If a multiparty system possesses centre parties, what kind of impact will they have on electoral competition, systemic polarization, governmental durability, and democratic stability? Will the presence of parties in the centre compel the other parties to moderate their positions in order to court the centre? Or, will the mere existence of centrally located parties push the remaining parties in an outward direction? In other words, are centre parties negatively or positively associated with party system polarization? If the answer is positive, then must a party system with an occupied centre impact negatively on governmental durability and democratic stability?

The intuitive conceptualization of the centre as a force for moderation pervades much of the literature that deals with party systems, is accepted almost universally by political scientists, and has achieved the status of an axiom when politics are commonly discussed. This perception of a

Resumé

In a recent study it has been argued that strong centre parties may lead to polarization, not moderation. The study showed that as the centre's share of parliamentary seats increased, this convergence of voters was off-set by two concurrent divergent party trends. As the centre parties expanded, either: (i) the extremist parties increased as well, or (ii) an outward movement of moderate parties took place. This article sets out to test these two trends in order to assess which is the more valid according to two case studies, because each pattern has a different impact on electoral competition, governmental durability and democratic stability. The two trends are appraised in Denmark and The Netherlands for all post-war elections until 1990. The results show that the centre is indeed related to systemic polarization, but that one of the two patterns is invalid. The trend that perceives the centre as a possible destructive force is not supported, while the tendency that does not jeopardize democracy is supported. In both countries the centre's potential coalition partners - the parties on the moderate left and right - attacked their centre-based party system by pulling away in an outward polarizing pattern. The goal was the creation of a bipolar system, with a vacant centre. In each case the centre parties were of a different size and adopted different tactics in order to combat the "moderate-induced" strategy of polarization. The centre's counterstrategies but the party systems were transformed.

moderating centre results from the belief that those who vote for a centre party - and the party, or parties, which represent these voters - are able to negotiate with either the left or the right, are open to compromise, and are liable to seek a broader consensus on divisive issues. What this approach fails to take into account, despite the unquestionably moderate characteristicspossessed the centre, is the effect that the centre has on other parties in the system.

This article sets out to assess and outline the impact that centre parties have on the remaining parties in the party system. It will present the theoretical debate on the centre, and a new study which positively relates the centre to polarization. In an attempt to appraise this recent study's contribution, the proposed relationships it posits between the centre and polarization will be assessed for all post-war elections, through 1990, in Denmark and The Netherlands. After analysing whether or not the centre is indeed a contributor to enhanced party system polarization, the role of the centre in attempting to either combat or exacerbate this trend will be examined. In the process, the intuitive perception of the centre's moderating will be refined in an attempt to contribute to improved party systems theory.

Centre Parties and Systemic Polarization

Proponents of centre ideology, from Aristotle in ancient Greece to the Doctrinaires in post-Napoleonic France and the Reform Whigs in Britain, have attempted to establish middle-class rule as a golden mean between the extremes. Yet, in the accumulated contemporary academic literature there are few explicit references to the concept or impact of the centre, and fewer attempts to define and study the centre. Apart from the works of Duverger (1959) and Sartori (1976), only Downs (1957) and Daalder (1984) have presented the centre as an important variable in their theoretical of party systems. However, quite recently, there have been a few researchers who have begun to devote attention once again to the centre and to centre parties (Van Roozendaal 1990; leraci 1992; Scully 1992; Keman 1994).

The theoretical debate on the impact of the centre has not changed or advanced in the last two decades, ever since Duverger and Sartori adopted their opposing positions. According to Duverger (1959, 215) the centre is a moderating factor.

The term "centre" is applied to the geometrical spot at which the moderates of opposed
tendencies meet: moderates of the Right and moderates of the Left. . . . For the Centre is
nothing more that the artificial grouping of the right wing of the Left and the left wing of
the Right. . . . The dream of the centre is to achieve a synthesis of contradictory aspirations.
(italics added)

Sartori (1976, 135) disagreed completely, pointing out that the physical occupation of the centre implies that the moderate electorate is no longer the floating electorate, thereby pointing the direction of competition away from the centre.

. . . the very existence of a centre party (or parties) discourages "centrality", i.e., the centripetal drives of the political system. And the centripetal drives are precisely the moderating drives. This is why this type is centre-fleeting, or centrifugal, and thereby conducive to immoderate and extremist politics. ... In the long run a centre positioning is ... a cause of polarization, (italics in original)

The existence of a centre is, therefore, a contributing factor, or even a causal
factor, for either moderation or polarization. The case for moderation is
intuitive, the case for polarization, counter-intuitive.

A recent study (Hazan 1995b) has uncovered empirical evidence which substantiates the counter-intuitive approach - the centre as a polarizing factor. A cross-national analysis of party systems in Western Europe produced unchanging polarization measures while the centre parties' share of parliamentary seats multiplied tenfold, from 5 percent to 50 percent. An examination into the reason for this uncovered two explanations, which were subsequently backed by empirical testing:

(1) The presence and growth of centre parties directly and positively impact on the parliamentary strength of extremist parties; but they impact only when the centre is large - defined as more than 20 percent of the parliamentary seats.

(2) The presence and growth of centre parties directly and centrifugally impact on the movement of parties along the left-right continuum; but they impact uniformly for both left and right only on the moderate parties.

In other words, simultaneous to the moderating impact of the centre parties' parliamentary growth there appears to be a pattern of "centre-induced polarization" which suppresses any decline in the measurement of systemic polarization.

The ramifications of these two explanations are very different. According to the first, the centre could be correctly perceived as a destructive force. In such a scenario the parliamentary strength of the centre and the extremes are positively related - both gain seats from the moderates. Empirical research has shown that the establishment and growth of extremist parties impacts negatively on coalition formation and governmental stability (Taylor& 1971; Dodd 1976; Sanders & Herman 1977; Powell 1981, 1982, 1986). The simultaneous rise of both the centre and the extremes, at the expense of the moderates, could lead to the creation of a tripolar system in which only one centrally located pole is moderate, surrounded by weak parties to both its right and left, with the remaining two extremist poles

occupying the polar ends of the party spectrum (Daalder 1971; Sartori 1976; leraci 1992). The centre would then become the pivot and backbone of every coalition, while only its minor and moderate peripheral partners would alternate in government - post-war Italy, until 1994, being a case in point. This centre-based core would thus hold a perpetual grasp on government, and maintain it by capitalizing on the fear of extremism. When the central core could no longer attract more voters than the extremist periphery, its dominant position would be lost and with it possibly the continued survival of the entire democratic system.

The second explanation is not as acute, and democracy is not threatened. It is the moderate parties here who are the culprits for enhanced polarization. moderate parties grow tired of having to bargain with the centre in order to form a coalition, and they adopt a strategy of polarization to force the centre voters to choose between right and left. True, cooperation extremist elements could be part of the strategy, but they are usually relegated to the role of junior partners and their parliamentary strength does not increase. The goal here is to switch from a tripolar to a bipolar party system. In other words, the moderate parties want to destroy the centre's disproportionate influence, but they do not want to destroy centrism. The moderate parties prefer to function in a bipolar party constellation, than in a multipolar one. Through electoral strategies they seek to attain a type of party system which other countries have achieved by electoral laws. The major difference between this second explanation and the first explanation is that here working democracy is not placed in jeopardy.

Strategy and Methodology

In order to test which relationship between the centre and polarization is valid, data were collected and analysed from all post-war elections in Denmark and The Netherlands, until 1990. The countries chosen are based on the findings uncovered in Hazan's study, which divided the parliamentary strength of the centre parties into small and large. Both Denmark and The Netherlands produced almost congruent levels of systemic polarization (4.4 and 4.5, respectively), and are also quite similar in many other aspects which allows for variables exogenous to the party system to be relatively controlled. Also, each represents one of the two centre categories - the average parliamentary share of seats for the centre parties in Denmark during the period of this study was less than 15 percent, whereas in The Netherlands it was almost 50 percent. These two cases, therefore, are appropriate for examining whether the fairly equal levels of systemic polarization, despite the centre's difference in electoral strength of over

300 percent, were due to the rise of the extreme parties or to the outward
movement of the moderate parties.

I employed a research strategy designed to submit the opposing explanations the centre's positive relationship with extremist parties or its centrifugal impact on moderate parties - to statistical analysis. The dimension to assess the degree of polarization of party systems, and the centre positioning of one or more parties, is the left-right continuum.¹ The location of parties adopted is the one elaborated by Castles & Mair (1984).²

For the purpose of this research, a measure of polarization was used which focuses on the parliamentary strength of parties based on the percentage of seats won in the lower house of parliament. Polarization was assessed by adapting the measure introduced by Taylor & Herman (1971), which is based on the statistical formula for variance, according to the following formula:

where N is the number of parties in the system, p_t is the percentage of the seats won by each party, x4x₄ is the respective left-right ideological position of each party, and x is the weighted system mean³ of the left-right seat distribution for that election. That is, each party's percentage of seats is multiplied by the square of its left-right position, which has the weighted system mean deducted from it, and summed to produce the polarization score for a specific election.

By centre parties this study refers to a party, or parties, which is located at a relatively equal geometric distance from each end of an ideologically competitive space (Hazan 1994 and 1995a). That is, parties of the centre are those parties along an ideologically defined, cross-national left-right scale, which occupy the metrical centre, or are near it.⁴

For the purpose of deriving theoretical explanations, and subsequently exposing them to empirical testing, the ten points of the left-right scale - on which the parties have been located - were collapsed into five ideological categories, presented in Table 1. The size of a specific category was

determined by the percentage of seats won by all the parties which fell
within its range on the scale, for each specific national election.5

Owing to the relative stability of parties along the left-right scale in both Denmark and The Netherlands, and the adoption of categories which condensed this scale into only five groups, it is difficult to find a party that has crossed the threshold between one category and another - even after more than four decades of policy changes and election strategies since World War 11. That is, a moderate left party such as the Danish Social Democrats (SD) might have been more centre-oriented at some period, or quite left-leaning at another period. However, the SD cannot be said to have ever been a party of the centre, nor one of the extreme left. The parties and their respective categories are presented in Table 2.

Centre Parties and Extremist Party Growth

Utilizing the five categories presented in Table 2 produces an extremely stable measure over an extended period of time. The subsequent statistical analysis will regress one category of parties against another, rather than specific parties; and therefore the movement of a particular party within its category will not impact on the results. Since there are no Danish parties which have crossed from one category to another, the use of these fixed categories for the analysis of 18 post-war elections is quite applicable.

Hazan's first explanation of the relationship between the parliamentary strength of centre parties and the level of party system polarization argued that there is a positive relationship between centre and extreme parties' share of seats, but only when the centre is large. In order to validate this explanation the Danish case should exhibit no relationship between these two categories, since all but one election awarded the centre parties less than 20 percent of the seats in the Folketing, thereby making it representative of the small centre category. The empirical evidence does support this explanation - a correlation of centre and extreme parties' share of seats produces an insignificant result.⁶ Moreover, regression analysis also produces relationship between the parliamentary strength of centre parties and the level of party system polarization in Denmark.⁷ While the centre's share of seats fluctuated between 6 percent and 22 percent - an increase of more than 350 percent - the level of polarization produced a horizontal cluster without any apparent slope. Both results are, therefore, as expected, and they validate Hazan's first explanation. However, in order to accept Hazan's first explanation, a positive relationship has to be exhibited by the Dutch case between centre and extreme parties' parliamentary strength.

These categories of parties are as stable during the 14 post-war elections of the Dutch case as they were in the Danish case. Only the Radical Party (PPR) crossed the threshold between two categories, but it is a minor party that secured an average of only 2 percent of the seats in the Tweede Kamer during the six elections it achieved representation. Owing to the large size of the Dutch centre, the relationship between the centre and the extreme parties is expected to be positive, since The Netherlands represents the large centre category. A correlation of centre and extreme parties' share of seats, however, does not exhibit the expected relationship and instead produces no relationship whatsoever.⁸ This is similar to the Danish result, which represented the small centre category. However, it is contrary to what Hazan found for the large centre category. Regression analysis also produces a different relationship from the one expected between the parliamentary strength of centre parties and the level of systemic polarization. relationship was expected to be positive, but instead the correlation is strongly negative.⁹

In summation, the small centre case study exhibited the lack of relationshipsexpected Hazan's first explanation of the association between centre parties and systemic polarization; whereas the large centre case study did not produce the expected relationships. Therefore, where no relationship was expected, none appeared. However, where Hazan's case is made, in the large centre category where a counter-intuitive relationship is expected, here, too, none was found. Despite the cross-national empirical evidence which Hazan used to substantiate his first explanation - based on election results from one decade - these two expanded case studies, both

of which produced no association between the centre and polarization, may point out that there actually is no relationship where Hazan found one. Moreover, the analysis of the Dutch case exhibited a relationship contradictoryto one Hazan found. In short, although the Danish case indirectly confirms Hazan's first explanation, the Dutch case contradicts it and thus places its validity in question.¹⁰

Centre Parties and Centrifugal Electoral Competition

The ability of my measures to test Hazan's second explanation - based on the movement of the moderate parties away from the centre - is somewhat circumscribed. The same characteristic that made it possible for me to use a mapping of parties for the entire post-war period also serves as a liability in attempting to assess the validity of Hazan's second explanation. My leftright of party locations is a static one, and does not allow for party movement along the continuum. Once a party has been located at a specific point along the scale, it remains there. However, at this stage the movement of parties assumes the highest level of significance. If the Danish SD has indeed changed from a centre-leaning moderate left party to one that is courting the extreme left, my measures will not be able to pick this up - and this is the crux of the second explanation. The only alternative is to check the movement of voters between two parties within one category, thereby assessing the radicalization or moderation of the entire category. This cannot be accomplished if there is only one party within the category, which is the case for several of the five categories in both party systems.

The solution is either to generate a dynamic left-right scale, which is unavailable and not within the scope of this research, or to assess the movement of parties along the scale based on scholarly literature and empirical analysis, which I have done. In order to validate Hazan's second explanation, I must uncover indicators that would show that a centrifugal movement of the moderate parties took place in both Denmark and The Netherlands. An analysis of the literature points in that direction.

In Denmark, when the Socialist People's Party (SF) broke off from the Communist Party (DKP) in 1958, it represented a more temperate version of an extreme left party and totally replaced the DKP in the Danish parliament.¹¹ The formation of SF on the extreme left helped bring about a "workers' majority" in the Danish Folketing for the first time after the 1966 election. SF and the Social Democrats (SD) together held 89 of the 175 parliamentary seats.¹² SF represented an extreme left, however, it was now more restrained ideologically and more powerful electorally, and could therefore no longer be maintained in a political ghetto by the moderate

left. SF was able to put an end to the ostracism of the extreme left which had doomed it to be looked upon as untouchable when it came to government formation. SD thus decided to reverse its previous policy of not collaborating with the extreme left, and moved leftward in order to cooperate with SF. A full-fledged coalition was still unlikely, but the two parties formed a "contact committee" and began close yet informal parliamentary cooperation on domestic policy. SF was even offered seats in the so-called "Red Cabinet", which they did not accept.

The possibility of an amalgamated socialist bloc, which would rely on extremist support, was anathema to the centrist Radical Liberals (RV). They were forced to ally with the two "bourgeois" parties in the attempt to undermine the minority SD government supported by the extremist SF. But, simultaneously with the outward movement of the moderate left, the moderate right also exhibited a centrifugal trend. In the early 19605, the Liberals (V) moved closer to their more reactionary partner in the moderate right category, the Conservatives (KF). Discussions concerning a closer alliance and even a fusion took place, but were never realized. The 1968 election placed these three parties in government, and forced KF to gravitate the centre in order to cooperate with V and RV. The supporters of KF, expecting to see their party reverse the policies of the previous 15 years of socialist governments, were disappointed. In response they punished party in the 1971 election when KF lost the largest proportion of seats since 1947, and was then halved again in the subsequent 1973 election. The centripetal trend was immediately reversed, and KF moved back to its original ideological position.

The leftward movement of SD, and the rightward movement of both RV and V (and subsequently KF as well) almost brought about the formation of a two-bloc party system in Denmark. Fitzmaurice (1981,122) wrote that "Denmark was moving towards, if not a two-party system, a two-bloc system on a classic right-left alignment." Borre (1980, 247) agreed, stating that "when the Socialist People's Party [SF] in 1966 grew to the status of a regular fifth member of the party system, a two-bloc system with a clear socialist-nonsocialist cleavage resulted."

Pedersen, Damgaard & Olsen (1971) found that a centrifugal movement of parties did indeed take place, more than just once, in the post-war Danish party system prior to the 1973 election. This conclusion was corroborated by Holmstedt & Schou (1987) who attempted to chart the movement of parties over time along the left-right scale basedon an analysis of party manifestos. The authors concluded that the moderate left (SD), and the moderate right (V) and the KF, moved away from each other in the late 1960s and early 70s, while the centrist RV maintained an almost constant middle position during the entire post-war period.

The Dutch case exhibits a similar, if not a stronger trend. The early 1970s

saw the introduction of a clear-cut strategy of polarization initiated by the moderate left Socialist Labour Party (PvdA). The Dutch Socialists are among the weakest of the European Socialist parties, and have traditionally faced a bloc of religious parties which straddles the centre and has held a majority in parliament. The PvdA, much like the moderate right Liberals (VVD), ended up as a junior partner in whatever coalition they were invited to join. Moreover, the religious parties usually played-off the PvdA and the VVD against one another, eliciting numerous concessions in exchange for participation in government. When the confessional parties fell below the majority threshold for the first time in 1967, both the moderate left the PvdA and the moderate right VVD saw the opportunity to achieve electoral gains. The breakdown of what had until then been an unassailable religious majority, coupled with an on-going process of deconfessionalization,motivated Socialists and Liberals alike to compete for the available voters in the centre. Irwin (1980, 216) stated that, "as the decline of the religious parties set in, the Socialists and Liberals found themselves joined in a common strategy of attempting to win over those voters who were leaving the confessional fold. This has led to a heightened polarization of Dutch politics."

A short-lived attempt at a centripetal strategy was tried, but quickly discarded. Immediately after the 1967 election the PvdA hoped either to establish a long-term alliance with the centrist Anti-Revolutionary Party (ARP), or to split the centrist Catholic People's Party (KVP). The former never materialized, and the latter resulted in the creation of the moderate left Radical Political Party (PPR) in 1968, which proved still too weak to provide the Socialists with a majority. The centripetal approach was subsequently abandoned largely due to two reasons: Schmelzer's Night and the subsequent creation of the New Left. Schmelzer's Night took place in October 1966, when the KVP dropped the PvdA from the coalition in midterm, continued to govern with the parliamentary backing of the VVD. The betrayal of the PvdA by the KVP stirred a group within the PvdA which crystallized into a faction called the New Left. This faction was composed of militant Socialists who sought to create a popular front with the parties of the extreme left - the Communist Party (CPN) and the Pacifist Socialist Party (PSP). Their influence could not be restrained and helped bring about the quick demise of the centripetal tactics. The growing strength of New Left had an added effect in that it alienated many moderate Socialists who then split from the PvdA and spurred the formation of the centrist Democratic Socialists '70 (DS'7O), which intended to continue the moderate line abandoned by the PvdA. This split left the PvdA in the hands of the New Left and its allies. According to Irwin & van Holsteyn (1989, 112), "instead of resigning itself to the rules of the 'politics of accommodation' it [the New Left] set upon a policy of polarization."

The PvdA's party executive began to pass resolutions committing the party to more leftist policies. Stringent attacks on the religious-liberal cabinet were mounted, which culminated just prior to the 1971 election, when the PvdA, the PPR, and Democrats '66 (D'66) formed a progressive alliance and presented a programme together with a slate of ministers. The same alliance presented itself when new elections were called the following year. This time the PvdA sought support from the extremist PSP and CPN as well, with which it had formed coalitions on the local level, but D'66 vetoed their incorporation. The focus of this polarization strategy was to shift from parliamentary to electoral politics. In other words, instead of the parliamentary parties forming a coalition based on election results, the parties would agree on coalition partners prior to the election and offer the electorate a clear choice. Implicit in this strategy was that the religious parties would become part of the right bloc; however, this attempt at polarization was also aimed at the middle-class progressive supporters of the confessional parties who, it was hoped, would switch their support to the left bloc if the religious parties moved to the right. The result would be a two-bloc system, where all non-bourgeois religious voters would defect to the progressive bloc.

The VVD, which also held hopes for an exclusive majority, began to consider turning to the fundamentalist parties of the extreme right for support. However, due to the large gap between the combined seats of the right and extreme right parties on the one hand, and an independent parliamentary majority on the other hand, the VVD had to tread more softly than the PvdA. They stood to gain from the growing alienation between Socialists and confessionals because they were now the only likely coalition partners for the religious parties. However, the VVD also wanted to capture the newly available centre voters, and thus adopted a tactic of convincing former confessional voters that only a vote for the VVD could prevent the re-emergence of a religious-socialist coalition. The VVD, therefore, also took part in the centrifugal trend based on a strategy of polarization (De Jong & Pijnenburg 1986). It deliberately sought to broaden its appeal to include not only middle-class and religious liberals, but also conservative voters who supported the extreme right, as well as those voters who were opposed to socialism. As Daalder (1979, 185) stated, "Socialists and Liberals thus became tacit allies in a non-zero sum electoral game at the expense of the religious parties. Although they remained dependent on a post-election coalition with the religious parties for the formation of cabinets, they had every incentive to polarize the vote at election time."

This pattern of polarization by moderate parties on both the left and the right is described also by Dittrich (1987), who has analysed the manifestos of the Dutch parties during the post-war period and assessed how their positions moved over the years on the left-right continuum. He generated

a scale that shows how in the late 1960s the PvdA broke sharply to the left followed by a sharp break to the right by the VVD, creating a gap between the moderate left and moderate right covering almost one-half the range of the entire scale - thereby more than doubling the usual distance between them. Survey data based on the issue positions of parliament members corroborated these findings and showed that by the late 1970s the PvdA and VVD placed themselves furthest apart on practically all issues (Daalder 1987, 207-208).

Combating "Moderate-Induced Polarization"

Hazan's second explanation of the relationship between the centre and polarization, based on a centrifugal movement by the moderate parties, has been corroborated by both the Danish and Dutch cases. The question that beckons is whether the strategy of polarization introduced by the moderate parties, aimed at the centre, succeeded or failed? And, did the centre play a role in the outcome?

The polarization of the Danish party system, with its roots in the 19605, came to an end after the "electoral earthquake" of 1973. The centrifugal movement of both moderate left and moderate right parties, not to mention the centre party, created the openings which were filled in the tumultuous election that year.¹³ The appearance, or reappearance, of five additional parties - thereby doubling the number of parties represented in parliament, raising fractionalization to its highest point ever, and ranking topmost of all Western European postwar elections in aggregate electoral volatility - was an obvious sign that the Danish voters were opposed to the two-bloc constellation that was emerging. Over one-third of the seats went to these heretofore absent parties - more than half of which went to extremist parties - while all five main parties lost seats. The incumbent parties took notice, and within a relatively short time the tripolar (left-centre-right) and centripetal pattern of interaction was reinstated.

Immediately after the 1973 election, the leaders of the major parties realized that their polarizing tactics had brought on not only volatility but instability as well. The spectrum of possible coalition partners was shortened, once again, to exclude the extremist parties. Both moderate right and moderate left party leaders recognized the limitations of coalition making, and their need to govern with the centre. The ensuing government was based solely on V, with less than 13 percent of the parliamentary seats, which built ad hoc legislative coalitions with each party that was willing to offer support. The 1975 election produced another single-party minority government, this time based on SD, which courted similar legislative coalitions. The emerging pattern exhibited moderate party movement away from the extremes and cooperation across the centre.

The moderate parties had come to acknowledge several factors: (i) they would not be able to build a majority on their own; (ii) they could not govern with their extremist neighbours without incurring a backlash; and thus, (iii) cooperation across the centre was essential. The logic of these factors brought about a rapprochement between SD and V, who together formed a coalition after the 1977 election, thereby signifying an end to their strategy of polarization and the adoption of a centripetal trend. The "cynics of the left and right", as Rusk & Borre (1974, 353) characterized them - those voters who traditionally supported the moderate left SD, or moderate right V and KF - switched to either extreme or centre parties. "These people were the vote-switchers who established, at least temporarily, cleavage lines in the system." They were the voters whom the moderate parties sought to recapture by initiating a centrifugal movement along the left-right continuum. The fact that this movement lost them even more voters stopped them from continuing their outward pattern and returned them to a more moderate course.

In The Netherlands, the "deliberate mutual polarization between the Socialists the Liberals", as Daalder (1979, 190) described it, placed the centrally located religious parties at a point where they faced competition for their previously stable electorate from both sides. After the 1972 election, the new Dutch government was made up of the PvdA, PPR, D'66, KVP, and the ARP. The progressive parties held ten ministries, while the religious parties were allotted only six. The severance of the Christian Historical Union (CHU) - which was forced into the opposition - from the other two religious parties in government elevated the Socialists' hopes that a split could be orchestrated between the left-oriented confessionals and their more conservative brethren. The result was exactly the opposite. The inferior status that the ARP and KVP received in the socialist-religious coalition provided the impetus for the three religious parties to discuss joining forces in future elections. Within less than a decade the religious parties had gone from possessing a majority to being a minor coalition partner. This rapid decline produced an atmosphere conducive to collaboration, after much consultation a joint list called the Christian Democratic Appeal (CDA) was presented for the 1977 election. This joint list successfully stopped the haemorrhaging of religious electoral support, and the three parties - which had previously begun discussing a possible merger as early as 1968 - received impetus and formally merged prior to the subsequent election.

The 1977 election was a landslide victory for the PvdA which gained 10 new seats- an increase of 23 percent- but their gains were not from where they had hoped they would come. The formation of the CD A had stopped the centre parties' loss of seats to the Socialists. The extreme left parties, on the other

hand, were diminished from 16 seats in 1972 to only six in 1977. The gains made by the PvdA thus appear to have come mainly at the expense of the extreme left parties, thereby not increasing the total strength of the left bloc. However, the PvdA still perceived the 1977 results as a victory and enteredthe negotiations with stringent demands. The 1977 election, though, brought gains not only to the PvdA and CD A, but to the VVD as well. The CDA and VVD now held a majority, and formed the next governmentcoalition. PvdA had won the election but lost the government.

Realizing that their gains were not from the centre and that both the liberals and the religious bloc had registered gains as well, the PvdA came to the conclusion that the polarization strategy it had adopted in three earlier elections had backfired: (i) the left bloc was nowhere near a parliamentary (ii) the right bloc had expanded its share of the seats by a larger ratio; and, (iii) the religious centre bloc had successfully confronted the polarization strategy and managed not only to hold on to both its pivotal position and central location, but to consolidate and augment electoral support. In short, the polarization strategy aimed at splitting or destroying the diminishing religious centre helped push it down the road to merger. The result was that by 1977 the CDA held the prime ministership, and that by 1981 it was the biggest party in the Tweede Kamer.

In the aftermath of the failed polarization strategy, the PvdA tried to regain the moderate left position it had forsaken a decade earlier. Its stated policy in the 1980s was a coalition with the CDA and possibly D'66, but not with the extremist PPR, PSP, or CPN. In the long run the centrifugal forces were defeated and the Dutch parties returned to their centripetal tactics. As Daalder (1989,14) stated, "there have been no lasting centrifugal drives; a deliberate attempt to 'polarize' the party system by setting up opposing coalitions of parties, has come to nought and has ended in all major parties returning to a policy of wooing votes at the centre of Dutch politics"¹⁴ (italics in original). The formation of a socialist-religious coalition after the election that followed in 1989 led Wolinetz (1990, 286) to argue that "it is possible that the 1989 election will have marked the definitive end of the polarization strategy which the Socialists pursued in one form or another from 1969 to the mid-1980s and the resumption of a pattern of centre-left cooperation characteristic of the 19505."

In both Denmark and The Netherlands the moderate parties introduced a strategy of polarization aimed at the centre, with the goal of either forcing it to align with one of the two emerging blocs or breaking it apart. In both countries polarization failed to achieve its desired goals. The centre parties' counter-strategy was, however, quite different in each country.

In Denmark, the centre's strategy involved tactical shifts between the
two polarizing blocs. The biggest and oldest centre party, RV, led the other

two new centre parties, the Centre Democrats (CD) and Christian People's Party (KRF), in the adoption and execution of a moderating role. At the outset of the polarization pattern, initiated by the SD, the centre allied with the moderate right in order to oppose the moderate left governing with the support of the extreme left. However, as soon as the moderate and extreme left fell below a parliamentary majority, the centre detached itself from the moderate right and once again adopted a central position agreeing to negotiate with both poles as long as the extremes were excluded.

Moreover, since 1971 all Danish governments have been of minority status. This has placed them in the precarious situation of relying on the support of the centre in order to avoid an "alternative majority" in the Folketing. However, the centre parties have not allowed this to destabilize the party system or the government. As long as the minority government cooperates with the centre, the centre parties oppose the adoption of policies that could be passed by the majority in opposition. In other words, the centre parties have taken steps to stabilize the centre-based party system. According to Bille (1989, 53), the centre parties "took a clear stand against extremist tendencies, constantly emphasizing the need to reach agreements 'across the centre'. For these parties, it became an ideology and a raison d'etre to compromise and to mediate."

These tactics, supported by the electoral results, played a crucial role in halting the polarization drive toward a two-bloc party system in Denmark, and the re-emergence of a tripolar and moderate system. The centre has therefore been able to achieve the following: (a) to maximize its political influence; (b) to achieve a renewed level of moderation in the party system; and, (c) to maintain stability despite the minority status and weak nature of the governments.

In The Netherlands the decline of the religious parties during the period of de-confessionalization produced two developments. The first was the emergence of additional non-religious centre parties. These parties made it possible for the traditional religious voter to avoid the choice between remaining a religious voter and becoming an anti-religious one. The new minor centre parties, D'66 and DS'7O, allowed the previously confessional voters to switch parties without immediately supporting those parties which carried a tradition of animosity toward their former affiliation. Moreover, these minor parties quickly adopted a centre position and functioned much like the religious parties in their advocacy of centre-based coalitions.¹⁵ Indeed, there is no direct relationship between the decline of the religious parties and the rise of the PvdA and VVD, while the best gains of the secular parties - the 1977 election - came at the point when the confessional losses had been successfully halted. Therefore, the shift of previously confessional supporters to the new minor centre parties allowed the centre to maintain its overall parliamentary strength.

The second development was the merger of the three religious parties into the CD A, which has attempted to reduce its confessional dimension, assume several characteristics of a true catch-all centre party, and increase its share of the secular vote. This merger, therefore, brought about the creation of a strong centre party that could not be split between the right and left blocs, and it produced a moderate point of attraction at the centre of the party system for both confessional and secular voters. The strategy of polarization was aimed at a declining religious centre which could compete only "defensively" to maintain its voter base. In response to this strategy, the dwindling confessional centre was replaced by a growing mixed religious-secular centre which now was able to undertake "expansive" competition in the search for new voters. As Daalder (1986, 530) pointed out, "a centre rapidly losing its traditional voting support, potentially replacing it by a plebiscitary appeal, is an ironic outcome of the process of change ... in which the left tried to introduce the principle of alternative government in a system where it did not command enough votes to make a success of its policy of 'democratic' polarization."

Discussion

The centres in both countries have enjoyed a level of political influence far beyond their relative size, but this is especially true in the case of the Danish centre. There are two main reasons for this: position and performance. refers to the location of the centre parties between the two opposing ideological blocs, neither of which is able to gain a parliamentary majority. Performance refers to the ability of the centre parties to negotiate with both blocs, and to form the pivot of alternating coalitions with either. However, the exaggerated level of political influence attained by the centre also made it a target for destruction.

Whereas the centre parties seem content to continue their pivotal role for the foreseeable future, neither of the major blocs to their right and left appears willing to do so. The moderate right and moderate left are forced to play by the centre's rules, and both must give up some of their political leverage in order to form a government with the centre. The dream of both moderate camps, though, is to govern alone. Their inability to pursue and enact their interests, and the constant concessions made to the centre, lead to a situation of enhanced competition from the extremes. The extremes capitalize on the impotence of either moderate pole to push through its agenda, as well as on the exaggerated influence awarded to the centre. However, this study has not found any relationship between the parliamentary of the centre parties and that of the extreme parties. The growth of the extreme parties, if such a trend is present, is not related to the centre - contrary to Hazan's first explanation.

The moderate parties, much like the centre parties, found themselves squeezed from both sides. Their only choice became the direction of competition - centripetally in order to attract centre voters and forsake the extreme ones, or centrifugally in order to recapture the more radical voters and attempt to split the centrist ones? Their decision supports Hazan's second explanation of the relationship between the centre and polarization - in both cases the moderate parties moved away from the centre. However, the centrifugal trend adopted by the Dutch moderate parties was more acute than that adopted by the Danish ones. The divergence can be seen by referring to the difference in party behaviour: pre-election electoral strategies in The Netherlands versus post-election parliamentary tactics in Denmark.

The Danish case showed that the moderate parties - SD, V, and KF - moved away from the centre. However, this centrifugal process began only after the 1966 election gave the moderate left SD and the extreme left SF a "workers' majority" for the first time. There were no attempts by SD to court its extremist fringe prior to the election. On the contrary, the extreme left had become accustomed to the cold-shoulder policy of the moderate left. When these parties did indeed begin to cooperate, the extremist SF never became a full coalition partner. It simply provided parliamentary support for a minority SD government. Only when the voters provided these parties with a majority in parliament, and thereby a credible opportunity to transform the tripolar and centre-based party system into a two-bloc system, did they adopt a strategy of polarization. When the electorate ceased to provide results that were conducive to two-bloc politics, the polarization tactics were abandoned by the parties.

The Dutch case, on the other hand, also showed the moderate parties - PvdA and VVD - moving away from the centre, but this was a predeterminedand pre-election tactic aimed at providing the voters with a clear choice between the two-blocs of party alliances. When the PvdA, D'66, and PPR presented a joint programme and a coordinated slate of ministers to the voters prior to the 1971 and 1972 elections, these parties were far from the threshold of a parliamentary majority. The polarization strategy in the Netherlands was not the consequence of specific election results, but was adopted in order to cause a particular electoral outcome. The moderate left actively courted the parties on the extreme left before the elections, and had the stated goal of pushing the centre to ally with either the left or the right - a bipolar system. The coalition formed after the 1972 election included a party that was quickly becoming extremist in nature, the PPR, and which remained part of the government well past the point when it crossed into the extreme left camp. In other words, not only was this a deliberate pre-election strategy, it also embraced extremist elements much more willingly than in the Danish case. When the centre

refused to align with one of the two emerging blocs, and decided to confront the polarization head-on by merging the three religious parties into a seemingly catch-all centre party, only then was polarization forsaken. Therefore, the strategy of polarization was enacted due to systemic reasons and goals, but was manifested by electoral tactics. It was abandoned for the same systemic considerations, regardless of the electoral gains it had accumulated for both the moderate right and the moderate left in the shortterm.This is almost the inverse of the Danish case.

In both Denmark and The Netherlands the centre found itself caught between two poles that no longer sought to cooperate with it, however, it still based its appeal on its continued ability to cooperate with both. Moreover, the centre refused to associate with the extremes, which further hindered its continued possession of the broker role during the period of moderate-extreme collaboration. The centre thus presented itself as the defender of the old system - the centre-based moderate and consensual party system. It promised to continue its policies of negotiating with either bloc, and of not joining any government based on the support of extreme parties.

At this stage the existence of a centre and the performance of the centre became two opposing and off-setting patterns. The existence of a centre in a multiparty system with neither a left nor a right majority - the definition of a pivotal party - elicited a pattern of centrifugal polarizing competition introduced by the moderate parties and meant either to force the centre toward one of the two poles, or to eradicate it. The performance of the centre, on the other hand, was always that of a moderating factor, and during the strategy of polarization this became the sine qua non of its tactics and appeals for centripetal moderating competition.

Both in Denmark, where the centre is small, and in The Netherlands, where the centre is large, polarization was the strategy adopted by the moderate parties in order to attack the centre and win over the centre voters. Even in the case of Denmark, the small centre was able to alter the competitive configuration of the party system at least twice - once as the stimulus for the adoption of polarization, and once as a force behind the return to moderation. In The Netherlands the impact of the centre was similar, with two caveats: first, the strategy of polarization was adopted only when the centre fell below majority status; second, the polarization was more extensive than in Denmark. This leads to the conclusion that in those cases with large centre parties the strategy of polarization adopted by the moderate parties will be more acute and divisive. That is, the strength of the attempt at polarization, initiated by the moderate parties, is positively to the size of the centre parties. However, when the centre passes the threshold of a majority the moderate parties will perceive it to be too strong and stable, and will not attempt to polarize the party system.

Polarization can, therefore, be related to the centre. However, it appears as though the mere existence of centre parties is not sufficient to produce polarization. As long as the centre parties in Denmark and The Netherlands were part of a stable and structured party system, there was little or no polarization. When instability was introduced into the party system, as a result of socio-economic transformations in the polity which produced enhanced electoral volatility, only then did polarization manifest itself. In short, when the centre no longer appears to be impenetrable, only then will the moderate parties attempt to attack it by polarizing the system. Furthermore, the larger the centre parties are at the moment of vulnerability harsher the strategy of polarization will be. But, if the centre is able to hold onto its pivotal position, while performing in accordance with a moderating doctrine, it can successfully counter the polarization attempt.¹⁶ As both Denmark and The Netherlands show, the moderate parties did not achieve their goal of a majority and abandoned their centrifugal strategies, reversing their movement back toward the centre.

Conclusion

The potential of polarization might be a constant presence in all party systems. However, these case studies lead to the conclusion that polarization manifested by the moderate parties, is not related to the extreme parties, but is nonetheless stimulated by the centre parties. The centre parties in Denmark and The Netherlands are thus partially responsible for the emergence of "moderate-induced polarization" aimed at bringing about a bipolar party system; but, the actions and performance of the centre parties prevented the permanent transformation of the party system into a polarized one. Through the emergence of new centre parties, the merger and adaptation of older ones, and the adherence to a moderating role the centre parties in Denmark and The Netherlands succeeded in returning their party systems to a less polarized, albeit less moderate, constellation.

This situation leads me to conclude that Sartori's counter-intuitive analysisof centre is more valid than that of Duverger's entirely intuitive assessment. Nonetheless, my findings show some inconsistencies in Sartori's framework.¹⁷ Sartori stated that a centre tendency can exist, but only when there is no centre party. A system is characterized by a centre tendency (centripetal electoral competition) when the centre is vacant, that is, when the moderate and centrally located electorate does not have a party to identify with. My findings have shown that it is not necessarily an either/ or phenomenon. A party system with centre parties can still exhibit a centre tendency. The party system does not need a vacant centre in order to be characterized by a centre tendency. A centre tendency, on the other hand,

does not necessitate a centrally located and unidentified floating electorate. Moreover, while according to Sartori, the existence of a centre discourages centripetal competition, my findings point out that a centre can combat this trend and become conducive to centripetal competition. More precisely, the presence of parties in the centre discourages other parties, i.e. the moderate parties, from adopting a centripetal strategy to gain new voters. In short, the very existence of a strong centre party, notwithstanding its impact on increased systemic polarization, does not discourage "centrality". Furthermore,in to Duverger, this research has found that the centre is not an artificial grouping of moderates. Had this assumption been true, then the strategy of "moderate-induced polarization" should have torn apart the centre, which it failed to do in both cases.

In summation, this article has shown that the centre can stimulate "moderate-induced polarization". However, government durability and democratic stability, on the one hand, and the existence of centre parties, on the other hand, are not mutually exclusive. "Moderate-induced polarization", opposed to extremist party growth, can be perceived as part of the natural mechanics, drastic as it might seem, of multiparty competition. Its most significant consequence would be the transformation of a multipolar system to a bipolar configuration, rather than the loss of government stability or the threat of a possible crisis of democracy.

---

NOTES

1. This research presupposes that polarization can be measured, and that political parties can be characterized in terms of their ideology and positions on a left-right socioeconomic That is to say that party system profiles can be generated and translated into quantitative scores basedon a spatial notion of party systems. Limiting spatial analysis to a single dimension has encountered criticism, yet scholars have come to widely endorse it as a viable method for political analysis. The case for a unidimensional left-right continuum, as advanced by Downs, is supported by a significant body of literature which provides evidence of a single dominant left-right dimension. For arguments in favour of unidimensionality, see: De Swaan (1973), Mavrogordatos (1987), and Laponce (1981).

2. "Country experts" were asked to place all the parties represented in their national parliaments on an ideological ten-point scale ranging from zero representing the ultraleft, ten representing the ultra-right.

3. The weighted system mean of seat distributions, for a particular election, is attained by multiplying each party's percentage of the vote by its position on the left-right scale and summing. I correlated both my scores of polarization and my weighted system means with other indices from the literature with the following results: Measure Index r R2R² SigT N Polarization Sigelman & Yough 0.601 0.361 0.05 11 Polarization Ersson & Lane 0.757 0.573 0.01 13 Weighted Mean Sigelman & Yough 0.841 0.708 0.01 11 Sigelman and Yough's (1978) measures were basedon the decade of the 19605, Ersson and Lane's (1982) measure was basedon the post-1945 period.

---

4. During my statistical analysis, all the parties located between four and six on the ten-point left-right scale will be considered centrally positioned parties. For the most comprehensive inventory of the different meanings of the centre, and a critique of the concept, see: Daalder (1984). I strongly agree with his concluding remark that, "We clearly need an explicit study of normative assessments of centre parties. We have frequently drawn attention to the many unspecified value judgments associated with the treatment of notions like the centre and centre parties. A careful review of such value judgments might help us to obtain clearer insights into alternative theories of good government, which after all lay a heavy hånd on our discussions and comparative research on the empirical functioning of party systems" (p. 108). For a new attempt at defining and discussing the concept of the centre, see: Hazan (1994, 1995a).

5. Castles and Mair divided their ten-point left-right scale with smaller spaces allocated to the extremes than to the moderates and centre. Their division resulted in parties such as SF and FRP in Denmark and PPR in The Netherlands all being classified as moderate parties, and not extremist parties, which would be more appropriate. My division allows each category to encompass a similar amount of space on the continuum while simultaneously reducing the amount of misclassified parties.

6. r = 0.237; R2R² = 0.056; b = 0.534; significant T = 0.345.

7. r = 0.152; R2R² = 0.023; b = 0.032; significant T = 0.546.

8. r = 0.104; R2R² - 0.011; b = 0.039; significant T = 0.724.

9. r = -0.690; R2R² = 0.476; b = -0.048; significant at the 0.01 level.

10. The empirical evidence used by Hazan (1995b) in his cross-national study covered the decade between 1979 and 1989, the data in this study cover the entire post-war period until 1990.

11. DKP returned to the Folketing later, as did the Left Socialists (VS) - another ultraleft established in opposition to the more restrained SF. However, both DKP and VS failed to survive and most of their seats were recaptured by SF.

12. This does not include the four seats allocated to the Faroes and Greenland, whose representatives have large ly remained neutral in the formation of Danish coalitions, thus bringing the majority threshold down to 88 seats.

13. In the vacant extreme right position the Progress Party (FRP) emerged; in the centre the Justice Party (DR), Centre Democrats (CD), and Christian People's Party (KRF) appeared, due to the rightward movement of RV; in the extreme left the DKP returned to occupy the ultra left, vacated by the less radical SF (in the following election VS reappeared as well).

14. The end of the strategy of polarization became apparent when the PvdA refused to enter into a tactical electoral alliance for saving extra votes with the extreme left parties prior to the 1986 election.

15. It is true that D'66 joined the progressive alliance prior to the 1971 and 1972 elections which included the PPR, but it was due to the objections of D'66 that the extremist CPN and PSP were not included in the alliance. Moreover, at the time the PPR was still considered by most to belong to the moderate left, and only since 1977 has the party become part of the extreme left.

16. Scully's (1992) recent analysis of the centre in Chile refiects similar tindings.

17. The only empirical analysis of Sartori's framework and its implications for the electoral dynamics of multiparty competition is Powell (1987), who found empirical support for several of Sartori's identifying properties of "polarized pluralism". However, Powell's analysis found that the centre occupation measure was the least satisfactory, and he failed to observe centrifugal electoral behaviour - defined as the loss of centre votes to the extremes. This is in line with my findings, but the exhibition of centrifugal electoral behaviour through an outward movement of the moderate parties was not analysed by Powell.

ACKNOWLEDGEMENTS

I thank Giovanni Sartori, Lewis Edinger, Douglas Chalmers, David Lawrence, Robert Shapiro, Donald Hancock and Herbert Kitschelt for their invaluable comments, suggestions and guidance during the numerous drafts of this article. An earlier version was presented at the 1992 Annual Meeting of the American Political Science Association.

REFERENCES

Bille, L. 1989. "Denmark: The Oscillating Party System", West European Politics 12, 42-58.

Borre, O. 1980."The Social Bases of Danish Electoral Behaviour", in Rose, R., ed., Electoral
Participation: A Comparative Analysis. Beverly Hills: Sage.

Castles, F. & Mair, P. 1984. "Left-Right Political Scales: Some 'Expert' Judgments", European
Journal of Political Research 12, 73-88.

Daalder, H. 1971. "Cabinets and Party Systems in Ten Smaller European Democracies", Acta
Politico 6, 282-303.

Daalder, H. 1979. "The Netherlands", in Henig, S., ed., Political Parties in the European
Community. London: George Allen & Unwin.

Daalder, H. 1984. "In Search of the Center of European Party Systems", American Political
Science Review 18, 92-109.

Daalder, H. 1986. "Changing Procedures and Changing Strategies in Dutch Coalition
Building", Legislative Studies Quarterly 11, 507-531.

Daalder, H. 1987."The Dutch Party System: From Segmentation to Polarization - And
Then?", in Daalder, H., ed., Party Systems in Denmark, Austria, Switzerland, The Netherlands,
Belgium. London: Frances Pinter.

Daalder, H. 1989."The Mould of Dutch Politics: Themes for Comparative Inquiry", West
European Politics 12, 1-20.

De Jong, J. & Pijnenburg, B. 1986. "The Dutch Christian Democratic Party and Coalition Behaviour in the Netherlands: A Pivotal Party in the Face of Depolarization", in Pridham, G., ed., Coalition Behaviour in Theory and Practice: An Inductive Model for Western Europe. Cambridge: Cambridge University Press.

De Swaan, A. 1973. Coalition Theories and Cabinet Formations: A Study of Formal Theories
of Cabinet Formation Applied to Nine European Parliaments After 1918. San Francisco:
Jossey-Bass Inc.

Dittrich, K. 1987."The Netherlands 1946-1981", in Budge, 1., Robertson, D. & Head, D.,
eds., Ideology, Strategy and Party Change: Spatial Analyses of Post-war Election Programmes
19 Democracies. Cambridge: Cambridge University Press.

Dodd, L. 1976. Coalitions in Parliamentary Government. Princeton: Princeton University
Press.

Downs, A. 1957. An Economic Theory of Democracy. New York: Harper & Row.

Duverger, M. 1959. Political Parties: Their Organization and Activity in the Modern State.
New York: John Wiley & Sons.

Ersson, S. & Lane, J. 1982. "Democratic Party Systems in Europe: Dimensions, Change and
Stability", Scandinavian Political Studies 5, 67-96.

Fitzmaurice, J. 1981. Politics in Denmark. London: C. Hurst & Co.

Hazan, R. 1994. "Partiti di Centro e Partiti Centrali: Una Chiarificazionne Concettuale",
Rivista Italiana di Scienza Politico 24, 333-348.

Hazan, R. 1995a. "Does Center Equal Middle: Towards A Conceptual Delineation, With
Application to West European Party Systems", Party Politics (forthcoming).

Hazan, R. 1995b. "Center Parties and Systemic Polarization: An Exploration of Recent
Trends in Western Europe", Journal of Theoretical Politics (forthcoming).

Holmstedt, M. & Schou, T. 1987. "Sweden and Denmark 1945-1982: Election Programmes in the Scandinavian Setting", in Budge, 1., Robertson, D. & Hearl, D., eds., Ideology, Strategy and Party Change: Spatial Analyses of Post-war Election Programmes in 19 Democracies. Cambridge University Press.

leraci, G. 1992. "Center Parties and Anti-System Oppositions in Polarized Systems", West
European Politics 15, 17-34.

Irwin, G. 1980."Patterns of Voting Behaviour in the Netherlands", in Griffiths, R., ed., The
Economy and Politics of the Netherlands Since 1945. The Hague: Martinus Nijhoff.

Irwin, G. & Van Holsteyn, J. 1989. "Towards a More Open Model of Competition", West
European Politics 12, 112-138.

Keman, H. 1994. "The Search for the Centre: Pivot Parties in West European Party Systems",
West European Politics 17, 124-148.

Laponce, J. 1981. Left and Right: The Topography of Political Perceptions. Toronto: University
Toronto Press.

Mavrogordatos, G. 1987. "Downs Revisited: Spatial Models of Party Competition and Left-
Right Measurements", International Political Science Review 8, 333-342.

Pedersen, M., Damgaard, E. & Olsen, P. 1971."Party Distances in the Danish Folketing
1945-1968", Scandinavian Political Studies 6, 87-106.

Powell, G. 1981. "Party Systems and Political System Performance: Voting Participation,
Government Stability and Mass Violence in Contemporary Democracies", American Political
Review 75, 861-879.

Powell, G. 1982. Contemporary Democracies: Participation, Stability, and Violence.
Cambridge: Harvard University Press.

Powell, G. 1986. "Extremist Parties and Political Turmoil: Two Puzzles", American Journal
of Political Science 30, 357-378.

Powell, G. 1987. "The Competitive Consequences of Polarized Pluralism", in Holler, M.,
ed., The Logic of Multiparty Systems. Dordrecht: Martinus Nijhoff.

Rusk, J. & Borre, O. 1974."The Changing Party Space in Danish Voter Perceptions, 1971-1973",
Journal of Political Research 2, 329-361.

Sanders, D. & Herman, V. 1977. "The Stability and Survival of Governments in Western
Democracies", Acta Politico 12, 346-377.

Sani, G. & Sartori, G. 1983. "Polarization, Fragmentation and Competition in Western
Democracies", in Daalder, H. & Mair, P., eds., Western European Party Systems: Continuity
and Change. London: Sage.

Sartori, G. 1966."European Political Parties: The Case of Polarized Pluralism", in La-
Palombara, J. & Weiner, M., eds., Political Parties and Political Development. Princeton:
Princeton University Press.

Sartori, G. 1976. Parties and Party Systems: A Framework for Analysis. Cambridge: Cambridge
University Press.

Scully, T. 1992. Rethinking the Center: Party Politics in Nineteenth- and Twentieth-Century
Chile. Stanford: Stanford University Press.

Sigelman, L. & Yough, S. 1978. "Left-Right Polarization in National Party Systems: A Cross-
National Analysis", Comparative Political Studies 11, 355-379.

Taylor, M. & Herman, V. 1971. "Party Systems and Government Stability", American
Political Science Review 65, 28-37.

Wolinetz, S. 1990."The Dutch Election of 1989: Return to the Center-Left", West European
Politics 13, 280-286.

Van Roozendaal, P. 1990. "Center Parties and Coalition Cabinet Formations: A Game
Theoretic Approach", European Journal of Political Research 18, 325-348.