KNOWING THE DEMOS

Concepts like voter and electorate are fundamental in all states practising democratic representation. However, the construction of these concepts following the introduction of universal suffrage is rarely studied. Swedish parliamentary debates on election statistics and sex-segregated ballots in 1921 and 1922 offer an illuminating opportunity to do that. Thus, this article argues that these key democratic concepts were in part constructed through the production of election statistics and in debates about what should be known about the respective political preferences of male and female citizens. Both sides in the debates emphasized sex as a fundamental category for understanding voters. But the debates also feature incompatible ways of representing the electorate – as individuals, as a unified whole, and as target groups – entailing conflicting visions of democratic politics. Thus, rather than being solely remembered as attempts to denigrate women’s votes and hence limit democracy, these debates should be understood as ways of dealing with the conceptual implications universal suffrage.

Concepts like voter and electorate are fundamental and ubiquitous in all states practising democratic representation. However, the construction of these concepts following the introduction of universal suffrage is rarely studied. 2 In this article, I argue that these concepts were in part constructed through the production of election statistics and in debates about what should be known about the respective political preferences of male and female citizens. I argue that this produced particular cultural representations of 'the people', or demos. This study does not concern itself with 'volkish' ideologies, or of the ways in which discourses of 'the masses' worked to exclude certain people from political representation. 3 Instead, I am interested in cultural representations of the electoral people, i.e. of that particular 'people' which is granted the right to vote and hence the capacity to authorize and inform its representatives in parliament through elections and referenda. The electoral people is seldom imagined and depicted in clear-cut and definitive ways; not even as statistics. 4 Here, Swedish parliamentary debates on sex-segregated ballots in 1921 and 1922 offer an unusually illuminating opportunity to explore dilemmas regarding the representation of this particular 'people'; dilemmas that women's suffrage did not solve but, rather, as in other European countries, made all the more urgent. 5 Thus, using the case of Sweden, this article asks: how was knowledge of sexual differences between voters debated after the introduction of women's suffrage? What were the consequences of these debates with regard to contemporary understandings of the democratic people? Following the wave of democratization around the end of the First World War, the Swedish constitution was rewritten in [1918][1919][1920][1921]. From then on, suffrage laws were neutral in regard to the voters' sex. 6 Subsequent descriptions and analyses of the electorate were not, however. As demonstrated by numerous scholars, nowhere in Europe did universal suffrage mean that politics lost its gender dimension. 7 The Swedish parliamentary debates on sex-segregated ballots show that the production of knowledge of sexual differences could be viewed as deeply problematic, and that ignorance was often deemed preferable. As will become apparent in the following, however, they also exemplify thatuniversal suffrage could, somewhat paradoxically, justify the production of knowledge of sexual differences.
In this context, Sweden is an interesting case because its democratization is generally considered quick and unusually successful. In both scholarly and popular works on Swedish political history, democracy is often taken for granted as definitivethat is, finally and successfully accomplishedwith the introduction of women's suffrage in 1921. 8 At the time, however, democracy was still thought of as a hope, an uncertainty, a continuing process of change, and something to fight for, or against. How to understand democracy and reinvent politics was an 'overriding problem', even in Sweden. 9 Perhaps it's easier today than ever since end of the Cold War to agree with Mark Mazower that we shouldn't think of democracy as 'rooted deeply in Europe's soil'. 10 It is not always remembered, however, that interwar European democracy was a novelty and a fragile experiment: 'Triumphant in 1918, it was virtually extinct twenty years on'. 11 It didn't go extinct in Scandinavia, but that didn't necessarily appear as a foregone conclusion. In addition, universal suffrage presented new difficulties for European democrats. It made politics unpredictable and uncertain in new ways. For that reason, getting to know voters and their political preferences became a crucial task for political parties. 12 Debates on knowledge production about 'the people' thus offer a way of investigating democracy as an ongoing and uncertain experiment, even in countries in which it was apparently very successful. Universal suffrage didn't suddenly unify nations. In terms of participation and influence, it is well known that nowhere did women's suffrage make men and women equal in the political sphere. 13 At the same time, in Sweden several limitations of the right to vote, apart from restrictions based on age and citizenship, still existed. 14 On a more theoretical level, historians and political scientists have pointed out that the demos can only exist through representations, which means that it is always unfinished, elusive, and potentially in dispute. This means that democracy is never definitive. 15 In the following analysis, I take 'gender' to denote, as Joan W. Scott puts it, 'the articulation and implementation of knowledge about the differences between the sexes'. 16 This analytical concept was obviously not available to the historical actors. To minimize confusion, I use the term 'sex' when referring to the historical actors' manner of speaking about men and women, which as a rule took for granted the existence of clear-cut and natural distinctions between them (kön is the Swedish term). The result of these discussions was, as will become evident, thoroughly gendered concepts of voters and of the people.

Studying productive debates: methodological considerations
The events referenced in this article may be summarized as follows. In January 1921, prominent Social Democratic politician and member of the second chamber of the Swedish bicameral parliament Arthur Engberg introduced a motion in parliament stating his regret that, at present, it was impossible to 'express statistically how the two sexes are distributed between the different political parties'. Women were soon to participate for the first time in elections to the second chamber, and it would be of 'no small interest' to get 'an exact image' of how men and women voted. 17 The separation of men and women in society was, of course, neither unusual nor in most cases controversial. It happened in homes, schools, associations, public spaces, and places of work. However, the political motives and consequences of the motion, including the alleged need to understand sexual differences in the electorate, were heavily criticized. Both chambers of parliament rejected Engberg's proposal, but the idea did not go away. Separate ballots for men and women were again discussed at length in parliament in 1922. This time, the question was whether it was important to know how men and women differed in their views on prohibition of alcohol, the subject of the first consultative referendum in Sweden. The answer, both chambers decided, was yes, and so separate ballots were produced.
Previous research has discussed how these debates can elucidate views of women as political actors, and they have explained the wish to use separate ballots as a political tactic in the prohibition question. Several historians have argued that politicians who long fought for women's right to vote now apparently wanted to depreciate their votes and in fact make them second-degree citizens. 18 In this article, judgments of that kind are only relevant as part of arguments made by the historical actors. Furthermore, in contrast to previous research, this article will not demonstrate how various positions in the debates can be explained by existing political ideologies, current political tactics, or personal motives. Instead, it will explore how the debates were productive; that is, how they contributed to the shaping and making of voters as objects of knowledge and how they generated representations of the people as political community. Hence, rather than seeking out what was 'behind' them, it is necessary to read the debates quite literally, and so clarifying the explicit arguments made. 19 Borrowing the vocabulary of Quentin Skinner, it may be said that while the motives of the debating MPs can, for the purpose of this article, be left out of the picture, their intentions to answer the question of whether the ballots should be segregated by sex were obvious, and these varying answers are examined in the following two sections. The debates, in their turn, formed interventions in discussions on how to know voters and represent the people, the topic of the concluding third section of the article. 20 While the first debate did not attract considerable public interest at the time, the second one did. Prohibition of alcohol was a long-standing women's issue to which the question of sex-segregated ballots added another contentious dimension. However, limiting this study to the parliament makes sense, as the aim is not to produce a comprehensive survey of the debates, but rather to provide a concentrated analysis of how the questions of knowledge of sexual differences and the issue of separate ballot papers were formulated and resolved. Finally, it may be pertinent to further clarify what this article does not do. It does not investigate the differences in opinions and attitudes between male and female voters; instead, the interest in understanding such differences is studied as a historical phenomenon.
A general interest? The first debate on sex-segregated ballots The stated aim of Engberg's motion in 1921 was to introduce sex as a new category in the official election statistics. 21 This would require changing the constitution. Hence, the parliamentary Committee on the Constitution presented a proposal, which was subsequently debated in parliament. In their position on the question, Social Democratic MPs differed from both the Committee majority and the rest of the participants in the debates. 22 The Committee majority rejected Engberg's motion: it would 'infringe on the secrecy of the ballot', especially in small districts, and result in 'practical difficulties' in the counting process. Furthermore, it was not clear why the suggested separation of ballots was of greater interest than other possible divisions based on, for example, age or occupation. 23 The separation of ballots would, the Committee suggested, mean 'the first step on a path, whose consequences cannot be foreseen'. Finally, the Committee majority pointed out that the motion had not demonstrated any need for the reform. 24 The objection that the reform would infringe on the secrecy of the ballot concerned the question of what could and should be known about voters. Nobody denied the importance of the secret ballot. But Social Democrats argued that, as it was already compromised in small districts, this reform would not make any difference in this regard, and the proposal would in any case not result in more unwanted information on women than on men. 25 The alleged practical difficulties were likewise dismissed. The motion's critics, however, underscored the point further, claiming that appeals and 'election troubles' would increase greatly, thus implying increased risk of disputed elections. 26 The questions about secrecy and practical difficulties were only two among many that dealt with the proposals' consequences for the system of democratic representation. Sooner or later, willingly or not, one cannot avoid 'estimating the respective value of male and female ballots', one conservative MP claimed. 27 The proposed investigations, another MP argued, would result in 'a kind of gradation' of the voters. 28 What kind of gradation this implied was spelled out explicitly: the reform would make it possible to value women's votes as less than men's. The reform would 'put women in a particular class of voters', in complete contradiction to the recently introduced principle of equality between men and women. 29 A system of unequal votes was nothing unknown or untried: in local elections, a system of this kind had just been abolished. 30 According to Engberg's adversaries, his seemingly harmless interest in reliable and useful numbers would undermine the very democracy Social Democrats themselves had worked long and hard for.
Of course, Social Democrats did not accept such a characterization of their position. The reform would not result in votes being valued unequally, they argued. According to Engberg, it was 'impossible to understand, why it would be a degradation to mark female ballots with a K, when it is not a degradation to mark male ballots with a M'. He also did not think that women would experience separate ballots as a 'violation of their independence or standing as political citizens', as suggested by a liberal MP. 31 In making these arguments, Engberg actually failed to counter his critics' main points. For them, it wasn't only or even primarily the voting procedure per se that would degrade women's votes, but rather the subsequent actions made possible by the knowledge this procedure would generate. The assessment was that if any votes would be degraded, it would be those cast by women. This was not the purpose, Engberg assured, while one of his fellow party members in the first chamber made it clear that 'the opinion, which wants to ascribe different values to votes in general elections' would surely be opposed, if it showed itself. 32 This does not mean that anyone argued that men and women were alike. However, the fact that men and women were different did not necessarily mean that they needed separate columns in the election statistics. Voters were quite unlike each other in many ways, so why was this particular categorization so important? On this point, Engberg's motion was rather vague, which made it easy for critics to speculate on its true intentions, and to attack it as undemocratic. To the troublesome consequences the Committee had hinted at but not specified, one liberal MP counted the risk that divisions of the electorate would multiply, and in the process change the meaning of the act of voting itself. Here, something resembling a society of estates was suggested as the ultimate result of the reform. 33 The reform would institutionalize a segregation and gradation of social groups in the very act of voting; that is, in the symbolically most important event in modern democratic societies. In opposition to this imagined outcome, well-known liberal and individualist arguments were formulated: men have 'diverse dispositions and political views', and so do women; women cannot be made to 'follow each other' and, just as men, they should be considered 'citizens without special interests'. 34 Social Democratic MPs insisted, however, that sex was a unique, and uniquely important, category. A number of them described this as simply common sense. As women were the newly enfranchised group, it was obviously interesting to seek 'information' on their views. 35 However, arguments explicitly grounded in perceived natural differences between men and womenthat is, differences that had nothing to do with individual choices or particular circumstanceswere also put forward. According to Engberg, the fact that people 'choose their occupations, but hardly their sex' was an argument in his favour. 36 A fellow party member maintained that the separation was important because men and women were 'essentially different' and because 'the female, both as wife and mother, holds such an exceptional place in society' that the categorization in question 'cannot even be compared to' those based on occupation or age. 37 In this view, there was nothing undemocratic or politically reactionary in introducing sex as a category in the official election statistics.
In order to clarify the arguments made by Social Democratic MPs, it is necessary to demonstrate in further detail how the purpose of the motion was subjected to debate. What was the utility of the numbers? In both chambers, opponents of the motion put pressure on Engberg and his fellow party members to clarify their reasoning: 'the very purpose is still obscure to me', one liberal MP claimed more than halfway through the debate. 38 This particular MP did not doubt that supporters of the motion were 'good democrats' and that they did not wish to 'disenfranchise women'. 39 The only explanation remaining, he maintained, was that the result of the separate count would 'have its value especially as a guidance for agitation '. 40 In drawing this conclusion he was not alone, and while this purpose was not described as fundamentally undemocratic (like a de facto gradation of the vote), it wasn't portrayed as completely innocent either: 'Women have enough suitors as it is, without one having to take special measures to encourage the fight over them.' 41 Hence, in their apparent defence of the need to consider men and women as equal citizens and not as separate classes -'it is not a question of men or women but of civil rights', the above-quoted MP continuedliberals did not keep themselves from mobilizing traditional views on sexuality and sexual differences in order to stop the reform. 42 Here, women voters were depicted as passive objects courted by apparently male parties, while respectable members of (the still all-male) parliament acted to protect them from the aggressive suitors. In short, women voters were not, in the end, put on an equal footing with male voters. Intensified agitation, not the unequal treatment of the sexes, was the fundamental problem. In the standard liberal view, of course, 'agitation' was a problem because it was understood to be the very opposite of a rational conversation between honest people, leading ever closer to the truth about matters of public concern. Agitation introduced an irrational dimension to politics, appealed to collectives rather than individuals and to emotions rather than the intellect, which made women and the lower classes more susceptible to it than men and the educated classes.
When Social Democrats subsequently offered more exhaustive justifications of the reform, they stressed the general need for more comprehensive knowledge of public opinion, and, when pushed a little further, their statements demonstrated that 'guidance for agitation' was not an altogether incorrect description of the numbers' intended use. 43 It was important to understand 'the effects of women's suffrage', and to know if women had differentperhaps more conservative?political preferences. 44 Gustav Möller explained that historians and everyone else 'interested in the study of our social conditions and political views in society', regardless of party affiliation, would find it to be of 'immense interest' to be able to 'follow the currents' among men and women. The burden of proof lay rather on the opponents of the motion, Möller implied. Why would anyone want to keep this information secret? Out of fear of being revealed to belong to a women's party? 45 In this particular line of argument, the opposition to the motion was based on political interests, while the interest in producing the numbers was more accurately understood as social scientific. It was, another proponent of the motion insisted, important to know how 'public opinion is formed', and in this matter, one did not want to 'be blindfolded' regarding women. The most 'sensitive instrument' possible was needed in order to be able to understand how 'the so-called will of the people' arises and how different camps 'react' to a variety of occurrences. 46 While this speaker only hinted at the use to which political parties could put the new numbers, Hjalmar Branting was, in the end, more direct. There was a need for 'an arrangement, which makes it possible to follow the mood among both men and women in the country, so that one may see how their views are laid out, and hence could give to each and every political group a sense of how it should aim its efforts to spread information in order to bring about a different view than the existing one'. 47 In the liberal interpretation, this was 'guidance for agitation'.
This wish to understand public opinion was also subjected to a conservative criticism that portrayed the separation of men and women as a great danger to fundamental social and political values. The aim of the proposed measure was to 'listen to the will of the people', one member of the Agricultural Federation argued, however 'not to the collective will of the people, but to the male and the female popular will separately'. This 'divorce', this 'great danger to the whole of our people' would 'split the people into two nations', so that the 'two sexes would be vehemently opposed to each other'. 48 Here, a classic conservative principle was invoked: the people as a political community should be understood as one, and the political leadership should transcend special interest in order to care for the whole, 'the common fatherland'. 49 The implication was that the proposed reform was a building block of a new kind of divisive self-interested class politics. If separating men and women at the ballot boxes meant further support to agitators, it would make the problem even worse, according to the same logic. Among conservatives, agitation was generally understood as a divisive political practice, intensifying conflicts and creating new onesfor example between the sexesand thus opposed to the common good. For this reason, even political parties could be considered a problem: they represented an effort to replace the public good with the interests of a particular group. In other words, preferring ignorance about the statistical differences between men and women was consistent with mainstream conservative political thought. 50 Reliable citizens or wilful consumers? The second debate on sex-segregated ballots The rejection of Engberg's motion was not the last word about sex-segregated ballots. In debates on the prohibition of alcoholic beverages, this was a major issue. In 1920, the governmental Temperance Committee, previously tasked with investigating the sale and serving of alcoholic beverages, suggested a referendum to confirm or reject a decision by parliament (i.e. an abrogative referendum), without separate ballots. Committee member Ivan Bratt did not agree with this, however, arguing instead that the referendum should be consultative, and that separate ballots should be implemented. This was ultimately the outcome of the parliamentary debates on the issue.
These debates repeated several themes from the previous one. There were, however, key differences. For one thing, opinions on neither separate ballots nor prohibition followed simple party lines. A main issue in regard to prohibition was whether a ban on alcohol would work, and, equally important, if it was possible to confidently know so in advance. In this particular context, a separate ballot was suggested as an important tool of knowledge. As only men were keen on drinking alcohol, maybe their views should be studied separately? When parliament debated a prohibition referendum, this was the centre of attention. 51 The referendum was explicitly discussed as a way of studying 'the people'. 52 But what kind of knowledge would it generate? The Temperance Committee held a positive view of people's understanding of the issue, but it did not expect voters to have a 'more correct view' than their representatives in parliament, and it didn't matter if the referendum 'would equal "an appeal from knowledge to ignorance"'. The result would not be interpreted as an informed decision on prohibition; however, it could tell whether a fundamental condition for its successful implementation was at hand; namely support from 'a clearly and unambiguously manifested majority of the people'. 53 With this, the Committee made it clear that the usual critique of direct democracy was not relevant, as the purpose of the referendum would be fulfilled simply by making people express their views, regardless of how ignorant or knowledgeable they were.
At the same time, the Committee argued that not only 'the size of the majority, but also its quality' mattered. 54 These considerations did not include separating voters according to sex; rather, such a move was condemned as undemocratic and unfair. 55 Instead it was a question of how opinions were formed: 'A majority resulting from a violent, emotional agitation is obviously of less value than a majority which bases its judgment on experience and reflection'. 56 The important thing to know was whether a majority could be relied upon to continue supporting prohibition. 57 An opinion formed as a result of careful consideration was assumed to be stable. The Committee also hoped the referendum would be a powerful force of change in case a prohibition law was accepted. Most of the people would, through its participation in the decision, 'shoulder the moral responsibility for upholding the law'. 58 This was the main argument for an abrogative referendum, rather than a consultative one. 59 That argument was brushed aside by the dissenting committee member Ivar Bratt. 60 He preferred a consultative referendum, and in arguing his case, he further underlined the referendum as an investigative tool. A consultative referendum, he stated, would be 'one of the investigations, whose results would be valuable for the state authorities to know before they finally make a decision'. 61 Bratt spelled out the standard argument for separating the ballots: because only men would frustrate the successful implementation of prohibition, it was imperative to know how many of them opposed it. This was the argument of someone who opposed prohibition. It was obvious to everyone that separate ballots would potentially make it easier to stop a ban. 62 In 1922, a motion to go through with a consultative referendum without separate ballots was submitted in parliament. 63 The motion was read by the first Committee on Laws and in its proposal for a decision, which included a number of reservations from minority members, the committee majority supported a referendum. However, it also proposed separating ballots according to sex, using the same arguments Bratt had made. The purpose of the referendum, the committee maintained, was to 'provide to legislators reliable and comprehensive material on which to judge the issue'. 64 Unsurprisingly, Engberg took an active part in the subsequent debates. His contribution is instructive, as it elucidates in detail several key contested aspects of the issue. A referendum, Engberg and other proponents of separate ballots emphasized, was 'only a tool for gathering information'. It did not provide a decision, and it did not shed any light on 'the will of the people' but only on 'the composition of that opinion, which is for and against prohibition'. 65 In other words, the referendum should not and could not contribute information on a single, homogeneous entity (the will of the people), but it could make known the constituent parts of heterogeneous units (the adherents of different opinions). Hence the need for qualitative, not only quantitative, knowledge. Because the referendum only informed parliament ahead of its decision, the majority of the people could not simply force its will on the minority, Engberg observed. On the contrary, not only the majority but the minority as well must be respected. Here, members of 'class parties on both sides'among whom Engberg counted himselfshould learn from 'classical liberalism'. 66 In short, advanced democracies adopted a 'different perspective than a purely quantitative one'. 67 The consequences of this argument were spelled out by Engberg himself, who joined others in wishing for more categories than sex to be used, including age and social position. Unfortunately, he conceded, this would mean too many technical difficulties in this case. He also insisted that categorizing voters in this manner was potentially relevant in other referenda as well. Perhaps in some future referendum, it would be especially important to know women's views rather than men's, he argued. 68 Engberg stated his view on separate ballots explicitly as a defence of democracy. His views clearly evoked a political order that could survive the well-known critique of democracy as characterized by the power of the masses, the levelling out of differences, and decision-making based on ignorance. He urged his audience to recognize that a democratic form of government had to be 'integrated with the will and courage to face reality and produce all the information that can be obtained' and he continued to make clear that he would not profess himself an adherent of a 'democracy of obscurantism, which deliberately cuts off the possibility to get a clear picture of this matter'. 69 In making statements of this kind, he was not alone. 70 How, then, did other MPs argue that no one needed to know or should try to find out how men in particular voted on this issue? One line of argument demonstrated the absurd consequences of separating the ballots. For example, shall potential criminals or those hurt by a proposed lawin this case male drinkersbe allowed to determine if it should be adopted? 71 More frequently and emphatically, it was claimed that separating the ballots would offend women, undermine their status as equal citizens, and fail to recognize the value of their opinions on social issues. 72 This was a question of democratic principles and equal rights; separating the ballots was in fact 'antidemocratic'. 73 One conservative MP explained that, while most women had had no desire to gain the right to vote, there was no reason now to 'tamper with their rights or make them unhappy'. 74 At issue was more than this particular referendum. A Social Democrat pointed out that, as an institutionalized practice, referenda had 'a much deeper importance than that of an arbitrator in the temperance question'. This made it all the more important that 'every individual', regardless of sex, had the same possibility to influence the outcome. 75 Several liberals argued along similar lines. The proposal could create 'a difficult precedent', Kerstin Hesselgren argued, while a fellow liberal underscored the need to view this question 'not as a prohibition issue, but as a matter of principle and a constitutional question' and to recognize that separating the ballots would 'ruin democracy'. 76 As in the previous debate, the proposal's enemies did more than invoke equal rights and constitutional principles. Several argued that if men and women were separated in the referendum, half of the electorate would not have a very good reason to participate in it. 77 If the proposal was accepted, one socialist argued, one would have 'incited an agitation, in which one told women that their participation in the referendum was less valuable', making the set-up comparable to the recently abolished 40-grade scale in local elections. For 'purely psychological' reasons, 'this must have a negative effect on their participation in the referendum'. 78 The implication was that the referendum in this case would not let MPs know what they needed to know; namely, as one of them put it, whether a 'clear and tangible majority' supported prohibition, regardless of its composition. 79 A related consideration dealt with what one could and should get information about. For Engberg and others, 'opinion' was the key concept and the referendum could only work as a 'tool for gathering information' if voters were separated and categorized according to sex. In contrast to this stood definitions of the object of study as an indivisible whole or a single actor. The referendum should express, according to one liberal MP, 'the collective will' of the people, and, as a Communist MP argued, a ban on alcohol would not be supported by this group or the other but by 'the Swedish people'. 80 If voters were separated, a Social Democrat explained, 'one would get a completely picked apart and destroyed image of the people's opinion '. 81 There were also concerns about far-reaching long-term consequences. Hesselgren maintained that prohibition was a 'common question for the whole people, and it should not have to be divided on male and female lines'. 82 This time around it was mainly Social Democrats who claimed that women and men would be made to confront each other, creating unnecessary conflict and a new 'issue to fight over', perhaps resulting in political parties breaking apart. 83 The consequence might be, one socialist warned, that in future parliamentary elections 'there will be a march on two lines, one male, the other female'. 84 Instead of a 'united and cooperating nation', a fellow party member predicted, 'we might have two fighting sexes', at a time when 'the nation's collective strength' was needed. 85 A conservative MP went further, underlining that sex was a completely irrelevant category (while at the same time conceding that sex made people unequal): 'From a political point of view, we do not stand as man and woman, no, we stand as equal citizens in a country we call ours and whose well-being we all want to promote'. 86 At the same time, however, the debate offered ample opportunities to oppose separate ballots using observations of women's peculiar nature and distinctive, evidently unchanging relations not only to alcohol but to men, children, and society at large. In this case, responsibility for future generations was a major concern. For example, when arguing against apparent attempts to devalue women's votes, Social Democratic MP Agda Östlund insisted that, while the referendum couldn't predict future generations' attitudes, women could strongly influence them. 87 Another speaker opposing the separation claimed that women's interests were not as 'egoistic' as men's. Instead, this MP continued, they speak for 'the family, the rising generation, and also for society'. 88 Women didn't drink, but they suffered from men's drinking, so why should they have less influence on the matter? 89 The implication of all this was clear: the fact that citizens of both sexes should be treated equally in the referendum didn't mean that they were alike or that differences between them didn't matter. Quite the opposite.
On one level, the referendum was a study of people's opinions, as it asked whether they favoured a ban on alcoholic beverages or not. According to some proponents of separate ballots, information on men's opinions reduced the possibility of interpreting the result in arbitrary ways. In other words, separate ballots made MPs more dependent on citizen's explicit opinions when deciding on the ban. 90 Ultimately, however, the main point was not getting information on opinions, but rather gaining knowledge about likely future actions. The fundamental idea behind the referendum was that someone who opposed a ban on alcohol was also very likely to violate future prohibition laws. In other words, the real question was: will (male) voters continue to drink with a ban in place? The existence of a strong connection between expressed opinion and future action was thus a premise in the debates, shared by opponents and proponents of separate ballots alike. 91 This further underscores that the element of direct democracy in the referendum was very modest at best, and it means that voters were primarily constructed as consumers rather than political subjects. The latter aspect was emphasized by the proposal to separate men and women, as this separation was based precisely on their assumed consumption patterns.

Productive debates: a concluding discussion
These were uncertain times in Europe. The questions of where to find and how to understand the people were pressing ones, but without obvious answers. As Masower tells us in his survey of interwar European politics, in many places fragmented legislatures didn't seem to reflect the popular will as much as reveal its absence, magnifying rather than resolving tensions in society, while political parties were accused of acting as intermediaries for sectional interests. In Sweden, as in other countries, the many short-lived minority governments did little to prove the stability and permanency of the democratic system. 92 The idea that democratic representation based on universal suffrage was a final stage of political development was disregarded by many politicians and activists from both the Right and the Left, and not only by extremists. 93 Furthermore, in the early 1920s, nobody could be quite sure of women's long-term roles as members of the political community. There was great interest in mobilizing their support as voters, but they were often compared to men and found lacking in political interest, education, and maturity. In the Swedish case, these widespread but not uncontested depictions, Kjell Östberg argues, were part of strategies to exclude women from politics, but they were also grounded in disappointments following women's lower turnout in general elections. 94 The debates studied in this article contributed to rather than counteracted the uncertainties regarding democracy and future political developments. However, at the same time, they raised important questions about how to understand and study voters, and in so doing constructed cultural representations of the people.
The two parliamentary debates differed in several ways, apart from the different outcomes. In the first debate, the manner of counting votes in all subsequent elections to the second chamber of parliament was at stake; the main interest was knowing how women voted; and the separation could be defended as simply producing knowledge of equal interest to everyone. In the second debate, the question of separating men and women at the ballot boxes had everything to do with a future referendum on prohibition; men's votes were of special interest; and the question was obviously politicized from the start. Furthermore, knowing how political developments changed voters' opinions was the main point in the first case, whereas knowledge of future consumption patterns was at stake in the second. While these differences are important, for the purpose of this article there are also more general points to be made.

Producing knowledge and ignorance
In the prohibition question, separating the views of men from those of women could make it easier to dismiss a ban on alcohol. In the end, such a move was not necessary, as the outcome of the referendum did not produce a clear majority in favour of prohibition. Out of all eligible voters, slightly more than 55% participated (among men the turnout was almost 63%, among women a little more than 48%), and 51% voted against a ban. The separate count demonstrated that 58.5% of participating women and 40.9% of men voted for a ban. 95 For those inclined, the result could unequivocally confirm that men and women were different and that this difference was important to consider as part of the political landscape.
Engberg's reform would have transformed certain election statistics from only giving information on men (before universal suffrage) to contributing politically useful knowledge about men and women and hence about sexual differences. If implemented, men would have been regarded as characterized by their sex just as well as women were (this move was of course resisted by those who claimed that sex had nothing to do with citizenship). Engberg's proposal was not accepted, but the election statistics did not remain unchanged. In other words, rejecting the proposed reform was not a non-decision preserving status quo; along with women's participation in parliamentary elections the statistics were inevitably transformed anyway. For instance, it was now harder to get information on male voters specifically, while in other circumstances they were explicitly understood in relation to female voters. 96 As scholars in the field of agnotology have argued, studying the production of ignorance is as important as studying the production of knowledge. 97 These cases demonstrate clearly how deliberate ignorance could be ascribed important political meanings by the historical actors themselves. MPs were aware of the fact that not only knowledge but also ignorance could make certain actions and practices possible, while making others significantly harder or impossible. Undermining the secret ballot would alter the way politics were carried out and change election results. Nobody wanted that. The question in the first debate was whether a new form of election statistics could have similar effects. With Robert N. Proctor's terminology, the outcome of the debate was a case of ignorance as a 'passive construct'; in other words, MPs chose deliberately not to find something out. 98 For opponents of Engberg's reform, ignorance about sexual differences assured equal treatment of men and women, at least in some circumstances. This ignorance made it harder for political parties to shape their communications to a specifically female audience, and that was indeed the stated purpose of the opposition to the reform. Women should be protected from agitation, hence in this case not treated the same as men.

Changing politics, altering the electorate
The debates dealt with how knowledge about voters could legitimately be produced and used by MPs and political parties. This did not mean, however, that different ways of counting and categorizing the electorate were assumed to be of no consequence for ordinary voters themselves. Perhaps Engberg's reform would not have had an immediate effect on the parliament's constitution, but the general understanding was that it would influence the way parties communicated with and engaged voters; women in particular. In the longer view, therefore, the reform would certainly have an indirect influence on voters' views and decisions, and hence their representation in parliament. Put differently, election statistics were understood as one component in a feedback loop between the people and their representatives. Underlining this further, separating men and women at the ballot boxes would, several MPs maintained, have the very serious consequence of creating divisions between the sexes and hence a divided people and a nation split in two. Classifying and representing voters as different from each other would create differences, in effect 'making up' the people. 99 In sum, MPs agreed that changing the ways voters were investigated and described would, in turn, change political practices and actors alike, for better or for worse. Whether correct or not, they put forth a view of the electorate as a moving target, oftentimes assumed to be a fairly unstable and easily influenced one at that. At the same time, the general agreement about the natural differences between men and women constructed the people in this important aspect as completely stable and unchangeable.

Gendering democracy
MPs in favour of separating men and women argued that sex mattered, even though suffrage laws no longer made any distinctions between the sexes. But very few, even among their opponents, claimed that sex was an irrelevant category. Nobody argued that women should not be understood and treated as individuals; this concept, however, implies differences as well as similarities and equality, and sex was generally understood to be by far the most important of these differences. In arguing that no distinction between the sexes should be made at the ballot boxes, several MPs themselves in fact made that very distinction: for women's sake, and because of their distinct social roles, votes should not be counted separately. Of course, nobody explicitly made the case for depreciating women's votes or, more generally, women as political actors. Both sides in the debates openly and emphatically embraced democratic representation, and argued that democratic considerations were in favour of their particular views.
As illustrated by previous research, it is easy to appreciate and agree with the democratic case against using separate ballots. In practice, if not formally, it undermined equality before the law and reintroduced a hierarchy between male and female citizens in the political sphere. According to proponents of sex-segregated ballots at the time, however, it was instead obvious that this fear of unequal treatment was unfounded. As explained by Engberg, separating ballots would instead produce relevant qualitative information about the people, which would otherwise be understood as an undifferentiated mass. As a practice, separate counting was assumed to support a modern democracy, built on group interests, parties, and mobilization of voters. It is well worth pointing out that MPs on both sides of the debates hadregardless of possible tactical considerations or obscure personal motivesrational reasons to consider their own arguments as true and relevant. By this I mean that their arguments were grounded in accepted ways of reasoning about political representation, democracy, and the nature of men and women. 100 Additionally, in making these arguments, they collectively made gender, i.e. 'the articulation and implementation of knowledge about the differences between the sexes', a fundamental part of democratic political discourse. 101

Competing representations of the people
In spite of this common ground, much was at stake in the debates. What kind of democracy had been created, and how should it be developed? How should the people be understood? As both history and the current rise of populism make clear, the latter question is fundamental for democratic political thought and practice, but it is not one that can be definitively answered, not even by statistics. In the first debate, the supposedly neutrally observed people invoked by social democratic speakers consisted of a number of segments or target groups that could be appealed to in different ways, depending on who they were and the kind of opinions and sentiments they already had. In other words, the people resembled the audiences of advertising. Similar considerations were important in the development of interwar European political practice. In Britain, for example, so-called targeted electioneering 'sought to overcome the anonymity of mass politics […] by breaking the electorate down into distinct sub-groups and interests'. 102 In the Swedish debate, opponents of the social democratic proposal instead invoked either a liberal view of the people as primarily individuals rather than groups, or a conservative vision of the people as an undividable community. The liberal and the conservative views converged in their argument that the reform threatened the democratic principles of equality and universal suffrage. Furthermore, the practice of sex-segregated ballots seemingly threatened to divide the people into irreconcilable groups, undermining much-needed cooperation. This, indeed, turned out to be a major political problem in interwar Europe. 103 As a practice of cultural representation, election statistics, however, were more closely aligned with the social democratic vision. Their task was neither to give information on individuals nor to characterize the people as a whole. On the contrary, they already presented the people as groups, making it possible to analyse voters by using and combining information on, for instance, places of residence and kinds of occupation. The debate, however, made it clear that when categorizations became openly politicized, as in this case, the position of election statistics as a neutral description of the people and its choices was undermined. From this perspective, it made sense for Engberg to initially omit a clear explanation of the reform's purpose.
In the second debate, it was a matter of dispute whether it was acceptable to show more consideration for one group over another when deciding on political issues. Those arguing that this was indeed the caseamong them Engbergdid not only defend the separation of men and women in the current prohibition question but also the more general idea that, in response to specific political questions, the people naturally divided itself into different groups. In opposition to this, the conception of the people as a whole, whose unitary will would reveal itself in the referendum and which must be adhered to, was put forth. Nobody articulated the liberal response to this more clearly than Engberg: the people is not, in fact, a single entity with a single will. In reality, accepting that idea meant allowing a simple principle of majority rule, and it meant excluding minorities from the concept of the people. This could not be accepted, Engberg emphasized.
Election statistics have so far been conspicuously absent from the historiography of statistics and the politics of numbers. Therefore, we still know very little about the kind of political or social scientific considerations that went into this form of knowledge production. Clearly, the debates studied in this article did not simply mirror contemporary views of women as political actors, and they cannot be fully accounted for by referring to political tactics or current party politics. As suggested in the Introduction, the debates were productive as well. In sum, this concluding section has made four interlinked arguments. First, the debates established the assumption that ignorance as well as knowledge of sexual differences in the electorate could support ways of making politics and reshape the relationship between the people and its representatives. Second, the electorate itself was constructed as a moving target, dependent on the very knowledge produced about it, thus creating a feedback loop between the people and its representatives. Third, both sides in the debates emphasized sex as a fundamental category for understanding voters, and in so doing used arguments recognized at the time as democratic. Finally, I have argued that the debates put forth different and incompatible ways of representing the electoral peopleas individuals, as target groups, and as a unified wholeentailing conflicting visions of democracy and democratic politics. The key question was not who to include or exclude from 'the people', but how the already includedthe votersshould be understood and treated. The wish to segregate ballots by sex countered the liberal logic embedded in the secret ballot; instead of identifying oneself as an individual citizenabstracted from group interest and undue influencewhile casting one's vote, one had to make one's choice as a man or a woman, thus identifying oneself as belonging to a pre-existing and clearly defined segment of the electoral people.
In the following years, a number of circumstances further emphasized democracy as an uncertain matter, not least the new European dictatorships. The interwar experiments in 'making peoples, remaking souls' continued in Scandinavia and elsewhere. 104 In this context, how to make sense of voters and represent the people was not a settled matter, the ideology of a class transcending 'people's home' and bipartisan cooperation notwithstanding. The debates on separate ballots anticipated later discussions on, and ways of conflating, citizens, audiences, and consumers, which were in part triggered by the increasing use of ever more sophisticated advertising techniques in all spheres of society, including politics. They also anticipated political hopes and fears associated with modern opinion research, first introduced in the late 1930s. Democracy, along with its demos and ways of studying voters, was in motion, controversial, uncertain, and would so remain. The question of how the democratic people can be represented in a politically legitimate way is still a part of political, historical, and social scientific discussions in our time. This should not in itself be seen as a problem for democracy; on the contrary, ending or rendering irrelevant discussions about how to understand and represent 'the people' is of course the real undemocratic move. 105  Rosanvallon, Democratic Legitimacy, 130-1. 5

Funding
Cf. Scott, Only Paradoxes to Offer, 69: 'Instead of eliminating the general problem of sexual difference, the vote drew attention to it with greater force'. The Swedish case is probably not unique, but similar debates appear to be rare. Scott quotes one later example in France: Scott, Only Paradoxes to Offer, 221n7. 6 The only exception (removed in 1922) was that male voters had to complete military service. 7 See e.g. Scott, Only Paradoxes to Offer, in particular ch. 6; Gottlieb and Toye, Aftermath of Suffrage; Borchorst and Dahlerup, Før og efter stemmeretten. 8 -Friberg, Demokrati bortom politiken, 11-12. Even in works that discuss continuing problems for or critique of democracy, it is often said that democracy was ultimately 'victorious' at this time: e.g. Nilsson, 'Högern 1900-1940 It also is telling that most accounts of democratization in Sweden end in 1921 (parliamentarism had been in place since 1917). Cf. Gottlieb and Toye, Aftermath of Suffrage. 9 Jonsson, Crowds and Democracy, 9. See e.g. Friberg, Demokrati bortom politiken; Tistedt, Visioner om medborgerliga publiker. 10 Mazower, Dark Continent, ix. 11 Ibid., 5. 12 Schröder, 'Wer ist Freund'. 13 Östberg, Efter rösträtten; see also note 7 . 14 These restrictions were based on, for example, enduring dependence on poor relief, personal bankruptcy, and criminal sentences; see also note 6 . 15 This is variously explored in, for example, Rosanvallon, Democracy Past and Future; Frank, Constituent Moments; Jonsson, Crowds and Democracy. 16 Scott,Gender,206. 17 Bihang … 1921: Motioner i Andra kammaren 1921, Nr 2, 3. Separating votes by sex was very unusual during the interwar years. According to Tingsten, Political Behavior, 37-8, 65, 72, it occurred in some parts of Germany in some elections to the constituent national assembly after 1918; in elections to the lower house of parliament in Austria after 1920; and in a few elections in the United States.