Skip navigation.

All-out war? Brexit, voting and the production of division

Cathy Elliott

It’s commonly assumed that the Brexit referendum exposed pre-existing faultlines in British society. But we need to take seriously the idea that voting produces divisions and identities, rather than simply measuring them. If we consider the sorts of subjects and identities our current modes of voting in elections and referendums produce, we might be prompted to embrace more reflective, and more deliberative, democratic practices, in order to bridge rather than entrench divisions in British society.

On 23 June 2016, Anne Perkins wrote in her Guardian column: ‘an unusually low tide occasionally reveals an ancient henge, the referendum campaign has exposed a chasm in British politics’.1  This trope of the chasm or divide in British politics has become extremely widespread, and the idea that ‘Leavers’ and ‘Remainers’ represent two distinct and irreconcilable groups in British politics is now rather commonplace. However, it is not the metaphor of a divided Britain that I want to discuss here, but rather the assumption in the article that the referendum result revealed something that already existed: like a henge buried by low tide.

This assumption that referenda and other elections function as ways of revealing the preferences of a population is likewise widespread. The day after the referendum, a range of newspapers from across the political spectrum2 referred to the revelation of a divided Britain, a narrative that also appeared in BBC coverage of the result.3 Academic accounts, such as the psephological study by Clarke et al, Brexit: Why Britain Voted to Leave the EU, similarly refer to the referendum as revelatory:

Britain’s decision to leave the European Union revealed as much about how its society had been changing for many years as it did about the impact of the short and bitter referendum campaign itself.4 




This study carefully traces the reasons why leave voters appear to have voted leave over many years through the rise of UKIP and increasing pressure on David Cameron to hold a referendum. The authors conclude that the referendum vote was by and large a question of ‘valence’ (or competence): voters felt that the EU was preventing effective action by their elected representatives on issues they cared about, such as immigration and the economy, and that is why they voted leave.

However, another way of thinking about the referendum is that, rather than revealing divisions that had been there all along, instead it produced those divisions. This analysis is also incipient in the same journalistic accounts, with the BBC’s Mark Easton suggesting immediately after a claim that the referendum had revealed ‘ancient’ divisions, that the result is ‘a scar that has sliced through conventional politics and traditional social structures’.5  This seems to be a contradiction: if all the referendum did was reveal existing decisions, surely it could not also ‘slice’ or be at odds with the structures and conventions that existed previously? To get to the bottom of this problem, we need to engage with a central concern in the philosophy of science as it features in debates about how we can know anything about the political and social world: are human individuals reasonably stable and knowable, such that their interests and preferences can be counted? Or does generating knowledge about them change and shape those human subjects to the extent that their individuality is formed and transformed in the process? In other words, by asking people to vote one of two ways, do we come to know and understand people’s pre-existing preferences, or do we shape those people and their preferences into two exclusive groups that did not exist before?

In what follows, I sketch out the difference it might make if we take seriously the notion of elections as productive not only of division, but also of some of the characteristics that we take to be most fundamental about the individual voter: their identity and their preferences. I want to suggest that such characteristics are produced through the very acts and practices that appear to be their expression. To return to the metaphor with which we started: what if the sea were not exposing the ancient henge, but rather the action of the water was carving the structure out?

What might it mean to think about elections as events that can fundamentally shape and change the societies in which they take place? How might such an understanding change the ways we account for the role of elections and voting in liberal democracies? And what might the consequences be?

What purpose does it serve?

In trying to understand any political practice or technology, it is often useful to ask what purpose it serves and what problem it is being used to solve. In the case of elections, why is the transformation of political preferences into numbers a useful thing to do, and for whom?

As Sally Engle Merry points out, ‘it is the capacity of numbers to provide knowledge of a complex and murky world that renders quantification so seductive’.6  Stephen Coleman assesses the attractiveness of quantifying preferences in elections by pointing to three related uses of voting and polling.7  First, the practice of counting votes is detached: each vote counts for one no matter who is casting or counting it. Second, by creating simple comparable categories, quantification smooths over complexity, including the messy and contradictory reasons why voters cast their vote in particular ways in the first place and the intensity with which they do so. Votes, unlike the people who cast them, are commensurable and comparable. Third, voting makes preferences visible. In an era of small-l liberalism, where political equality and freedom are the taken-for-granted background assumptions and the role of government is to ensure the flourishing of the population, the logic of elections is that they solve particular problems of knowledge: what should be done to enable the population to be optimally governed and who should take on the role of governing in the name of the population. For referenda, the point is even more straightforward. For a given policy position – what should the voting system be, say, or should we remain in the European Union – the practice of getting the eligible population to vote will yield an answer that can be understood as legitimate, enabling a decision to be made. As David Cameron’s own comments make quite clear, this is often the explicit, as well as implicit, reason for calling a referendum. In his famous 2013 Bloomberg speech, pledging to hold the referendum, he said: ‘It is time to settle this European question in British politics’.8

Certain assumptions underpin this logic. For example, each voter must count as one and must be individually responsible for their decision, which is to be arrived at through the free exercise of their own individual conscience. When I went to the polling station on referendum day to cast proxy votes on behalf of my sister and brother-in-law, I was carefully informed which ballot paper belonged to whom, so that there be no mistake about whose vote was whose: this was important because it was their personal conscience and individual choice at stake in their vote. Not mine and not one another’s. Nevertheless, although each individual vote matters, it matters  predominantly at the level of the population as a whole: it is the tally of each individual vote, each small act of individual conscience that is the crucial thing. The logic of elections, then, is both the logic of individualism and of the aggregate. The power of the election as a technology is its ability to arbitrate neutrally amongst the many competing claims of individuals and to offer a result that enables us to move forward.

It should be noted that this idealised logic of the election is not necessarily closely related to the actual experience of voting, which may have little to do with the rational exercise of individual freedom. As Stephen Coleman has pointed out, elections are frequently anxious, intimidating and emotional for their voting participants, who may cast their ballot without any clear understanding of what they are voting for or what the broader consequences might be.9 Nevertheless, this is in some sense precisely the point: the conflicting emotions and anxieties of voters do not help us move forward or make decisions, whereas the cross in the box, and the assumption of political freedom that underlies it, does. So, the process of how those anxious, contradictory and emotional subjects become the useful, knowable citizens who go and cast their vote is important to look at in more detail.

Michel Foucault’s work has shown us that the logic of the production of citizenship in late liberalism is part of a much broader set of power relations that likewise privilege both individual freedom and the flourishing of whole populations. His earlier work on the prison explained the emergence of ideas about the individual as a measurable unit of analysis, to be understood, quantified and measured through practices of the drill, the examination, the dossier and the report, amongst others. This objectivisation of human beings into individual subjects that could be measured, known and understood in relation to others is, he argues, a recent phenomenon coming at the time of the Industrial Revolution when bodies needed to be known, understood and put to use.10 His later work on governmentality examines the ‘conduct of conduct’ of individualised subjects, showing how in conditions of late liberalism, the individual is made responsible for his or herself, taught to exercise their freedom in ways that serve the interests of the flourishing of the population more broadly.11 

Mitchell Dean gives the example of healthy eating to show how diffuse this form of power is and how it is decentred from the institutions of the state. Children may be taught at a young age to take personal responsibility for their diet and although advice may be dispensed by government (such as in ‘five a day’ campaigns), the habits and sense of responsibility that undergird the daily practice of eating to promote good health are taught by schools and, above all, families. Any risk from eating unhealthily is understood to be borne primarily by the individual concerned, who must regulate her or his habits, understand the consequences of that third doughnut and behave accordingly.12 Freely-choosing individuals, such as healthy eaters or good citizens who vote responsibly, are not born but rather are produced through processes of socialisation and reproduced and moulded through (often conflicting) power relations that direct, encourage and disincentivise free choices in particular ways.

Miller and Rose are similarly interested in the ways particular sorts of human subjects are produced. Their starting point is that there is no such thing as a human essence, which various forms of power attempt to crush or turn to their own ends. Rather, it is power that produces any kind of subjectivity, in modest, everyday practices that tend to elude the attention of political scientists. Thus, ‘[o]ne could examine, not subjectivities, but technologies of subjectivity’.13 What I want to suggest is that voting  is a technology of subjectivity, one which enables us to understand that ‘our own idea of the human subject as individuated, choosing, with capacities of self-reflection and a striving for autonomy, is a result of practices of subjectification’.14 As we have seen, voting as a practice works on the assumption that individual voters will enter a polling station as ‘choosing’ subjects who have made use of their ‘capacities for self-reflection’. However, the desire to be a choosing, autonomous, self-reflective subject who will make an informed individual decision is not something that is already fully-formed in human beings at birth, but rather is a historically specific, learned attribute of modern citizens conditioned by their participation in, and understanding of, the activity of voting the first place. Thus voting is learned at school and home through citizenship programmes, mock elections, student participation in school councils and the election of prefects and through parental guidance, children’s television, even a bit of illicit joining in at polling stations.

So, if voting is part of a larger set of practices involved in the ‘conduct of conduct’ in which individuals are produced as individuals and in which their free participation is required in order that decisions can be made and the flourishing of the community promoted, what sorts of individuals do elections produce?

Kind making and looping effects

The role of exams, metrics, reviews and other numerical ways of assessing, judging and managing human beings and their actions have been the subject of many studies aiming to understand how counting, classifying and measuring people changes them and produces particular forms of behaviour and subjectivity.15 Nevertheless, the practice of voting, despite being one of the most politically significant practices of quantification, has not come under much scrutiny as one of the ways in which individuals, their preferences and identities are produced. In other words, whilst the epistemological function of voting – the way in which it enables us to know populations and their preferences – is widely taken for granted, its ontological function – the way in which it makes worlds and the people who inhabit them – is neglected.

To understand this, it is useful to consider Ian Hacking’s discussion of ‘kind-making’. He argues that the ways in which people interact with the types of categories to which they are assigned is important and worthy of study. Classifications ‘when known by people [...], and put to work in institutions, change the ways that people feel about themselves’.16  He also points to a ‘looping effect’ within practices of classification.17 The way this works is that people may reflect on the ways in which they have been classified and this may change them and the behaviours that produced the classification. Their reflections, the action they take as a consequence, perhaps the activism they engage in to change perceptions of people like them, who have been classified in this way, may then change the way the classification itself is defined, understood, used and so on. Hacking is most interested in forms of classification that tend to be stigmatising or the focus of social or medical intervention, such as homosexuality or being the victim of abuse, for example. However, this way of thinking about classification and the ways in which people interact with it, is nevertheless useful for thinking about forms of classification that emerge from electoral practices.

To explain what this means in terms of Brexit, Clarke et al  point to the rise of UKIP, with its slowly increasing vote share from its formation in 1993 onwards, as a central harbinger of the Brexit vote.18 UKIP have never performed especially well in parliamentary elections under the First Past the Post system and even at their highest point won only 4 million votes in the 2015 General Election, a fraction of the eventual.17 1 million Leave votes cast the following year. In the previous year’s European Parliament elections, they won approximately the same number of votes, though with a relatively low overall turnout and under the proportional representation electoral system this translated not only into first place in terms of votes cast but also twenty-four of the UK’s seventy-three seats.19  Whilst clearly this shows that it matters how  you count things, it is worth noting that it also matters that  you count them. What emerges from the long process of UKIP’s rise is that it becomes increasingly apparent to a broader section of the population that a UKIP voter is a kind of voter it is possible to be.

The EU referendum similarly creates a type of person understood to be a ‘Leaver’ or ‘Remainer’. Our opinions on the EU might not have mattered to us much before the referendum and the huge amount of media attention paid to its perhaps surprising and certainly dramatic result. Nevertheless, now we find ourselves labelled as Leavers or Remainers, we may give some thought to what that label means, and debate about the precise meaning of these terms has ensued. Concomitantly, a ‘Remainer’ or ‘Leaver’ becomes ‘a ‘kind’ of human being about which specialised knowledge [is] possible’, to quote Hacking again.20 A whole apparatus of knowledge production – itself generally reliant on quantification – seeks to elucidate the question, requiring or at least encouraging people to continue reflecting on the attributes, beliefs or behaviours that they hold that make them identify with one ‘side’ or another.

The purpose of counting voters in an election is clearly not to understand each individual conscience in detail. As has been widely noted, there were many reasons why individuals voted in particular ways in the referendum, but what is useful and, therefore of interest to policy-makers and politicians, is the tally. A majority voted to leave the EU – and therefore that is what politicians have a mandate to enact. It is becoming increasingly obvious that this leads to further problems of knowledge, though. If the usefulness of an election is to allow governments to understand how to proceed, then the open-ended question of what Brexit might mean, as well as the uncertain ways in which voters give meaning to their own votes, raises questions. It becomes important to find out exactly what a ‘leave’ voter believes and what ‘leave’ voting means. How can this be assessed? For Clarke et al  the answer is more polling to obtain yet more quantifiable information, which in turn gives rise to more counting. However, any attempt to render this complex reality into the simplicity of comparable numbers has its own effects.

As ethnographers of quantification and classification have shown about other forms of counting, Coleman explains, counting votes simplifies a hugely complex set of contexts, histories and meanings. He shows in detail, on the basis of sixty interviews with voters, the excitement, cynicism, confusion, shame, hope, fear and so on that citizens may bring with them to the polling station. Voters, he says, know what they stand for but do not quite know how to link where they stand with what is promised by politicians and their campaigns.21 However, this is not to say that voters ‘know what they stand for’ or have a set of fixed preferences that can be polled separately from the act of voting itself . It is through voting that they come to a sense of what they stand for, even as the campaign promises and policy prescriptions remain mysterious and elusive to them: as Coleman rightly points out, drawing on the work of Ernesto Laclau, being counted is always in some sense constitutive of the thing that is counted, producing rather than revealing what voters stand for, including the relevant divisions between identity groups.

Thus the kind of scrutiny involved in further counting intensifies the ways in which the classification and the individual interact. The Clarke et al study starts out from a particular set of attitudes and attributes – such as levels of education or attitudes towards immigration – and uses polling to judge whether these factors predict voters’ likelihood of having voted to leave the EU. However, they do not envisage that self-identification as a Leave voter would tend to interact with the sorts of attitudes that tend to be ascribed to Leave voters. We nevertheless need to consider that ‘the way new knowledge about [the Leave voter] becomes known to the people classified, changes the way these individuals behave, and loops back to force changes in the classification and knowledge about them’.22 Ongoing contestation about what it means to be a Leave voter and in favour of Brexit is not only reflected in polls and surveys but also produced through evolving understandings of the meaning of a Leave vote. Thus what it means to be a Leave voter becomes the sort of questions that can only be answered by rich, qualitative studies that privilege interpretation over measurement with careful empirical attention to the kinds of looping effects that may be taking place.


The final point worth making here is that there is danger inherent both in technologies of voting and polling, and in the ways we interpret their results. This is because there is an incipient violence inherent in the division of a whole country into just two different types of people. Nigel Farage understood this point when he remarked that ‘violence is the next step’ if, by voting, the electorate were not able to curb immigration, but he did not acknowledge the way in which such a statement itself may contribute to divisive looping effects in which the self-understandings of Leave voters are formed and reproduced.

Farage was later widely criticised for claiming that the Leave campaign had won ‘without a single bullet being fired’ just days after Labour MP, Jo Cox, was murdered by gunshot. Whilst there is no particular evidence that Thomas Mair, her murderer, was motivated by the part she played in the Remain campaign, he appears to have been a far-right, racist extremist. The murder was therefore understood by some commentators to have been incited by the anti-immigration rhetoric of the Leave campaign and by Farage’s rhetoric. The ‘sharp increase’ in racially motivated hate crimes in the period immediately after the referendum also underlines the violence of electoral production of difference and division.23

The claim that elections can and do provoke violence may seem counter-intuitive, as we are so frequently told that democracy is the opposite of, even the remedy for, violence – a process for the peaceful resolution of differences. Some objections to this account of the peaceful nature of democracy, though, have been well remarked on: insofar as ‘democracy’ or ideas about the people ‘having spoken’ are used as ways of marginalising the judiciary, dampening further debate, shutting down opposition or forcing through doubtful agendas, it is quite clear that ideas about democracy are being used in ways that run counter to broader notions about what   democracy is supposed to do and be for. A proliferation of statements that forcefully invoke ‘democracy’ as the ultimate guarantor of legitimacy themselves contain undercurrents of violence. However, what is less well remarked and what I want to draw attention to is the fact that an account of electoral democracy as a peaceful way of resolving difference is also partial insofar as it does not also attend to the ways in which voting produces division in ways that are all or nothing, for or against, yes or no, leave or remain. It is little wonder that some people now have a sense of a country in which we suddenly, seemingly out of nowhere, all hate each other.24

What is also worth noticing is that the divisions electoral democracy creates are individualised in the sense that an act of voting requires a private act of self-scrutiny and decision, the casting of a vote that is yours and yours alone. This very logic militates against careful consideration of the thick tissue of relationships and compromises in which we are all, always, deeply imbricated and asks us to put those considerations to one side in order to ask what we personally believe to be right, abstracted from those relationships. The privacy of the ballot box similarly seems to absolve us from accountability to those others in our communities and our lives who may be affected by what we do. This is by no means to say that everyone votes selfishly: we might draw on examples of elderly people who refrained from voting altogether on the basis that they shouldn’t have a say in a decision that would in all probability not affect them because of its long-term effects, and others who voted in particular ways because they cared what happened not only to themselves and their children, but also to the others whom they live around and amongst. Nevertheless, the individualising and privatising nature of voting in itself may very well curtail the peaceful resolution of difference as much as promote it, by taking the decision-making process away from those spheres of life where compromise and accommodation are ordinarily demanded of us and made.

The UCL Constitution Unit’s work on asking citizens to deliberate together in ‘citizens’ assemblies’ on what Brexit should look like may well be an instructive exercise in what difference it might make if we were to fully involve others in our decision-making.25 Indeed, the literature on deliberative democracy seems to demonstrate that asking people to deliberate and make decisions together makes a big difference to possible outcomes.26 However, as many writers have pointed out, it would be naive to assume that there is no undercurrent of violence in deliberative settings, no matter how much effort is put into creating ‘ideal speech situations’.27

The ability to make the sort of argument that counts as good or persuasive is stratified by class, race, and gender, and it is important not to be complacent about the pain, the violence, in being shouted down, ignored or otherwise marginalised in even the safest of safe spaces. Deliberative spaces, too, unavoidably produce difference and division. Therefore a continued ethical focus on how others and divisions are produced through democratic processes, as well as an ongoing commitment to treating disagreement as fundamental to the human experience and an attentiveness to treating others as adversaries, not enemies, is fundamental to any democratic process that can really be called peaceful.28 What’s more, some emphasis needs to be taken away from formal institutions, in order to consider how practices of self-formation and the ‘conduct of conduct’ can enable more peaceful and democratic conversation in everyday social locations such as workplaces, streets, and homes. As Romand Coles rather beautifully puts it: ‘democracy […] hinges on not being in complete possession of itself: it must not know too well what it is, but rather remain significantly a question to itself so that it can be opened to others and a future that is neither yet known nor yet realised in being’.29  By focusing on the promise that democracy makes to value lived experience without closing down the possibilities of all it might mean, we could begin to think beyond the dreary divisiveness of yet more voting and polling as the answer to questions raised by voting and polling.

Of course, there is no wishing away the social power relations that condition any form of decision-making, but paying attention to the ways in which democratic processes produce violent division might make us reflect on the kind of democracy, and democratic engagement, we want. This attention might further encourage us to be careful of how we interpret those differences: the unrelenting militarism of the language in books with titles like All Out War30  and in Clarke et al ’s emphasis on ‘battle’, ‘rival armies’ and so on31  is not helpful if we wish to avoid looping effects of violent division. However, I want to suggest that understanding the productive effects of everyday practices like voting, and paying careful attention to the consequences of quantification, may have virtuous looping effects of their own, enabling us to ask questions of the categorisations we are caught up in in ways that are kinder, more curious, less divisive. In so doing, we may transform our understanding of how the henge was formed, once the tide allows us to see it.

Cathy Elliott  is Senior Teaching Fellow in Qualitative Research Methods and International Development at UCL.

You can listen to a podcast Cathy made with her students on Jewish student life by searching for ‘The JewCL podcast’ on your podcatcher of choice or stream it at:


1. A. Perkins, ‘The Referendum Campaigns have Revealed a Fractured Country’, Guardian, 23.6.16,

2. Just a few examples: J. Ebner and J. Anderson, ‘I’m an Austrian in the UK’, Guardian, 24.6.16; J. Tapsfield and T. Sculthorpe, ‘Brexit splits Britain: HUGE vote for Leave in Sunderland Exposes Scale of Rift in Country’, MailOnline, 24.6.16; J. Samuel, ‘The EU Referendum Reveals a Nation Utterly Divided’, Telegraph, 24.6.16. All articles sourced through the Nexis database.

3. M. Easton, ‘A Less than United Kingdom’, BBC website, 24.6.16,

4. H.D. Clarke, M. Goodwin and P. Whiteley, Brexit: Why Britain Voted to Leave the EU (Kindle Edition) Cambridge, 2017, loc. 230.

5. Easton ‘A Less than United Kingdom’.

6. S. Engle Merry, The Seductions of Quantification, Chicago, 2016, p1.

7. S. Coleman, How Voters Feel, Cambridge, 2013, pp29-30.

8. David Cameron’s speech at the London headquarters of Bloomberg on 23 January 2013 quoted in T. Shipman, All Out War: The Full Story of Brexit, London, 2016, p5.

9. Coleman, How Voters Feel, p16.

10. M. Foucault, Discipline and Punish: The Birth of the Prison, trans. A.Sheridan. London, 1991.

11. Foucault’s ideas on governmentality are elaborated in his 1977–8 and 1978–9 lectures at the Collège de France, published in English as M. Foucault, Security, Territory, Population, trans. Graham Burchell. London, 2007; and M.Foucault, The Birth of Biopolitics, trans. Graham Burchell, London, 2008.

12. M. Dean, Governmentality: Power and Rule in Modern Society, 2nd Edn, London, 2010, p20.

13. P. Miller and N. Rose, Governing the Present: Administering Economic, Social and Personal Life, Cambridge, 2008, p7.

14. Ibid., p8.

15. For example, Engle Merry, Seductions of Quantification; the essays in V. Adams, ed., Metrics: What Counts in Global Health, Durham, N.C., 2016; P. Andreas and K.M. Greenhill, eds., Sex, Drugs and Body Counts: The Politics of Numbers in Global Crime and Conflict, Ithaca, 2010; G.C. Bowker and S. Leigh Star, Sorting Things Out: Classification and Its Consequences, Cambridge, M.A., 2000.

16. I. Hacking, The Social Construction of What?, Cambridge, M.A., 2009, p104.

17. Ibid., p105.

18. Clarke et al, Brexit, see particularly Chapter 6.

19. BBC, Vote 2014, See also the account of UKIP’s electoral fortunes in Clarke et al, Brexit, Chapter 6.

20. Hacking, The Social Construction of What?, p135.

21. Coleman, How Voters Feel.

22. Hacking, The Social Construction of What?, p105.

23. H. Corcoran and K. Smith, Hate Crime, England and Wales, 2015-16, London, 2016, p1.

24. This formulation comes from a joke that I believe I heard on BBC Radio 4’s News Quiz when it was on in the background of my day: ‘You know that referendum we had that made us all hate each other? The Lib Dems want us to have another one!’ Sadly, although it stuck in my mind, I have been unable to re-locate the quotation in order to acknowledge it properly.

25. The Constitution Unit, ‘Citizens’ Assembly on Brexit’, 2017,

26. R. Goodin and S. Niemeyer, ‘When Does Deliberation Begin? Internal Reflection Versus Public Discussion in Deliberative Democracy’, Political Studies, 51, 2003.

27. For excellent discussions of these issues see, for example, B. Flyvberg, ‘Habermas and Foucault? Thinkers for Civil Society?’ British Journal of Sociology, 49, 1998; C. Mouffe, ‘Deliberative Democracy or Agonistic Pluralism?’ Social Research, 66, 1999; A. Norval, Aversive Democracy: Inheritance and Originality in the Democratic Tradition, Cambridge, 2007.

28. The notion of an agonistic democracy in which the other is always treated as an adversary and disagreement cannot be wished away is from Mouffe ‘Deliberative Democracy’. I find Norval’s development of Mouffe’s work in Aversive Democracy particularly inspiring.

29. R. Coles, ‘Democracy and the Radical Ordinary: Wolin and the Epical Emergency of Democratic Theory’, in S.Hauerwas and R. Coles, Christianity, Democracy and the Radical Ordinary: Conversations between a Radical Democrat and a Christian, Eugene, O.R., 2008. I have written at more length about radical ordinary possibilities for democracy, drawing on Coles’ work, in C. Elliott, Democracy Promotion as Foreign Policy: Temporal Othering in International Relations, London, 2016, pp170-5.

30. Shipman, All Out War.

31. For example, Clarke et al, Brexit, loc. 532, 590, 901, 2730.

Privacy policy