I. Introduction

Comparative law grapples with formidable methodological problems. These problems make it difficult to formulate richly detailed general theories that have broad explanatory power across the diverse array of legal systems that exist in the world. In the field of comparative federalism, the successful formulation of comparative theories is especially challenging due to the context-dependent nature of the negotiated compromises that ordinarily underlie the formation of a federal constitution. Because federal bargains are a function of the unique conditions of each country, comparison between federations is difficult and a general explanatory theory is elusive.

Helpful comparative studies of the formation and amendment of constitutions have been undertaken. However, the criteria used in such studies do little to illuminate the specific issues that arise in federal systems. This is partly because the criteria used in such analyses, although ostensibly generic to all constitutions, presuppose the unitary state to be the archetypical case and the federation to be nothing more than a particular kind of state organisation. This gives rise to a set of analytical criteria that are not well-suited to drawing out the special issues that arise when the formation and amendment of federal constitutions is being considered.

Thus, in most studies of constitutional change the subject matter is analysed in terms of a series of questions: How difficult is it to change a constitution? By what mechanisms can it occur? How frequently does it occur? Which political actors are able to secure it? Is the process democratic? How much of it occurs through judicial interpretation? Can it occur through informal or extra-legal means? In these studies, the particular features of each constitutional system, such as whether it is federal or unitary, presidential or parliamentary, written or unwritten, are recognised as relevant, but are not the focus of attention. Rather, what tends to be of special interest is whether, as consequence of various factors, change in a particular constitutional system is relatively frequent or infrequent, incremental or episodic, elitist or popular, formal or informal, legal or extra-legal, and so on. As a consequence, while many studies of constitutional change recognise that federalism poses unique issues, only a few grapple closely with the exact reasons why this is so. Rather, the terms in which constitutional change is analysed are usually generic to the study of constitutional change generally.

Why are federations different? While in a unitary state questions about constitutional change may focus, for example, on the evolution of the political relationship between the government and the people, in a federation there is also the question of the constitutional relationship between the federation and the states to be reckoned with. Instead of one government and one people to be considered, in a federation there are at least two sets of governments and two sets of people. As a consequence, as Olivier Beaud has argued, federations cannot adequately be understood by recourse to the state-centred concepts of modern public law; they need to be understood in their own terms. Constitutional change within federal systems can certainly be analysed in terms of concepts such as rigidity and flexibility, formality and informality, and so on, but these issues are generic to all constitutional orders. The specific problem that federalism raises concerns a different question, the question of the constitutive relationship between polity and polities.

In many studies of constitutional change this special question of the relationship between the federal polity and the constituent polities is analysed in terms derived from the study of unitary states. Thus, it is often observed, following the lead of James Bryce and Albert Venn Dicey, that federal constitutions are necessarily rigid, as opposed to flexible, in order to ensure that the distribution of powers between the federation and the states cannot be altered by either the federation or the states acting unilaterally. Recognising the significance of rigidity in federal constitutions is an important insight. Federations tend, indeed, to be more difficult to amend than the constitutions of unitary states. However, focussing only on this point distracts attention from the equally important question of precisely how and exactly why constitutional rigidity is secured in federations. For constitutional rigidity can be secured in many different ways, such as by requiring special majorities for the passage of constitutional amendments or by referring proposed constitutional changes to the people in a referendum. But these mechanisms, common to both federal and unitary constitutions alike, are not of themselves distinctly federal in character. They may be used in federations to contribute to the rigidity of the constitution, but the amendment clauses of federal constitutions typically prescribe special procedures that not only secure rigidity, but are themselves characteristically federal in structure and design. For example, if special majorities are required for constitutional amendment in federations, they usually involve special majorities of a bicameral federal legislature, the second chamber of which consists of representatives of the constituent states; or if a referendum is required, it is a referendum of both the people of the federation as a whole and the peoples of the constituent states. Constitutional rigidity is thus secured through characteristically federal procedures and mechanisms, the amendment clause of each federal constitution expressing a particular understanding of the exact nature of the constitutional relationship between the federation and its constituent states.

This article offers an account of constitutional amendment within federations that focusses on this underlying question of the relationship between the federation and its constituent states. Its starting point is the proposition that we cannot begin to understand constitutional change within federal systems without grappling with the fact that a federation is a polity composed of polities. The further premise of the article is that amendment clauses are closely related to, although not simply identical with, the constituent power underlying a constitution. It is argued that mechanisms of constitutional change within federal systems are best understood when related to the constitutive power upon which each federation is founded, and continues to operate. In federal systems, however, the location of constituent power is complicated by the existence of two sets of polities. Where in a federal system is constituent power located? Is it in the federation as a whole, the component states, or somehow both? This is both a theoretical and an empirical question. However, if close attention is given to the way in which constitutive power is effectively located and organised at the critical moment when a federal constitution is brought into being, it is possible to identify certain systemic consequences for the design of its formal amendment procedures. This, it is suggested, is because the exact way in which effective constitutive power is configured operates as a kind of presupposition in constitutional reasoning, as well as a locus of political power, which tends to have a systemic effect on the way in which constituted power is distributed within the resulting federal system. For, if a federation comes into being through an aggregation of previously independent states, there is reason to expect that many of the structural features of the federation will be in some important respects different from those of a federation that comes into being through a devolution of power within an originally unitary state. The way in which constituent power underlying the federation is configured is significantly different in these two cases, and given the close relationship between the power to make a constitution and the power to change a constitution, there is good reason to expect that the amendment procedure will reflect the aggregative or devolutionary origin of the constitution. By approaching the question of constitutional change in this way, the uniquely context-dependent nature of the bargains that underlie the formation of federal constitutions can be assimilated, it is argued, into a theory that offers a generalised explanation of the amendment of federal constitutions understood comparatively.

II. Constitutional Change

While it is recognised that federations pose unique questions for constitutional change, most of the discussion implicitly accepts, as its starting point, the proposition advanced more than a century ago by Albert Venn Dicey, that the most important function of a federal constitution is to set out the distribution of powers between the federal and state governments in a manner that cannot unilaterally be changed by either of those governments. According to Dicey, to function effectively in this way a federal constitution must possess three important features: it must be « written », it must be « rigid », and it must be « supreme ». Dicey derived this concept of constitutional rigidity from his colleague, James Bryce, who had influentially argued that the most important distinction to be drawn between constitutions generally is whether they are rigid or flexible. For Bryce, a constitution is rigid if the authority by which it can be amended is different from the authority by which ordinary laws are enacted. Like Dicey, Bryce thought that a federation requires the establishment of a rigid constitution, for otherwise the constituent states will not have any assurance that their rights will be preserved under the federation. Dicey added that the specific sense in which a federal constitution must be rigid is that, at the least, neither level of government must be able alter the fundamental terms of the federal arrangement unilaterally. Bryce and Dicey were aware that federations often use characteristically « federal » mechanisms for the amendment of the constitution in order to secure the necessary degree of rigidity, such as the involvement of the constituent states in the prescribed process of constitutional amendment. However, they did not consider such mechanisms to be essential, because there were examples of federations that relied on other mechanisms to secure rigidity, such as super-majorities or appeals to the people.

This mode of analysing federal constitutions has been very influential. Ronald Watts, for example, maintained that the existence of a supreme constitution is a necessary prerequisite for the effective operation of a federation, and that this required that the constitution of a federation must not be amendable unilaterally by either level of government, for some element of rigidity is necessary to safeguard the rights of the constituent states. Watts drew attention to the kinds of mechanisms typically used in federal constitutions to secure the requisite rigidity, including provisions that in some way require the involvement of both orders of government for the amendment of the constitution, especially in relation to its most important or fundamental features. However, Watts thought that the most important question was not the exact way in which particular federal constitutions provide for their own amendment, but rather the extent to which such constitutions are either rigid or flexible.

Kenneth Wheare similarly considered it was essential that if there was to be an amendment power in a federal constitution, that power could not be conferred exclusively on either the general or regional government, at least in relation to provisions of the constitution that regulate the respective powers of those governments. Wheare observed that it might be practically wise to involve the general and regional governments, or their peoples, in the amendment process, as was the case in the United States, Switzerland and Australia, but he denied that this was a necessary requirement of federalism as such. The implications of his approach were made clear in his remarks about Canada, where the legal power to amend the British North America Act 1867 at the time rested exclusively with the British Parliament (noting that this power was exercised in conformity with the Statute of Westminster 1931, which provided that the Parliament would not legislate for Canada without its request and consent). Wheare argued that the very fact that no express power was given to either the federation or the provinces to amend the constitution, but rather made them dependent on the British Parliament, took the logical requirements of federalism to their most « extreme conclusion ». The British Parliament, he observed, had amended the British North America Act several times, sometimes on the request of the federal and provincial governments, but often only on the request of the federal government. Wheare considered that this did not contravene the federal principle so long as the convention surrounding the amendment procedure did not develop to the point that the British Parliament was required to act as the « agent » of the Canadian federal government in doing so.

Although the question of constitutional rigidity is undoubtedly of profound importance in federal systems, focussing on this question alone has a tendency to distract attention from deeper reflection on the exact reasons why different methods of constitutional change are adopted by particular federal systems. In any such analysis, the specific question that federalism raises concerns the precise relationship between the federation and its constituent states. Studies that use generic categories (such as rigidity, formality, etc.) to analyse constitutional change fail to grapple with this latter question. The dual character of federations—as simultaneously consisting of a federalised polity composed of constituent polities—requires a mode and method of analysis different to that which is usually applied to the study of constitutional change in unitary states. A federation is not only one state, it is also many states, and this complicating two-fold character of federations is deeply embedded in the processes by which federations are formed, the institutions through which they operate, and the mechanisms by which they change and evolve. Federalism « interacts differently with the rest of the factors that influence constitutional change », and poses its own unique set of questions about how and why constitutional change can or should occur. For, the specific way that federal constitutions are made difficult to amend typically involves some formal recognition of the plurality of polities of which the federation is composed. We cannot begin to understand the dynamics of constitutional change in federations without grappling with the problem of this relationship between the federalised polity and its component polities.

III. Constituent Powers

Constitutions that claim to be democratic frequently appeal to « the people » or « the nation » as the constitutive power on which the constitution is founded. On the kind of analysis initiated by the Abbé Sieyès, a constitution derives its existence from an exercise of constituent power by the people of the nation. However, the constituent power of the people has to be institutionalised in order to give rise to a written constitution. It has to be exercised through some kind of constitution-making process, and the particular process that is adopted will unavoidably constrain and shape the constitution that emerges. In federal systems, moreover, there is a plurality of peoples that has to be reckoned with. In federations, therefore, the process through which the constituent power is exercised is constrained and shaped by the emergence of two sets of peoples: the people of each component state and the people of the federation as a whole.

Constituent power is distinct from the constituted powers contained within a constitution. Amendment clauses appear within the texts of constitutions and are therefore a kind of constituted power. However, amendment clauses offer a means by which constitutions can themselves be altered, which is virtually the exercise of a kind of constituent power. And yet, many amendment clauses prescribe procedures that are different from those used when the constitution was created. When this is the case, do they thereby displace the constituent power, or somehow co-exist with it? Merely attending to the written text of a constitution alone will not answer this question conclusively, because the constituent power is the presupposition of the constitution, and not something contained within it. And yet the text of the constitution is a product of the constituent power, and thus contains within itself suggestions as to its nature.

This is complicated in federal systems by the question of polity and polities. Here, the fundamental question is whether the constituent authority that lies behind the federation is best conceived in unitary or plural terms, or somehow both. The preambles of many federal constitutions testify to the importance of this question, often by using language that expresses the mixed character of the federation. Compare, for example, the preambles of three very different federations, namely India, the United States and Ethiopia, which respectively begin with the following words: « We, the People of India », « We the People of the United States » and « We, the Nations, Nationalities and Peoples of Ethiopia ». The use of the singular and the plural in these formulas is very deliberate. Each expresses a subtly different account of the constituent power on which the constitution was founded. It is true that the language does not specify the particular institutional processes that were used, and it is possible that is said in a preamble may not quite match reality. However, a more detailed analysis can be undertaken of the institutional procedures that were actually used, and conclusions can be drawn about the exact way in which constituent power was institutionalised in each case. The precise form in which this institutionalisation takes place is of utmost significance.

The amending clauses of federations have to grapple with essentially the same problem of polity and polities. To take India, the United States and Ethiopia again as examples, it is evident that the amendment clause in each country points to a particular configuration of power, more or less reminiscent of the way in which the constituent power was institutionalised in the framing of the constitution. All three constitutions in some way and to some extent involve the constituent states in processes of constitutional amendment, but the significance and degree of that involvement varies. In India, the main bulk of the constitution can be amended by two-thirds majorities of the two houses of the Parliament, and only particular articles of special importance require the approval of at least half of the state legislatures. In the United States, the entire constitution can be amended only when ratified by the legislatures or in conventions of three-fourths of the states, except that, in addition, no alteration to the suffrage of a state in the Senate can occur without that state’s consent. In Ethiopia, the constitution can be amended only when approved by a two-thirds majority in a joint sitting of the two houses of the federal legislature and by two-thirds of the Councils of the member states, except that, in addition, the rights and freedoms protected by the constitution cannot be altered except with the approval of all state councils and a two-thirds majority in each of the two houses of the federal legislature. Moreover, the Ethiopian Constitution affirms the right of every Nation, Nationality and People to secede from the federation. This is in stark contrast to the Indian Constitution, which grants the Union Parliament power to create new states by amalgamating or dividing the territories of existing states and to alter the territories of the states generally, and which affirms the right of the federal government to intervene in the affairs of the states in cases of emergency. The U.S. Constitution falls somewhere between: it is silent on secession, but the attempted secession of the southern states during the civil war was repulsed by the Union, and the Supreme Court subsequently held that the Constitution had established « an indestructible Union composed of indestructible States ». On the face of these provisions, the Indian constitution offers the least opportunity for state-input into constitutional amendments, the American offers substantially more opportunity, and the Ethiopian constitution offers the most. In India, it seems, the very identity and existence of the states is subject to the authority of the Union; while in the United States both the Union and the States are deemed indestructible; while in Ethiopia, the territorial bounds of the Union itself can be altered by unilateral withdrawal of a state. These differences correlate in general terms with the way in which the constituent power is described in the preambles of each of these constitutions.

This apparent correlation between constituent power and amendment clauses gives rise to a wider question. To what extent does the unity or plurality of the constituent power shape the amendment procedure adopted in a constitution? As the examples of India, the United States and Ethiopia suggest, federal constitutions can be the result of very different conceptions of constituent power. For one of the commonplaces in the literature is that federations can be formed, not only through aggregation of formerly separate and independent states, but also through devolution of powers of self-government to nascent political communities within a formerly unitary state. To what extent, it may further be asked, does the aggregative or devolutionary origin of a federation influence its amendment clauses? If each federal system is unique in the way that its constituent power is conceived and institutionalised, what influence does the exact location of constituent power have on the processes for constitutional alteration in each constitution? If the configuration of the constituent authority lying behind a federal system is not simply unitary, but also fundamentally plural, there is good reason to expect that this will shape not only the mechanisms by which a federal constitution was brought into being, but also the mechanisms by which it can be changed in the future.

IV. Formative Processes

Federal systems can be established through constitutive processes that are aggregative or devolutionary. A plurality of relatively independent states may agree to form themselves into an aggregative federation, or a unitary state may find it necessary or convenient to federalise itself by devolving independent powers of governance upon a plurality of component polities. However, the line between aggregation and devolution is not always clear: a recognisably federal system may come into being through processes intermediate between these two possibilities. In addition, a state external to the federation, such as an imperial power or an occupying military force, may coerce, guide or facilitate the formation of a political system that is federal in structure. As such, the formation of a federal system may be more or less shaped by factors that are either endogenous or exogenous to the federation. The exact location or locations of effective constitutive power in any particular federal system will therefore be a function of the extent to which the federation is exogenous or endogenous, as well as aggregative or devolutionary, in its political origins and constitutional presuppositions. This is significant. As Alfred Stepan has observed, federations formed through processes that are either aggregative, devolutionary or coercive display a « completely different historical and political logic ».

When a federation comes into being through processes that are predominantly aggregative, the constituent authority on which the constitution is founded is by its nature relatively plural: it is distributed, more or less equally, among the constituent states. In the most absolute or pure cases of aggregation the constituent states will be acknowledged to possess the kind of sovereign equality attributed to independent nation-states in contemporary international law. A union of such states will be predicated upon the exercise of this independent sovereignty expressed in the agreement of each state to its membership in the proposed federal arrangement. On the other hand, when a federation comes into being through predominantly devolutionary processes, sovereignty is concentrated within the original unitary state. In the most absolute cases of devolution, the decision to federate is taken by the relevant institutions of the parent state to devolve self-governing power to a plurality of nascent political communities within its territorial borders. The devolution of power is therefore made through processes prescribed by the constitution of the parent state and the initiative and cooperation of the governing institutions of the parent state is usually required.

However, pure or absolute cases of aggregation and devolution are rare. On closer examination, it becomes clear that the formative processes by which most federal systems have come into being lie somewhere between these two ideal types. Thus, on one hand, even the federal system of Switzerland, although built up by the successive addition of new cantons over the long course of its history, is not quite a case of pure aggregation, not least because the modernising constitution of 1848 was adopted by only a majority of the cantons, breaking with the rule of unanimity that had existed under the federative pact of 1815. Similarly, the formation of the United States Constitution broke with the rule of unanimity established by the Articles of Confederation by requiring ratification to occur in a minimum of nine of the thirteen states; and yet, the Constitution, like the Articles of Confederation, only bound those states that ratified it, and was therefore aggregative in that respect. Likewise, on the other hand, devolution does not normally occur unless nascent political communities within the parent state agitate for measures of local self-government and degrees of constitutional self-determination. Even in the United Kingdom, for example, where devolution was established and has become progressively deeper through the enactment of statutes by the British Parliament, all of this would not have occurred if there had not been effective agitation for it, in particular by Scottish nationalists, often with claims that, notwithstanding the doctrine of parliamentary sovereignty, political sovereignty was vested in the Scottish people. Similar observations can be made about other devolutionary federal systems, such as Belgium and Spain, where pressures from « below » led to concessions from « above ».

Whether a federal system is predominantly aggregative or devolutionary, if the constitutive procedures are democratic, the crucial decision to federate will involve appropriately representative or participatory decision-making processes within the constituent polity or polities. To the extent that legislatures, assemblies, conventions or referendums are involved, the decision-making principle within the relevant body will therefore usually be one of majority-rule. However, it is in the nature of aggregative federations that, while the decision to federate made within each constituent state may be made by majority vote, as regards the relationship between the constituent states, the foundational decision-making rule is the consent of every state that agrees to join the federation. Unanimity among the component states is thus a crucial indicator and measure of the aggregative character of a federal system. Constitutive decisions taken by less than the agreement of all of the states indicates, by comparison, a federative process that is relatively less aggregative; and the closer the constitutive decision-making principle is to majority-rule, the closer the system is to being devolutionary. However, so long as constitutive decisions are taken by each of the constituent states as such—even if by a simple majority of those states—the location of constitutive authority is not yet entirely unitary and the logic of the federative process is not yet fully devolutionary. This is because a unitary state, if democratic, makes decisions with the support of representatives of a majority of its citizens, not a majority of its states.

These propositions can be illustrated by further examples. The European Union, for instance, despite its pursuit of an « ever closer union among the peoples of Europe », continues to rest on a series of international treaties established strictly upon the international law principle of the sovereign equality of the member states. Article 1 of the Treaty on European Union thus declares: « [b]y this treaty, the high contracting parties establish among themselves a european union… ». By contrast, the Spanish Constitution, despite its far-reaching federalisation, explicitly derives its authority from « the Spanish nation », proclaims the « sovereignty » of the « Spanish people », and asserts the « indissoluble unity of the Spanish nation ». On the other hand, the Australian Constitution, although enacted into law by the British Parliament, was drafted by delegates of the constituent colonies and approved by referendums held in each of the constituent colonies. The Preamble to the Australian Constitution accordingly recites the « agree[ment] » of « the people » of each of the constituent colonies to be united in an « indissoluble federal commonwealth under Crown » and « under the Constitution ». And, to take yet another example, despite the supervisory role of the Allied forces, the German Basic Law was promulgated by a Parliamentary Council consisting of representatives of the participating German Länder and came into force only after it was ratified by their respective parliaments. The Basic Law declares that « [a]ll state authority is derived from the people » and that the « Federal Republic of Germany is a democratic and social federal state ».

In these ways, the texts of federal constitutions, especially their preambles, articulate prevailing conceptions of the constitutive authority on which the federation is founded, even if the language of a preamble may sometimes conceal as much as it discloses about the formative processes that led to the adoption of the constitution. The United States Constitution, with its language of « We the People of the United States », embodies something of this equivocation in its famous reference to « the People » in the singular but the « United States » in the plural.

V. Federative Transformations

There is reason to expect that the institutional decision-making processes through which a federal constitution comes into being will have an impact on the processes adopted for its amendment. This is in part because prevailing normative perspectives that exist at the founding not only shape conceptions of the constituent authority lying at the foundation of the constitution but are also likely to have a profound influence on a constitution’s institutional design. It is also due in part to the fact that those who hold institutional power at the founding are likely to use that power to ensure that they will continue to exercise power under the new dispensation. A kind of path-dependency thus often seems to be at work, operating at both an institutional and normative level. This does not mean that constitutions cannot be transformative, but transformative constitutions are only likely to come into being if a cultural or political transformation has already occurred, or is in the process of occurring, and is somehow embedded into the constitutive process itself. In this way, constitutions can certainly give effect to transitions from autocratic or military rule to democracy, for example, as occurred in South Africa. But the formative conditions have to be right.

The formation of a federal constitution involves a particular kind of transformation: either the aggregation of a plurality of relatively independent states into a more unified constitutional relationship, or the disaggregation of a unitary state into a relatively more plural constitutional arrangement. Because federative processes involve shifts in the balance between unity and plurality of polity and polities, the most important measure of that shift concerns the relative balance of powers of the constituent polities and the federal polity. This balance of powers is multi-layered, involving the scope and operation of the ordinary legislative, executive and judicial powers of the two sets of governing institutions, as well as the constitutional provisions regulating the power to alter or revise the constitution. Accordingly, the array of powers that are conferred upon or retained by a federal polity and its constituent polities varies significantly from federation to federation, as does the design and composition of the executive, legislative and judicial institutions in which these powers are vested. However, there is often a structural relationship between these two aspects of « self-rule » and « shared rule ». The balance of power between polity and polities is affected by a trade-off between the distribution of powers between the federal polity and the constituent polities and the extent to which the constituent polities have an effective voice in the decisions of the federal polity. In Germany, for example, the balance of power between the Federal Republic and the constituent Länder is shaped by the competences conferred respectively upon the federal and Länder parliaments and the power exercised by the Länder governments in the federal Bundesrat, so that the negotiation of a shift in one of these parameters must often be met by a corresponding shift in the other.

A similar balance of power exists in the relationship between constituent power and amendment clauses in a federal constitution, except that what is at stake is not the enactment of ordinary legislation, but changes to the constitution itself. As such, amendment clauses in federal constitutions are an especially informative indication of the transitional character of a federative process. The more closely an amendment clause reproduces the constituent authority on which the constitution is founded, the less significant is the transformation, whereas the more an amendment clause departs from the formative process, the greater the transformation. For example, so long as changes to the founding treaties of the European Union must be made unanimously by the member-states, there remains a real sense in which the member states continue to be « masters of the treaties », whereas the introduction of an amendment procedure that does not require unanimous member-state consent involves a significant change in the configuration of constitutive power. Likewise, in America, one of the most important differences between the Articles of Confederation (1781) and the United States Constitution (1788) was that under the former the constituent states had insisted that changes to the terms of the confederation would need to be approved unanimously, whereas under the latter changes to the terms of union could be approved by a majority, albeit a special majority, of the state legislatures or conventions held in each of the states. A similar shift from unanimity to qualified majority rule occurred in Switzerland when the Federal Pact (1815) was replaced by the Federal Constitution (1848), and in Australia when the Federal Council of Australia Act (1885) was replaced by the Commonwealth of Australia Constitution Act (1900).

As such, the transformative pattern that characterises the formation of an aggregative federal system is a movement from plurality to a degree of unity expressed in a shift from unanimity among the constituent states towards a form of majority-rule within the federal-state as a whole, but in a manner that preserves a self-governing and participatory role for the constituent states in some form or other. On the other hand, the transformative pattern that characterises a devolutionary federal system is a movement from unity to a degree of plurality expressed in a shift from majority-rule within the original unitary-state towards a degree of unanimity of decision-making within the federalised state, and yet in a manner that preserves the governing authority of the institutions of the originally unitary state. For example, the system of devolution in the United Kingdom has involved not only the establishment of self-governing legislative and executive institutions in Scotland, Wales and Northern Ireland, but also the development of the constitutional convention that changes to the parliamentary statutes establishing those institutions shall not be altered by the British Parliament without the consent of the devolved governments and legislatures. As a consequence, although the Parliament retains the power, as a matter of law, to amend or repeal the statutes of devolution as it thinks fit, it is constrained in practice by the recognised convention that it will not do so without the consent of the devolved community. Something similarly remarkable has occurred in Spain, where the Autonomous Communities first secured their rights to self-government by exercising newly-conferred quasi-constitutive powers to draft their own statutes of autonomy for approval by the central government, and have exercised an ongoing capacity under the constitution to secure additional measures of autonomy through amendments to their statutes of autonomy in a similar manner. As a consequence, although formal amendments to the national constitution still need to be approved by special majorities of the Spanish Parliament and, in many cases, a national referendum, substantial federalisation of the system has occurred upon the initiative of the Autonomous Communities with the consent of the central government.

While the common element in the formation of all federal constitutions is a federative transformation involving shifts in the constitutional configuration of power between the federalised polity and its component polities, the matter is complicated in cases where a federation is formed out of colonies that are, or once were, subjected to the overriding authority of an imperial power. In these cases, careful attention needs to be given to two sets of relationships: those between the states and the federation, and those between the states, the federation and the imperial power. If a federation of a group of former colonies comes into being after the colonies secure their independence from the imperial power, then the constituent power effectively underlying the federal constitution will be local to the federation and its constituent states. The American and Indian federations are examples. However, if a group of colonies are federated while still lawfully subject to the authority of an imperial power, then the constituent power will, as a matter of strict law, be located at the imperial rather than colonial level. In such cases, a genuinely functioning federation can be formed, but the powers and functions exercised by the federal and state orders of government, including the negotiation of the terms on which the federation will occur, and any power to amend the constitution, will involve the exercise of constituted powers, not constitutive ones. The self-governing powers exercised by both orders of government under the federation will, in a sense, be powers devolved from the imperial power. Nonetheless, it is at the same time very possible that the federating colonies may exercise very substantial measures of self-determination in the formation of the federation. If the constituent colonies have been granted powers of local self-government, their elected political leaders may take a leading role in negotiating an agreement upon which the colonies wish to federate; the proposal may be submitted to the people of the colonies for their approval; and the colonists may insist that the agreed terms be given full effect by the imperial parliament. Canada and Australia are, to varying degrees, examples of this complex interweaving of imperial, federal and state (or provincial) elements in the formation of a federal constitution.

In such cases, whether the federation is formed before or after independence, the terms on which independence is secured tend to have a significant influence on the way in which constitutive power is configured as between the federalised state and its constituent states. For example, although some have argued that the American states secured their independence from Britain as an already-formed union, this is difficult to square with the terms of the Articles of Confederation, under which each state expressly retained « its sovereignty, freedom, and independence », noting that the Articles had to be ratified separately by each State before they were regarded as binding upon them. The continuing sovereignty of the states was presupposed, in turn, by the manner in which the Constitution was negotiated by delegates of the states at the Philadelphia Convention in 1787 and ratified by each of the states over the next two years. It was recognised that the Constitution needed to be acceptable to each of the states, and an important part of the bargain was the role guaranteed to the legislatures of the states, or conventions held in each of the states, in the prescribed processes for amendment of the Constitution. On the other hand, the terms on which independence were secured in India reflected the British tendency to administer the country as a single territory, albeit one divided into numerous administrative provinces, alongside hundreds of princely states which, from a British point of view, although internally autonomous, were considered subject to the suzerainty of the British Crown. While the autocratic and centralised nature of British rule in India was somewhat moderated by a succession of Government of India Acts, passed in 1909, 1919 and 1935, the government did not cease to be essentially unitary. Correspondingly, although Indian activists, particularly through the Indian National Congress, called for independence, they did so with a view to establishing an all-India constituent assembly to enable the Indian people « as a whole » to assert and exercise their inherent right to constitutional self-determination. The Indian Independence Act 1947 (UK), by which the British Parliament eventually agreed to Indian independence, relinquished suzerainty over the princely states, partitioned British India into two independent dominions of India and Pakistan, brought to an end the unilateral authority of the British Parliament to legislate for India and Pakistan, and conferred authority upon separate Constituent Assemblies in India and Pakistan to establish new constitutions for the two dominions. Succeeding, in a very real sense, to the concentrated powers of governance formerly exercised by the British authorities, but now acting in the name of the Indian people, the Constituent Assembly embarked on the challenging task of drafting a new constitution for India. Its claim to act in the name of the people, although contestable, was supported to the extent that its members had been chosen by the provincial legislative assemblies or nominated by the princely states, each roughly in proportion to its population. From the outset, the Assembly described its objectives as being the establishment of India as an « Independent Sovereign Republic » which derived its « power and authority » from « the people ». It proceeded to draft and ratify a constitution which, although federal, reflected its unitary origins in numerous ways, including, as noted earlier, the procedures for its amendment.

Similar observations can be made about Canada and Australia. While in both countries the federation was brought into being as a matter of law by a statute enacted by the British Parliament at Westminster, the formation of the two federations was an initiative of the self-governing colonies in each country. What especially distinguished the two countries was that the Australian colonists were relatively more insistent on their independent right to make their own constitution. The Australians accordingly sought, and the British granted, a local power of constitutional amendment from the very beginning. The Australian federation was also premised on the constitutional equality of the constituent states. This presupposition of equality resulted in the construction of a highly symmetrical federation, in which the powers and rights of the states in virtually all respects were preserved equally under the constitution, including their continuing powers, their rights to representation in the federal senate, and their role in the amendment of the constitution. This latter capacity included not only a power to make formal changes to the constitution, but also to exercise the powers of the British Parliament as they applied to Australia. This meant that, when the Australian Constitution was finally patriated in 1986, the Australian federal and state parliaments were able to play an essential legislative role alongside that of the British Parliament. While one version of the Australia Act 1986 was enacted by the British Parliament upon the request and consent of the Australian federal government, a second version was enacted by the Australian Commonwealth Parliament upon the request and consent of the Australian States.

The Canadians, by contrast, were less insistent than the Australians. They did not call for, and the British authorities did not grant, a local power to amend the federal constitution. Canadians would therefore have to petition the Imperial Parliament to make changes to the constitution until it was patriated in 1982, at which time the British Parliament abdicated its capacity to legislate for Canada and a local power of constitutional amendment was conferred. Moreover, although the Canadian Constitution was the result of a negotiated agreement between elected representatives of the constituent colonies, it was not premised on their constitutional equality. Each colony had a full say in the negotiations, and each was allowed the same basic set of provincial powers, but the maritime colonies of Nova Scotia and New Brunswick were granted less notional representation in the federal senate, and when the provinces of Manitoba, British Columbia, Prince Edward Island, Alberta, Saskatchewan and Newfoundland were later added to the federation, their rights to representation in the senate were not treated equally either. Consistent with this approach, when a local power of amendment was finally secured in Canada, the role accorded to the provinces in the amendment of the constitution was again not strictly equal. The general procedure for constitutional amendment gives more weight to the provinces that have larger populations. Among other things, it requires supporting resolutions in the legislative assemblies of at least two-thirds of the provinces that have, in the aggregate, at least fifty per cent of the population of all the provinces. This, however, is subject to the important qualification that any proposed amendment that derogates from the legislative powers, proprietary rights or other rights or privileges of the legislature or government of a province will not have effect in a province the legislative assembly of which has expressed its dissent thereto by resolution supported by a majority of its members prior to the issue of the proclamation. In addition, amendments concerning particular matters of intrinsic interest to all of the provinces require the support of the Senate, the House of Commons and the legislative assembly of every province. Despite the protection these procedures gave to the provinces, they remained unacceptable in some quarters, especially in Quebec. In response, the Dominion Parliament enacted a law that prevents a Minister of the Crown from proposing a constitutional amendment without first obtaining the consent of each of Ontario, Quebec, British Columbia, two or more of the Atlantic provinces and two or more of the Prairie provinces.

VI. Multi-layered Configurations of Power

The configuration of power within a federal system is typically expressed in a multi-layered fashion, beginning with the constitutive authority on which the constitution is based, extending to the amending power by which the constitution can be altered, and including the distribution of ordinary powers of legislation, administration and adjudication between the federalised polity and its component polities. The constitutional balance of power between polity and polities at any particular point in time is a function of the composition of the institutions that exercise such powers, the decision-making rules they are required to follow when exercising those powers, and the scope or extent of the powers themselves. Every time constitutional power is exercised, a particular aspect of the federal balance is affected, sometimes with relatively specific and limited significance, sometimes with more general and systemic significance, depending on the effect of its exercise on the powers of the other institutions within the constitutional system.

These last observations can be made more precise by conceiving the conceptual space as a kind of matrix defined along one axis by the distinction between constitutive authority, amending power and the ordinary constituted powers of legislation, administration and adjudication, and defined along the other axis by the distinction between institutional composition, decision-making rules, and field of competence. There is generally more flexibility in a federal system as regards the operation of the ordinary constituted powers of legislation, administration and adjudication than there is in relation to the amending power and especially the constitutive authority upon which the whole constitution rests. Moreover, constitutional change tends to occur more easily and frequently in relation to the competences of the institutions than in relation to their composition and decision-making rules. As a consequence of these general tendencies, the greatest flexibility tends to be located in the bottom right-hand corner of the matrix, whereas the greatest stability tends to be located in the top left-hand corner.

Figure 1: Constitutional change in federal systems

Constitutive authority is generally the most stable element of a constitutional system. The way in which constituent authority operated in the formation of a constitutional system is a matter of historical fact, which does not change, even if prevailing interpretations of the evidence may possibly change. Moreover, the legitimacy of the entire constitutional order depends on how the constitutive authority underlying the constitutional order is conceived, which also contributes to its stability. A change in the locus or nature of the constituent authority is nothing less than a constitutional revolution. The amendment clauses of written constitutions are also relatively stable, largely due to their association with the constitutive fundamentals of the constitution. Amendment clauses stipulate the rules by which the text of the constitution can lawfully and legitimately be changed. Changes to the rules of change within a constitutional system are profound events. By contrast, changes in the distribution of legislative, executive and judicial power are relatively more frequent, although their frequency varies from one constitutional system to another.

Thus, in all of the cases considered in this article, the amendment power has changed only very infrequently, and the likelihood of change seems to be related to the extent to which the foundations of the federal system are aggregative or devolutionary. For example, the amendment clause of the U.S. Constitution, which requires a significant level of state consent, has never been altered. The amendment clause of the Australian Constitution, which similarly requires significant State consent, has been amended only once, to enable citizens of the two mainland federal territories to vote in referendums to amend the Constitution. In Canada, a very major change occurred when the power to amend the Constitution was transferred from Britain to Canada. However, the constitutionally-prescribed methods introduced at that time, which also require significant provincial consent, have not been altered since. In the United Kingdom, the doctrine of parliamentary sovereignty means that constitutive authority, amendment power and ordinary legislative power are exercised by the same institution, the Parliament at Westminster. This has remained fundamentally the case at least since the Act of Union 1800, and is traceable to the Revolution of 1688. As a consequence, the substance of the British constitution is highly flexible, except for the principle of parliamentary sovereignty itself. The procedures of the Parliament have been changed fundamentally on two occasions by the Parliament Acts 1911 and 1949, which limit the capacity of the House of Lords to delay bills and enable them to be passed without its consent. The entire system of devolution in the United Kingdom accordingly still depends on statutes enacted by the British Parliament, although convention requires the consent of the devolved legislatures for the enactment of changes to constitutional arrangements affecting them. But even subject to this stricture, the devolution statutes have already been amended on numerous occasions. In India, which has the most flexible of all the written constitutions considered in this article, not only have a very large number of changes been made to the Constitution without any need for the consent of the states, but the amending clause has itself been altered twice, while the constitutionality of these alterations has also been subject to judicial scrutiny, with one set of changes being found contrary to the basic structure doctrine. The protections given to the states under the Indian amendment clause are not as extensive as those in the United States, Australia or Canada, and many constitutional changes have been made to the federal balance of power as a result. All of this reflects the relatively more unitary configuration of constitutive authority in India.

It is also important to consider the composition, decision-making rules and competences of the political institutions established by federal constitutions. The constitutions considered in this article tend to entrench the compositional features of their federal institutions more deeply than the competences exercised by those institutions, and the depth of such entrenchment once again appears to be influenced by the extent to which the federal system is aggregative or devolutionary. Thus, in the United States and Australia the composition of the federal legislature, and especially the representation of the constituent states in at least one of the houses of the legislature, is not only constitutionally entrenched, but receives special protection in the form of the requirement that the affected states, or the people of those states, must approve of any proposed change. In such countries, changes can be made to the distribution of legislative powers and other aspects of the federal system by ordinary constitutional amendment, but not changes to the representation of the states in the Senate. Although the Canadian Constitution now offers some protection for the representation of the provinces in the House of Commons vis-à-vis their representation in the Senate, the representation of the Canadian provinces in the Senate is not as strongly protected as in the United States and Australia. In India, certain federal features of the Constitution are more deeply entrenched than other aspects, including the representation of the states in the Parliament and the distribution of legislative powers between the Union and the states. But the depth of this entrenchment is less far-reaching, as changes to these particular features can still be made if ratified by one half of the State legislatures. In form at least, these specific aspects of the Indian Constitution only receive as much protection as the entire text of the Australian Constitution. In this way, the nature and scope of constitutional entrenchment in India seems to reflect its unitary origins and devolutionary character, and the basic structure doctrine, although still controversial, becomes more comprehensible when understood in this context. Indeed, the tendency of federal constitutions to protect the composition of their federal institutions more than the competences of those institutions corresponds interestingly with the fact that the Indian basic structure doctrine is especially concerned to protect various structural features of the Constitution, such as federalism and democracy, and not the distribution of competences between the Union and states. In the United Kingdom, similarly, the most recent round of constitutional reforms focus on the proposition that the Scottish institutions of government ought to be recognised as permanent parts of the British constitution, while the distribution of powers between the British and Scottish Parliaments is treated as a matter for readjustment from time to time, either by Act of Parliament or Order in Council.

V. Conclusions

The federal (or federal-like) systems considered in this article illustrate how the configuration of constitutive power at the founding of a constitution influences its text and structure. The pattern in which constitutive power is configured tends to be replicated, in modulated form, in the distribution of powers, representative institutions and amendment clauses of the resulting constitutional arrangement. Federal constitutions are usually rigid in order to ensure that the terms of the federal compact cannot be altered unilaterally by either the federation or the constituent states. But rigidity is only one part of the equation. Federations also display characteristically federal ways of securing that rigidity. Typically, this involves some formal recognition of the plurality of polities of which the federation is composed. Federal constitutions thus involve a kind of plurality-in-unity which simultaneously recognises the plurality of states of which the federation is composed as well as the unity into which those states are combined.

All federal constitutions are in an important sense transformational. They either involve the transformation of a plurality of states into a federation or the transformation of a unitary state into a federalised state consisting of component polities. However, the extent of the federalising transformation can, in each case, vary. The more closely the procedures prescribed by an amendment clause replicate the constituent power on which a constitution is founded, the less significant is the transformation, while the more an amendment clause departs from the formative process, the greater the transformation. However, here it is important bear in mind the distinction between federations that begin as unitary states and become federalised and federations that begin as a plurality of independent states which agree to federate. An aggregative federal constitution, premised on the independence of the constituent states, will be more or less transformative depending on the extent to which the states retain or else relinquish control over the amendment of the federal constitution. A devolutionary federal constitution, premised on the sovereignty of the original unitary state, will be more or less transformative depending on the extent to which control over the amendment of the constitution is retained by the parent state or else relinquished to the newly-created component polities. In both cases, there is a movement, more or less extensive, in the direction of a kind of plurality-in-unity or unity-in-plurality, but the starting point is different.

The difference in starting point is significant because it shapes the way in which constituent power is configured at the time that a federation comes into being. In the most pure cases of aggregation, the constituent states are acknowledged to possess a kind of sovereign equality. When every potential member state is treated as a constitutive equal any federation of those states will be founded on the unanimous consent of them all. In such cases, moreover, the constituent polities are in a position to insist upon being treated as equals within the federation, and tend to do so, especially in relation to those matters they regard as most vital to their interests. However, because federal constitutions are transformative, it is possible that the constituent polities may agree to form a federation in which they are not treated as equals in all respects. On the other hand, where a federation is devolutionary, this tends to be reflected in amendment clauses that allow the legislative institutions of the original unitary state to retain an essential role in constitutional amendment, even if at the same time an important or essential role is also given to the newly-established component polities. However, in devolutionary systems, the equality of the component polities is not necessarily a presupposition of the federating process and so the status and powers conferred upon the component polities may often be asymmetrical rather than symmetrical.

Similar analyses apply to federal systems that emerge out of territories under the imperial or military control of a state external to the federation. Once again, the exact configuration of the constituent power at the establishment of the federal system is of crucial importance, the major difference being that there are three political orders to be considered: the imperial polity, the federal polity and the component polities. In such cases, the federating process may be more or less aggregative or devolutionary as regards the relationship between all three political orders of government.

While each constitutional system has its own particular constitutive logic and evolving balance of polity and polities, there seem to be certain general patterns of constitutional ordering evident in each of the cases considered in this article. The formative character of each federal system, whether aggregative or devolutionary, contributes to the development of assumptions that underlie constitutional reasoning in each country, as contested as that reasoning so often is. For example, no matter how much the principle of federalism may have come to be recognised and applied in the interpretation of the Indian Constitution, the Supreme Court appears to remain wedded to the proposition that the Constitution derives its authority from a constitutive act of the people of India, conceived in fundamentally unitary terms, and uses this idea in its decisions to very practical effect. Likewise, no matter how centralised judicial interpretation of the American and Australian Constitutions may have become, the courts in both countries continue to conceive of their respective systems as constitutively federal in ways and to an extent that simply cannot obtain in countries like India and the United Kingdom. The basic settings of the constitutional systems of these countries are fundamentally different. It would take a constitutional revolution in the fullest sense of the word to change this.

Placing federal constitutions into comparative perspective offers a perspective from which the relative plausibility of particular interpretations of each constitution can be assessed. This is one of the very important things that comparative inquiry can do, especially when pursued in a manner that Sujit Choudhry has called « dialogical ». Rather than depend solely upon debatable theses about normative universality or convergence among constitutional systems, comparative analysis can be directed to highlighting differences, as well as similarities, and to tracing the systemic effects of these points of difference and similarity in each constitution. By taking account of what Cheryl Saunders has called the « multiple layers of meaning » that may potentially be discerned in constitutional systems, comparative constitutional law can engender « a keener awareness of the particular », and help « to expose the factual and normative assumptions underlying … [a particular] constitutional order », thus opening for « discussion and contestation those characteristics which had remained invisible to domestic eyes ».

The alternatively imperial, aggregative and devolutionary foundations of a federal or quasi-federal system have enduring and very significant implications for the distribution of powers, institutional design and amendment procedures adopted thereunder. Comparative constitutional law has lacked an explanatory theory of the formation and amendment of federal constitutions which grapples sufficiently with the characteristic issue that arises in federal systems, namely the relationship between the federated polity and the constituent polities. Explanatory theories, in the context of constitutional law, offer accounts of the textual content of constitutions (or the doctrinal content of judicial decisions) that aim to render them rationally intelligible. However, one of the prime difficulties encountered by attempts to formulate explanatory theories in the field of comparative federalism concerns the context-dependent nature of the negotiated compromises that ordinarily accompany the formation of a federal constitution. Because federal bargains are a function of the unique conditions of each country, comparison between countries is especially difficult and general explanatory theories of federalism seem elusive. However, if close attention is given to the way in which effective constitutive authority is configured at the critical moments when a constitution is brought into being and undergoes fundamental change it is possible to identify certain systemic consequences for the textual and structural features of the constitution, including its formal amendment procedures. This is because the exact way in which constitutive authority is configured operates as a kind of presupposition in constitutional reasoning, as well as a locus of effective political power, which tends to have a systemic effect on the way in which constituted power is distributed within the resulting federal system. By approaching the question in this way, the uniquely context-dependent nature of the bargains that underlie the formation of federal constitutions can be assimilated into a theory that offers a generalised explanation of the formation and amendment of federal constitutions understood comparatively.

Nicholas Aroney

Visiting Professor, Institut Michel Villey, Université Paris II (Panthéon-Assas); Professor of Constitutional Law, T.C. Beirne School of Law, University of Queensland.

The support of Australian Research Council grant FT100100469 is gratefully acknowledged.