Movements and groups abound in modern society. Sometimes, a movement or group succeeds in mobilizing a large section of a country’s population and thoroughly knitting it together, by building a pervasive subculture and by setting up a vast interrelated network of organizations, resulting in a segmented and powerful bloc. This happened to different degrees in most Western countries, including the United States and Canada. It is also occurring now in the non-Western world. Belgium and the Netherlands were particularly affected by extensive bloc building. As a consequence, the phenomenon has been studied more thoroughly in these countries than it has elsewhere, and under a specific label: ‘verzuiling’ (pillarization). In both countries, Catholic, Socialist, and Liberal pillars – plus a Protestant pillar in the Netherlands – divided society and determined political and social life from the late nineteenth century up to the late twentieth century. Reviewing research of these pillars and of pillarization in Belgium and the Netherlands and of similar phenomena and perspectives can help not only to advance the study of the formation of segmented and organized sections of the population elsewhere in the world, but also to acknowledge the significance of this specific type of collectivity in modernity.

In the West, bloc building of sections of a country’s population was common throughout the past two centuries. As a result, many labels exist to refer to these blocs – pillars, camps, fortresses, ghettos, milieus, worlds, political cultures, subcultures, sub-societies… Before entering into the pillarization debate proper, let me first give three impressionistic, older descriptions of these blocs to convey a feeling for the subject. In 1910, the British sociologist Seebohm Rowntree depicted Belgian society and politics in the following terms: “There is extraordinarily little social intercourse between Catholics and Liberals, and practically none between Catholics and Socialists. Politics enter into almost every phase of social activity and philanthropic effort, and it is the exception, rather than the rule, for persons holding different political opinions to co-operate in any other matter. Thus, in one town there will be a Catholic, a Liberal, and a Socialist trade union, a Catholic, a Liberal, and a Socialist co-operative bakery, a Catholic, a Liberal, and a Socialist thrift society all catering for similar people, but each confining its attentions to members of its own political party. The separation extends to cafes, gymnasia, choral, temperance, and literary societies, indeed it cuts right through life” (Seebohm Rowntree 1910: 24). Another description of what became known shortly afterwards as the Dutch Catholic ‘pillar’ came from the historian Rogier (1956: 613): “Thus the curious… constellation of an almost autarchic Catholic religious community has emerged within which one not only votes in political party discipline, subscribes to a Catholic newspaper, a Catholic women’s magazine, a Catholic illustration and a Catholic youth magazine, allows one’s children to enjoy purely Catholic education from nursery school to university, but also listens to the radio in a Catholic context, travels, insures one’s life, engages in art, science and sport”. Herberg (1955: 168–169) wrote about Catholics in the United States: “The Catholic Church in America operates a vast network of institutions of almost every type and variety. The social and recreational activities in the Catholic parish – from baseball teams to sewing circles, from bowling leagues to religious study groups – are only a beginning. Every interest, activity, and function of the Catholic faithful is provided with some Catholic institution and furnished with Catholic direction. There are Catholic hospitals, homes, and orphanages; Catholic schools and colleges; Catholic charities and welfare agencies; Catholic Boy Scouts and War Veterans; Catholic associations of doctors, lawyers, teachers, students, and philosophers; Catholic leagues of policemen, firemen and sanitary workers; Catholic luncheon clubs and recreation fellowships; there is the Catholic Youth Organization, with some six million members; there is even a Catholic Audio-Visual Educators Association. This immense system constitutes at one and the same time a self-contained Catholic world with its own complex interior economy and American Catholicism’s resources of participation in the larger American community” (see also Greeley 1962 for a description of the “invisible ghetto” of a US Catholic parish around 1960). The similarity between these three quotes is striking – and many more could be given (for Austrian, Belgian, and German Socialist blocs, see Diamant 1974: 151; Vandervelde 1925: 456; Evans 1982: 19 respectively). Of course, these are all ideal-typical descriptions. In practice, most of the different blocs’ constituencies did not fit so easily into the mold. As Luyckx (2000) observed in respect of Dutch Catholics, there were many ‘other Catholics’. It is also true that the societal segregation of these blocs was less absolute than terms like ‘pillar’, ‘world’ or ‘sub-society’ seem to suggest – the economy, science and technology were less affected and class divisions and types of occupation had a greater impact on daily life than the cultures of the blocs. Nevertheless, the formation and maintenance of these blocs were impressive achievements. They had major effects on people’s lives and on society. It is no wonder therefore that they were perceived by friend and foe alike as comprehensive ‘self-contained worlds’ (to use Herberg’s expression).

In Belgium and the Netherlands – and in a number of other countries (cf. infra) – these worlds extended into the political sphere (with national political parties) and the socio-economic sphere (with trade unions and a host of professional organizations). This distinguishes them from the blocs that appeared in the United States. Comprehensive bloc building of groups and movements on a national scale which also extended into the political and socio-economic realms – this is what we will be analyzing under the term of pillarization. The terminology and theory first emerged in the Netherlands. In essence, pillarization theory is based on the socio-scientific exploitation of a metaphor, the image of the façade of a classical temple: a small number of columns or pillars, which support a triangular pediment. The pillars represent segregated population groups, the pediment represents parliament and the government. The metaphor thus focuses attention on two important issues: why and how did segregated worlds emerge and flourish, and second, how did these worlds, with political parties as their representatives in the political world, result – or not – in stable governments? These are crucial issues in modern society. It is no surprise therefore that these questions were discussed in many European countries and that the pillarization perspective also spread beyond the Netherlands and Belgium.

The first section of this article will offer a review of pillarization theory (cf. also the encyclopedia entry by Maussen 2015). It will first more or less chronologically review the emergence and evolution of the theory in its home country, the Netherlands, and in its adoptive country, Belgium. This is followed by an overview of historical and international research of pillarization, as well as of research of de-pillarization and late pillarization. It will end with a note on the end of pillarization in the West. In the second part, so as to advance the study of organized blocs across the world, I will argue for a broad, international perspective on pillarization against the particularistic tendencies of many pillarization researchers, especially in the Netherlands. It comprises four major controversial issues: the geographical range of the pillar phenomenon, its relationship to modernity, the definition of pillars and pillarization, the absence of a generic term and the excess of idiosyncratic labels. In a shorter third part, I will address the isolation of pillarization theory from general sociological theory. I will contend that pillarization needs to be placed in a wider theoretical framework. As pillarization is focused on the segmentation, organization and mobilization of large population groupings in the modern world, I specifically have in mind theories of group formation, social movements and subcultures on the one hand, and theories of modernity on the other. Indeed, pillarization theory offers an antidote to all too individualistic interpretations of modern society: self-reinforcing group processes were and still are a common feature in the modern world and they have resulted on more than one occasion in divided societies, and continue to do so.

Part I: Pillarization Theory in the Netherlands, Belgium and Beyond

The Netherlands: Birthplace of Pillarization Critique and Pillarization Theory

Pillarization theory first emerged in the Netherlands in the 1950s. The perspective it developed subsequently spread to Belgium and to other countries. We will begin our inquiry with the emergence of pillarization theory in the Netherlands.

It is important to note that the public debate came first, and that scholars became involved only subsequently. The public debate began shortly after the Second World War. The swift post-war reconstruction of the pre-war blocs, and the new heights of success which the corresponding organizations attained in the 1950s caused great disappointment among Liberals and Socialists, who, in the wake of the alliances and friendships forged across the divides during the war, had hoped for a breakthrough in the old divisions. Moreover, more and more people within the Protestant and Catholic pillars began to feel uneasy vis-à-vis the constraints, now experienced as outdated, of living within their confines. The new mood led to hot-tempered public debates about the disadvantages and, to a far lesser extent, advantages of these blocs. The invention and rise to prominence of ‘zuil’ and ‘verzuiling’ (‘pillar’ and ‘pillarization’) as concepts and as an analytical perspective must be seen in this context. In fact, they were used as weapons in a kind of culture war over what was labelled the ‘de-pillarization’ (‘ontzuiling’) of Dutch society (cf. the title of an influential booklet by Blokker et al., Verzuiling, een Nederlands probleem (‘Pillarization, a Dutch problem’, 1959). From the start, therefore, pillar and pillarization, both as phenomena and as concepts, carried the negative connotation of a closed and ossified world, enforcing hard, unyielding borders with the outside world.

Dutch sociologists deserve credit for having transformed this metaphor used in public debate into a scientific concept and perspective in the mid-1950s. The most important author in this first period was Jakob Pieter Kruijt (1957; see also Kruijt and Goddijn 1962). He devised a carefully considered definition (Kruijt 1957), gathered data, and invented indices to describe the phenomenon empirically (Kruijt and Goddijn 1962: 233–247). As early as 1957, he stated that “the spell has been broken” with regard to pillarization, pointing to the post-war rise of mental ‘discomfort’ at a time when ‘structural pillarization’ of pillar organizations was still increasing (Kruijt 1957: 12–13; Kruijt and Goddijn 1962: 247–249). Characteristically, he sided with the critics of pillarization in regarding worldview (religious or atheist) rather than ideology or subculture in general as the main criterion for deciding whether a bloc qualifies as a pillar. This led to awkward consequences, such as negating the existence of Socialist or Liberal pillars just because they stopped propagating atheism or humanism in their party programs after the war, or affirming the existence of a miniature Humanist pillar even though it lacked socio-economic, cultural, and political organizations. Unsurprisingly, given the critical orientation of pillarization research and the first stirrings of de-pillarization, the bulk of subsequent publications dealt with the Protestant and Catholic pillars, the two most extensive blocs. These studies were often written by authors who had themselves been raised within the pillars they were describing. Johannes M. G. Thurlings’ book, De wankele zuil (‘The unsteady pillar’) (1971, 1978 for the 2nd extended edition) on the origins, growth, and decline of the Catholic pillar can be regarded as the most important of these sociological publications.

In the course of the 1960s, sociologists were joined in their pillarization research by political scientists, in particular Daalder (1966) and, above all, Arend Lijphart (1968, 1969, 1977; Van Schendelen 1984 discusses his work). Lijphart afforded pillarization a pivotal place in his answer to a central question in political science at the time: are stable democracies the prerogative of countries with a homogeneous and competitive political culture, such as the United States, or are they also possible in deeply divided countries, such as Austria, Belgium, Switzerland, or the Netherlands? In his 1968 book The politics of accommodation. Pluralism and democracy in the Netherlands, Lijphart argued that stable democracy was indeed possible in divided countries, provided that the political elites of the blocs in question were prepared to compromise on the most contentious issues. According to Lijphart, Dutch party elites had been doing this ever since the Pacification Pact of 1917. In this Pact, Protestant, Catholic, Liberal, and Socialist parties, representing the main sections of Dutch society, agreed not only to a compromise on denominational schools, but also on universal suffrage. But the interest of political scientists in pillarization was short-lived. Their primary focus was on the conditions and workings of consociational politics, not pillarization as such. Many countries were divided without being pillarized. Their interest in pillarization consequently evaporated after 1975, with the exception of Huyse’s work on Belgium (cf. infra).

Belgium: The (‘Amazing’) Persistence of Pillarization

Leaving aside a few scattered references in the years following 1957, the pillarization perspective only really caught on in Belgium in the mid-1960s. But whereas it was eagerly adopted in Flanders, it never really gained much of a foothold in French-speaking Belgium, even though this region was just as pillarized as Flanders was. At the time, Flanders, unlike French-speaking Belgium, was undergoing strong cultural influences from the Netherlands.

To my knowledge, the American political scientist Val Lorwin (1966a) was the first to analyze Belgian politics from a pillarization perspective. He did this in a paper published by the ‘Courrier Hebdomadaire CRISP’ in French, in which the term ‘verzuiling’, despite its centrality to the argument, was not translated, but consistently rendered in Dutch! In Flanders, De Clercq (1968) gave the first systematic exposition – his overviews of the history of the concept and of the literature in and beyond the Netherlands (p. 13–54 resp. p. 86–102) are still worth reading. The most substantial analyses of Belgian pillarization were published in the 1970s and the 1980s, by Huyse (1970, 1984, 1987) on the one hand – analyzing the political system of pillarization – and Jaak Billiet and Karel Dobbelaere on the other (Billiet and Dobbelaere 1976; Billiet 1976, 1982; Dobbelaere 1979, 1982), focusing on the Catholic pillar. In the Netherlands, researchers were mesmerized by the rapid collapse of the pillars in their country in the 1960s. In Belgium, on the contrary, scholars were wondering why pillarized politics and the pillars themselves, in particular the Catholic pillar, remained so firmly in place despite the cultural revolution of the long 1960s. The explanations advanced were twofold: subcultural changes within the Catholic pillar, more specifically the process of value generalization (‘from churched Catholicism to socio-cultural Christianity’ – according to Billiet and Dobbelaere) and a shift from subculture to the political parties as the main binding mechanism within the pillars (‘from pillars to political holdings’ – according to Huyse).

In Belgium, as in the Netherlands, the Catholic pillar received most of the attention. In spite of the formal resemblances – the presence of a subcultural labor milieu and of an extended Socialist organizational network, which both supported a single political party – few studies were published that analyzed Socialism in terms of pillarization. In Belgium, I have found only one, Van Haegendoren and Vandenhove’s inventory of Flemish Socialist organizations (1983). Researchers who study Socialism, and not just in Belgium, have always been very hesitant to apply the pillarization framework to Socialism – and many have openly rejected it. Instead, they used less forbidding alternatives, such as ‘world’ (see the Belgian ‘Courrier Hebdomadaire du CRISP’ 1972), ‘sphere’, ‘bloc’, or ‘movement’ (a very clear Dutch example is Stuurman 1983: 60). Moreover, these researchers had other, specific concepts available to them to make sense of the rise of the Socialist labor movement (e.g. capitalism, social classes, bourgeois hegemony, and working-class culture). In an outburst of irritation, Dierickx (1986: 510) criticized the prevailing politically inspired attitude with the words, “I belong to a movement, you to an organization, and he belongs to a pillar”.

Historical Research

From the start there was much interest in the history of pillarization. Lijphart (1968) devoted a great deal of attention to the pacification of the political system in the Netherlands in and after 1917. Thurlings (1978) devoted a third of his book to the origins and flowering of the Catholic pillar before 1960. Historians took over in the 1980s, with a special interest, once again, in the Catholic pillar. In part, this was due to the foundation of two major research centers, the Catholic Documentation Center (KDC) in the Netherlands in 1969, and the Catholic Documentation Center (KADOC) in Belgium in 1976 – the identical names are no coincidence. This resulted in a flurry of excellent studies (e.g. for the Netherlands, Roes 1985; Duffhues et al. 1985; Luyckx 2000 and, for Belgium, Lamberts 1984; Billiet 1988; Pasture 1992). Another group of historians, including a number of social geographers, embarked under the leadership of Hans Blom on a series of local studies of Dutch cities and municipalities. They concluded that the intensity of pillarization was far less clear-cut in reality than the theory had postulated. And, as expected, there was a lot of variation in the extent of pillarization on the local level, up to the total absence of anything resembling it at all. The main finding was the confirmation that pillars, i.e. pillarized organizational networks, were indeed formed in the last decades of the nineteenth century, although they were built upon pre-existing religious and class divisions (see Pennings 1991, with a series of interesting figures, and Blom 2000 as a qualifying synthesis of the whole project).

Over the course of time, this interest in history resulted in the formulation of a number of hypotheses in the Netherlands and in Belgium to explain the emergence and stubborn endurance of what were generally viewed as being spectacular formations. The so-called ‘emancipation hypothesis’ is the oldest of these explanations (Verwey-Jonker 1962). Regarded as an obvious explanation for the rise of the Socialist labor movement, it was also used to interpret the confessional pillars in the Netherlands. Protestant ‘kleyne luyden’ (‘small people’) of the middle and lower classes had, it was thought, emancipated themselves through pillarization from the dominance of the Liberal and modernist upper classes (Hendriks 1971). Something similar was believed to be true for Dutch Catholics, who, from the mid-nineteenth century onwards, had successfully challenged their subordination in the Dutch Protestant nation (see already Rogier 1956). The second explanation stressed defense against modernization and secularization as the main driving force behind confessional pillarization. This hypothesis became popular from 1970 onwards, in particular in studies on the Catholic pillar (Billiet 1976; Thurlings 1978; Righart 1985). Strong arguments in favor were the emergence of the Catholic labor movement in reaction to the rise of the Socialist labor movement, and the general policy of the bishops and the Catholic Church to forbid Catholics from joining Liberal and Socialist organizations and to discourage them from voting for other parties than the Catholic party. A third hypothesis interpreted pillarization as a convenient instrument of control in the hands of the bloc elites (Van Doorn 1956; Huyse 1984), political leaders (the Austrian Steininger 1977), and the bourgeoisie (Stuurman 1983). The fourth explanation was based on the fact that pillars only emerged from the mid-nineteenth century onwards. Modernity for the first time created the opportunity to organize larger sections of the population into tightly-knit networks of supra-local organizations (Ellemers 1984; Bax 1988; Hellemans 1985, 1990; Vanderstraeten 1999; the Swiss historian Altermatt 1989, 2004). Often, combinations of these four hypotheses were used to explain the rise of pillars and their evolution.

International Research

The international spread of the pillarization perspective began in the 1960s, but on the whole remained limited. Political scientists were instrumental in the dissemination of this perspective, as their ideas on ‘consociational democracies’ and ‘cooperating elites’ who bring political stability to divided societies were quickly taken up by authors outside the Netherlands (see, in addition to Daalder and Lijphart, Lorwin 1966a, b; Lehmbruch 1967; Huyse 1970; Mac Rae 1974). By contrast, international research on the formation, growth, and decline of the pillars themselves was slower to develop. Before 1970, references remained sparse and isolated (see De Clercq 1968: 86–102); unsurprisingly so, since many Dutch sociologists, historians, and journalists portrayed ‘verzuiling’ as ‘a typically Dutch phenomenon’ (see Schöffer 1956; Blokker et al. 1959; Stuurman 1983: 61–62 and 307–343; Te Velde 2009 for a critical review). Only neighboring Belgium enthusiastically welcomed the pillarization perspective.

Elsewhere, especially in Germany, Switzerland, and Austria, the issue of pillarization also arose, but a different, homegrown idiom was used, even though contacts with the Netherlands were significant and references to the pillarization perspective and the Netherlands were made, e.g. in a book on Catholicism edited by Gabriel and Kaufmann (1980), which included a contribution by Leo Laeyendecker from the Netherlands. Damberg (1997; see also Arbeitskreis 1993), whose mother was Dutch, more systematically compared Catholicism in Germany with its Dutch counterpart. But he preferred the concept of ‘Catholic milieu’ to that of ‘Catholic pillar’. Lepsius (1966) had already introduced the concept of ‘socio-cultural milieu’ to explain the persistence of political divisions in Germany since the end of the nineteenth century. The milieu concept has generated a flood of historical studies in Germany, most of which focus on Catholicism, as is customary also in the Netherlands and Belgium (for overviews, see for instance O’Sullivan 2009 and Henkelmann 2018). Gangolf Hübinger (1994) is the only author I know of who has used the concept of pillarization to analyze progressive liberalism in Wilhelmine Germany – and he uses it indiscriminately alongside concepts such as milieu and subculture. In Switzerland, Altermatt (1972, 1980), a good friend of the Dutch historian Jan Roes, has described the history of Catholicism in terms of ghetto and ‘Sondergesellschaft’ (sub-society). In Austria, ‘Lager’ – faction or camp – has become the usual designation for the three large, comprehensive political movements, Catholics, Socialists, and Nationals, since the work of Wandruszka (1954). Steininger (1975, 1977) is an exception; he favored the terminology and perspective of pillarization over the ‘Lager’ theory in his comparison of Austria and the Netherlands. In France, as in French-speaking Belgium, the preference is for ‘monde’ (‘world’), for example in Emile Poulat’s L’Eglise, c’est un monde (1986).

In the Netherlands itself, the international dimension of pillarization research was promoted mainly by the historian Hans Righart’s book (1985) on the formation of Catholic pillars in Belgium, the Netherlands, Austria, and Switzerland between 1880 and 1920. Against the particularistic tendencies of a number of Dutch pillarization researchers, I myself, at that time still working in Belgium, tried to demonstrate that pillars in fact occurred in many places in 19th- and twentieth-century European modernity, from Italy to Sweden. They were particularly common for Catholicism and Socialism, but Protestant, Communist, farmers’ and ethnic pillars could occasionally also be found (Hellemans 1990, 1993).

De-Pillarization Research

Nearly all studies in all Western European countries show that the main tendency after 1960 has been for de-pillarization, the dissolution of the subcultures and organizational networks. In the Netherlands, reflection in terms of pillarization only began in the 1950s, when the pillars were already being contested. Very soon, in the 1960s, they started to collapse. In Belgium, Austria, Switzerland, and Italy, however, they initially seemed to survive despite a decline in subcultural cohesion. Nevertheless, after 1990, it gradually became clear that they were also irrevocably declining, albeit very slowly in Belgium, Austria, and Switzerland. We can now confidently state that the era of pillarization is over, at least in Western Europe. It lasted there from about 1850 at the earliest to about 2000 at the latest.

Although the pillars in the Netherlands still looked strong in the 1950s, they collapsed rapidly after 1965. The de-pillarization of organizations occurred in three forms: through mergers, disappearance, and identity generalization. As several pillars were undergoing the same processes of de-pillarization, merging with parallel organizations in other pillars was an obvious strategy. The three confessional political parties – two Protestant and one Catholic – began to work closely together in 1967, transforming into a CDA (‘Christian-Democratic Appeal’) federation in 1973, and merging into a single CDA party in 1980. Also in 1967, the two main trade unions, Socialist and Catholic, began to join forces, a process that resulted in a merger in 1976, with the smaller Protestant Christian National Trade Union continuing separately. Yet the decline of the subcultural bonds within pillars also led to wholesale disappearance of organizations. This happened to many smaller pillar organizations, in particular in the youth, cultural, and leisure sectors – examples are the Socialist AJC (‘Workers Youth Movement’) as early as 1959; most Catholic publishers, and, in 1974, the Catholic newspaper ‘De Tijd’ (‘Time’); and the Reformed (‘Gereformeerde’) youth organizations. A third strategy, often chosen by well-established pillar organizations whose existence was not immediately under threat, consisted in loosening the organization’s subcultural identity by adopting a more general profile. This was the preponderant strategy in Belgium (cf. Billiet and Dobbelaere, supra). Examples in both the Netherlands and Belgium include Catholic and Protestant schools, as well as hospitals and healthcare organizations.

In Belgium, where the pillar networks remained in place for longer, the persistence of pillarization despite the loss of subcultural cohesion became the object of research. This focused on three levels: the political system level, the (inter)organizational level, and the membership level. As has been seen, Huyse (1987), in analyzing the first level, proposed the term ‘political holdings’ to describe the successors to ‘pillars’. Considering the declining strength of the subcultures, he emphasized the importance of political parties and their mediating role in providing pillar-friendly legislation and governmental subsidy policies as causes of the continuing presence of pillar-like networks of organizations. In any case the members of the old pillar parties generally still represent the old societal segments (Van Haute and Wauters 2019). Nevertheless, it was clear that the pillars themselves were, at the same time, slowly disintegrating. With the decline of, first, the Catholic Church and, after 1990, the Christian Democratic Party, several major Christian organizations attempted to retain and increase the number of their adherents by generalizing their identity. Thus they adopted a different name without reference to Christianity and/or continued their activities but at a greater remove from the church and the party. The National Christian Alliance of Small Entrepreneurs (NCMV), for instance, became the Union of Independent Enterprises (UNIZO) in 2000; the Christian Workers’ Movement turned its exclusive relationship with the Christian Democratic Party into a preferential one around 2000 and changed its name in 2014 into ‘beweging.net’ (‘movement.net’). Yet strong organizational ties continue to exist between these organizations. At the third level, research has indicated that the integration of ordinary members into pillar organizations was also fading. However, Hooghe (1999) showed that the links between people’s integration into a pillar – with membership of Christian, Socialist, or Liberal trade unions and health insurance funds as indicators – and electoral behavior and ideological attitudes had by no means disappeared. Statistical links between church involvement, membership of a Christian, Socialist, or Liberal health insurance fund or trade union on the one hand, and party choice on the other persisted even into 2010 and beyond. Nevertheless, the overall effect had diminished substantially, and fewer and fewer people were involved (Quintelier and Hooghe 2010; Billiet and Abts forthcoming). We can conclude therefore that subcultural and organizational de-pillarization is clearly also occurring in Belgium, albeit very slowly, as a long-drawn-out process, the history of which still needs to be written (Billiet 2006 gives a brief overview).

Elsewhere in Western Europe, the phenomena that pillarization theory calls pillars have also eroded. In France, the Communist ‘world’ has strongly contracted. In Italy, the old polarized system of Christian Democratic and Communist blocs exploded in 1992, and a whole new party landscape was created. A ‘Catholic world’ of the church and Catholic associations can still be discerned, but Catholics are now divided among several political parties, while the Catholic Church operates as a pressure group without exclusive links with any single party (Diamanti and Ceccarini 2007; Santagata 2014). Similarly, leftist associational networks can still be identified in the ‘red regions’, but here too members are spread across several left-wing parties. Relations between the network organizations and the left-wing political parties have become much looser as well (Ramella 2007). The situation in Austria can be compared to that in Belgium. The two remaining major ‘Lager’ – Catholic and Social Democrat – still exist, but they have lost much substance in terms of political clout, subcultural cohesion, and organizational strength since the 1980s (Plasser et al. 1992). Switzerland has followed the same route of slow de-pillarization (Altermatt 2012). In all these cases, histories of de-pillarization remain to be written.

Just as pillars emerged and developed in very divergent environments and at different times – from the second half of the nineteenth century up to the postwar era, the latter in Italy for instance – there has also been much diversity in the patterns and periodization of de-pillarization. The Catholic and Socialist pillars in Germany had already been largely dismantled by Hitler and were only reconstructed in minimal fashion after the war, without a party or trade union. The pillars in the Netherlands collapsed in the 1960s and 1970s. In Italy, de-pillarization came suddenly with the collapse of the party system in 1992. In Belgium, Austria, and Switzerland, de-pillarization took the form of a long-drawn-out process of erosion and disintegration. History rarely plays the same tune in different times and places.

Late Pillarization in a Time of De-Pillarization

From the 1980s onwards, some scholars in the Netherlands raised the possibility of fresh orthodox Protestant and Muslim pillarization. As far as I know, these are the only two cases for which possible late pillarization in a time of de-pillarization has been discussed in Western Europe. They were interpreted using the format of the earlier Catholic and Protestant cases of pillarization, as examples of defensive withdrawal in combination with emancipatory ambitions (cf. the first two explanatory models in 1.3).

Orthodox Calvinists in the Netherlands themselves re-interpreted the evolution of their churches and movements – with a flurry of new organizations being set up from the 1960s onwards – in terms of pillarization. Two small pillars were identified (see, in particular, Massink 1993). One is informally known as the ‘refo-zuil’ (Reformed pillar) and was built around the pietist ‘(Oud-) Gereformeerde Gemeenten’ (‘(Old) Reformed Congregations’, °1907), the ‘Staatkundig Gereformeerde Partij’ (‘Reformed Political Party’, °1918), a network of schools and educational institutions, a newspaper (‘Reformatorisch Dagblad’, °1971) and the ‘Reformatorisch Maatschappelijke Unie’ (°1983). The other pillar is somewhat larger and was built around the ‘Vrijgemaakt-Gereformeerde Kerken’ (‘Liberated Reformed Churches’, °1944), the ‘ChristenUnie’ (‘Christian Union’), born in 2001 out of two previous small Protestant parties, a newspaper (‘Nederlands Dagblad’, °1967 in its present form) and social, healthcare, broadcasting, tourist, and other organizations. The ‘ChristenUnie’ participated in Dutch coalition governments from 2007 to 2010 and has once again been part of the government since 2017. Although both pillars predate 1960, they significantly strengthened their identity and extended their organizational reach in the 1960s and thereafter. Nevertheless, even these pillars are not immune to de-pillarization, as subcultural dilution and organizational erosion have made some inroads over the last two decades (Janse 2015).

Non-European migrants were the second group thought to be susceptible to pillarization after 1960. Their subcultural ghettoization and organization reminded some observers of the earlier segregation of Catholics and Protestants. There are two aspects to this debate. First, from the bottom up: are migrants in different areas likely to organize themselves in pillars on the basis of their subcultures? Second, from the top down: is the government’s policy on migrants conceived in terms of pillarization, of segregation and emancipation of disadvantaged cultural minorities? The debate, if such there was, on these two issues seems in the meantime to have been resolved in favor of a negative answer. There is no migrant or Muslim pillar in the making in the Netherlands (Maussen 2012). Migrants and Muslims are too divided, culturally and organizationally, and there is certainly no exclusive connection between organizations in society and any political party (Sunier 2000; Scheffer 2007:170–178). The same is true for Muslims in Belgium (Torrekens 2019). In addition, at least in the Netherlands, government policy with regard to migrants, which was aimed in the 1970s and 1980s (to the extent that it existed at all) at promoting the cultural self-organization of migrants, has subsequently been reoriented towards socio-economic equality, individualization, and public security (Vink 2007; Hoogenboom and Scholten 2008).

Does the End of Pillarization in Western Europe Also Mean the Repudiation of Pillarization Theory?

As pillarization processes – subcultural densification in major sections of the population combined with the emergence of a network of domain-specific organizations – rarely if ever occur anymore in Western Europe, pillarization analysis there has also come to a halt. With the demise of the pillars, the drive to criticize or analyze blocs and movements in terms of pillarization has also waned. Only in places where the remnants of former pillars are still visible, like Belgium, does the power they still wield continue to interest researchers (e.g. Bruyère et al. 2019).

The question consequently arises whether pillarization was really as comprehensive and divisive as the theory presumed. Was it worth all the fuss? And what is the added value of the pillarization perspective? It is to the historian Peter van Dam’s credit (2011) that he raised these kinds of questions very vigorously with regard to the Netherlands in his book ‘The state of pillarization’, appropriately subtitled ‘About a Dutch myth’. He boldly stated there that “The term pillarization does not help us to better understand either the past or the present” (p. 21). According to Van Dam, the pillar perspective exaggerates the separation between ‘heavy communities’ (his preferred term for pillars). It proposes an overly static picture of society. It incorrectly views the phase of de-pillarization as a radical break with the past. And finally, it wrongly presents all these developments as ‘typically Dutch’. Van Dam’s presentation of the theory may hold true for the views of the politically inspired critics, but it is exaggerated with regard to the more complex pillarization theories that have been proposed. Be this as it may, Van Dam’s call for a critical assessment of pillarization theory and for alternative perspectives on history and society is justified.

Part II: Towards a Worldwide Study of Pillarization

Part I has given an overview of the literature on pillarization. In order to stimulate pillarization analysis globally, I will focus in Part II on the major difficulties that social scientists and historians face when they attempt to study the phenomenon outside Belgium and the Netherlands, the only two countries where the terminology and perspective of pillarization are common. It was questioned from the very beginning of pillarization research whether it is actually legitimate to do this. A number of authors, in particular in the Netherlands, emphasize the particular and exceptional character of pillarization. They deny that the phenomenon and the theory have any wider significance. Others, including me, regard pillars and pillarization as a widespread phenomenon. The opposition between what could be called particularists and universalists comprises several closely linked issues: the geographical range of pillars, their relationship to modernity, the definition of pillars and pillarization and the idiosyncratic labeling of the many forms of segmented, organized blocs. Anyone who intends to analyze pillars globally – or in countries other than the Netherlands and Belgium – will have to address these issues.

Widening the Geographical Range

The first task is to show that pillars can indeed be found in many places across the globe. From the start, the geographical range of pillarization has been a bone of contention: is it a uniquely Dutch phenomenon, is it limited to the Low Countries or to a few small Western European democracies, or can it be found in many European countries, and even outside Europe?

Most early and many subsequent Dutch pillarization researchers restricted pillarization in two ways from the outset: to confessional pillars, and/or to the Netherlands. As we have seen, confessional pillars were targets of criticism in the Netherlands from the start. The restrictive definition that emphasized religious identity was instrumental to this. By definition, this excluded pillars on social, political, or ethnic grounds. The second restriction, the geographical limitation to the Netherlands, was never the result of comparative analysis. It was a gut feeling, often legitimized by reference to the fact that the concepts of ‘zuil’ and ‘verzuiling’ were invented in the Netherlands to refer to what many regarded as uniquely Dutch phenomena. But this is to ignore that similar concepts were also invented in many other countries (cf. part I). Whenever such authors did acknowledge the existence of similar phenomena elsewhere, they confidently contended that the Dutch case was by far the most developed (e.g. Kruijt and Goddijn 1962: 229).

The Dutch-centrist view was undermined as soon as political scientists began to contribute to the debate in the 1960s. They saw from an early stage that elite cooperation between antagonistic blocs happens frequently in situations where all parties lack the prospect of enduring victory (cf. Daalder and Lijphart). Moreover, researchers on segmented bloc building from several Western European countries, sometimes influenced by Lijphart’s work, came to realize that (something like) pillars could also be found in (the history of) their own countries. Huyse, Billiet and Dobbelaere, Altermatt and Steininger are all good examples. This widening of the perspective led to a new view on pillarization, albeit one that still included geographical restrictions: pillars, it was thought, could only exist in a few small countries – the Netherlands, Belgium, Austria, and Switzerland were usually mentioned. It stimulated Righart (1985) to focus his research on the origins of the Catholic pillars on these four countries. I see no reason to restrict the incidence of pillars and pillarization a priori to one country or to several specific countries. Nor should the investigation be limited to Western Europe. The occurrence of pillarization should be decided on the basis of empirical research. It can occur in any country.

To support my thesis that segregated and organized blocs of large population groupings can also be found elsewhere in the world, I will now give a few examples. In Bosnia and Herzegovina, one of the successor states of former Yugoslavia, the three population groups – Muslim Bosniaks (50% of the population), Orthodox Serbs (30%) and Catholic Croats (15%) – set up segregated subcultures and networks of organizations after 1995, when the civil war ended that had ravaged the country from 1992 onwards. These segregated worlds had not existed in the Communist era or before. Kapidzic (2019) has called them ‘pillars’, rightly so in my view. The three population groups lead segregated lives and have relatively few contacts with one another. They have established quite a number of separate organizations along ethnic lines, including political parties. The pillars in Bosnia and Herzegovina also have a number of particular traits. Policymaking does not generally happen at the national level, and the three pillars enjoy “semi-sovereign status” (2019: 1). The political parties are the dominant forces within each pillar, while other organizations are weak and unstable, with the exception of war veterans’ groups and some business associations. The running of each of these pillars is assured by informal contacts within and between key families that are dominant in the party and the pillar organizations rather than by formal negotiations between organizational elites. As was the case in the Netherlands and Belgium, pillarization has also fostered consociationalism in politics, although in this case this was forced upon the country by the international community. According to Kapidzic, Bosnia and Herzegovina should therefore be regarded as “an unconsolidated consociational democracy” (2019: 11).

Outside Europe, too, phenomena can be detected that bear much resemblance to what are called pillars in the Low Countries. Take, for example, the Hindu nationalist movement – which labels itself the ‘Sangh Parivar’ – in India. It comprises a whole network of organizations in several domains. The mother organization is the RSS (‘Rashtriya Swayamsevak Sangh’ – the ‘National Volunteer Corps’). Founded in 1925, it endeavors to organize Hindus in every village and city quarter for the promotion of ‘Hindutva’, ‘Hinduness’. The RSS is closely connected with the ‘Bharatiya Janata Party’ (‘Indian People’s Party’). Founded in 1980 out of the predecessor organization ‘Bharatiya Jana Sangh’ (° 1951), this party has regularly been in government since the end of the 1990s. Other related organizations include a student movement (° 1948), an Indian Workers’ Association (° 1955), which has been India’s largest trade union since the 1990s, a Center for Tribal Welfare (° 1952), the World Council of Hindus (° 1964), the semi-religious branch that is adept at launching mass campaigns (e.g. the Ayodhya temple controversy, especially virulent in the early 1990s), the Vidya Bharati (° 1977) which coordinates its developing school network, and the Seva Bharati (° 1979) which organizes social activities, particularly in the slums of the major cities (Jaffrelot 2005; Nair 2009). The studies I have found focus on some of the organizations of this tentacular network and not on the segregation that it achieves among its followers.

In the Middle East, Islamist movements like Hamas in the Gaza Strip, or Hezbollah in Lebanon are, in my opinion, also examples of pillarization. Of course, they both act primarily as political parties, linked in their case to influential military wings. But they also have the backing of ‘a comprehensive self-contained world’. I quote Pascovich (2012: 126):“Both the Hamas Sunni movement in the Palestinian Territories and the Hezbollah Shi‘ite organization in Lebanon separately operate a vast civilian infrastructure with a budget of tens of millions of dollars annually that comprises numerous activities and institutions that assist the population in every aspect of life: health (hospitals, clinics), education (schools, kindergartens, community and sport centers, and even universities), religion (mosques, Qur‘an memorizing institutes), and welfare (delivery of financial and material aid)”. The description reminds us of the quotes given in the introduction to this article…

I have to admit that my Google search for examples of ethnic pillars in Africa did not produce any clear results, but perhaps this is due to my lack of acquaintance with this continent. There may also be substantive reasons, such as the lack of strong national organizations other than of a political, military and religious nature and the absence of long-standing, multiparty democracies, as these exist in the West, but also in India and Lebanon. If these conditions improve in the future, I think we can also expect to see pillarization processes – attempts to segregate large population groupings and to organize them in separated networks of organizations, including political parties – in Africa, as in other parts of the world. I am not at all sure, however, that pillars will become as common and as impactful in the non-Western world as they were for a time in Western Europe.

From ‘Exceptions in’ to ‘Expressions of’ Modernity

The preconception that pillars are limited to the Netherlands and/or a few small countries is related to and supported by the ill-conceived view that pillars are anomalous, even illegitimate phenomena that should not normally exist, that they are exceptional. Between the lines of quite a few scholarly texts on the subject, it is possible to sense the incomprehension of their authors with regard to pillarization: how is it possible that people submissively accepted isolation within their subculture, and guidance from elites with interests that conflicted with their own? This view that pillars are anomalous in modern society, and that they are or should be prone to swift disintegration, can be traced back to the motives behind the original debate, the intention to criticize pillarization. The exceptionalist view was further strengthened by the tendency to reduce pillarization to confessional pillars. The three types of reductions (that pillars were anomalous in modernity, were exclusively confessional, and were limited to one country or a few small countries) were sometimes combined (e.g. in Stuurman 1983). The exceptionalist view was particularly widespread in the Netherlands in the 1950s and the 1960s and continues to find support there even today.

As has been seen, the exceptionalist view came under increasing pressure when political scientists and scholars from other countries began to contribute to the debate from the 1960s onwards. Yet, it was not until 1977 that the great Norwegian scholar Stein Rokkan (1977) launched an appeal to construct a general theory of pillarization, based on comparative analysis. Responding to Rokkan’s call and the openings already made by other scholars, and using information gathered on Catholic and Socialist pillars before 1933 during a long research stay in Germany, I myself endeavored to develop a general view on pillarization (Hellemans 1985, 1990, 1993). My express intention was to strip pillarization theory of its implicitly and often explicitly exceptionalist character: pillars should be regarded as relatively frequent phenomena in Europe before 1960, they continued to exist after 1960 in several countries and, moreover, they could be found elsewhere in today’s world. They could be based on a variety of foundations – not only of a religious or philosophical nature, but also of class, ethnic, nationalist or cultural identities. They could emerge in many countries, both big and small, sometimes as a single pillar, sometimes alongside other pillars, resulting in a number of cases in ‘consociational democracy’. This perspective could accommodate a lot of variation and it demanded a more complex explanatory strategy than frameworks that stress emancipation, defense, or perfidious elite control could offer. Given the fact that they are so frequent, pillars surely could not be depicted as exceptions to modernity. On the contrary, they should be seen as phenomena that are at home in modernity. Far from waging peripheral rearguard battles against modernity, they were important actors in modernity, operating at their center. Pillars should accordingly be interpreted not as anomalies, but as products and exponents of modernity. They were crucial players capable of mobilizing large constituencies in a ‘war over modernity’.

Decentering the Netherlands and Belgium in Defining and Circumscribing Pillarization

Because pillarization was studied first and foremost in the Netherlands and Belgium, the definition of what can properly be called a pillar and the circumscription of pillarization research is geared to the particular set of circumstances that prevailed in these two countries. It is necessary to remove these particularities if we are to detect and compare pillars of all shades and hues across the world.

First, we need an analytical definition of pillars that is not hampered by unnecessary restrictions, such as limiting pillarization to religious pillars. Yet opening the door to a very wide range of pillars runs the risk of ending up with an imprecise definition – and of regarding any major group or subculture at all as a pillar. It is imperative therefore to distinguish as clearly as possible between pillars and other, less comprehensive types of blocs and subcultures. I have attempted to devise an analytical definition that can do this. In my view, a pillar consists of (a) a major section of the population (b) that is held together by a pervasive subculture, (c) and by a network of functionally differentiated organizations, (d) including a political party (Hellemans 1990: 19–27). The Dutch pillars are used here not in an exclusive sense, but as prototypes that can be helpful in drawing up the definition. To take the case of Belgium: Catholics, Socialists, and Liberals can each be said to have formed a pillar – each to a different degree. Communists and Flemish nationalists were borderline cases at certain stages of their development. And according to this definition, Greens and Muslims in Belgium have never been pillarized. Of course, the debate about definition will never be solved definitively. Each researcher has to make definitional decisions for him- or herself – this is perfectly acceptable, so long as the definitions are reasoned, clear, and not too restrictive or inflationary.

The second problem, closely linked to the very metaphor of pillarization, is the precise circumscription of what pillarization research is. It is often easy to forget that the single image of the façade of a temple comprises three distinct elements: a pillar, the fact that several pillars exist side by side, and the overarching pediment. Pillarization in the first place refers to the erection of a pillar, the subcultural and organizational bloc encompassing a major section of the population. If the majority in a country’s population is segregated into pillars, then the country in question has become pillarized. If, on top of that, the pillars’ elites work together with each other in the political system, this highlights the significance of the overarching pediment. All three elements can occur together – as they did in the Netherlands and Belgium – but often they do not. The Communists became pillarized in France, but not the other social groups, or they did so to a much lesser degree. Several pillars were established in inter-war Austria, but instead of elite cooperation civil war broke out (as, at first, it did in Bosnia and Herzegovina). Conversely, elite cooperation is possible without the ingredients of pillarization in the first sense (e.g. the so-called ‘polder model’ in the Netherlands after the 1970s). It is necessary therefore to distinguish clearly between the three elements or levels of the metaphor. In my opinion, it is best to restrict the concept of pillarization to the first level, the analysis of the construction of a segregated and highly organized bloc and of its preservation within a major section of the population. If there are several pillars in a country, this could be regarded, as Daalder (1966) and Mac Rae (1974) have argued, as ‘segmented pluralism’ or ‘a segmented society’. If the elites nonetheless cooperate, it is possible to use the concepts of elite cooperation, ‘consociational democracy’ or ‘consensus democracy’, the latter two introduced by Lijphart (Hellemans 1993: 136–140). Research of consociational democracy is still thriving, but this concept does not require that the underlying divided societies should be split into several pillars existing side-by-side (Bogaards and Helms 2019). In the same way, we should study pillarization of a section of the population without requiring that a whole country is pillarized or that national politics are organized along consociational lines.

Acknowledging and Sorting Out the Terminological Chaos

The last problem that needs to be discussed when widening the study of pillarization outside the core countries of Belgium and the Netherlands regards terminology. Though I want to promote pillar and pillarization as the overall concepts, it must be conceded that the terms have never become dominant outside the Netherlands and Belgium. Neither did other attempts at literal translation of ‘zuil’ and ‘verzuiling’, such as ‘compartment’ and ‘compartmentalization’. I have tried to explain why: the metaphor of pillarization itself is in some ways deficient… In addition, the neologism ‘verzuiling’ (pillarization) at first sight also seems awkward to outsiders. The image of a pillar for them has connotations such as fossilization and closure, but not the segmented organization of a large section of the population.

The terminological problem is aggravated by the fact that there are many concepts that are close to ‘zuil’ and ‘verzuiling’ that are used elsewhere to analyze very similar phenomena: ‘milieu’ (Germany), ‘Lager’ (Austria), ‘parti-société’, ‘monde’, ‘famille spirituelle’, ‘conglomérat communiste’ (France), ‘family of the Sangh’ (India), ‘ghetto’, ‘organized subculture’, ‘political subculture’, or just ‘subculture’ in the English-speaking world, ‘Sondergesellschaft’ (Altermatt), ‘Catholic universe’ (Diamanti, Ceccarini), ‘bloc’, ‘sphere’... But each of these concepts carries slightly different connotations, so that neither the concepts nor the underlying phenomena are mutually interchangeable. Not every ‘monde’ or ‘milieu’ has the hard boundaries and organizational viability of a pillar. Not every pillar relates to its rivals in the way hostile ‘Lager’ did (as in Austria, where polarization between the ‘Lager’ led to civil war in the 1930s). To call blocs built around churches ‘political subcultures’ is to deny the inner, religious dynamic of confessional pillars. As a result, no international consortium of researchers was ever created to investigate these kindred phenomena comparatively, employing a shared terminology and connected perspectives (Hellemans 2019a). By contrast, political scientists have managed to do just that with concepts such as ‘divided societies’, ‘cooperating elites’ and ‘consociational democracy’ – unlike ‘pillars’ and ‘pillarization’, these were broader and more neutral constructs. Even within Belgium and the Netherlands, a whole range of competing notions referring to the formation of socio-political blocs has emerged from the later part of the nineteenth century onwards (Verleye 2014). First, before and around 1900, the terms ‘party political partisanship’ (‘partijzucht’, ‘esprit de parti’) and ‘sectarianism’ were proposed, above all in Belgium – cf. the quote by Seebohm Rowntree in the introduction of this article. After the First World War, there were occasional references to ‘pigeonhole thinking” (‘hokjesgeest’) in the Netherlands. From the 1930s onwards, again in the Netherlands, ‘pillar’ and ‘pillarization’ started to emerge. From this point of view, the success of the latter metaphor and terminology was a late and essentially contingent development – some other notion could just as easily have become dominant.

We must therefore conclude that no term – neither pillarization nor any alternative – has so far been adopted as an unequivocal denotative label for our research object, the segmentation and organization of large population groupings. This gives ammunition to ‘particularists’, who stress the uniqueness of the blocs in their country using one of the many home-grown concepts. It is true that the unease associated with the pillarization terminology evaporates as soon as it becomes an accepted part of the idiom in a country – as happened in Flemish-speaking Belgium. It is also true that the term pillarization has been the most widely disseminated of all the terms proposed. Above all, we need a common terminology to study this specific type of collectivity. Most of the terms are too indeterminate, such as milieu, subculture, world, or universe. Some are too absolute, such as sub-society, or too negative, such as ghetto. All in all, I think the term pillarization, despite certain flaws, appears to be the most appropriate. I am aware, however, that these arguments are not conclusive. We have reached an impasse on this issue. The absence of an agreed terminology is, in my opinion, the single most important obstacle to the global study of what is called ‘pillarization’ in Belgium and the Netherlands. At the same time, the near-absence of pillarization studies globally and of the establishment of pillars as a specific type of impactful collectivity prevents us from reaching terminological consensus. I think the deadlock can be broken by acknowledging the many empirical cases that can be observed both in Europe and elsewhere.

Part III: Desegregating Pillarization Theory

A final and crucial problem – one that has unfortunately not often been dealt with – is the isolation of pillarization theory from general sociological theory. Admittedly many studies of pillarization have been informed by sociological theories. Almost all studies of religious pillars after the Second World War rely, more or less, on the now ‘classic’ secularization theories of the 1960s and 1970s. Stuurman’s (1983) view on pillarization in the Netherlands is extraordinary in that he uses feminist and Marxist theories. Modernization theories are also often quoted. Nevertheless, the interaction is limited and is in any case one-sided, that is, pillarization studies draw from a few, selected sociological theories, but they themselves do not aspire to contribute to general theory.

Three reasons can explain this. First, from a particularistic perspective, pillarization by definition has no wider significance, as this perspective believes that pillars emerged only in a few small states or even in the Netherlands alone and that they are historical exceptions doomed to disappear due to advancing modernization, and, in the case of religious pillars, secularization. Second, most pillarization studies are case studies of a particular pillar in a country that focus on describing their case or illuminating one aspect of it (e.g. emancipation, defense, persistence). There are relatively few attempts to compare cases for the purpose of generalization and of establishing broader theories. But the major and underlying reason is that no one has succeeded in placing pillars alongside other concepts that describe comparable social phenomena. This gives rise to the impression that pillarization analysis has no relevance for the study of other collective phenomena whatsoever. It would help if a general conceptual classification of large-scale collectivities were to be developed. But the diversity of such collectivities is so huge that this has deterred anyone from attempting what risks becoming an endless and sterile undertaking. The reason for this is that there is no obvious candidate for any generic concept immediately above the species of ‘pillar’ that could link pillars to kindred collectivities (e.g. the Catholic ‘self-contained world’ in the United States or segregated ethnic ‘groups’) and that could form the beginnings of the integration of pillarization theory into general sociological thinking. The analysts of pillarization are certainly not the only ones to blame for this. The isolation of pillarization research is also due to the relative neglect with which large-scale, hard-to-define collectivities other than, for example, movements and institutions have been treated in general sociological theory.

For the moment, I see three possible generic candidates that could fill the void above the species ‘pillar’: group, movement and subculture. First, pillars may be regarded as a particular type of group, that is, as large-scale groups that have become highly segmented and more heavily organized. The problem is that group theories are, in general, concerned only with small or medium-sized groups. Maybe the growing literature on the formation of ethnic groups, in particular in the non-Western world, could offer new perspectives in the future, in particular if pillars were to develop in a number of ethnicities. Theory building on the formation and maintenance of large-scale groups could be the fruit of such a development. In the past, I have proposed ‘social movement’ as another more encompassing framework. Pillars can then be viewed as exceptionally well-organized and very potent movements that vied with each other for some time in the context of a ‘war over modernity’. Conceptualization in terms of social movements on the one hand stresses the constructive and fragile character of pillarization and, on the other, the active and mobilizing involvement of pillars and similar types of blocs in modernity (Hellemans 1990). A third possible generic taxon centers on the notion of ‘subculture’. Subcultures are, in general, less successful in binding large numbers of people, and their organizational networks are looser and less extensive – this is why they are called ‘subcultures’, with the stress on the segregating cultural dimension. Pillars, on the other hand, stand out by highly organized bloc building and by the power they wield. Relating pillarization and other extensive forms of bloc building to subculture theory might help broaden subculture theory, moving it away from its tendency to focus chiefly on deviance and youth cultures, and towards a greater appreciation of the role of organizations. The Catholic ‘self-contained world’ in the United States, to give only this example of a social phenomenon that is cognate to pillars, can then be analyzed as a large-scale collectivity that closely approximated pillars in the European countries of the time. That this group, social movement, subculture did not scale the last hurdle towards pillarization – the erection of a national Catholic party and of Catholic trade unions – can be explained by its smaller size, in particular before 1900, and by the American electoral system (two major parties and first-past-the-post voting), which vitiated from the start any attempt to set up a national political party. I am sure that there are many more examples of pillar-like phenomena whose analysis could benefit from cross-analysis with pillarization theories. In turn, this could lead to theories of (aspects of) large-scale collectivities.

Theories of modernity, which conceptualize the historical environment in which pillars thrive, constitute a second route to desegregating pillarization theory. Pillarization and modernity are closely connected. In the West, the heyday of both pillars in Western Europe and of the Catholic ‘self-contained world’ in the United States occurred during a particular period in modern history: they all flourished between 1880 and 1960. This is no coincidence. This was a time of increasing mobilization of the lower strata of the population and a time of mass organizations. Pillars and large-scale bloc building in general epitomize the intensifying ties between people and organizations. Liberals always had hesitations about the prospect of ‘mass society’. From the 1950s onwards, they were joined by the new, better-educated middle classes. But in the decades before that, organizations were regarded by most of their members as useful carriers that enabled them to participate in modernity – to practice the faith, to participate in politics, to foster their socio-economic interests, to enjoy cultural activities, to go on vacation… Later on, other channels became available, in particular commercial culture and the leisure industry, and ‘individualized individuals’ (Luhmann) began to view these organizations and, even more so, the subcultures, as oppressive and as restrictive instead of life-enhancing. It was at this juncture that pillarization theory and pillarization critique emerged – as did the theories and critique of totalitarian societies, to mention an adjacent field. A more reserved attitude towards organizations is currently affecting all main organizations, churches, unions, banks and big companies, political parties… There are now many organizations of all sorts and types and people have learnt to switch between them. Conditions for segregation have similarly changed and are now more favorable to small organizations (cf. sects) as well as to informal groups of individuals (cf. lifestyle groups). Unlike in the past, people can easily depart when the organization they belong to no longer suits them. In late modern Western society, major sections of the population can no longer be organized in pillars, ‘Lager’, ‘self-contained worlds’, or whatever term one chooses to use. The context was only propitious for pillarization in the West during a particular phase of modernization that can be called high modernity or organized modernity. The question now is whether the improving conditions for pillarization in Eastern Europe and the non-Western world will prove superficial and ephemeral or whether they will last long enough to enable the formation of a fair number of pillars, as was the case in the West during high modernity.

This brings me to three main conclusions. First, pillars – or kindred phenomena – were important and impactful in Western Europe from approximately 1850 up to 2000. Yet, with the decline of pillarization in that part of the world due to changing conditions in late modernity, the interest in the study of pillarization has waned. Second, in Eastern Europe and in the non-Western world, conditions are emerging that make pillarization processes possible there. We will have to wait and see, however, whether they will be as frequent and impactful as they were in Western Europe – it will depend on the evolution of modernity there. If cases prove to be few, pillarization research will soon come to an end. Third, the lack of theories of large-scale collectivities in general sociological theory and the concomitant segregation of pillarization research and theory must be deplored. It is symptomatic for the uncertainties surrounding a more general theory that includes pillarization that it proved difficult to find a good subtitle for this contribution. In the absence of an accepted generic term, I considered ‘blocs’ (neutral, but vague), ‘population groupings’ or ‘sections in a country’s population’ (but to avoid reification, pillars must be distinguished from their target population), ‘mega-subcultures’, ‘highly organized group formation’, etc. As the reader can see, I eventually decided to use a casual expression, ‘self-contained worlds’ (as a tribute to Will Herberg), until a better, generic term presents itself…