1In 1951, Maurice Duverger (1917-2014) published his seminal work Les partis politiques, [1] which has since been translated into ten languages [2] (with a Czech translation forthcoming). In 1963, Hannah Arendt (1906-1975) wrote that this book “supersedes and by far excels all former studies on the subject”. [3] A quarter of a century would have to pass before a work appeared that was comparable in stature to Duverger’s and which extended the insights of works on parties and party systems published in the interim: this work was Parties and Party Systems, [4] in which the Italian political scientist Giovanni Sartori (1924-) set out his famous typology of party systems. Vincent Lemieux (1933-2014) remarked in 1985 that:

“Sartori’s work on parties and party systems is undoubtedly the most important contribution to the study of political parties since Duverger’s book, published twenty-five years earlier.” [5]

Duverger’s refined typology of pluralist party systems

2According to Duverger, the main variables in the typology of pluralist party systems [8] are as follows: the number of parties (Sartori would add that what counts is the number of “relevant parties”); whether they are disciplined (rigid) or undisciplined (flexible); whether they are dependent or independent (dependent parties being those that establish electoral alliances); whether they are practically and technically oriented (or “pragmatic”, as Sartori would have it) or theoretically and doctrinally, or “metaphysically” oriented (Sartori uses the terms “ideological” and “rationalist”); [9] and whether the party system is “dualist” (later on, Duverger uses the term “bipolar”) or not (in this case, the term later used is “multipolar”).

3If we limit ourselves to pluralist party systems and take into consideration only the two extremes, leaving aside intermediary types, we arrive at the following typology: bipartism or multipartism; rigid (disciplined) or flexible (undisciplined) parties; dependent or independent parties; “technical” or “metaphysical” party systems (from 1951 onwards, Duverger distinguishes “technical dualism” from “metaphysical dualism”); [10] and dualist (bipolar) or non-dualist (multipolar) party systems, with this last distinction, according to Duverger, being the decisive variable, even more important than the number of parties – by the 1950s, Duverger no longer uses the terms “bipolar” and “multipolar” (although he would use these terms again later, these fluctuations in vocabulary altered nothing fundamental).

4Thus, in 1951, Duverger identifies four principal pluralist party systems: [11] true two-party systems (of the British type, with rigid, independent, and technical parties); “pseudo-two-party” systems (as in the United States, with flexible parties); apparent multi-party (dualist – that is, bipolar) systems [12] with either two sufficiently homogeneous, durable, and disciplined alliances or one “dominant” rigid party set against a long-lived alliance, with technical and dependent parties; and true multi-party (multipolar) systems with independent parties.

5True two-party systems (as in the British model) and apparent (bipolar) multi-party systems are both dualist (bipolar) and disciplined in nature, which is why Duverger groups them together – they function similarly. In the same way, “pseudo-two-party” (as in the US) and “true multi-party” systems are both multipolar systems, and Duverger groups them together as such.

Table 1

Main pluralist party systems according to Maurice Duverger

Rigid, bipolar, and technical systems True (rigid) two-party systems Apparent (dualist – also known as bipolar) multi-party systems Multipolar systems Pseudo (flexible) two-party systems True multi-party systems

Main pluralist party systems according to Maurice Duverger

6Schematically, Duverger associates these party systems on one hand with the corresponding electoral systems (in his famous “laws” or, more precisely, “tendential regularities”), and on the other hand with different types of democracy. [13] Therefore, as Jan-Erik Lane and Svante Ersson have perceptively remarked, in Duverger’s Political Parties we can glimpse the outlines of an important theory of democratic politics in general, which can be contrasted with the well-known theory of Arend Lijphart. [14]

Main steps in the development of the refined typology

7In 1950, Duverger edited the collective work L’influence des systèmes électoraux sur la vie politique. His contribution, which bears the same title, contains the most systematic definition of his so-called laws. [15] It also contains an outline of his more fully developed typology of pluralist party systems. The first passages in the text concern the effect of a second-ballot majority system on alliances:

8

“It is thus possible to achieve a stable, well-organized political system rather like the two-party one: instead of two large unified parties, there are two ‘federations of parties’ facing one another, their strength depending largely on the state of discipline and organization in the constituent parties.” [16]

9In examining the relations between the various electoral systems and their sensitivity to variations in opinion, Duverger comes closer still to the idea of a “bipolar multi-party” system.

10

“In fact, if the many parties to which the system gives rise coalesce to form two great coalitions in which discipline is strict, and which are clearly separated from each other, there is a definite resemblance to the two-party system; while it is possible for the moderations of opinion to find expression within each of the two tendencies, the distribution of votes among them is amplified by the electoral system as it is under a two-party regime.” [17]

11According to Duverger, then, the second-ballot majority system certainly favours flexible parties; but this is of little importance, since the influence of electoral systems upon the flexibility or rigidity of a party is minimal. [18] This means that, even under the second-ballot majority electoral system, fairly solid alliances can be formed and maintained.

12This second-ballot majority system may combine the advantages of proportional representation and single ballot majority systems:

13

“From this point of view, the second-ballot system has a certain advantage, in that it makes it possible at the same time – as a result of the alliances formed at the second ballot – to reflect the fundamental division into two camps as well as the minor conflicts within each of the major groups of opinion. It should be noted, incidentally, that a two-party system would produce the same result if the structure of each party remained so flexible as to permit the development and co-existence of diverse factions.” [19]

14Thus, in 1950, Duverger shows that, setting out from a “true” multi-party system, one can arrive at an “apparent” multi-party system that is in fact dualist (or bipolar) via the formation of alliances (if those alliances are sufficiently homogeneous and disciplined). Similarly, a “true” two-party system (of the British type, with disciplined parties) can develop into something more like a multi-party system if the parties become less disciplined. There may of course be intermediate situations, which Duverger associates above all with the second ballot majority system.

15The next – decisive – step in Duverger’s development of a refined typology of pluralist party systems was taken in his 1951 masterwork: Political Parties. Here, his analysis of the relationship between electoral systems and their sensitivity to variations in opinion leads him to the following conclusion:

16

“If very close electoral alliances prevent any see-sawing between Right and Left according to the constituency, such as was practised by the French Radical party, then the situation very nearly approaches that of the two-party system: variations in opinion continue to be toned down within each alliance, but the distribution of votes between the two is exaggerated by the electoral system as in a two-party regime.” [20]

17Duverger addresses this “dualism of alliances” in more detail in his discussion on the geographical localization of coalitions:

18

“If two rival coalitions, Left and Right, are set up and if they are sufficiently strong, a multi-party system may become very much like a two-party system.” [21]

19On one hand, he continues, “a two-alliance system is not so sound as a two-party system”, [22] but on the other, “[i]n the last resort […] all depends on the degree of cohesion manifested by the respective alliances or parties”. [23] Explaining the role of parties in political regimes, Duverger then writes:

20

“A system of close and stable alliances may also transform the foregoing schemas and reduce the gap between multipartism and bipartism; in so far as a true dualism of alliances is established, the situation resembles a dualism of parties.” [24]

21While it is possible to come close to dualism while remaining within the framework of the multi-party system, the contrary is also possible:

22

“Weak internal party structure […] brings dualism nearer to multipartism.” [25]

23The same is true for the role of opposition parties:

24

“Solid and homogeneous coalitions may give a multi-party system a close resemblance to a two-party system and make the opposition more coherent, more moderate, and more distinct. Conversely a two-party system in which the parties are lacking in discipline, centralization and organization may have an opposition often nearer in its operation to the multi-party than to the two-party pattern.” [26]

25Thus, whether analysing the sensitivity of electoral systems to variations in opinion, alliances, the role of parties in political regimes, or the role of opposition parties, Duverger unfailingly highlights the fact that a dualism of alliances, provided the alliances are homogeneous and disciplined, leads to a system that is very close to a “true” two-party system such as the British one. And in the same way, a dualism with undisciplined, decentralized, or poorly organized parties results in a system that is strongly akin to a true (non-dualist) multi-party system.

26Duverger formulates his refined typology more systematically in 1960 in “Sociologie des partis politiques”, one of the two chapters he contributed to the Traité de sociologie edited by Georges Gurvitch (1894-1965). Sartori in fact knew and cited this text. [27] With regard to two-party systems, Duverger distinguishes two principal categories:

27

“The fundamental distinction between two-party systems resides in the degree of discipline of the parties. In this regard Great Britain and the United States are typical examples […]. From the point of view of the political consequences, the American-style ‘flexible two-party system’ is actually closer to multipartism than the British-style ‘rigid two-party’ system. This is why we have proposed calling such systems ‘pseudo-two-party systems’ […]. As far as the classification of political regimes is concerned, the tendency today is to contrast rigid two-party systems on one hand to multi-party systems or pseudo-two-party systems on the other. In the first, the parliamentary mechanism functions very well […]. In flexible multi-party or two-party systems, on the contrary, the parliamentary regime functions very poorly.” [28]

28Just as one can contrast the rigid, British-style two-party system with the US-style pseudo-(flexible) two-party system – especially from the point of view of political consequences – multi-party systems can also be divided into different types:

29

“Alliances can totally change the aspect of a multi-party system. If two large permanent coalitions form, which present to the electorate a common programme and act in concert in parliament, then one is very close to the conditions of functioning of a two-party system. Under the external appearance of a multi-party system, we find in reality a profound dualism. But everything obviously depends upon the solidity of the alliances and the discipline of those in the coalition.” [29]

30The consequence of all of this is clear enough. The distinction between bipartism and multipartism may be an important one, but it is by no means the last word in the typology of pluralist party systems.

31

“The distinction between two-party and multi-party systems is therefore very important. But its importance should not be exaggerated. The formation of stable alliances, going into the electoral battle on clear platforms written together, and then applying them in government, brings the multi-party system close to the two-party system. Inversely, when each of two parties has a flexible structure, when there is no discipline in parliamentary voting, governmental majorities become incoherent and unstable, and the two-party system looks more like a multi-party system.” [30]

32In later works such as his 1966 Sociologie politique and his 1964 Introduction à la politique, Duverger reprised his formulations from 1960, sometimes to the letter. [31] Finally, in his late writings from the 1980s and 1990s, such as the most recent versions of the first part of his textbook Institutions politiques et droit constitutionnel, the typology of party systems remains the same, except for a change in vocabulary (the adjective “dualist” is often replaced by the adjective “bipolar”, for example).

33The similarity of Duverger’s refined typology to Sartori’s famous later typology is evident, above all in its treatment of pluralist party systems. Before comparing the two systems in detail, let us open up a parenthesis on Duverger’s category of systems with a “dominant party”. Sartori objects to Duverger’s use of this term, emphasizing that his expression “dominant party” fails to define any particular party system, either with regard to its class, based on the system’s “format”, or its type, based upon its “mechanics”.

34

“In short, the notion of dominant party establishes neither a class nor a type of party system.” [32]

35Whereas “predominance” as defined by Sartori unequivocally determines a particular system, namely the “predominant party system”, Duverger’s dominant party is, in my view, compatible with three of the systems defined by Sartori, if we limit ourselves to the “types”: namely, the “predominant party system”, “moderate pluralism”, and “polarized pluralism”. It is ill-judged to introduce the category of the “dominant party” as Duverger did in 1951, and then to shift into talking about a “dominant party system” as if one were implied by the other:

36

“Dominant party is a category that confuses party (in isolation) with party system.” [33]

37In 1960, Duverger writes the following on the dominant party system:

38

“This system is midway between pluralism and the single party system.” [34]

39He distinguishes two forms that such a system might take:

40

“Among the countries where this system is now in operation, we can distinguish approximately two categories. On one hand, India appears as a nation where the opposition parties have a real existence, where they bring together a number of important voices, where one is closer to a multi-party system than to a single party. On the other, certain African republics come closer to the latter, since there the opposition is very much reduced, and the dominant party manifests quite clear authoritarian tendencies.” [35]

41It is clear that the first category, represented by India, is close to Sartori’s predominant party system, while the second resembles Sartori’s hegemonic party system. [36] With this second form of dominant party system, though, we are moving beyond the pluralist party systems that are the subject of this analysis. Let us simply flag that the notion of a “dominant party system”, sometimes supplemented by some additional adjective, continues to appear frequently in today’s literature, especially in the Anglophone world. [37]

Duverger and the typology proposed by Jean Blondel

42According to “conventional wisdom”, passed on from one author to another – in France, in particular, Jean Charlot (1932-1997), and in the Anglophone literature Peter Mair (1951-2011) – Duverger went no further than a banal classification of party systems, based solely on the number of parties, and which was developed years before him; after him, it was Jean Blondel whose contribution to the field represented a great step forward. As Jean Charlot writes:

43

“In the 1930s, the typology of party systems according to the number of parties, as well as the problems of each type of system, were already known. Thus Arthur N. Holcombe, in the Encyclopedia of Social Sciences, distinguishes between single-party, two-party and multi-party systems – as does Maurice Duverger, twenty years later.” [38]

44Fourteen years later, Jean and Monique Charlot stepped back into the fray:

45

“The classic typology of party systems according to their degree of fractionation – single-party, two-party, multi-party (Duverger, 1951) – was further developed by Jean Blondel with the distinction between perfect and imperfect two-party systems (the latter being really two-and-a-half party systems, as in the German Federal Republic), and that between perfect and imperfect multi-party systems (the latter being those with a dominant party, as in Sweden) (cf. Blondel, 1968; […]).” [39]

46Without wishing to minimize Blondel’s considerable contribution, we must recognize, and add to this, the fact that Duverger had himself fine-tuned the primitive classification system, and that he had done so in 1951, some time before Blondel.

47What is more, as far as Blondel’s category of “imperfect two-party system” (or “two-and-a-half-party system”) is concerned, in 1966 – that is, shortly before Blondel’s crucial article – Duverger had already defined the “para-dualist system” in Sociologie politique:

48

“We call ‘para-dualist’ systems in which two major parties dominate political life, but usually without either of them being able to obtain the majority alone. They must therefore either form an alliance with the third party, or with each other. These para-dualist systems should not be confused with the ‘pseudo-two-party’ or flexible two-party system […] of which the US furnishes the model: a two-party system without internal discipline, where a majority held by one of the parties means nothing, since the party members hardly ever vote in the same way […]. Thus defined, para-dualism is characteristic of Federal Germany, Belgium, and Austria.” [40]

49But there is more. In his book La démocratie sans le peuple, written in 1966 and published in 1967, Duverger explains:

50

“In Great Britain an almost perfect two-party system is in operation, the liberals having gained for thirty years no more than a minimal number of votes and seats, insufficient to prevent Labour or the Conservatives from achieving a majority in the Commons themselves. In Austria, in Federal Germany, and in Belgium, on the other hand, the liberal party is strong enough to often hold the balance of power between the two major parties, and to oblige them to form an alliance either with them, or with each other: this is a two-and-a-half-party system.” [41]

51In 1968, Blondel published an article entitled “Party systems and patterns of government in Western democracies”, [42] in which he set out his typology of party systems. This text was presented in the form of an illustrated report to Brussels, for the Seventh World Congress of the International Political Science Association (IPSA) in 1967. The French adaptation of the report, prepared by Blondel himself, was published by Jean Charlot in 1971. [43]

Comparison of Duverger’s and Sartori’s typologies of pluralist party systems

52In examining two-party systems, [44] Sartori distinguishes the two-party “format” (or “class”) on one hand, from the two-party “mechanics” (and “type”) on the other. I will apply this distinction of Sartori’s to Duverger’s typology. Thus, a rigid or true two-party system (of the British type) would, in Sartori’s terminology, have two-party “mechanics”, and thus would be a two-party-“type” system, whereas the US “pseudo-two-party system”, with flexible parties, would have only a two-party “format”, and would be two-party only in “class”.

53I shall then take a different tack: in anti-chronological fashion, I will seek approximate equivalents of Sartori’s categories in Duverger. It is quite possible to do so since the two typologies are sufficiently close to each other. [45]

54If we take Sartori’s “moderate pluralism”, it corresponds roughly to Duverger’s “apparent [dualist, or bipolar] multi-party system”. Both moderate pluralism and the apparent multiparty system are bipolar (dualist), non-polarized (“technical”, in Duverger’s terms); and moreover Duverger emphasizes their rigidity (discipline in party voting). Certainly Sartori also highlights the role of discipline, but only under the parliamentary regime:

55

“Without parliamentary discipline, whether parties are two or more does not make much difference.” [46]

56Here, Duverger goes further than Sartori: he considers it important to have disciplined parties even under a presidential system, but insists that in order for the political system to function properly, legislative and presidential elections must be synchronized.

57Now we come to “polarized pluralism”, one of Sartori’s types which has been abundantly discussed in the literature. Duverger’s “true [multipolar] multi-party system” is rather similar to this type. Polarized pluralism and the true multi-party system are both polarized (in Sartori’s sense) and multipolar; Sartori particularly emphasizes polarization, Duverger non-dualism and multipolarity.

58Let us now leave behind the “types”, defined on the basis of the “mechanics” of the system, and move on to the “classes”, based on the system’s “format”. We remark that the class of “limited pluralism” in Sartori corresponds to Duverger’s 1951 “limited and orderly multiparty system”. And likewise, the counterpart of Sartori’s “extreme multi-party system” is Duverger’s “anarchic and disorderly multipartism”, again in 1951:

59

“The tendency towards multipartism is obvious. It seems to assume two somewhat different patterns. In Switzerland and in Holland the multi-party system is limited and orderly; in Italy it is anarchic and disorderly…” [47]

60As for Sartori’s “atomized system”, it corresponds quite closely to Duverger’s “polypartism, or the tendency to extreme multiplication of parties”. [48] Within the latter he differentiates many types: nationalist or ethnic polypartism, the polypartism of the right, and the polypartism of Latin peoples.

61From polypartism, it is only one further step to the absence of any true parties:

62

“Multipartism is often confused with an absence of parties. A country in which opinion is divided amongst several groups that are unstable, fluid and short-lived does not provide an example of multipartism in the proper sense of the term: it is still in the pre-historic era of parties; it is to be situated in that phase of development in which the distinction between bipartism and multipartism is not yet applicable because there are as yet no true parties.” [49]

63Sartori also considers that “the concepts that apply to the structured and differentiated polities cannot be carried over, as such, to diffuse and embryonic polities”. [50]

Terminology used by Duverger and Sartori

64Sartori sometimes uses the term “polypartism”, but with implications different from those it has in Duverger. [51] The term polarization is of greater consequence: it plays a considerable role in Duverger, and for Sartori it is crucial, but the latter uses it in an entirely different sense than Duverger. In the volume Political Oppositions in Western Democracies, [52] which he edited, Robert A. Dahl (1915-2014), who passed away in the same year as Maurice Duverger and David Easton (both of the latter were born and died in the same year: 1917-2014), dissects the various significations of the term “polarization” in a most enlightening way. The second sense of the word given by Dahl is more or less Sartori’s:

65

“It also contains the idea of a dimension of distance between the two principal categories. The greater the distance, the more polarized a society is.” [53]

66As for polarization in the sense that Duverger understands it, we find it in Dahl’s third category:

67

“In the simplest case, a country where everybody voted either for Party A or Party B would (by definition) be dualist with respect to voting.” [54]

68Duverger introduces the concept of “polarization” in 1950 (in “The influence of electoral systems on political life”) and returns to it in 1951 (in Political Parties). Polarization is closely bound to his idea of a psychological factor at work in electoral systems. As he explains, in the case of a single-ballot majority system, electors take account of the fact that their voices will be lost if they continue to give their vote to the third party; consequently, they may decide to resort to the least unacceptable of the two major parties. [55] This is precisely the phenomenon of “polarization” around two major parties as Duverger describes it. Its counterpart is the phenomenon of “under-representation” of the third party, which is the consequence of the mechanical factor of the single-ballot majority electoral system, just as polarization (or “bipolarization”) is the consequence of the psychological factor at work in the same voting system.

69Sartori once again constructs his own terminology. What he calls the “reductive effect” is the approximate equivalent of the “mechanical” factor in Duverger: the approximate equivalent of the “psychological” factor in Duverger is the “constraining effect” in Sartori. [56] The reductive effect acts on the number of parties, the constraining effect on the electorate. Reading the contemporary Anglophone literature, one notes that it is Duverger’s vocabulary that is most often used since it is more neutral, whereas Sartori’s terminology is associated with the somewhat arguable idea that proportional representation (PR) in its pure state simply has “no effect”. [57] Sartori defends this point of view as follows:

70

“To be sure ‘no effect’ on the number of parties, which is the dependent variable under consideration. For, surely, PR has effects in other domains and, in general, electoral systems always are of consequence on something.” [58]

Divergences between Duverger and Sartori: real or imaginary?

71Sartori criticizes Duverger’s use of the term “polarization” in the following terms:

“It should be noted that while I hold that a political system may be bipolar and not polarised, Duverger identifies (or confuses) polarisation with ‘bipolarity’ (see Les partis politiques, p. 279).” [59]

Table 2

Summarized glossary of pluralist party systems and their factors in Duverger and Sartori

Duverger 1951 Sartori 1976 True two-party system Pseudo-two-party system Two-party type Two-party class Types Apparent multi-party True multi-party Moderate pluralism Polarized pluralism Classes Limited and orderly multi-party Anarchic and disorderly multi-party Limited multi-party Extreme multi-party Residual category Polypartism Atomized system Factors or effects Psychological factor Mechanical factor (Duverger, “The influence of the electoral system”, 318) Constraining effect Reductive effect (Sartori, “The influence of the electoral system”, 58ff) Duverger-Sartori “False friends” Polypartism Extreme multiplication of parties (Duverger, Political Parties, 240) Equivalent in Sartori: “atomized system” Polypartism 1. Multi-party systems in general (Sartori, Parties and Party Systems, 116) 2. Anything apart from single-party system (Sartori, Parties and Party Systems, 230) Polarization Concentration of votes in two major parties (Duverger, “The influence of the electoral system”, 317) Polarization Significant ideological difference (Sartori, Parties and Party Systems, 117-18)

Summarized glossary of pluralist party systems and their factors in Duverger and Sartori

Note 1. The equivalences between Duverger’s and Sartori’s terms are only approximate.
Note 2. The principal points of comparison are Duverger’s 1951 Political Parties and Sartori’s 1976 Parties and Party Systems, but some of Duverger’s categories were developed in 1950 and some of Sartori’s were conceived in the 1960s.

72Secondly, in 1951 Duverger adopts practically the same idea that Sartori will later profess, namely “that a political system can be bipolar and not polarized”; except that instead of the expression “not polarized”, Duverger writes “technical”; and in the 1950s, when he is no longer using the word “bipolar”, he instead says “dualist”.

73Here it is a question of the “centre” in politics, and it is in this context that Sartori castigates Duverger. Sartori tries to explain why, in his view, the characteristics of “polarized pluralism” have not been understood by specialists:

74

“One [of the two reasons] is the use of dualistic blinders, that is, the tendency to explain any and all party systems by extrapolating from the two-party model. These dualistic blinders have been proposed by Duverger as an almost ‘natural law’ of politics: ‘We do not always find a duality of parties, but we do find almost always a dualism of tendencies […] This is tantamount to saying that the centre does not exist in politics: We may have a centre party, but not a centre tendency […] There are no true centres other than as a cross-cutting of dualisms…’.” [61]

75Sartori discusses this quotation from Duverger in the following terms:

76

“I will argue, contrariwise, that when we do not have a centre party, we are likely to have a centre tendency. For the moment let it just be pointed out that Duverger’s dualistic blinders lead him – as subsequent developments have abundantly confirmed – to astonishing mis-preconceptions, as when he finds that Germany and Italy are the two European countries that ‘display a rather marked tendency’ towards ‘bipartism’.” [62]

77Sartori’s diatribes against Duverger do not stop there. With a certain satisfaction, the Italian political scientist cites the famous fifteen-page critical review of Political Parties by Aaron B. Wildawsky (1930-1993), [63] in which the latter opines that, among other deficiencies, Duverger is the victim of an “eminently superstitious impression that phenomena occur in pairs”. [64]

78With regard to the dissimilarities between Germany and Italy, Sartori adds the following:

79

“This misapprehension of the entirely different mechanics of the two systems has been maintained to this day. See esp. Giorgio Galli, Il Bipartitismo Imperfetto, II Mulino, 1966.” [65]

80And finally, still with Sartori:

81

“Duverger’s thesis that ‘the centre never exists in politics’ (Les Partis Politiques, op. cit., p. 245) confuses the various aspects of the problem and should be reversed: a centre ‘tendency’ always exists; what may not exist is a centre party.” [66]

82Reuven Hazan, who systematized Sartori’s critiques in 1995, holds that Duverger cleaves to an “intuitive” approach according to which centre parties exert a moderating influence, whereas Sartori adopts a “counter-intuitive” approach according to which centre parties, on the contrary, amplify “polarization” (in Sartori’s sense). [67]

83In reality, Duverger had always opted for a conception that is very close to Sartori’s. He distinguishes between governing at the centre and governing from the centre:

84

“One can only really govern at the centre in the absence of a centre majority, that is to say in the absence of centre parties, alliances, and majorities, that is, in the absence of any means to govern from the centre. When the centre is in power, it is condemned to impotence and it slowly wears away the foundations of the regime. When the centre is split by bipolarity, right and left are obliged to moderate their policies, that is to say orient them towards the centre, unless they want to lose the next election. Such is the paradox of the centre: it only really governs when it does not exist.” [68]

85Duverger had already put forward this idea in 1951, when he explained the functioning of the British two-party system:

86

“Both [Labour and the Conservative Party] will have to draw up policies clearly aimed at the Centre and therefore profoundly similar. We arrive at the paradoxical situation that the Centre influences the whole of parliamentary life in the very country in which the electoral system prevents the formation of a Centre party.” [69]

87What is striking in this question of the centre is not the divergence but the convergence between Duverger and Sartori. Hazan is mistaken in believing that he can oppose Sartori’s “counter-intuitive” approach to Duverger’s “intuitive” one: it is undeniable that both of these outstanding political scientists are partisans of the “counter-intuitive” approach.

88When Sartori objects, against Duverger, that “a centre ‘tendency’ always exists; what may not exist is a centre party”, [70] this bespeaks a misunderstanding on his part. What Duverger has in mind here is a party or coalition of parties that does not have (or did not have) its own “centrist tendency”, whereas Sartori takes him to mean a partisan system that presents “centrist tendencies” in the sense of “centripetal competition”. Duverger’s and Sartori’s theses are not fundamentally contradictory.

89Let us now come to the criticism according to which subsequent developments have abundantly confirmed the existence in Duverger of “astonishing mis-preconceptions, as when he finds that Germany and Italy are the two European countries that ‘display a rather marked tendency’ toward ‘bipartism’”, and that Duverger – like Galli – did not understand that the two-party systems of Germany and Italy (under the First Republic) had “entirely different mechanics”. Once more, Sartori’s attacks are, on both points, unjustified.

90Firstly, Duverger does not claim at all that the evolutionary trajectory of Germany and Italy leads (or must lead) to a two-party system.

91

“[T]here are six parties in Germany and eight in Italy, and their number tends to increase rather than decrease.” [71]

92According to Duverger, in the Italy of the First Republic as in Germany, the voting system does not favour bipartism. Although from a formal point of view the German electoral system is “mixed”, in reality it is a “proportional system of compensation”, and has a “clearly stated aim of proportionality”. [72] Duverger formulates this quite clearly:

93

“Mathematically, the electoral system is purely proportional, even though half of the deputies are designated in an entirely different way.” [73]

94Secondly and most importantly, Duverger does not say that Germany and the Italy of the First Republic function in a similar way. He explains that it would not be hard to establish a single-ballot majority electoral system in Germany which, in the long run, would bring about a two-party system. On the other hand, if a single-ballot majority electoral system were to be adopted in Italy, the result would be “catastrophic” [74] because one of the two parties – the Communist Party – is totalitarian.

95This leads Duverger, on the one hand, to introduce alongside the “dualist oppositions” the opposition between the Communist Party and a “Western” party – in other words, the “East-West” line of demarcation; [75] and, on the other hand, to highlight the distinction between the “technical” two-party system which would appear sooner or later in Germany following the introduction of a single-ballot majority electoral system, and the “metaphysical” two-party system that would be the long-term consequence of the adoption of the same system in Italy:

96

Only the first type [technical dualism] is viable. This is equivalent to saying that the two-party system is inconceivable if one of the two parties is totalitarian in structure. [76]

97So Duverger is fully aware of the profound differences between the “mechanics” of the German and Italian systems. What he writes agrees with the theses that Sartori later professes, apart from in the terminology – and one can hardly blame Duverger for having failed to adopt in 1951 the terminology developed later by his Italian colleague.

The tables turned?

98According to Duverger, then, it is the proportional electoral system (which in Germany is personalized but remains “fundamentally proportional”) [77] that implacably blocks the two-party “tendency” of German public opinion from being transposed into the actual party system: [78]

99

“In any case Germany and Italy have multi-party systems, like all other countries with proportional representation.” [79]

100Sartori objects that Ireland, with its single transferable vote system “which is a pure proportional system”, [80] has a two-party format:

101

“And there are additional countries that are, or have been, two-party with PR: Malta and Austria.” [81]

102Sartori argues that Ireland, because of its single transferable vote system, in fact belongs to the category of proportional electoral systems. However, Ireland also has a small number of seats per constituency (between three and five), which means that it has a significant majoritarian element. Specialists have established that smaller numbers of seats produce majoritarian effects. [82] The lower house of the Irish parliament comprises 166 members for 43 electoral constituencies, which does not even amount to four seats per constituency (3.86). Here I am limiting myself to arithmetic alone: Rein Taagepera has certainly shown that to make the argument scientific, it would be better to use an “effective magnitude” (“M”), [83] but this is of more importance when the number of seats varies a great deal from one constituency to another, which is not the case in Ireland. Between 1933 and 1989, the country had a “predominant party system”, and after 1989, something more like a “moderate pluralism”.

103Malta may certainly be considered as having a two-party system, but like Ireland it has a single transferable vote system with small numbers of seats. Its unicameral parliament is composed of 65 seats for thirteen constituencies, each with only five mandates.

104On that note, let us recall here Sartori’s complaints against his detractors when he defends his own formulation of Duverger’s laws:

105

“The authors who deny the reductive effects of plurality systems almost inevitably make reference to unstructured party systems, thereby challenging a ‘law’ in situations where the law does not apply.” [84]

106Certainly: but what does Sartori do when he takes up the counterexamples of Ireland and Malta? For his part, Duverger is clear on this subject, for example when he discusses his first “law” in the 1955 first edition of his Droit constitutionnel et institutions politiques: [85]

107

“1/ Proportional representation leads to the formation of multiple and independent parties. Of course, this is only the case for a wholly proportional system of representation, with closed lists and redistribution of votes to the remaining candidates, within the national framework. In so far as PR is implemented in attenuated form, its consequences are attenuated in parallel.” [86]

108The case of Austria is more interesting. In his works from the beginning of the 1950s, Duverger leaves Austria aside. Throughout the 1960s, he considers that the country does not have a two-party system, but a “para-dualist” system (1966) – also known as a “two-and-a-half party system” (1967). [87] The “two-and-a-half party system” (“imperfect two-party system”) is part of the typology constructed by Blondel, who, writing at almost the same time as Duverger, nonetheless classes Austria in another category. For Blondel considers the country to have a “two-party system” (i.e., a “genuine two-party system”). Lastly, Sartori asserts that Austria’s system is not of a two-party “type”, but only a two-party “format”.

109Duverger defines the “two-and-a-half-party system” differently from Blondel. Just like Sartori, or even more so, Duverger bases his classification on the functioning of the system: there are two large parties, but usually neither of them are capable of gaining an absolute majority of seats:

110

“So they must either ally themselves with a third party, or form an alliance together.” [88]

111But a true two-party system excludes alliances in principle. [89] Sartori’s reasoning is almost identical: a two-party system necessarily means a single-party government; it excludes coalition governments. Nothing of the sort is to be found in Blondel, who defines the “genuine” two-party system in operational terms as the system where “90 per cent or more of the seats are distributed between the two major parties”. [90] Whereas countries with a “two-and-a-half party system” (“imperfect two-party”), according to Blondel, “give between 75 and 80 per cent of their votes to the major parties”. [91]

112I would argue that the approaches of Duverger and Sartori are more pertinent than that of Blondel, in spite of the unquestionable merits of the latter. Blondel is to be commended for not being satisfied with the usual variable, the number of parties: he supplements it with the variable “relative force (importance and size)”, to mention only one. [92] From this point of view, Laakso and Taagepera’s index of “effective number of parties” is an opportune contribution. [93] I do not, however, follow Blondel in taking the percentage of votes rather than percentage of seats as the basis for his calculation. In doing so, he glosses over the effects of the electoral system, in particular the effect of “mechanics” (or the “reductive” effect). This is something that is unthinkable for Duverger and Sartori alike. If one operates with Laakso and Taagepera’s “effective number of parties”, what counts is the effective number of parliamentary (legislative) parties, not simply the number of electoral parties.

113There is another weakness in Blondel’s typology, which it has in common with other (not necessarily all) quantitative approaches. Sartori explains the problem of such approaches very well, concluding as follows:

114

“[I]t is fair to say, in general, that the nominal route affords – for all its limits – far more theoretical and predictive mileage than the quantitative and mathematical routes.” [94]

115This is perfectly clear in the case that interests us here: Duverger is right in his objections to Blondel when, in the 1960s, he proposes that Austria does not have a two-party system but a two-and-a-half party system. Even in the 1960s, the party system in Austria functioned according to a logic far removed from two-party “mechanics”. And even if in 1967 there was still room for hesitation between the two interpretations, after 1986 there can no longer be any doubt. To paraphrase Sartori, in theoretical and predictive terms, Duverger’s approach turns out to have more mileage than Blondel’s.

116To sum up, of the three counterexamples Sartori offers against Duverger’s thesis on the link between proportional representation and the multi-party system, two (Ireland and Malta) are inadmissible because proportionality is supplemented by a clearly majoritarian element (small number of seats); and the third (Austria) is not even considered by Duverger as a two-party system – and for good reason.

Dualism and the sociology of conflict

117Let us now examine those “dualistic blinders”. Duverger had already written in 1950:

118

“It may be wondered whether public opinion does not in fact tend to split into two great rival factions, each, of course, containing a multitude of shades of opinion but each fairly sharply outlined. It is interesting, in this connexion, to find that much the same conclusions have been reached in widely differing studies. Some sociologists suggest that a distinction should be made between two fundamental political temperaments (the ‘radical’ and ‘conservative’); the Marxists conceive of the dynamics of society as a struggle between two great opposing classes; and the founders of electoral geography in France recognize, behind the apparent multiplicity of political opinions in our country, an enduring fundamental opposition between the Right and the Left, between order and movement.” [95]

119Duverger develops and refines this aspect of his thought further in 1951. We will limit ourselves to a significant extract:

120

“[P]olitical choice usually takes the form of a choice between two alternatives […] Every policy implies a choice between two kinds of solutions: the so-called compromise solutions lean one way or the other.” [96]

121What is really at stake here? Duverger belongs to a long line of “sociologists of conflict”, whereas Sartori, influenced by the structural functionalism of Gabriel A. Almond (1911-2002), is more on the side of the “sociologists of consensus”. The sociology of conflict is logically linked to dualism. One of the last great figures in this lineage, Ralf G. Dahrendorf (1929-2009), formulates this very clearly:

122

“Whatever criticism may be required of the Marxian theory, any theory of conflict has to operate with something like a two-class model. There are but two contending parties – this is implied in the very concept of conflict.” [97]

123Moreover, with his idea of “cross-cutting” and the superimposition of dualisms, [98] Duverger is one of the precursors of the theory of fundamental divisions, [99] developed in 1967 by Stein Rokkan (1921-1979) with Seymour M. Lipset (1922-2006). [100]

124***

125Throughout the literature, we read that Maurice Duverger was satisfied with a banal classification of party systems based only on the number of parties, and which had been developed many decades before him. This is simply not the case. Duverger develops a far more sophisticated typology. He partly outlines it in 1950 in “The influence of electoral systems on political life”, and completes it a year later in Political Parties. However, the exposition of this typology is dispersed throughout his work. It is in 1960, in his “Sociologie des partis politiques”, that Duverger first sets it out in an ordered and systematic form. Sartori knows it, and cites – in the French version, even – all three aforementioned works by Duverger. The similarity of Duverger’s refined typology and Sartori’s famous later typology is striking, especially with regard to pluralist party systems.

126What is more, with the terms “para-dualist party system” in 1966, and “two-and-a-half-party system” in 1967, Duverger introduces a category similar to that proposed by Blondel in an article published in 1968. Duverger’s approach is closer to Sartori’s than to Blondel’s, though, since it focuses on “function”. Sartori tries to demarcate himself from it: firstly by almost systematically replacing Duverger’s terminology and by introducing distinct terms, sometimes even at the price of unfortunate confusions (as with the term “polarization”); and secondly, by asserting that his approach is profoundly divergent from Duverger’s, which it is not. [101] Sartori’s criticism of Duverger is most often unfounded, at least in the case of pluralist party systems. Similarly, Hazan is mistaken, on the question of the centre in politics, to oppose Duverger’s “intuitive” approach to Sartori’s “counter-intuitive” approach. In reality, in 1951 Duverger opts for a conception of the political centre that is very similar to that which Sartori will later profess. The political scientist Daniel-Louise Seiler observed as much in 1993:

127

“Maurice Duverger’s contribution to the study of party systems thus proves decisive.” [102]

128Consequently, we must revise our understanding of the pattern of development of political science. Duverger is closer to us than we commonly think. [103]