Skip to main content

Conventions, Constituent Assemblies, and Round Tables: Models, principles and elements of democratic constitution-making


The article presents the Round Table form, elsewhere post-sovereign multi-stage constitution making as an independent democratic type superior to the alternatives. It locates the form along with Convention and Constituent Assembly both in a comprehensive typology based on models of regime transformation, as well as historically. After making a set of normative arguments comparing the three forms, focusing on the issue legitimation, I make a case for the synthetic nature of the Round Table in relation to the two important democratic predecessors. Finally, I reluctantly admit the path-determined nature of the Round Table that strictly speaking seems relevant only 1) in the transitions from dictatorships, if 2) new forces do not have the power to accomplish revolutionary change. Nevertheless, I argue that the principles of the Round Table (inclusion, consensus, publicity, legality and veil of ignorance) are relevant to other paths, from the point of view of their legitimation. I further claim with reference to Iraq, Turkey and the European Union that elements of the Round Table can be adopted even under conditions of revolutionary change, as well as constitutional reform.

Corresponding author
Hide All

1 Below I will use the term Round Table or Round Table form to indicate a process or paradigm of constitution making that I elsewhere also call multi-stage post-sovereign. When speaking about the central institution of negotiation within this form, I refer to round table forum or round table institution. I owe this important clarification to Julian Arato.

2 I understand that in South Africa the term round table is sometimes avoided, perhaps because of unsuccessful and rather different but similarly named London fora concerning India in 1931, or because there were proposals for a final constitution to be made by such a body. That fact should not disguise the near identity of the type of negotiating body, involving regime and opposition forces, that worked out interim arrangements for democratic transitions in at least seven countries (Poland, Hungary, Czechoslovakia, GDR, Bulgaria, South Africa and Nepal, in that order). See: Elster J (ed), The Roundtable Talks and the Breakdown of Communism (Chicago University Press, Chicago, 1996); Bozóki A (ed), The Roundtable Talks of 1989. The Genesis of Hungarian Democracy (Central European University Press, Budapest, 2002).

3 For the impressive literature I relied on: Atkinson D and Friedman S (eds), The Small Miracle: South Africa’s Negotiated Settlement, (Raven Press, Johannesburg, 1994); Faure M and Lane JE (eds), South Africa: Designing New Political Institutions (Sage, London, 1996); Ebrahim H, The Soul of a Nation: Constitution-making in South Africa (OUP, Cape Town, 1998); and Andrews P and Ellman S (eds), The Post-Apartheid Constitutions: Perspectives on South Africa’s Basic Law (Witwatersrand University Press, Johannesburg, 2001).

4 Strictly speaking there were two such fora: Codesa and the Multi-Party Negotiating Forum or MPNF. See for example H Ebrahim (n 3).

5 This requires that we do not follow Kelsen in identifying legality and legitimacy, but those who differentiate between these domains. See: Kelsen H, General Theory of Law and State [trans. Wedberg A] (Harvard University Press, Cambridge, 1945); Weber M, Economy and Society [trans. Fischoff E et al. ] Roth G and Wittich C (eds), (Bedminster Press, New York, 1968); Schmitt C, Legality and Legitimacy [trans. Seitzer J] (Duke University Press, Durham, 2004). Recently: Janos Kis, ‘Between Reform and Revolution’, East European Politics and Society Spring 1998; Arato A, Civil Society, Constitution, and Legitimacy (Rowman and Littlefield, Lanham, 2000) Chapter 3 and Constitution Making under Occupation: The Politics of Imposed Revolution in Iraq (Columbia University Press, New York, 2009) Chapter 1.

6 Thus he can only have two major types of transitions, reform and revolution. Coups in my version need not involve breaks in legitimacy. However, revolutions contain coups, logically in my scheme, and generally empirically as well. This is the case not only because of legal breaks, but because after ruptures a new organized force must take power if the state is to survive. We have seen this from the Russian to the Islamic Revolution. Alas, the recent revolution in Egypt confirms the insight. There was not really a choice between revolution and coup, because the first entailed the second. That it was in the end a military coup against an originally military regime does not vitiate the fact that there was a break in legitimacy as well as legality. But because of the coup within the revolution, the new democratic legitimacy replacing the old national-plebiscitary one has not (yet) found democratic institutionalization. Learning from the Round Table model would have helped. See my short piece ‘Egypt’s Transformation: Revolution, Coup, Regime Change, or All of the Above?’ available at:

7 I also locate in this fourth cell, left open by Kis in his early analysis, a self-coup carried out by an organ of government. There is a world of difference of course between a presidential imposition like Louis Napoleon’s coup, and the convention form discussed here, and the relationship is merely formal. But: interestingly, in Ackerman’s account American constitutional moments subsequent to the founding have some elements common with such presidential self-coups. Presidential leadership in constitution making is indeed what he seems to have favored for East and Central European countries, if not stressing illegality. On this see the Future of the Liberal Revolution (Yale University Press, New Haven, 1992). Yeltsin’s self-coup has been a realization of this suggestion, though not in a way Ackerman probably intended.

8 Ackerman B, We the People; Transformations Vol. 2 (Harvard University Press, Cambridge, 1998).

9 These pervasive authoritarian attempts at reform do not belong in my typology. Regarding them, Przeworski A, Democracy and Market: Political and Economic Reforms in Eastern Europe and Latin America (CUP, New York, 1991) maintained that ‘liberalization’, a broader category, always fails. I consider the Mexican reforms under the PRI, focusing on electoral rule change, successful for a very long period. See Arato (n 5) Chapter 1.

10 Or it can succeed, as did the initial top-down reform in Spain, by turning into a multi-stage process with an interim constitution, the law of reform. Spain for this reason is a marginal case of the post-sovereign model, achievable because of the rare democratic commitment of some of the initial regime protagonists. Prime Minister Suarez proceeded to establish forums of negotiation later than customary in this model.

11 Such a weighing is not involved, whereas in many American states, the normal method of significant revision presupposes extra-ordinary institutions like a convention. When there is a choice, as in the U.S. Article V there is a need to weigh the risks and benefits, and in the actual case the risks of the convention formula were obviously always deemed to be greater.

12 It had a distinguished history before, from deficient parliament in 17th century England to extra-parliamentary assemblies in America, and then to various forms of constitution-making bodies, including normal parliaments, constituent assembly and eventually the doubly differentiated type. This is reconstructed by Wood G, The Creation of the American Republic, 1776–1787 (2nd edn, University of North Carolina Press, Chapel Hill, 1998; and Morgan Edmund, Inventing the People: The Rise of Popular Sovereignty in England and America (Norton, New York, 1998) as well as Ackerman in We the People (n 8).

13 Here the original American terminology is of course different than mine, with conventions being the constitution-making bodies in reforms like Virginia (drawing Jefferson’s objections), revolutions as in Pennsylvania, and conventions in my sense as in Massachusetts and New Hampshire. Even the idea that a convention only recommends a draft was not established; for example the 13 ratifying conventions of the Federal Constitution had ratificatory but not drafting power. The idea that conventions were a democratic instrument of the people themselves, well documented by Wood, was also violated by the Philadelphia Convention that was neither elected nor popularly generated.

14 See Palmer RRThe Age of the Democratic Revolution: A Political History of Europe and America, 1760–1800 (Princeton University Press, Princeton, 1959) and Gordon Wood (n 12).

15 The exceptions all involve legal ruptures, as in Iraq and Nepal. Even in Iraq a subsequent confrontation between top-down reform, led by the occupying power’s Coalition Provisional Authority (CPA), and radical democratic challenge led to the type.

16 Revolutions and revolutionary constitution-making has now returned during the Arab Spring and its aftermath. It is too early to say at the time of the final conclusion of this article (November 1, 2011) whether the constituent assemblies in three countries, Tunisia, Egypt and Libya will be influenced by the normative model of constitution making here, as it happened in Iraq, in an externally imposed revolution. The article’s prediction would be that unless there is such an influence, based on the principles stressed below, the result would be authoritarian or highly conflictual. In Tunisia currently it may be that the parliamentary arithmetic will favor consensual solution. The fact that the Islamic party there, as well as in Egypt resisted all prior, negotiated constraints on the constituent assembly is however a bad sign.

17 While this was said most clearly when substituting state conventions for state legislatures, Madison’s Federalist 40 makes a clear case also for the unauthorized propositions instituted by patriotic individuals at the Federal Convention.

18 Later, during the discussion, and the ratification debates, the analogous idea of a second national convention was unsuccessfully recommended by framers like George Mason and Edmund Randolph.

19 The alternative understandings continued till the civil war. They had their bases in different passages in the Federalist Papers e.g. 39 on the state sovereignty side, and 15 that attacks the idea of imperium in imperio. Much of the Federalist however follows a model of dual sovereignty, involving a distribution of the powers of sovereignty, without admitting that ultimate sovereignty cannot be so divided, the view of Hamilton at least. Later, while the main opinions by James Wilson and John Jay in the Supreme Court decision Chisholm v Georgia in 1793 supported the nationalist interpretation, the overriding 11th amendment did the opposite. Under the Marshall Court judges like Storey and Marshall himself reaffirmed the perspective of Chisholm, the counsels for the states of Virginia, Georgia and Maryland re-articulated the sovereignty of the plurality of states.

20 And occasionally such an idea of sovereignty was clearly rejected, as in Federalist 22 where Hamilton calls the accumulation ‘in a single body all the most important prerogatives of sovereignty’ an execrable tyranny. The implications of this view for the constituent process were not however thought through.

21 This is pointed out by both Carré de Malberg and Hauriou. Had these amendment rules been actually used, the cases would have fallen on the borderline of reform and convention in my terminology. The same could be true for the convention formula of the U.S. Article V. if it were ever used. Perhaps there is a lesson in this: established governments do not like to deliver the constitution-making process to bodies out of their control, and generally succeed in doing so whatever problems this may create.

22 See for example Fritz Christian, American Sovereigns: The People and America’s Constitutional Tradition Before the Civil War (CUP, New York, 2009).

23 The case of Columbia in 1991 is an interesting mixed type. As in Venezuela later, the constitution in force allowed only revisions by Congress. After a series of popular initiatives, and a presidential decree, the Supreme Court, as in Venezuela too, allowed the calling of a constituent assembly, a constituyente. This was at first doubly differentiated as a convention, operating side by side with Congress. But: 1. It moved to dissolve Congress prematurely, upon the enactment of the Constitution (with the agreement of the president); and 2. As in the case of some constituent assemblies, but no conventions, the product of the constituyente was not a recommendation; it was the constitution and involved no external ratification. See DT Fox et al ‘Lessons of the Colombian Constitutional Reform of 1991 in Miller L (ed) Framing the State in Times of Transition: Case Studies in Constitution Making (USIP Press, Washington D.C., 2010) The ability of the constituyente to achieve this unlimited status was grounded in a prior Supreme Court decision ‘While the Presidential Decree that called for the convocation of a Constituent Assembly attempted to limit the assembly’s discussions to certain topics [and explicitly stated: may not modify the period in office of those elected this year], the Colombian court expressed that the constituent power could only be subject to self-imposed limits and that its actions could not be revised by the constituted powers.’ Joel I. Colón-Ríos ‘Carl Schmitt and Constituent Power in Latin American Courts: The Cases of Venezuela and Colombia’ (2001) 18 Constellations 3, 365–88. In Venezuela, on the contrary, argues Ríos, the Supreme Court when granting a similar authority to a constituent assembly denied its sovereign or unlimited nature, but was powerless to prevent it subsequently. On this case see also AR Brewer-Carías’ essay ‘The 1999 Venezuelan Constitution-Making Process as an Instrument for Framing the Development of an Authoritarian Political Regime’ in Framing the State.

24 By binding also the minority of states that did not ratify, the makers of the Grundgesetz did not produce, even in the first instance a treaty, as arguably in the case of the Federal Convention. See Madison in Federalist 39.

25 Arato A., ‘Post-Sovereign Constitution-Making in Hungary: After Success, Partial Failure, and Now What?’ 26 South African Journal of Human Rights: Constitution-Making as Learning Process (2010) 1, 19–44.

26 See my Constitution Making under Occupation (n 5).

27 The largest-scale empirical study by Elkins, Ginsburg and Melton The Endurance of National Constitutions (CUP, Cambridge, 2009), as important and sophisticated as it is, does not make a distinction among the three forms discussed here.

28 Klein C, Théorie et pratique du pouvoir constituant (PUF, Paris, 1997).

29 On this see my third chapter, ‘Sistani v Bremer’ in Constitution Making under Occupation (n 5).

30 Federalist 63 speaks of ‘total exclusion of the people in their collective capacity’ but specifically in relationship to government. This Ackerman takes as a reason to reassert embodied popular sovereignty for the constituent process. I believe the argument, logically, applies to both constituent and constituted powers. Obviously there are passages in the Federalist that support either of these interpretations. On the one hand there are ringing declarations that the constitution is the work of the people; on the other hand this is said to be impossible.

31 Namely: Sieyès, Condorcet, and Lafayette criticizing Sieyès, as well as Laboulaye, Carré de Malberg, but omitted by Schmitt and those who follow him; and by Ackerman in Future of the Liberal Revolution (n 7) where he pronounces the two forms more or less equivalent.

32 On this see de Malberg CarréContribution à Théorie générale de l’État II (1922) reprint (Dalloz, Paris, 2004).

33 Arguably the presence of the figure of legislator in original constitution making could be such a difference in Rousseau as long as we assume, without textual support, that when the sovereign dissolves and remakes a constitution there will no longer be need for the authority and expert advice of such an external law-giver. That figure also anticipates the American notion of 1787 that the actual drafter only recommends or proposes. See Madison in the Federalist especially # 41 and John Marshall in McCulloch v Maryland, 4 Wheaton 316 (1819).

34 Die Diktatur (Duncker and Humblot, Berlin, 1921) and less frequently but still emphatically in Verfassungslehre (Duncker and Humblot, Berlin, 1926).

35 Böckenförde, ‘Verfassunggebende Gewalt des Volkes – Ein Grenzbegriff des Verfassungsrechts’ in Staat, Verfassung, Demokratie (Suhrkamp, Frankfurt, 1992).

36 The degree of radical break likely when using the Round Table form is however likely to be greater than in the case of the convention that relies on the legitimacy of inherited institutions that continue to function through the constitution-making process.

37 Under a liberal democratic regime in place, the maximum level of rupture would not be a normative desideratum. Here the convention would be a superior vehicle for democratic change. The Round Table, as I will show below is difficult to justify where free elections are possible. This path-determined form emerges only under authoritarian regimes without free, competitive elections.

38 Admittedly this is so from my normative point of view. From another point of view, the transition from a secular to a religious dictatorship (as in Iran) can be deemed equally fundamental.

39 It may seem that there is no point in comparing these two forms, since they apparently belong to different paths, the convention to already existing electoral regimes, while the Round Table to authoritarian ones. As the Russian case shows however, a dictatorship can be transformed in several steps with something like an elected parliament being put in place through reforms, keeping the matter of constitutional change open. While the Soviet Union or its republics could have had round tables initially, followed by a democratic second stage, in Russia, initial reform was followed by an unstable convention model that led to dual power and legal rupture. In any case the comparison here does not mean that a choice between the forms is generally available.

40 A case could be made, that with the help of the North Carolina ratification process, and the making of the Bill of Rights (that was the most important demand at several ratification conventions) by the first Congress, the idea triumphed at least in part.

41 In the U.S. it was also five: the convention, Congress, the electors, the ratifying conventions, and, I would add, the new Congress that produced the Bill of Rights. The last step was not formally part of the process, but significant, obviously, as well as the result of ratification documents in several states.

42 Actually two, see note above. But this was formalized in the Round Table model.

43 See Arendt H, On Revolution (rev. 2nd edn, Viking, New York, 1965).

44 Arato A and Cohen JBanishing the Sovereign? Internal and External Sovereignty in Arendt’ (2009) 16 Constellations 2, 307–30.

45 Lefort Claude, ‘Permanence du theologico-politique’ Essais sur le politique (du Seuil, Paris, 1986).

46 Condorcet ‘Discours sur les conventions nationales’ (1791) 213–14.

47 From Massachusetts and New Hampshire in the revolutionary period to many individual states later on.

48 From the Federal Constitution to the Grundgesetz and potentially even the convention for the Future of Europe. It is another matter that the creation of such federation alters, as Schmitt noted, but only in the sense of limited reform, the constitutions of the unit states. This can happen implicitly, or through formal amendment.

49 Elsewhere I have argued that because of its two stages, the method has also the virtue of being able to separate state formation from constitution making proper that will no longer keep open the structure of the state. This possibility, realized in South Africa where it was enforced by the Constitutional Court, answers Schmitt’s objection that the U.S. Federal Convention conflated the two processes. See my ‘Conclusion’ to Constitution Making under Occupation (n 5).

50 There is also learning, beyond the makers of the German Grundgesetz, and the German failure of 1990, in clearly requiring and delineating the conditions under which the final constitution is to be made, without leaving loopholes for the interim becoming permanent as also in Hungary.

51 Interestingly, referendum is used in the most pathological case, that of Iraq, where the constitutional assembly was indeed marginalized, not even allowed to take a final vote.

52 We see the second option clearly operating in Egypt currently, where the military high command simply imposed the initial rules.

53 And was argued in France, by the Gaullists after 1946.

54 Gouverneur Morris at the Convention July 23, 1787 (James Madison Notes of Debates in the Federal Convention of 1787; 351; Farrand Max (ed) The Records of the Federal Convention of 1787 (Yale University Press, New Haven, 1911) vol 2, 92.

55 Elster for example did not notice the decisive difference, already clear in Hungary in 1989 as well as the contemporary failed case of the GDR, that the round table forum does or can make a constitution, an interim one if serious legitimation problems are to be avoided. See his ‘Introduction’ to The Roundtable Talks and the Breakdown of Communism (University of Chicago Press, Chicago, 1996).

56 It was this very tendency that facilitated the conclusion of the process by the Fidesz government in 2011 in a pathological form of an illegitimate but legal reform.

57 I cannot review here the distorted process by which a new constitution, called a basic or fundamental law came into being in Hungary in 2011. If the results of this process stand, Hungary will turn into a pathological case of the Round Table model, rather than an incomplete case I analyzed in ‘Post Sovereign Constitution Making in Hungary’ (n 25). Here I only note that the Hungarian Constitutional Court accepted to review a petition concerning the constitutionality of amendments prior to the making of the Basic Law that anticipated some of its contents, and turned down the petition on the bases of its previous denial of substantive amendment review. For the first time, the Court asserted its right to invalidate procedurally incorrect amendments. [AB 61/2011. (VII. 13.)] The Basic Law of 2012 (as its predecessor, the constitution of 1989–1990) was enacted on the basis of the amendment rule of the current constitution, but I believe did involve other procedural errors. See Arato, Halmai and Kis (eds) submission to the Venice Commission. (‘Opinion on the Fundamental Law of Hungary’ published in Hungarian in 1 Fundamentum (2011)). We have decided not to submit this petition to an already dramatically packed (first two, then five more judges in addition to one replacement) Hungarian Constitutional Court. The constitutional case here referred to is analyzed in terms of its comparative background by G Halmai in ‘Constitutional Violation of the Constitution’ forthcoming (in Hungarian) in Fundamentum.

58 This is similar to the major role of these courts in the separation of powers, in the context of parliamentary governments.

59 For example, currently in Libya they are not likely to do so. Collapse, produced by civil society and a plurality of political actors as in Nepal is another matter. In that case a plurality of actors, including conservative ones not directly tied in the old regime, can enter into a process of Round Table negotiation and compromise even after a legal break.

60 Austin’s GWorking a Democratic Constitution: The Indian Experience (OUP, New Delhi, 1999) is still incredibly impressive and balanced concerning this history

61 Kis (n 5).

62 Arato ACivil Society, Constitution, and Legitimacy (n 5) 230–1; Lindahl H ‘Constituent Power and Reflexive Identity: Towards an Ontology of Collective Selfhood’ in Loughlin M and Walker N (eds) The Paradox of Constitutionalism: Constituent Power and Constitutionalist Form (OUP, Oxford, 2007).

63 Such a learning can be achieved also through the examination of the principles involved in a very large number of empirical cases. It is interesting that the normative lessons drawn by Elkins, Ginsburg and Melton (n 27) converge to an extent with my analysis, even though their concern is more stability than democracy. The concept of legitimation I think links these two concerns.

64 Constitution Making under Occupation (n 5).

65 This can be also put in the reverse way, namely that it was only under the pressure of the Shi’ite movement led by the Grand Ayatollah Sistani that would-be American top-down constitution makers conceded the post-sovereign model. Thus, as I tried to document in Sistani v Bremer (n 29), the double origins of the Round Table model were also reproduced in Iraq’s rather different and so far, unique context.

66 See my ‘Democratic Constitution-making and Unfreezing the Turkish Process’ (2010) 36 Philosophy and Social Criticism 3/4.

67 Following Carl Schmitt Verfassungslehre, Chapter 29 (n 34) and M Forsyth Unions of States: The Theory and Practice of Confederation (Leicester University Press, Leicester, 1981), as well as the recent work of Jean Cohen, I conceive of ‘federation’ to indicate a type of polity other than either a federal state or a treaty organization. While it is hardly contested anymore that the EU falls between the two types, it is of course controversial whether it is a federation in Schmitt’s sense, having a constitution but without the state. Indeed there are important differences with the American federation before 1865, Schmitt’s prototype, raising the question whether there is a plurality of forms between treaty and state, as Forsythe maintained. Another equally convincing approach would be to treat Europe, currently, as a hybrid, as does Weiler JosephThe Constitution of Europe: Do the New Clothes have an Emperor? and other essays on European Integration (CUP, Cambridge, 1999), as a federal state on the constitutional, and treaty organization of what he later called the institutional level. I do not believe that alternative choices here would affect my argument.

68 See my ‘Europa und Verfassung’ in Brunkhorst H et al. (eds) Habermas-Handbuch (Metzler Verlag, Stuttgart, 2009).

Recommend this journal

Email your librarian or administrator to recommend adding this journal to your organisation's collection.

Global Constitutionalism
  • ISSN: 2045-3817
  • EISSN: 2045-3825
  • URL: /core/journals/global-constitutionalism
Please enter your name
Please enter a valid email address
Who would you like to send this to? *



Full text views

Total number of HTML views: 2
Total number of PDF views: 87 *
Loading metrics...

Abstract views

Total abstract views: 199 *
Loading metrics...

* Views captured on Cambridge Core between September 2016 - 17th January 2018. This data will be updated every 24 hours.