1. Introduction

Constitutions change with time. Such change can take place in various ways: according to a procedure stipulated within them or outside of the formal amendment process, for instance, through judicial interpretations or practice. A modification of a constitutional text’s meaning through interpretation may often carry a greater effect than its formal modification. For some, the issue of amendments is less interesting than that of informal transformation. Nonetheless, formal amendments remain an essential means of constitutional change and raise imperative questions which are far from being tedious.

Amendment procedures are nowadays a universally recognised method. They are important since “the ultimate measure of a constitution is how it balances entrenchment and change”. Indeed, since the modern era of constitutionalism, the amendment formula is considered important as the “healing principle” that allows the constitution to stand the test of time; it is “the keystone of the Arch”. However, the “rule of change” is not merely a technical mechanism of balancing stability and flexibility. It directly implicates the nature of the constitutional system, as it is “the space in which law, politics, history and philosophy meet”.

Whereas the definition of the nature of the amendment power is among the most abstract questions of public law, the question of its scope raises important practical questions. Are there any constitutional principles so fundamental that they carry a supra-constitutional status in the sense that they cannot be amended? Does a radical constitutional change brought about through an amendment become a revolutionary act? There is an increasing trend of unamendability in global constitutionalism. Unamendability describes the (explicit or implicit) resistance of constitutional subjects to their amendment. Whereas between 1789 and 1944, only 17% of world constitutions included unamendable provisions, between 1945 and 1988, 27% of world constitutions enacted in those years included such provisions, and out of the constitutions which were enacted between 1989 and 2013 already 53% included unamendable provisions. Unamendability is not merely declarative. In various jurisdictions, such as India, the Czech Republic, Turkey and Brazil, amendments which violate those unamendable subjects may be considered unconstitutional and invalidated by courts. The idea that amendments that were enacted according to the procedure could be declared unconstitutional on the grounds that their content is at variance with the existing constitution is perplexing. After all, is it not the purpose of amendments to change the constitution?

Indeed, at first glance, the very idea of an “unconstitutional constitutional amendment” seems puzzling. The power to amend the constitution is a supreme power within the legal system, and as such, it can reach every rule or principle of the legal system. If this power is indeed supreme, how can it be limited? If it is limited, how can it be supreme? This is the legal equivalent of the “paradox of omnipotence”: can an omnipotent entity bind itself? Both positive and negative answers lead to the conclusion that it is not omnipotent. Moreover, if the amendment power is a kind of constituent power, then it remains unclear why a prior manifestation of that power prevails over a later exercise of a similar power. Quite the reverse: according to the lex posterior derogat priori principle, a later norm should prevail over a conflicting earlier norm of the same normative status. Finally, the constitution, which expresses the people’s sovereign power, binds and guides parliament’s ordinary law-making power. The common meaning of “unconstitutionality” is that an ordinary law, inferior to and bound by the constitution, violates it. How can “unconstitutionality” refer to an act carrying the same normative status as the constitution itself? Therefore, the idea of an unconstitutional constitutional amendment seems paradoxical. This article argues that clarifying the nature and scope of the constitutional amendment power is the first step for undoing this apparent paradox.

The question of unconstitutional constitutional amendments has recently attracted increased attention. Yet it suffers from the lack of a comprehensive and coherent theoretical framework. The framework which contextualises the theoretical approach of this article is constitutional theory which aims to explain the character of existing constitutional arrangements and practices. This article thus develops a general theoretical framework that addresses unamendability and explains the doctrine of unconstitutional constitutional amendments. It does not focus on any specific jurisdiction and confronts the research questions from a more general perspective since its enquiries transcend any specific boundaries insofar as they present phenomena common to all contemporary constitutional democracies.

This article progresses as follows: section 2 addresses the thorny problem of the nature of the amendment power: is it an exercise of constituent power or constituted power? Reviving the old French doctrine distinguishing between original constituent power and derived constituent power, it argues that the amendment power is sui generis: it is neither a pure constituted power, nor an expression of original constituent power. It is an exceptional authority, yet a limited one. I term it a secondary constituent power and apply a theory of delegation in order to illuminate its unique nature. While section 2 explains why the amendment power is limited, section 3 explains how it is limited. Following the delegation theory presented in section 2, it is argued that the primary constituent power may explicitly limit the inferior secondary constituent power. Moreover, any organ established within the constitutional scheme to amend the constitution, however unlimited it may be in terms of explicit language, nonetheless cannot modify the basic pillars underpinning its constitutional authority so as to change the constitution’s identity. A constitution, according to this section, has to be read in a foundational structuralist way – as a structure that is built upon foundations. Section 4 concludes.

2. The Nature of Amendment Powers

The theoretical path for comprehending any limitation on the amendment power must commence by explaining the nature of that power. The manner in which we grasp the nature of the amendment power affects our thinking about its scope. The section begins by illuminating the theoretical distinction between constituent power and constituted power. It then explores possible understandings of the amendment power, both as a constituent and a constituted power. It proposes that the amendment power has to be regarded as sui generis, a unique power situated in a grey area between the two powers. It is distinguished from constituent power in that it ought to be comprehended in terms of delegation, but it is also a distinctive form of a constituted power. Understanding the exceptional nature of the amendment power as a secondary power serves as the theoretical starting point for understanding its limited scope.

A. Constituent and Constituted Powers

In the modern era, a nation’s constitution is regarded as receiving its normative status bottom-up; from the political will of “the people” to act as a constitutional authority, and through which “the people” manifest itself as a political and legal unity. This notion is now explicitly stated in various constitutions.

The procreative principle of modern constitutional arrangements is constituent power, understood as the power to establish the constitutional order of a nation. Whereas the idea of the people’s constituent power begins in early modern legal thought, the concept of constituent power finds its first articulations in English revolutionary debates of mid-seventeenth century, and has been more fully articulated during the French and North-America eighteenth century revolutions. In order to understand the concept, one has to return to Abbé Emmanuel Joseph Sieyès, who distinguished in his ‘What is the Third Estate?’ between constituent power (pouvoir constituant) and constituted power (pouvoir constitué): “in each of its parts a constitution is not the work of a constituted power but a constituent power”. The latter is the extraordinary power to form a constitution – the immediate expression of the nation and thus its representative. It is independent of any constitutional forms and restrictions. The former is the power created by the constitution, an ordinary, limited power, which functions according to the forms and mode that the nation grants it in positive law. Hence, contrary to constituted powers, constituent power is free and independent from any formal bonds of positive law created by the constitution. “The nation”, Sieyès wrote, “exists prior to everything; It is the origin of everything. Its will is always legal. It is the law itself”. The constitution, as a positive law, emanates “solely from the nation’s will”. For Sieyès, the nation is free from constitutional limits as the sovereign people are exterior to their institutions; the constituent power was unlimited for “it would be ridiculous to suppose that the nation itself could be constricted by the procedures or the constitution to which it had subjected its mandatories”.

The conceptual relationship between constituent and constituted powers is that of subordination. Constituted powers are legal powers (competence) derived from the constitution (and are limited by it). They owe their existence to the constituent power and depend on it; thus, constituent power is superior to them. In contrast to constituted power, constituent power is unlimited – at least in the sense that it is not bound by previous constitutional rules and procedures. On that account, the distinction between constituent and constituted powers is imperative for any investigation regarding possible limitations on the amendment power, since if this power is conceptualised as a constituted power, it is subordinated to the constitution, whereas if it is conceptualised as constituent power, then it should be regarded as unlimited and unbound by prior constitutional rules. However, as demonstrated in the next section, this classification seems extremely thorny when one has to assess the amendment power.

B. The Amendment Power as Sui Generis

The constituent power establishes the constitution, which in turn regulates the ordinary constituted powers, such as the executive, legislative, and judiciary. Once the constituent power has fulfilled its extraordinary constituting task, it “retire[s] into the clouds”, and from that moment public authority is exercised under the constitution. Thus, by establishing a constitution, the constituent power is “digging its own grave”. At the backdrop of this story, the amendment power is an extraordinary authority. It is “peculiar and not fully understandable in terms of the hierarchical model of the legal pyramid”. The reason for that is because, as Stephen Holmes and Cass R. Sunstein observe, it “does not fit comfortably into either category. It inhabits a twilight zone between authorizing and authorized powers. […] The amending power is simultaneously framing and framed, licensing and licensed, original and derived, superior and inferior to the constitution”.

On the one hand, if “the people” control the government (qua constituted powers) through the constitution, then arguably, constitutional amending power is “the highest power in the nation’s political life”. Viewed in that respect, the amendment process is a mechanism for constitution-makers to ‘share part of their authority’ with future generations. Ostensibly, if it is permissible for “the people” to re-shape their constitution, amending a constitution, like constitution-making, is part of the people’s ultimate constituent power. This is the prevailing approach of American constitutionalism. “Americans”, as Gordon Wood wrote, “had in fact institutionalized and legitimized revolution”. Several arguments support this approach:

Supremacy argument: constituted powers are bound by the constitution. By means of amendments, “the people” may alter constituted powers. Therefore, this power differs from and superior over ordinary constituted powers and must be of a constitutive nature. Not only can it modify other constituted powers, but it may also, arguably, change its own boundaries since it possesses Kompetenz-Kompetenz.

Procedural argument: most constitutions provide different procedures for ordinary legislation and constitutional amendments, which emphasize the exceptional process of constitutional amendment. This distinction strengthens the argument that the amendment procedure is not an ordinary constituted power; but different from and more unique than ordinary law making.

Consequential argument: from a juridical perspective, constituent power is “the source of production of constitutional norms”. If constituent power produces constitutional laws that govern constituted powers, then amending those constitutional laws is an exercise of constituent power. Amending a constitutional provision creates the same legal product as writing a new provision. Therefore, amending the constitution is arguably an exercise of a power similar to that which created the constitution in the first place – constitute power:

On the other hand, the mere stipulation of an amendment procedure points to its instituted and thus constituted – rather than constituent – nature. True, it has a remarkable capacity to reform governmental institutions; yet it is still a legal competence defined in the constitution and regulated by it. Even if one applies here the concept Kompetenz-Kompetenz, the constituent power declares the constituted power competent to define its competences, but only within the limits set in the constitution. If all powers derive from the constitution, then the amending power must be a constituted power just like the legislative, judicial, or executive powers. As a legally defined power originating in the constitution, it cannot ipso facto be a genuine constituent power.

As a result, the amending power is multi-faced. It carries dual features of both constituent and constituted powers, hence the question of its nature is a knotty on. Asem Khalil writes that the amendment power is “constituent power in nature and a constituted power in function”. In contrast, Grégoire Webber claims that the amendment formula is constituted by nature, but functions as constituent authority. Since this power does not fit comfortably into any of these categories, it should neither be regarded as another form of constituted power nor equated with the constituent power; it is a sui generis power.

C. The Secondary Constituent Power

(i) The Distinction between ‘Original’ and ‘Derived’ Constituent Powers

 “To know how the constitution of a given State is amended”, A.V. Dicey wrote, “is almost equivalent to knowing who is the person or who are the body of persons in whom, under the laws of that State, sovereignty is vested”. Note that Dicey is not stating that sovereignty is vested in the amendment authority but uses a terminology of “not quite” – but “very nearly” – sovereignty. This resembles Max Radin’s two notions of “sovereignty”. Radin distinguished between real sovereignty, which can materialise only in revolutions, and “minor or lesser sovereigns”, created by the real sovereign. The amendment power, created by the “original sovereign”, is a lesser sovereign, “almost sovereign”, situated between the real sovereign and lesser sovereign, such as governmental functions. The basic presupposition underpinning Radin’s argument, and the one this article advances, is that the amendment power is a special power, weaker than the constituent power but greater than the ordinary constituted powers. This proposition revives and relies upon the French doctrine that distinguishes between original constituent power (pouvoir constituant originaire) and derived (or derivative) constituent power (pouvoir constituant derivé). The first is a power that is exercised in revolutionary circumstances, outside the laws established by the constitution, and the latter is the legal power exercised according to rules established by the constitution.

This distinction between original and derived constituent powers was developed during the debates of the French National Assembly on the 1791 Constitution, albeit with different terminology. At the assembly, debates took place on how the Constitution ought to be amended in light of the fragility of the constitutional project. The adopted process was that the Constitution would be unamendable for ten years, after which amendments could take place through an Assembly of Revision, and after approval of three successive legislatures. During the debates of the National Assembly, some argued that the Assembly could not limit or even procedurally frame the constituent power, while others sought to minimise the likelihood of future constitutional changes. Frochot proposed that there be a differentiation between partial and total change to the Constitution, believing that each involves a fundamentally different power. Thus, he proposed a certain procedure for partial change and another (more complex) for a total change. While his proposal was rejected, the distinction he made allowed others to justify the ability to limit and frame potential constituent power without forfeiting the idea of an unlimited constituent power. Barnave explained that the total change of the Constitution could not be predicted or controlled by the Constitution, because it is an unlimited power belonging inherently to the nation. However, the possibility of amending the Constitution is of a somewhat different nature, which may be limited and circumscribed. Barnave’s discourse reveals the distinction between original and derived constituent power.

This idea was evident in Title VII, Art. 1 of the 1791 Constitution, which, while acknowledging the nation’s “imprescriptible right to change its constitution”, limits the amendment power procedurally “by the means provided in the constitution itself”, and substantially by allowing amendments only to “the articles of which experience shall have made the inconveniences felt”. Additionally, Title VII, Art. 7, required members of the Assembly of Revision to take an oath, “to confine themselves to pass upon the matters which shall have been submitted to them […] [and] to maintain […] with all their power the constitution of the kingdom […]” Thus, the amendment power was conditioned by preserving the constitution.

Explaining this special, yet legally defined, power, Oudot wrote that some constitutions have organized aside the constituted power, a regular constituent power; they have settled the form by which the nation could change its political mechanism. As Claude Klein explains, the original constituent power is the power to establish a new legal order (ordre juridique nouveau). It is an absolute power, which may set procedural and substantive limits for the exercise of amendments. The derived constituent power acts within the constitutional framework and is therefore limited under the terms of its original mandate.

(ii) The Formal and Substantive Theories

Kemal Gözler recognised two schools of thought, formal and substantive, as the basis for the distinction between the original and derived constituent powers. According to the formal theory, original and derived constituent powers are distinguished by the form of their exercise. Constituent power is exercised outside the forms, procedures, and limits established by the constitution. On the other hand, a juridical concept of constituent power is exercised in according to rules established by the constitution. The formal theory can be summarised as follows: original constituent power is exercised in a legal vacuum, whether in the establishment of the first constitution of a new state or in the repeal of the existing constitutional order, for instance in circumstances of regime change. In this theory, the nature of the original constituent power is extra-legal, a pure fact. This is traditional positivist approach as expressed by Hans Kelsen, who does not tackle the question of the constituent power, but rather claims that the question of the basic norm or obedience to the historically first constitution is assumed or presupposed as a hypothesis in juristic thinking. Likewise, for political scientists such as Carl Friedrich, constituent power is not a de jure power but a de facto power which cannot be brought under “four corners of the Constitution”. In contrast, derived constituent power is a constraint power that acts according to the formal procedures as established in the constitution. Gözler makes an important clarification: for him, original constituent power does not have to be exercised for revising the entire constitution; it may be exercised even for amending a single provision (outside of the constitutional amendment process). Similarly, the exercise of the derived constituent power may cover the entire constitution.

For the substantive theory, the main criterion distinguishing between original and derived constituent powers is the different scope of their ability to influence the substance of the constitution. This theory is best represented by Carl Schmitt who distinguished between ‘the constitution’ (Verfassung) and “constitutional laws” (Verfassungsgesetz). The constitution represents the polity’s constitutional identity, which cannot be amended, and constitutional laws regulate inferior issues. The amendment process is designed for the textual change of constitutional provisions, but not of fundamental political decisions that form the substance of the constitution. Thus, for Schmitt, an amendment cannot annihilate or eliminate the constitution. It cannot abolish the right to vote or a constitution’s federalist elements, or to transform the president into a monarch. These matters are for the constituent power of the people to decide, not the organs authorized to amend the constitution. Thus, an amendment that transforms a state that rests on the power of the people into a monarchy, or vice versa, would be unconstitutional.

(iii) Integration: A Theory of Delegation

Gözler argues that these two schools of thought are fundamentally irreconcilable on the grounds that according to the formal theory, as opposed to the substantive one, the derived constituent power is limited only by the formal conditions under which it operates. Contrary to Gözler, I argue that the two theories should be regarded as mutually reinforcing, rather than exclusive. In order for the formal and substantive theories to coexist, the amendment power needs to be comprehended in terms of delegation.

Delegation affords the legal framework, even if not always consciously articulated, to rationalize this state of affairs surrounding the nature of the amendment power. Through the amendment provision, “the people” allow a constitutional organ to exercise a constituent authority – the authority to constitute constitutional laws. When the amendment power amends the constitution, it uses a legal competence delegated to it by the primary constituent power. As Alf Ross explains “in the concept of delegation is implied a vague idea that the entrusting of competence is in the nature of something exceptional in that it permits the delegatus to ‘appear in the role of legislator’”. The amendment power is a delegated authority, where the delegatus exercises a function of a constituent authority. But why does this infer limitability? Surely, one may claim that this is a “clear case of a non-sequitur” since it does not follow from the distinction between original and derived constituent power that the amendment power is limited, “for it is conceptually possible for the derivative constituent power to observe the procedural requirements and, at the same time, derogate the Constitution or replace it with a new one”. Nevertheless, modern studies of delegation now adopt the model of the “principal-agent” in order to define acts of delegation. The one who delegates authority (the original constituent power) is the principal, while the one whom the authority is delegated to (the amendment authority) represents the agent. The amendment power is a delegated power exercised by special constitutional agents. When the amendment power amends the constitution, it thus acts per procurationem of the people, as their agent. Having a principal-agent relationship, the delegated amendment power is subordinated to the principal power from which it draws its legal competency. Hence, contrary to the original constituent power, the delegation of the amendment power inherently entails certain limitations, as the legal framework of delegation is by itself characterised by constraints.

Since the amendment power is delegated, it ought to be regarded as a trust conferred upon the amendment authority: “All delegated power is trust, and all assumed power is usurpation. Time does not alter the nature and quality of either”, Thomas Paine reminds us. True, the amendment authority has the “supreme” amendment power, but it is only a fiduciary power to act for certain ends. If the amendment power is delegated, it acts as trustee. Trustee of whom? Of “the people” in their original constituent power.

Delegation and trust are conceptual keys to the nature (and consequently the scope) of amendment powers. The trustee (the amendment authority) has a legal right of possession of the trust corpus (the amendment power), conditional on his fiduciary obligation to comply with the terms of the trust (procedural or any explicit or implicit substantive requirements) and pursue the ends it established to advance (“amend the constitution”). Due to its nature, the trustee is always conditional and thus the fiduciary amendment power necessarily entails limits. As Akhil Amar has argued, within Art. V of the U.S. Constitution, the people delegated the amendment power to ordinary government, and limitations on the amendment power, as stipulated in Art. V, exist only when it is exercised by delegated powers following from the people. Likewise, William Harris correctly claims that when the sovereign constitution-maker acts as sovereign, “the notion of limits on constitutional change is inapposite”; however, “when the machinery of government is acting as the agent of the people in its sovereign capacity, the notion of limits not only makes sense; it is necessary”.

However, one may claim that even though the amendment power is delegated, it is still limitless since it represents the unlimited sovereign. The representation of an unlimited constituent power must logically result in a similar unlimited amendment power. Such an argument should be rejected. There is always a hierarchical relationship between the grantor and the receiver as “the agent is never equal of the principal”. This is precisely the distinction between original and derived constituent powers.

How does the theory of delegation manage to integrate the formal and substantive theories? First, delegation theory is not restricted to the substance of amendments. The amendment power must obey the procedure as prescribed in the constitution. Similarly, it is required to observe those explicit (not necessarily procedural, but also substantive) limits set upon it, as formally stipulated in the constitution. Explicit limits on constitutional amendments express the idea that exercise of the amendment power – established by the constitution and deriving from it – must abide by the rules and prohibitions formally stipulated in the constitution. Second, delegation theory is not restricted to form, but also concerns substance. The delegated amendment power, as a rational understanding of that delegation, must be substantively limited, whether these limits are explicitly stated in the constitution or not. Therefore, rather than being exclusive, the formal and substantive theories distinguishing between the constituent power and amendment power mutually reinforce one another.

(iv) Terminological Clarification: Primary and Secondary Constituent Powers

Due to the complexity of the concept of the amendment power and its relations with the constituent power, various versions have developed in the literature to describe these concepts. In the American literature, it was often common to distinguish between framing power and amending power. The German often term the amending power verfassungsändernden Gesetzgeber, the secondary constitutional lawgiver or amending legislature. In French constitutional discourse, the amending power is described by terms such as pouvoir constituant dérivé, pouvoir constituant institué, pouvoir de révision constitutionnelle, or pouvoir constituant constitué. Some of these terms are oxymoronic or “farfetched”. In order to elude any confusion, some plainly reject the use of the term constituent to describe the amendment power. I agree that the oft-used terms are imprecise. Both the constitution-making and constitution-amending powers are constitutive in the sense that it these are powers to constitute constitutional rules. Nonetheless the two are not identical. As for the constitution-making power, I reject the use of the term original constituent power. A constitution always bears a ‘relational account’, and never acts in a pure vacuum. Every constituent process must be based upon a certain prior “constitution-making moment”. Additionally, constitution-making takes many different forms. Some constitutions were formed in revolutionary circumstances, breaking the previous constitutional order. Others were constituted through international efforts or imposed by foreign and external forces, such as the cases of Japan and Germany after 1945 or post-2003 Iraq. Often, the constitution-making process is exercised in continuity with existing laws or in accordance with pre-determined rules (Post-1989 Eastern Europe and South Africa). Finally, the exercise of constituent power itself requires a certain representational form. Since constituent power is never purely original, I use the term primary constituent power instead. It is primary not only because it is the initial action, but also because it is principal in its relations with the amendment power. Congruently, instead of derived constituent power, I use the term secondary constituent power to describe the amendment power. It is secondary not merely because it necessarily comes (chronologically) after the constitution-making process, but because it is subordinated to the primary constituent power and inferior to it. This terminology manifests more properly these powers’ unique nature and sharpens the delicate distinction between them.

D. Conclusion

To sum up the argument thus far, the amendment power is a constitutional power delegated to a certain constitutional organ. Since it is a delegated power, it acts as a trustee of “the people” in their capacity as a primary constituent power. As a trustee, it possesses only fiduciary power; hence, it must ipso facto be intrinsically limited by nature. Put differently, a vertical separation of powers exists between the primary and secondary constituent powers. As in the horizontal separation of powers, this separation results in a power-block. The holder of the amendment power may be restricted from amending certain constitutional subjects. Identifying the amendment power as a delegated authority is the first step in understanding its limited scope. I now move on to explain how – according to this theoretical presupposition – the amendment power is limited.

3. The Scope of Amendment Powers

Based upon the previous section, this section provides the theoretical ground that elucidates various explicit and implicit limitations on the amendment power.

A. Explicit Unamendability

(i) The Validity of Unamendable Provisions

The idea of constitutional entrenchment is debated extensively in the literature. However, unamendability takes constitutional entrenchment to its extreme, hence it is often described as “absolute”. Ferdinand Regelsberger argued that “there is no law which cannot be changed. A legislator […] cannot control the unchangeability of a legal norm”. For this reason, the validity of unamendable provisions is often disputed, and critics describe them “a bit of useless verbiage” or “an empty phrase”. Notwithstanding such criticism, Kelsen’s view was that there is no reason to suppose that a norm cannot stipulate that it cannot be repealed. For Kelsen, a norm could be declared as unamendable, yet such a declaration cannot prevent the loss of its validity by a loss of efficacy. Moreover, since a provision prohibiting any amendments is not invalid by its very nature, in the case of unamendable provisions, it is not legally possible to amend the protected provisions. Indeed, nowadays unamendable provisions are commonly considered valid.

The theory hereby presented supports the validity of unamendable provisions, but relies on questions concerning the sources of constitutional norms. The secondary constituent power which is a delegated power may be restricted by the primary constituent power from amending certain principles, institutions, or provisions. The motives for such restrictions and the aims those are designed to accomplish vary. What is clear is that the amendment power, which is established by the constitution and subordinate to it, is exercised solely through the process established within the constitution. It is bound by any explicit limitations that appear in the constitution, if those are set by the primary constituent power.

Viewed from the perspective of the formal theory, explicit unamendability reflects the idea that any exercise of the amendment power must abide by the rules and prohibitions stipulated in the constitution, including substantive limits. In that respect, unamendable provisions “can be seen as a procedural constraint which can be surmounted by an entirely new constituent act”. From the perspective of the substantive theory, unamendable principles are an example of the fact that the amendment power may be limited with regard to the content of certain amendments, and can amend the constitution “only under the presupposition that the identity and continuity of the constitution as an entirety is preserved”, to use Schmitt’s words. However, the substantive theory can only explain those unamendable provisions that aim to prevent fundamental changes in an effort to ensure the constitution’s integrity and the continuity of its constitutive principles. But unamendable provisions may simply derive from constitutional compromise and contingency and cover a wide range of topics, not necessarily the basic principles of the constitutional order. These cannot be supported by the substantive theory. The theory of delegation explains all types of unamendable provisions. The secondary constituent power, as a delegated power, acts as a trustee of the primary constituent power. It must obey those “terms” and “conditions” stipulated in the “trust letter” – the constitution. The delegated amendment power is limited according to the conditions stipulated in the constitution, including various substantive limits.

What are the legal implications of a conflict between a new constitutional amendment and an unamendable provision, according to the delegation theory? Unamendable provisions create a normative hierarchy between constitutional norms. Just as the constitution prevails over an ordinary law, a constitutional provision established by the primary constituent power prevails over constitutional provisions established by the secondary constituent power. When resolving conflicts between constitutional provisions (unamendable provisions contrasted with later amendments), the paramount factor is not their chronological order of enactment (lex posterior derogat priori), but rather, the sources of these constitutional norms. Thus, the constituent power is divided conforming to a hierarchy of powers – primary and secondary – governed by the principle lex superior derogat inferiori; the constitutional rule issued by a higher hierarchical authority prevails over that issued by a lower hierarchical authority. Just as ordinary legislation retreats when it conflicts with constitutional norms, so do constitutional amendments retreat when they conflict with unamendable provisions. In other words, since the primary constituent power is a superior authority to the secondary one, the normative creations of the latter should withdraw when conflicting with that of the former. Unamendable provisions may lose their validity when they face a conflicting valid norm that was formulated by the same authority. Therefore, unamendable provisions cannot limit the primary constituent power; rather they “invite” it to be resurrected in order to change unamendable subjects.

(ii) An ‘Unamendable Amendment?’

A unique difficulty is arising when an amendment stipulates by its own terms that it (or other provisions) are unamendable. This is not a hypothetical scenario. The original French unamendability of the republican form of government was inserted into the 1875 Constitution through an amendment in 1884, stimulating lively scholarly debate. In 1861, the original proposed 13th Amendment to the U.S. Constitution, known as the “Corwin Amendment”, ‘eternally’ prohibited Congress from abolishing slavery. In Bangladesh, the constitution was amended in 2011 to insert an eternal clause, article 7B, that declared that “basic provisions of the Constitution” are amendable.

The distinction between primary and secondary constituent power provides a relatively unassuming solution to this conundrum. As only the primary constituent power can limit the secondary constituent power, unamendable amendments lose their validity when they face a conflicting norm formulated by the same authority. Accordingly, provisions created by the amendment power could subsequently be amended by the amendment power itself. For that reason, I disagree with the argument that as an unamendable amendment, the Corwin Amendment could not have been altered. An “implicit limit” exists, according to which “an amendment cannot establish its own unamendability”. Limitations upon the delegated amendment power can be imposed solely by the higher authority from which it is derived – the primary constituent power.

(iii) Amending ‘Unamendable’ Provisions

Most of the world’s unamendable provisions establish the unamendability of certain constitutional subjects but they are themselves not entrenched. Can non-self-entrenched provisions be amended? As a matter of practice, the answer is positive. In 1989, the unamendable provision in the Portuguese Constitution of 1976 (Art. 288) was itself amended and the unamendable principle of collective ownership of means of production was omitted. Da Cunha notes that this amendment “has always shocked us because it undermines the standard meaning and thus causes the Constitution to lose all of its enforceability”. Importantly, the court was never asked to review the validity of this controversial amendment.

There are three approaches for solving this challenge. According to the first approach, if unamendable provisions are non-self-entrenched, unamendable principles or provisions may be amended in a double amendment procedure. The first stage is to repeal the provision prohibiting certain amendments, an act that is not in itself prohibited. The second stage is to amend the previously unamendable principle or provision, which is no longer protected from amendments. According to the second approach, there is no need for a two-stage process as the unamendable provision and the protected subject could both be repealed in the same act since the outcome is similar. The third approach advanced in this article rejects such attempts to circumvent unamendability. Even non-self-entrenched provisions of unamendability should be implicitly recognised as unamendable. Whereas the double-amendment procedure may be formally tolerable, from a substantive perspective such a legal manoeuvre may be regarded as “fraud upon the constitution”. The famous maxim according to which “what cannot be done directly cannot be done indirectly. The Constitution deals with substance, not with shadows”, equally applies with regards to the amendment power. Therefore, unamendable provisions should be given a purposive interpretation according to which they are implicitly self-entrenched.

From a practical point of view, if unamendable provisions could be amended by means of the same procedure required to amend other provisions, they would almost be devoid of meaning. The declaration of unamendability remains important even if conceived as eventually amendable because its removal would still necessitate political and public deliberations regarding the protected constitutional subject, which assign the unamendable provision important role. Moreover, the unamendability adds a procedural hurdle – and thus, a better procedural protection. Lastly, the unamendability of a provision might have a “chilling effect”, leading to hesitation before repealing the so-called unamendable subject. The double-amendment procedure should therefore be rejected on both theoretical and practical grounds.

Of course, from a purely practical point of view, a clever constitution-maker would draft self-entrenched unamendable provisions. True, the unamendable provision cannot, as Vedel puts it, “be given to a jailer who will guard its intangibility”, but it could be self-entrenched which would then block the aforementioned loophole.

B. Implicit Unamendability

The previous sub-section focused solely on explicit unamendable provisions. However, in some jurisdictions, courts have ascertained a certain constitutional core which cannot be abrogated through the amendment procedure – an implicit unamendability. In this section, I argue that this global trend rests on a solid theoretical basis.

(i) Foundational Structuralism

The first implied limitation derived from the theory of delegation is the most basic: the constitutional amendment power cannot be used in order to destroy the constitution. Since, “the Constitution is not a suicide pact; and, consequently, its provisions should not be construed to make it one”, amendment provisions should not be construed as to embody the constitution’s “death wish”. The authority entrusted with the amendment power cannot use it in order to destroy the very same instrument from which its authority streams. Thomas Cooley wrote that the U.S. Constitution’s framers abstained from explicitly forbidding changes that would be incompatible with the Constitution’s spirit, simply because they did not believe that those would be possible under the terms of the amendment process itself. His metaphor is remarkably clear: “The fruit grower does not forbid his servants engrafting the with-hazel or the poisonous sumac on his apple trees; the process is forbidden by a law higher and more imperative than any he could declare, and to which no additional force could possibly by given by re-enactment under this orders.”

The delegated amendment power is the internal method that the constitution provides for its self-preservation. It cannot be used in order to abolish the constitution, even in the absence of any explicit limitations to that effect, since by destroying the constitution, the delegated power subverts its own raison d’être. Put differently, alongside the legal constitutional amendment power rests the liability not to undermine the same constitution itself. To amend the constitution as to destroy it and create a new constitution would be an action ultra vires; a usurpation of the amendment power that “the people” have not delegated to the amendment authority.

The second limitation derives from the first one, but it is one logical step forward: the constitutional amendment power cannot be used in order to destroy the basic principles of the constitution. Each constitution has certain fundamental core values or principles, which form the “the spirit of the constitution”. This is what I term the foundational structuralist perception of constitutions, according to which, constitutions are not merely instruments of empowerment and restrictions or “power maps” that reflect the political power distribution within the polity. They reflect certain basic political-philosophical principles, which form the constitution’s foundational substance, its essence. The constitution is structured upon these basic principles and it is no longer the same without them. That is, when the amendment power alters the basic essential principles of the constitution, it no longer amends the constitution but replaces it with a new one. Since an amendment cannot destroy the constitution, amending its basic elements is prohibited just as eliminating the constitution is prohibited. This is the basic rationale behind the Indian basic structure doctrine and the Colombian constitutional replacement doctrine. Thus, Schmitt was right to argue that the amendment process is not designed for modifying the fundamental decisions forming the constitution’s substance, since such modification results in a new constitution and such constitutive acts are for the people’s primary constituent power, not the delegated organs.

The third limitation is that the amending power, like any governmental institution, must act in bona fides. The amendment power is not the power to destroy the constitution. Constitutional destruction can also occur “by using the form of amendment to directly exercise other constitutional functions in given cases, disregarding constitutional limitations and upsetting the constitutional disposition of powers”. Even Richard Thoma, who otherwise opposed any notion of implicit unamendability, maintained that parliament could not, for example, dissolve itself in violation of normal prescribed procedures, or pass a bill of attainder. A “government with limited powers of legislation and at the same time, with unlimited powers of legislation, would be an absurdity”, Holding wrote, adding that “no enactment, in substance purely legislative, should be permitted to become a part of the Constitution”. If the material of an amendment is not commonly “constitutional”– i.e. it is ordinarily legislative in nature – this raises suspicions that the provision is being given a constitutional status solely in order to ‘shield’ it from judicial review. Implicit unamendability is a method to protect the constitution against the possibility that “the legislature of the day, hijacked by individual, group and institutional interests and temporary impulses or permanent passions may use its authority to inflict torture on the Constitution”.

Foundational structuralism necessitates an acknowledgment of two notions: a hierarchy of constitutional values and a constitutional identity.

(ii) Hierarchy of Constitutional Values

A constitution is “a rich lode of principles”. But not all principles are equally basic. The German jurisprudence on this idea is instructive. The German Basic Law is regarded as having an integrated structure and a hierarchical scheme of principles with human dignity at the apex. This was recognised by the German Federal Constitutional Court early in 1951 in the Southwest case: “A constitution has an inner unity, and the meaning of any one part is linked to that of other provisions. Taken as a unit, a constitution reflects certain overarching principles and fundamental decisions to which individual provisions of the Basic Law are subordinate”.

Drawing from German jurisprudence, Walter Murphy argued that constitutions in constitutional democracies present not simply a set of values, but rather an ordering of values. This system of values precludes the possibility of adopting an amendment that would infringe human dignity. A similar view, according to which amendments are not intended to disassemble the constitution’s structure or repeal constitutional essential was defended by John Rawls, Samuel Freeman, and Stephen Macedo. These leading scholars seem to share with Schmitt the essential notion of substantive implicit unamendability. Even Laurance Tribe, who calls for a reserved judiciary role with regard to constitutional amendments, seems willing to embrace the notion that some principles are so fundamental to the constitutional order that they can be regarded as indispensable to the system’s legitimacy. Likewise, Richard Albert, who frequently opposes the idea of unamendability for democratic reasons, recently claimed that in order for the US Constitution to remain internally coherent, First Amendment’s democratic rights must be regarded as informally unamendable. This article defends a similar view based on the distinction between primary and secondary constituent powers. As aforementioned, being a delegated authority, the amendment power must be conceived as inherently limited.

The claim for recognition of a hierarchy of constitutional values is not immune from criticism. Gözler, for example, argues that even if there might be a moral difference between constitutional norms, there is no hierarchy, since they do not derive their validity from one another. More recently, Albert criticised any attempt to create a hierarchy of constitutional norms which “threatens to deplete the text of its intrinsic value as an institution whose authority applies equally, fairly and predictably to citizens and the state”. Such criticism seems to be based on a misapprehension of the idea behind the hierarchy of constitutional values within a foundational structuralist analysis. A foundational structuralist analysis of the constitution does not necessary require the picking of a certain secluded constitutional provision, as ‘an isolated island’; rather, it urges us to look at the constitution as an organic whole. It is an exercise to find those systematic principles underlying and connecting the constitution’s provisions and which make the constitution coherent. In his early writings, which were so influential on the Indian endorsement of the basic structure doctrine, Dietrich Conrad used the metaphor of pillars to explain the unamendability of basic constitutional principles: “any amending body organized within the statutory scheme, however verbally unlimited its power, cannot by its very structure change the fundamental pillars supporting its constitutional authority”. This sentence was quoted verbatim by Khanna J. in the famous Kesavananda case in which the basic structure doctrine was developed, and was persuasive in the adoption of the basic structure doctrine in Bangladesh. Conrad later remarked that, “the graphical appeal almost by itself has the force on an argument”, highlighting the power of metaphors within legal argumentation. The metaphor of the pillars that hold the constitutional structure is powerful and corresponds with the foundational structuralism perspective endorsed in this article.

Even to those who do not regard the constitution as a structure but as an organic instrument, the argument of unamendable basic principles, which provide meaning for the greater whole, remains coherent. The metaphor of a living constitution is usually used to imply that the language of the constitution should evolve according to the changing environment of society. The amendment process provides another mechanism for such evolution, as a “built-in provision for growth”. Prima facie, the view that a constitution must develop over time supports a broad use of the amendment power. Nevertheless, even if we conceive of the constitution as a living tree, which must evolve with the nation’s growth and develop with its philosophical and cultural advancement, it has certain roots that cannot be uprooted. The metaphor of a living tree captures the idea of certain constraints: “trees, after all, are rooted, in ways that other living organisms are not”. These roots are the basic principles or structure of a given constitution, even if conceived as a living system.

Therefore, it is not an exercise of “ranging over the constitutional scheme to pick out elements that might arguably be more fundamental in the hierarchy of values”, William Harris correctly claimed, adding that: “a Constitutional provision would be fundamental only in terms of some articulated political theory that makes sense of the whole Constitution”. The idea of a hierarchy of norms within foundational structuralism is to examine whether a constitutional principle or institution is so basic to the constitutional order that changing it – and looking at the whole constitution – would be to change the entire constitutional identity.

(iii) Constitutional Identity

 “A constitution”, Peter Häberle states, “is not merely a juridical text or a normative set of rules, but also […] a mirror of cultural heritage and the foundation of its expectations”. Constitutions are designed to reflect society’s identity. A constitutional identity is defined by the intermingling of universal values with the nation’s particularistic history, customs, values, and aspirations. It is never a static thing, as it emerges from the interplay of inevitably disharmonic elements. But even major changes to the constitutional identity seldom end in a wholesale constitutional transformation. A nation usually aims to remain faithful to a “basic structure”, which comprises its constitutional identity. “It is changeable”, Jacobsohn writes, “but resistant to its own destruction”.

The identity, for foundational structuralism theory, is “the normative identity of the Constitution, supported by a coherent interpretation of its core constitutional principles or basic features”. Each constitutional system has its own basic principles and features, its “genetic code”. Changing this identity would result in the formation of a new constitution. This idea may extend back to Aristotle, who believed that a polis should be identified with its constitution, and that a change in identity of the polis cannot be considered a mere reform, but a birth of a new regime. True, one should not confuse constitutional preservation with constitutional stagnation. Conversely constitutional changes should not be tantamount to constitutional metamorphosis. Imagine Joseph Raz analogy between a constitution and a house built two hundred years ago is convenient to explain this: “[the] house had been repaired, added to, and changed many times since. But it is still the same house and so is the constitution. […] the point of my coda is to warn against confusing change with loss of identity[…]”.

Why is it not the prerogative of the amendment power to change even the basic foundations of the system? James McClellan, for example, asserted that even foolish amendments which violate the spirit of the constitution are still “the prerogative of the American people under Article V to make fools of themselves and to abolish their form of government and replace it with a new system if that is their wish”. McClellan is correct. It is the prerogative of the people to change their system of government, only not through the amendment procedure. A re-creation of the constitutional identity is an act which lies beyond the authority of those governmental institutions created by the people. It should be “the people exercising their constituent power, not the old constitution’s benediction that validates the new order”. This is precisely the distinction between the primary and secondary constituent powers, to use Jacques Baguenard’s metaphor; the primary constituent power is the power to build a new structure and the secondary constituent power is the power to make alterations to an existing building. As the constitution’s core cannot be altered without destroying the whole constitution, the delegated amendment power cannot use the power entrusted to it for quashing the constitution or its essential and pivotal principles so that it loses its identity. This may be viewed as a “constitutional breakdown”.

C. Conclusion

The formal and substantive theories distinguishing between primary and secondary constituent powers are not mutually exclusive, but rather mutually reinforcing. Being a delegated authority, the amendment power may be explicitly limited. It must abide these limitations. However, even a “blank cheque” which leaves everything to the judgment and discretion of the constitutional amendment authority has to be for the achievement of a certain objective – amending the constitution and not destroying it, or replacing it with a new one. It is thus implicitly limited by its nature. This conclusion is an indispensable consequence of the organization of the amendment power within the framework of a limited government.

4. Conclusion

In 1948, Kurt Gödel, the famous Austrian logician, applied for naturalisation as an American citizen. Preparing for the citizenship examination, Gödel thoroughly studied the American history and Constitution. One day, Gödel called his friend, Princeton University mathematician, Oskar Morgenstern. Years later, Morgenstern described the conversation that he had with Gödel:

[Gödel] rather excitedly told me that in looking at the Constitution, to his distress, he had found some inner contradictions and that he could show how in a perfectly legal manner it would be possible for somebody to become a dictator and set up a Fascist regime never intended by those who drew up the Constitution.

Morgenstern told him he should not worry since such events were unlikely to ever occur. Since Gödel was persistent, Morgenstern and another mutual friend – Albert Einstein – tried to persuade Gödel not to bring this issue up at the citizenship examination. On the examination day, Einstein and Morgenstern both accompanied Gödel to his interview at the Immigration and Naturalization Service as witnesses. After the examiner questioned both witnesses, the following exchange occurred, according to Morgenstern’s own account of the hearing:

Examiner: “Now, Mr. Gödel, where do you come from?”

Gödel: “Where I come from? Austria”.

Examiner: “What kind of government did you have in Austria?”

Gödel: “It was a republic, but the constitution was such that it finally was changed into a dictatorship”.

Examiner: “Oh! This is very bad. This could not happen in this country.’
Gödel: “Oh, yes, I can prove it”.

Examiner: “Oh God, let’s not go into this…”

Einstein and Morgenstern were horrified during this exchange, but the examiner swiftly quietened Gödel on this point until Gödel finished his interview. What was the “inner contradiction” that Gödel discovered within the U.S. Constitution? This will remain a riddle as Gödel left no clues. Scholars suggest that Gödel realized that an unlimited amending power possessed the risk of a tyranny as the amendment power might be utilised to subvert the democratic institutions designated in other provisions of the Constitution, including the amendment provision itself. Is the amendment power sufficiently broad so as to destroy the very basis of the constitution? Richard Kay notes that “the core notion […] that there is something wrong with the idea that an ‘amendment’ might alter the essential character of a constitution while simultaneously invoking its authority – has been embraced by many modern constitution-makers”. Indeed, a large percentage of world constitutions include formal unamendable provisions and there is a matching growing tendency of courts to acknowledge a set of implicitly unamendable core.

Unamendability is a powerful mechanism which would be used carefuly. It is often criticised as being undemocratic, perpetuating the ‘dead hand’ of the ancestors, enhancing judiciary’s power, ineffective or even dangerous since it invites the use of extra-constitutional means. While maintaining such challenges in mind, unamendability is compatible with the nature of amendment powers. Charles Howard McIlwain‏ wrote that “a constituted authority is one that is defined, and there can be no definition which does not of necessity imply a limitation”. The amendment power is not an ordinary constituted power, but a sui generis one. Yet, it is still a defined constitutional and limited authority. The delegated secondary constituent power, which acts as a trustee of the primary constituent power, cannot destroy the constitution or replace it with a new one. This is the role of the people who retain the primary constituent power; and through its exercise they may shape and reshape the political order and its fundamental principles. Understood in this way, the theory of unamendability can be seen as a safeguard of the people’s primary constituent power. Accordingly, even judicial enforcement of unamendability may be regarded as a mechanism for ensuring the vertical separation of powers between the primary and secondary constituent powers, making sure that certain changes take place via the proper channel of higher-level democratic deliberations. The theory of unamendability thus restricts the amending authorities from amending certain constitutional fundamentals. Underlying it rests the understanding that a constitution is built upon certain principles, which grant it its identity: “Every constitutional arrangement is based upon a set of core principles which cannot be changed and which can be regarded as intrinsic to its specific identity […] These superconstitutional provisions could be referred to as the genetic code of the constitutional arrangements […]. Accordingly, and in contrast with Kelsen’s conception of revolution as a replacement of the constitution in a way that is incompatible with the amendment procedure, a constitutional change may also be revolutionary substantively, even if adopted according to the prescribed constitutional procedures, if it conflicts with unamendable constitutional provisions, or collapses the existing order and its basic principles, and replaces them with new ones thereby changing its identity.

True, the thin line between primary and secondary constituent power is blurred. As Giorgio Agamben writes, within the current trend of legalisation, “constituent power is more and more frequently reduced to the power of revision foreseen in the constitution”. Indeed, constitutional practice demonstrates that amendments are often used in order to establish in effect a new constitution. This is a constitutional break concealed by continuity. For instance, the Hungarian transformation from communism was employed by way of constitutional amendments to the 1949 Constitution. Whereas such a transformation may well carry various benefits, this complete reform, which brought about a new constitution, suffered “legitimacy problems and clashes of identification”. By the same token, the authoritarian regime in Chile was transformed into a democratic one in the early 1990s through a series of constitutional amendments. While this experience demonstrates how an authoritarian constitution can change to a democratic one, the use of amendments of the previous constitution in order to achieve the transformation, created an element of continuity with the previous authoritarian regime, which hindered the democratization and liberalization process. When amendment provisions are used for creating new constitutional regimes, not only important issues of legitimacy are raised, but there is also a difficulty in clearly breaking with the past regime’s constitution. As Bruce Ackerman urged post-communists countries not to conduct a series of constitutional amendments, rather “if the aim is to transform the very character of constitutional norms, a clean break seems desirable […]”. Thus nations may favor completely replacing an old constitution.

“At first blush”, William Harris comments, “the question of whether an amendment to the constitution could be unconstitutional seems to be a riddle, a paradox, or an incoherency. This problem is accentuated when one asks whether there is an agency that could make the determination”. As this article demonstrates, the statement “unconstitutional constitutional amendment” does not entail a paradox, but merely a misapplication of presuppositions. Once the nature of the amendment power is correctly construed, the paradox disappears.

Yaniv Roznai

Senior Lecturer, Radzyner Law School, The Interdisciplinary Center Herzliya (IDC).