Class Divisions in Use: The Swedish Social Group Taxonomy as Difference Technology, 1911–1970

Abstract

This article investigates an important but understudied phenomenon: the bureaucratic class division, which is analysed as a difference technology for envisioning, studying and managing the population. I examine a long-lived and widely spread taxonomy of the Swedish population into three social groups (Socialgrupper). Specifically, I look at how it influenced the production of statistics and knowledge about the voter during the first half of the twentieth century and higher education in the post-war welfare state era. The article understands the effects of the taxonomy as a ‘scientisation of the social’, using Lutz Raphael's term, in which fuzzy conceptual class boundaries were turned into exact classification, making it possible for different actors to act and calculate through them. The division was at the same time contested among social scientists and politicians. However, because of lack of alternatives and because it was well established, actors continued using it.

During the twentieth century, a number of actors and institutions across the global north set out to develop hierarchical social taxonomies of their national populations. This article investigates a long-lived taxonomy of the Swedish population into three social groups (Socialgrupper), which was devised around 1900 but soon travelled and became productive in a plethora of other contexts. It grouped those with a higher degree, civil servants, businesspeople and landowners in ‘socialgrupp I’, farmers, lower civil servants and craftsmen in ‘socialgrupp II’, and workers of other kinds in ‘socialgrupp III’. My study connects statisticians, social scientists and politicians, all using the taxonomy for their own ends, to show the versatility and mobile nature of bureaucratic division during the twentieth century. I follow the taxonomy from social scientific attempts in pre-democratic Sweden to make hierarchy visible to its place as a welfare tool within post-war Social Democratic Sweden. To do this, I have chosen to look at how the division came about, how the taxonomy was used by the Central Bureau of Statistics (Statistiska Centralbyrån; SCB) and political parties alike to create knowledge about voters in the first half of the twentieth century, and, lastly, its place in envisioning and managing higher education in the post-war period.¹ This widespread use of the division would lead to a point where many Swedes described society through social group terms.² The division has ‘broken through so completely, that many seem to be inclined to take it more or less for granted that it makes sense to divide . . . the contemporary Swedish population into the social groups 1, 2 and 3’, wrote the SCB in the 1960s.³ Although losing popularity among social scientists and other types of experts in the latter part of the twentieth century, the division lives on within the cultural imagination. For example, since the 1980s a popular fans’ chant at the football club Hammarby goes ‘Alléz! Alléz! Socialgrupp tre!’, in which the supporters classify themselves as pariahs of society.⁴

Remarkably few have written about social taxonomies as historical phenomena. Several historians have studied ideas of social class in the social sciences.⁵ Simon Szreter and Margo Anderson have done interesting work outlining genealogies of British and American social taxonomy within statistical bureaus, ordering the population into, respectively, five and six classes based on social standing, constructed to address specific national concerns like fears of working-class fertility rates overtaking the higher classes in Britain. Szreter and Anderson both note how the divisions eventually became like ‘second nature’ for how social scientists understood the social structure of their countries, but without these historians studying how this wider circulation happened and to what effect.⁶ We also know of particular social taxonomies being created within official statistics in France and Norway.⁷ What is similar to the Swedish case is how these bureaucratic class taxonomies did not arise out of theoretical engagements with who are usually placed as the important figures in the history of class thinking, such as Marx or Weber, but were created by civil servants without much discussion of their design.⁸ They make up a forgotten but influential lineage.

The circulation and effects of the human sciences have recently been analysed as a ‘scientisation of the social’, a term coined by the German historian Lutz Raphael. With this, he means the intended and unintended effects of the presence of experts and expertise from the human sciences within society. The perspective is useful as it highlights the human sciences as productive forces within different contexts, shaping how people and institutions make meaning and act on phenomena by offering concepts, trained experts and techniques like statistical calculation, classifying and counselling. As Raphael has proposed, one way to study scientisation is through the focus on specific technologies.⁹ In Sweden in particular, the Social Democratic politicians and social scientists Alva and Gunnar Myrdal have been seen as embodying a spirit of social engineering. Through them and other experts, the Social Democratic Party reorganised Swedish society from the 1930s onwards, enacting welfare reforms but also darker biopolitical tendencies such as forced sterilisation.¹⁰ However, investigations into social taxonomies are absent from the scientisation and the social engineering research fields, an omission this article seeks to address. I show how Social Democratic social engineering was connected to earlier population knowledge. It is not just a lineage of ideas but also the continuous use of the same technology to envision and manage society. Importantly, these usages of the social group division can be understood as a scientisation of class and social belonging in twentieth century Sweden. Class boundaries went from fuzzy to solid and exact as the social group division enumerated and classified the Swedish population.

My article studies social taxonomies in use – what they do and what they engender when put into different practices and contexts.¹¹ The analytical concept of difference technology captures the active work social taxonomies perform in the making of communities when actors structure empirical material through its classifications. I have developed the concept from the term ‘social technology’, which has been understood as tools to intervene in the social, such as social insurance schemes, scenario-thinking or urban planning.¹² My use of difference technology posits it not only as a technology to envision and study a given population, but also one to manage and intervene in society. Difference technologies quantify and standardise concepts into precise classifications, bringing them together in a taxonomic order that relates them to each other. They enable statistical production and overview of populations, establishing one way of understanding society out of many possible ways. Social taxonomies reproduce conceptualisations of a particular social structure, its hierarchies, boundaries and communities, while at the same time constructing new knowledge when used, often in the form of statistical differences between the classifications. They work through a meeting between taxonomy and the empirical material.¹³

The choice of looking at voter knowledge in the first half of the twentieth century and higher education in the post-war period has been made for several reasons. Firstly, this is where the social group division was especially present. It was first used within election statistics, and therefore it was close at hand for the political parties to be influenced by it in their conceptualisations of the voters. Later, from the 1940s and onwards, educational research and state commissions would take up the taxonomy in their studies of and policy proposals for an expanding sector of higher education. Secondly, it makes it possible to examine the taxonomy over time – from its beginnings in pre-democratic Sweden to the post-war welfare society – as well as to compare different actors’ ways of creating knowledge through it. The article therefore combines different types of sources: debates in Social Democratic, Liberal and Conservative journals spanning 1911 to 1970¹⁴; the political parties’ election campaign material from 1911 to 1940¹⁵; educational research and state commissions on higher education covering 1945 to 1970¹⁶; as well as press material from the large newspapers, which I have searched through keywords in the digitised database created by the National Library of Sweden.¹⁷ Overall, I have looked at what is communicated outwards – what is public knowledge, using Andreas W. Daum's term – about the social structure of Sweden.¹⁸ Left out, for example, are debates within the parliament, a task for future research. The analysis will firstly explore how the taxonomy came about, then move on to how it became effective in knowledge production about voters and higher education. Through this, I will also try to answer why the division remained in use for so long, even though mounting criticism was directed at it.

Taxonomic Origin and Order

To understand the social group division we need to go back to the late nineteenth and early twentieth centuries. At the time, Sweden was one of the most unequal societies in Europe, although the labour movement – the Swedish Social Democratic Party (Sveriges Socialdemokratiska Arbetareparti, SAP, founded 1889) and the workers’ unions – was gathering momentum, threatening the old landed and industrial elites.¹⁹ What can we from the upper class do to ‘exorcise the storm’ that is raging across Western civilisation, asked the conservative politician and professor in political science and statistics Pontus Fahlbeck in one of the first scientific publications on class in Sweden, Estate and Class (Stånd och klass, 1892). The answer was a new taxonomy.²⁰ The example of Fahlbeck, following people like David Ricardo and Karl Marx in making class an object for theoretical and empirical consideration, shows how the formation of class language was a political project and a scientific endeavour intertwined²¹ – not just for socialists, but also for conservatives and liberals as well as state actors, who all grappled with the ontology and epistemology of class as a new demarcation of society. People from all political leanings around the turn of the century in Sweden used class terms.²² Within the Swedish government's substantial report on migration (Emigrationsutredningen, 1906–13), for example, social scientists discussed class as an important factor in understanding Swedes’ choice to leave for America.²³ Fahlbeck's thinking, however, proved to be influential in the Central Bureau of Statistics’ design of the social group division, and that is why I have chosen to focus on him in the section below.

For Fahlbeck, having been trained as a historian but becoming professor of history and political science at the University of Lund in 1889 (switching again in 1902 to become a professor of political science and statistics), society needed to be understood historically and sociologically. During the 1880s, he increasingly oriented himself towards political economy and statistics, influenced mainly by the German historical school of economics, arguing for the introduction of import duties and the protection of the national economy. In general, Fahlbeck was against the laissez-faire approach of modern economics and what he saw as its dangerous individualistic tendencies. He propagated for the state to intervene in society, alleviating the ills of industrialisation.²⁴

Fahlbeck argued for understanding class as a new modern order of society, in dialogue with socialist, communist and social democratic literature but presenting his own vision. What decided the stratification throughout history – between slaves and masters, later the estates and now the classes – was, according to his theory, culture. An elite group, free from the need to do manual labour, always upheld Western culture. Whenever they stopped performing their function, a new order of society would overthrow the old. The present-day elite consisted of the free professions, artists, scientists, civil servants, and the owners and the captains of industry. In between the upper class and the working class was a middle class, which both oversaw manual labour but themselves toiled, such as the large group of freehold farmers, craftsmen and shopkeepers. Fahlbeck could also calm the readers with his analysis: the working class was in no way ready to take over the task of upholding culture. For now, the upper class ruled. However, the state should support the labour movement such as the unions in trying to raise workers’ material and cultural standards, and even include them in the political sphere by expanding voting rights.²⁵

The historian Sabine Hake has described the tendencies of German academics in the decades around 1900 to label the workers as masses, that is, irrational and formless.²⁶ Fahlbeck's taxonomic project was in a way the opposite. He wanted to make visible and count the workers, to bring them into being part of the same order as the elites, all working to ensure the survival of culture but at the same time differentiating people into a hierarchy. However, the statistics also had practical functions. At a statistical conference in Paris in 1909, he concluded that class statistics were missing within Western countries’ statistical bureaus. Their adoption could help states deal with commonly faced problems such as socialism, emigration and social mobility, Fahlbeck argued.²⁷

Bureaucratic, Political and Scientific Transformations of the Social Groups, 1911–1970

The type of statistical production Fahlbeck pushed for was to be instigated in 1911 in the context of the first election with an expanded franchise for all men – except those receiving poverty aid or those who were lagging behind with their tax payments – a reform enacted by a Conservative government. To learn about the new electoral situation, the SCB was asked to classify voters.²⁸ To do this they needed a taxonomy, and in this meeting between Fahlbeck's political-scientific vision and the bureaucratic workings of the SCB the social group division was born. The social group statistics would later be taken up by Swedish political scientists, like Herbert Tingsten in his international classic Political Behavior: Studies in Election Statistics (1937),²⁹ but here I want to understand how the division was created and how political parties used it. Overall, political parties are understudied as users of population knowledge. Wim de Jong and Harm Kaal have investigated what they call a scientisation of the political in twentieth century Netherlands. Specifically, they have pointed towards discussions on mass psychology in the 1930s and the introduction of opinion polling in the post-war period.³⁰ However, as this Swedish case will show, the parties also found much use in the classed election statistics for understanding and acting on the voters.

The SCB described what they first called groups I, II, and III – in 1928 given the names socialgrupper – as synonyms for the upper class, middle class and working class. However, they gave little clue to the reasoning behind their taxonomy, only stating that the division was not based on ‘technical’ or ‘economical’ principles but a ‘social’ assessment of people's social standing, based on their occupation.³¹ The classifications map largely onto Fahlbeck's taxonomy, except that the SCB puts lower civil servants and office workers in the middle class, while Fahlbeck places them in the upper class.³² The concept of the social (Det sociala) was discussed throughout the nineteenth century in Sweden, marking a hybrid sphere between the private and the state as well as interconnecting them. By the early twentieth century, the word social was widely used as a synonym for society.³³

In 1912, the Social Democratic theorist and politician Rickard Sandler – serving briefly as prime minister in the 1920s as well as the translator of Karl Marx's Das Kapital into Swedish – hailed the statistics as a scientific and political breakthrough. For him, it helped to see the social structure behind the political parties – the material interests driving politics. ‘Every deeper study of the political battles must build on the knowledge of the different classes.’ Sandler did, however, put together a new taxonomy, more fitting to a Social Democratic worldview, with two dominant class interests: the capital class versus the ‘underclass’, but with a middle class in between these forces. The SCB's taxonomy in contrast merely reflected classes based on standards of living, according to Sandler, and therefore gave a static view of society. That there only were two classes was a ‘quasi-Marxist’ simplification not found in Marx, Sandler argued.³⁴ Since the beginning of the SAP in the late 1880s, its party theorists had debated the existence of a middle class; some said it was disappearing, while others wanted to revise those ideas in light of a changing social reality where a middle stratum was solidifying and becoming a permanent part of society.³⁵ Here, with the SCB and Sandler's statistics, the existence of the middle class was open for everyone to see for themselves, expressed in statistical tables.

Sandler's early deconstruction of the social group taxonomy proved in the longer run to be an exception. From the 1920s onwards, the Social Democrats instead started using the SCB's division unchanged when discussing voters.³⁶ Laboriously changing the statistics must have been too much work; it was easier just to go with the numbers the SCB produced after each election. Similarly, the Liberals’ Free-minded National Association (Frisinnade landsföreningen, founded 1902) and the Conservatives’ General Electoral League (Allmänna valmansförbundet, founded 1904) saw value in the taxonomy. This was at a time when nationwide political parties with the aim of mobilising voters were established in Sweden.³⁷ The Social Democrats, the Liberals and the Conservatives competed with each other, taking turns in leading the government through unstable coalitions up until 1933, when the Social Democratic Party entered into a partnership with the up-and-coming Farmers’ League (Bondeförbundet). The social group division and the voter statistics gave the parties tools to make representative claims on the population as well as knowledge on how to proceed with national and local election campaigns, showing where voters from their imagined constituency lived and how many of them had or had not voted in the last election. Statistics made it possible to survey society from above and to map out differences in order to act on that knowledge.

Here a new logic becomes visible – a change from Fahlbeck's emphasis on qualitative differences between the classes. For the political parties, numbers mattered most. The classes and their proportions of the population, statistically imagined and created with the social group division as a difference technology, now signalled democratic legitimacy to rule. The Social Democrats repeatedly expressed how social group III, the working class, constituted around 55 per cent of the voters according to the SCB, that is, a majority. The only obstacle to power was getting the non-voters within the class mobilised.³⁸ Social group III reserves were presented as possessing an inherent interest. The identification between the SCB's voter categories and the parties’ representative claims was taken to its extreme in the SAP's election analysis of 1924. The party translated the three classes into the spheres of interest of the parliamentary parties. The Conservatives had received hundreds of thousands of votes ‘from the main social groups, whose interests they cannot be considered to represent’, hundreds of thousands of lower-class votes were polled by other parties. The Liberals and the Farmers’ League, identified with group II, were thought to have lost over 230,000 votes ‘to parties which in politics represented other interests’.³⁹ That is, the SAP received votes from people they should not have been representing.

The Liberals were here outwardly more against linking voting to social belonging, instead framing voting as an individual act that should be based on political convictions.⁴⁰ However, internally they urged party functionaries to seek out people from group II, the middle class, as their base. One party official, for example, sent out a memo urging local party members to ‘focus on those professional categories who, according to the statistics, show little participation in the 1911 elections and who through their social status’ were inclined to vote for them, clarifying that these could ‘preferably be found in the occupational categories listed under group II’.⁴¹ The Conservatives in turn at first discussed becoming a party for the middle class.⁴² But the party leadership soon became harsher in its criticism of such ideas. Winning elections as a class party was an impossible equation because the SAP possessed far more class reserves, they argued.⁴³ An article in the membership magazine Medborgaren in 1920 pointed out that organising the middle class was ‘just the same as saying to all workers: nowadays all political groupings are to be made according to class, so you, who are workers, should join the SAP’. Since the working class was in absolute numerical preponderance, all would be lost for the Conservatives. The party should instead ‘follow the old rule of divide and conquer’ and ‘win over ever larger numbers on the basis of bourgeois opinions’.⁴⁴ The Conservatives challenged the Social Democratic claim on social group III and argued that many of the workers in reality voted for them.⁴⁵ They used the statistics to show that they had voters from all the classes, making them a party for the whole population. A Conservative politician and statistician calculated, for example, that their voters consisted of 15 per cent upper class, 65 per cent middle class and 20 per cent working class, which, according to him, gave ‘a pretty harmonious balance between different social classes’.⁴⁶

Interestingly, the social group division also found new use within the political parties’ navigation of the voters through Swedish Gallup (1941) and its embrace of the division as a difference technology to structure social differences within Swedish opinions. From the 1940s onwards, political parties started to probe the population about the effectiveness of their political campaigns and the popularity of suggested reforms.⁴⁷

In conclusion, during the interwar period the social group became an influential and dominant way to map out and understand social differences among the voters. It was a technology for seeing society from above. Class went here from fuzzy categories into precisely numbered ones – the parties suddenly knew how many members of the different classes lived in each voter district. However, a shift from Fahlbeck's emphasis on qualitative differences between the classes was that in the representative political sphere numbers mattered most. Within the expanding post-war welfare state that the SAP was building, the focus would increasingly be on social group III as the targets of welfare reforms – a development that the article will now turn to.

The Social Group Division within Higher Education

At the same time as opinion polls created and presented differences amongst opinions within the population, the social group taxonomy became entangled in knowledge production, political debate and policy making around higher education. Part of the welfare programmes for the Social Democrats in power during the post-war period – alongside maintaining low unemployment and high economic growth, redistributing resources through taxes and high public spending on housing and healthcare – was reorganising public education in Sweden.⁴⁸ The focus on education was something that had gained traction within the party in the mid-1940s. Importantly, schools were now seen as key in fostering democratic and civic capabilities among the population. Nine-year comprehensive schools were tried out in different school districts during the 1950s and enacted all over the country in 1962. Higher education was now also expanded for a greater number of students.⁴⁹ These developments resemble what was happening in other European countries around the same time, where higher education went from elite to mass institutions. In Sweden, similar to the United Kingdom, the debate was couched in reference to addressing class inequalities.⁵⁰ In order to envision and argue for these educational reforms in Sweden, knowledge production was needed, and here the social group division came to play an important part as a difference technology.

The Institute for Racial Biology (Rasbiologiska institutet, 1922–58) had already used the social group taxonomy in the years around 1930 to divide Swedish students in order to study racial degeneration at universities. The small increase in students with a working-class background that was noticeable since the turn of the century meant, for these population experts, a deterioration of higher education.⁵¹ Soon, similar numbers would come to mean quite different things. Overall, up to the end of the 1940s, there were only a couple of thousand students at any time, as only a small portion of the population was seen as having the intellectual qualifications to study.⁵²

Starting from the mid-1940s, a few young social scientists and politicians would challenge this doxa. The psychology student Torsten Husén (1916–2009) had begun working for the Swedish Army in 1942 to help set up intelligence tests for the draftees, mainly to sift out ‘imbeciles’ and fast-track those with high test scores for officer training. A side effect of this endeavour was the creation of a large database covering the new generation of young men each year. Husén here saw his chance to compare intelligence test scores, school grades, social backgrounds and current life paths. His conclusion in 1946 was that many of those not in higher education had the same IQ as the student population, so there was room for expansion.⁵³ Around the same time (1947), the social scientist Gunnar Boalt (1910–2000) was writing a dissertation following a generation of schoolchildren in Stockholm. By grouping them through the social group taxonomy, he could show the ‘social handicap’ of those in social group III at every new level. Not only did they get lower grades – something Boalt theorised as an effect of their upbringing and not genetics – but a far higher percentage within social group III left school than pupils from the other social groups, even when they had similar grades.⁵⁴ These results came to inform a state commission, led by the political scientist and Social Democratic politician Ragnar Edenman (1914–98), in its statistical mapping of all students at Swedish universities. The commission showed that only around 5 per cent of the students came from social group III – a number that would be circulated in public debate in the following years. Edenman launched the soon popular concept of a ‘pool of talent’ (Begåvningsreserv) to put focus on this societal waste. Here talent was centred as the most important ‘natural resource’ for a country and education as the main means for economic and social progress, in need of modernisation.⁵⁵

In public discourse, the social group statistics were seen as illuminating the profoundly unjust nature of higher education, which had a pressing need to be ‘democratised’.⁵⁶ ‘The numbers are grotesque’, one of the members of Edenman's commission wrote in the Social Democratic debate forum Tiden ⁵⁷ – a sentiment also mirrored by the Liberals.⁵⁸ The Prime Minister Tage Erlander brought up the statistics at the party conference that same year as proof of the need for reforms.⁵⁹ For the SAP, who governed during all of the post-war period up to 1976, broadening higher education now came to lie at the heart of the new welfare society that they were engineering.

Before delving into this process, I want to bring to light how contentious the social group taxonomy really was in these early post-war years. By the late 1940s, the SCB in their voter statistics actually discarded it, arguing internally that the socioeconomic differences between the occupational groups within the classifications were so big that it did not make any sense to keep using it. They started searching for a replacement, and with it, an ‘incontestable definition of a group’ – a quest not fulfilled until 1982, when a new taxonomy was put in place.⁶⁰ In the interregnum, the social group division continued to thrive as a difference technology even though the state administration had phased it out, since no other contender was around to take its place. In the social sciences – the newly founded subject sociology (1947), and the expanding psychology and pedagogy departments – the social group division was picked up as a standard for mapping and studying social difference in attitudes, behaviour and life patterns among the population. In the turn to statistical analysis as a scientific ideal, the taxonomy was an excellent fit.⁶¹

However, others within academia took up sometimes quite public critiques against it. One political scientist commented on the vicious circle social research was trapped in, where the social group division was kept in use because of a lack of alternatives, and by being used it became even more valuable. He stated that it lacked a theoretical foundation and was arbitrarily put together in a jumble of different variables, such as social status, income and education level.⁶² Similarly, Torsten Husén argued that social science needed to operate with pure variables such as income or educational background, to understand specific correlations between phenomena. According to him, it was a mistake to see society as moving under one common status hierarchy. Instead, groups valued different things: pupils good grades, for example, businesspeople their income and so on.⁶³ This plurality of orders was, however, never investigated, most likely because of the complexity of the task. Even Husén talked about the social groups as real and existing entities, for example in his inaugural professorial lecture in 1953.⁶⁴

Social scientists were rarely explicit in explaining why they had chosen the division; many probably saw the convenience of a well-tried taxonomy, in which all existing occupations had already been sorted into their classifications. Using it also offered many possibilities to compare results with other studies. It gave an overview of the social structure and, with it, different phenomena such as crime, urban planning and urban life, religious attendance and health could easily be mapped and studied. Gunnar Boalt took to its defence in the first handbook for sociology students, naming precisely those pragmatic reasons mentioned above for continuing to use the taxonomy. If it incorrectly represented the social structure, he argued, it would show in the statistical outcomes of the different studies.⁶⁵

The social group division remained a difference technology for imagining and intervening in education up until the 1970s. It showed up in all the major state commissions on the subject, for example.⁶⁶ In national education budgets, the cost of the state's stipends to those in need was calculated based on the social group composition of the student population, since it was mostly social group III students that were eligible. However, this was constantly in need of readjustment because of the increased presence of social group III – what was named an ongoing ‘social group alteration’ (Socialgruppsförskjutning) – at the universities.⁶⁷ ‘We are moving away from the distorted distribution of students in different social groups’, said Ragnar Edenman as the new Minister of Education in 1958, proudly announcing that social group III now made up 17 per cent of the students.⁶⁸

Educational democratisation came to mean equal participation between the social groups. We see here how the division became a welfare technology for measuring and evaluating progress within the sector, which came in handy in political debate and policy making. For the Social Democrats in power, the social group division remained valuable because social group III was easily translatable to their main political constituency, the working class, as we also saw in the section on their discussions about the voters. ‘The social groups have been forsaken by the Central Bureau of Statistics but not by me . . . It is difficult to describe the problems that different social backgrounds encounter without using the social group terms.’ Minister of Education Ingvar Carlsson's statement in 1970 illuminates how dependent the political and bureaucratic language around education was on the social group taxonomy at this point.⁶⁹

Not all were happy about the political project the social group statistics helped to constitute. Liberal and conservative editorials denounced the taxonomy as a ploy by the Social Democrats to keep Swedes thinking and voting according to their class belonging and not their beliefs. In addition, they argued that it was outdated and did not reflect the current differentiated society.⁷⁰ In 1950, a big debate broke out, mainly in the conservative press, decrying the plight and crisis of ‘the middle class’. In part, this seemed to be a reaction to the discussions around the pool of talent. Their message was that the academics, in large part framed by the Social Democrats and the social group division as part of the elite, belonged in reality to the middle class. In the welfare society, being seen as upper class was not always advantageous. According to some Conservatives, it meant less help from the state and society in general.⁷¹

For some Social Democrats, on the other hand, the focus on higher education sat uneasily with their vision of society. They wanted a re-evaluation of occupations – those working with their hands should not be valued less than those in other professions were.⁷² Some warned of a scenario where all with academic potential continued on to higher education, creating a new ‘class society based on the selection of talent’.⁷³ A prevalent way of addressing this issue was to distinguish between theoretical and practical talent as two separate entities that should be valued the same among Swedish children.⁷⁴ However, the fear lingered on: ‘massive reserves of intelligence from the working class is pumped up into higher strata of society and reinforces these’, with the result that ‘social group III is depleted of intelligence’, a Social Democratic politician and education expert wrote in the 1950s.⁷⁵ With the translation of Michael Young's satirical novel The Rise of the Meritocracy (1958), the debate continued throughout the 1960s and 1970s.⁷⁶ Clearly people felt, despite official emphasis on valuing those not in higher education the same, that higher education was the better option, supported by the fact that the university educated were placed at the top of society according to the social group division while workers were down at the bottom.

There is not a single point in time to mark the end of the social group division. From the 1970s onwards it was phased out. The Central Bureau of Statistics came up with a new, ‘official’ taxonomy in 1982: ‘the socioeconomic division’, which aggregated up to six classes, is still in use in Sweden today. This can be seen as attempting to capture societal developments of a larger service and public sector. It divides white-collar occupations into three separate classifications, for example.⁷⁷ Another influential division today is through education, seeing uneducated and educated people as social communities, in line with societal self-images of being a knowledge society.⁷⁸

Conclusion

The article contributes to the literature on scientisation of the social, the history of the social sciences as well as the history of social engineering by following a social division between different actors and contexts, something that has previously not been done. The social group taxonomy in Sweden was a part of quite diverse political and social visions. By studying it, I have tried to show the productiveness of social division during the twentieth century in creating imagined social differences. It was not just a mirror of an unequal society but contributed to creating visions of a shared Swedish social structure. I introduce the term ‘difference technologies’ to understand what these taxonomies do and what they engender when put to use in knowledge production. Actors, by sifting empirical data through social group classifications, repeatedly reproduced these classifications, although always with new distributions of numbers such as voting patterns after an election. In this way, the division always generated the social differently.

Using Lutz Raphael's perspective, this can be understood as a scientisation of how the social structure was perceived, by turning fuzzy conceptual class boundaries into exact classification, encompassing and enumerating the population. These classifications, after having been created, became real in the sense that people and institutions thought and acted through them. We can understand this as the reality effects of division, connecting people in new constellations and creating novel ways of seeing and being in the world.

I want to emphasise how the social group taxonomy was never one thing, expressing one idea about society. As a difference technology, it shifted its meaning and its effects by its use by different actors. This article investigated its career from pre-democratic and conservative hierarchy-making practices to being fused into the post-war Social Democratic welfare society. The historian and political scientist Pontus Fahlbeck, in his taxonomic advances, wanted to make social hierarchies evident and natural, and with this, stop socialists overthrowing the societal order. The SCB reworked it in their election statistics from 1911 onwards, putting it to use in their mission to enumerate society for the benefit of administration and the citizens. The political parties then took its statistics to heart for their own political ends. In this context, the upper class became the smallest group. Here, quantity meant the right to rule. The SAP claimed that the majority of social group III members among the voters would lead to their inevitable ascent to power, the Liberals sought to win over middle-class voters while stating outwardly that voting was an individual act, while the Conservatives underlined the mix from different classes that voted for them, making it in their eyes a people's party. Lastly, the taxonomy moved to become part of welfare reforms in the post-war period through state commissions and the social sciences, who used it for mapping, engineering and democratising higher education. The division became in effect a welfare technology. This was not a smooth process, however. One of this article's findings is the scientific and political conflicts arising out of technologies for scientising society.

One can point to several factors behind the taxonomy's longevity. It was a powerful template, standardising population knowledge for different types of actors as well as quantifying languages of class into exact and precise concepts, reinforcing them not just as a part of political discourse but making them legible for action. There is an aspect of path dependency at work here. For many actors, the fact that it was already established was a major reason for sticking with it, as they could therefore compare their results to other studies. Another key feature of the division was how it reduced the complexity of society down to three classifications. The simplicity made it easy for different publics to understand the produced knowledge. The social group taxonomy was translatable – different actors interpreted it to their specific goals. The SAP, for example, could convert the classification social group III to the class it had a representative claim on – i.e. its main political subject – the workers. Knowledge from the social sciences and bureaucracy through the social group division could then be used in election campaigns and policy making.

Historians have noted how important class taxonomies within official statistics have been, in Britain, Norway and France for example, in how actors created knowledge and talked about society. More in depth studies and comparisons of how this took specific national forms, with different effects, remains to be done. I argue that these kinds of social divisions provide promising entry points to grasping ways of thinking and moulding societies in the twentieth century. Moreover, they alert us to similar classificatory operations today, such as dividing the populations into the educated and the uneducated, which frame how we understand and discuss the social.