected as many as three deputies. And primaries were rarely held in the Liberal/PLP because so few of its associations elected even two deputies. The link between the selection procedure and the number of deputies elected may be explained in terms of the dynamics of the selection process in the various associations. In those associations which have no hope of electing even one deputy, the designation of candidates engenders little interest among party members and even less competition among prospective nominees. Often association leaders must search for party members willing to have their names placed on the list. Given the general indifference among party members and aspirants about who will occupy the hopeless places on the list, party leaders are understandably not eager to expend the time, money and energy to poll the members’ preferences. When candidates have no chance for electoral success, a primary becomes superfluous. The reluctance of Belgian parties to hold primaries under such circumstances is roughly equivalent to the uncontested primaries among minority parties in the United States. It is only when an association expects to elect at least several deputies that there is sufficient competition among aspirants and concern among the rank-and-file to warrant the expense and time required to hold a primary.

¹² Seats are distributed among the candidates on each of the party's lists in the following way. Those candidates who first reach the ‘eligibility figure’ (calculated by dividing the party's total vote with the number of seats it won plus one) are elected. Normally, party officials add the number of personal votes won by the candidate at the head of the list to that proportion of the list vote necessary to assure that he reaches the ‘eligibility figure’. This procedure is repeated for the candidate situated in the second place on the list and so forth until all the party's seats have been allocated. If the list votes are depleted before all the seats have been assigned, the next seat is accorded to the candidate with the most personal votes. According to this procedure, if no list votes are cast, the personal preferences indicated by the voters will determine who among the candidates offered by the party are elected. As the proportion of the list votes increases, the influence of the party over the recruitment of legislators is enhanced.

¹³ The relationship between a list electoral system and the role of parties in the selection of candidates was developed by Duverger, Maurice, Political Parties, trans. Barbara, and North, Robert (New York: Wiley, 1963), p. 368.Google Scholar Dividing the positions on the list among the safe, combative and hopeless places is especially common when the allocation of the popular vote among the parties remains fairly stable over an extended period. This was the case in Belgium for the first seven legislative elections following the War. But in the past three elections, it has been somewhat more difficult for the parties to determine which of the positions are safe since they have been compelled to anticipate substantial erosions in their popular support. Nevertheless, they have continued to make such estimations.

¹⁴ According to a survey conducted just after the 1968 legislative elections, the subjective class identifications of CSP voters were as follows: working class 52.6 per cent; middle class 39.1 per cent; bourgeois 3.8 per cent; not identified 4.6 per cent. The same survey indicated the following distribution of occupations among CSP voters: workers 32.6 per cent; whitecollar employees 24.5 per cent; farmers 15.0 per cent; artisans and shopowners 14.1 per cent; top level executives 8.7 per cent. See Delruelle, Nicole, Evalenko, Rene and Fraeys, William, Le comportement politique des electeurs Belges (Brussels: Editions de l'institut de Sociologie, 1970), p. 50.Google Scholar

In addition to the affiliates who belong directly and solely to the party, each constituency association draws its leaders as well as its members from (a) agricultural groups (the Boerenbond in Flanders and the Alliance Agricole Beige in Wallonia); (b) middle-class groups (Nationaal Christelijk Middenstandverbond NCMV and Mouvement Chretien des Independants et des Cadres MIC in Flanders and Wallonia respectively); (c) working-class groups (Algemeen Christelijk Werkersverbond ACW in Flanders and the Mouvement Ouvrier Chretien MOC in Wallonia); and (d) several other business, youth and female groups. Most of these organizations are quite autonomous with many having their own members, officers, headquarters, and publications. These social organizations have a larger following and are organizationally more developed in Flanders than in Wallonia. Because these groups are so well organized in Flanders, conflict among these groups has not been especially severe. In contrast, conflict among the various factions, and especially between groups representing the working and middle classes, has been particularly bitter among the Walloon associations. For discussions of the factions in the PSC see Debuyst, , La Fonction Parlementaire, pp. 211–33Google Scholar; Meynaud, Jean, La Décision Politique en Belgique (Paris: Fondation Nationale des Sciences Politiques, 1965), pp. 49–58Google Scholar; and Goffart, Vincent, ‘La Démocratie Chrétienne en Wallonie’, La Revue Nouvelle, XLVIII (15 02 1966), 113–30.Google Scholar

¹⁶ Just as Belgian political observers have been concerned about whether primary electorates are microcosms of the party's supporters in the general electorate, so American commentators have toyed with this problem. Key carefully explored the biases and their possible consequences in primaries for state offices; see American State Politics: An Introduction, pp. 145–64. Subsequently, several challenges to some of Key's conclusions have been offered; see Ranney, Austin, ‘The Representativeness of Primary Electorates’, Midwest Journal of Political Science, XII (1968), 224–38CrossRefGoogle Scholar and ‘Turnout and Representation in Presidential Primary Elections’, American Political Science Review, LXVI (1972), 21–37.Google Scholar

¹⁷ This analysis of the Brussels primary is based on interviews with the secretaries of the constituency associations, electoral data provided by the Ministry of the Interior and ‘Les polls préparatoires aux élections législatives du 26 mars 1961’, Courrier Hébdomadaire, 98 (3 March 1961) and ‘Les polls préparatoires aux élections législatives de mai 1965’, Courrier Hébdomadaire, 279 (26 March 1965).

¹⁸ The gap between the francophones and Flemings was widened even further after the 1965 elections. In 1968, the CSP presented two lists to the voters; one included moderate francophone and Flemish candidates and the other only the more militant francophones. Neither list was prepared by a primary.

¹⁹ Winning an unusually large number of personal votes, Gilson still managed to win a Chamber seat in the 1965 elections. He was, however, completely excluded from the CSP bilingual list in the 1968 elections.

²⁰ Similar problems have been common in Verviers, the only constituency which includes a substantial number of German-speaking Belgians. In relation to their support for the CSP in the general elections, German-speaking voters have been traditionally underrepresented in the primary turnouts. Perhaps due to this distortion in the primary electorate, Germanspeaking aspirants have usually secured only the third position on the list. Since this has not been a safe position, party leaders have intervened to reserve the second (and safe) place for a German-speaking candidate regardless of the primary results.

²¹ For a discussion of events in Dinant see La Libre Belgique, 22 April 1965 and ‘Les polls preparatoires aux élections législatives de mai 1965’, Courrier Hébdomadaire, 285–6 (20 May 1965). For those in Liège see La Libre Belgique, 25 March and 2, 7, 12 and 25 April 1965.

²² The following account of the model list is drawn from Debuyst, Fonction, LaParlementaire pp. 185–91.Google Scholar

²³ As seen in Table 3, support for the model list has varied considerably from one region to another. Among the Walloon associations, where factional disputes have been especially severe, the mean vote for the model list was 37 per cent and only in Soignes was it approved by a majority (50.2 per cent) of the voters. In Flanders, where relations among the various factions have been far more congenial, the model list attracted an average of 59 per cent of the vote and it was backed by a majority in nine of fifteen instances. This suggests that party members in Flanders have been more willing to go along with the preferences of the factional leaders who see the model list as an acceptable allocation of the best positions among the competing groups. In contrast, the Walloon factional leaders have been rarely pleased with the compromise reached through the model list and have therefore often exhorted their followers to ignore this compromise by supporting aspirants appearing on the provisional list. This discord has led the factional leaders to seek support among the voting members which, in turn, has made the primary an instrument of conflict resolution.

²⁴ In 1968, 81.0 per cent of BSP voters identified with the working class and 81.3 per cent were either blue-collar workers or white-collar employees. Delruelle, , Evalenko, and Fraeys, , Le comportetment politique, p. 50.Google Scholar

²³ Militant elements within the BSP were hostile to coalitions with the Catholics and Liberals and were more reluctant to accept gradual, piecemeal social reforms. They have been especially influential in the Provinces of Hainaut and Liege, while the moderates have controlled the party in most Flemish constituencies as well as those in the Provinces of Namur and Luxembourg. For a careful description of the factional disputes within the BSP see Debuyst, , La Fonction Parlementaire, pp. 233–42.Google Scholar

²⁶ The following discussion of the 1968 BSP Brussels primary is based on ‘La préparation des elections du 31 mars 1968: la stratégic des candidatures dans les partis (I)’, Courrier Hébdomadaire, 400 (29 03 1968), 21–31.Google Scholar

²⁷ In preparing for the 1971 elections, the linguistic factions tried unsuccessfully to mend their fences. Abandoning the primary, the Flemish and francophone factions once again established and then presented separate electoral lists.

²⁸ This age qualification eliminated such well known Socialist incumbents as Leo Collard, former National President of the BSP, and Camille Huysmans, one of the founding fathers of the Belgian Socialist movement and a leading figure in international Socialist circles. Collard offered no resistance when in 1971 the association in Mons asked him to step down. But Huysmans was far less co-operative when the leaders in Antwerp did not nominate him in 1965. Instead of retiring quietly, the 94-year-dd Huysmans decided to head a dissident list and to make a direct appeal to the older generation of Socialists in the constituency. Much to the dismay of BSP officials, Huysmans drew 14,937 votes (about 15 per cent of the BSP vote in 1961) from the party, which, though not enough to win a seat for himself, was enough to deprive the BSP of one seat. See Le Soir, 24 May 1965.

²⁹ Though formal, open campaigning is officially prohibited, informal campaigning is quite common. To maximize their support, aspirants frequently contact members in the local cafes and meeting halls. They also try to gain the support of some of the more influential party figures who often informally control the votes of several hundred members.

³⁰ See ‘Les polls préparatoires aux élections législatives du 26 mars 1961’.

³¹ This possibility was explored by Jean Stengers and André Phillipart in a survey of BSP members in Brussels after the 1958 primary. They asked the respondents to select candidates from a list identical to the one that had been presented to the primary voters. But while the latter had to choose thirteen candidates, the survey respondents were free to select as many as they pleased. This alteration of the voting method led to markedly different results. While only two non-incumbents managed to secure safe places in the primary, seven won safe places in the survey. Moreover, only a tiny fraction of the respondents cast as many as thirteen ballots, with the great majority indicating their preferences for five or six candidates. See Une Expérience D'Enquête Eléctorate (Brussels: Insitut Universitaire D'Information Sociale et Enonomique, 1959).Google Scholar

³² See Obler, Jeffrey, ‘The Role of National Leaders in the Selection of Parliamentary Candidates: The Belgian Case’, Comparative Politics, V (1973), 169–74.Google Scholar

³³ Debuyst, , La Fonction Parkmentaire, pp. 257–9Google Scholar and ‘Le PLP et les elections du 23 mai 1965’, Courrier Hebdomadaire, 289 (2 July 1965), 5–12.

³⁴ See ‘La préparation des elections législatives du 31 mars 1968’, pp. 31–7.

³⁵ See ‘La préparation des elections législatives du 7 novembre 1971: la stratégic des candidatures dans les partis (I)’, Courrier Hébdomadaire, 537 (29 10 1971), 12–17.Google Scholar