Positivism and Unity

Abstract

This article examines the grappling of modern positivists with the question of legal unity. It presents and contrasts two antagonistic positivist strands—naturalist and normativist—epitomized in the works of Austin and Kelsen, respectively. The two strands correspond to two contrasting models of legal authority—criterial and coherence-based—and they accordingly diverge on the proper explanation of unity. Naturalist, criterial models purport to explain the unity of law based on extra-legal facts alone; normativist, coherence-based models resort strictly to the interrelation of legal elements themselves. Against this backdrop, the article argues that Raz’s work on the subject is torn between Austin and Kelsen: While his naturalist ancestors accounted for legal unity externally, Raz’s prominent works are captivated by the Kelsenian realization that the unity of law must be accounted for internally. Two central upshots follow. First, the analysis provides a litmus test for the (in)ability of naturalist positivism to explain legal unity. Second, Raz’s strategic reliance on Kelsen distorts his work: Both the Grundnorm and validity chains are upended to fit Raz’s naturalist commitments.

Introduction—Legal Unity: The Question

The unity of legal systems has long tormented legal positivists. The facts seem plain enough: Legal systems exist; they each consist in the various rules and principles comprising them, and in these alone; and the belonging of a certain rule or norm to a certain system normally entails that it is not part of a separate, different one. Yet what is it exactly that unifies conglomerates of rules to legal systems? Indeed, while the question of unity has drawn the attention of luminaries such as Jeremy Bentham and John Austin, as well as more modern thinkers such as Hans Kelsen, HLA Hart, and Joseph Raz, it is not an easy one to articulate clearly, let alone satisfactorily answer.

First, it is not immediately clear what the notion of ‘unity’ means. What exactly is one after, when one attempts to explain the unity of legal systems? Positivist jurisprudence commonly holds that the question is what transforms an assemblage or a batch of norms to a normative system or order. In his formative The Concept of a Legal System, ¹ Raz argued that the problems of identity (viz. of how we can tell if legal element n is part of a system s) and structure (viz. of the patterns relating various ns in s) are two of the four central pillars of the analytical study of legal systems. ² The useful distinction notwithstanding, the two problems are related: Since Raz believes that laws can validate each other to thereby interrelate juridically, the structure of a legal system potentially affects its identity—as in the case of an English rule mandating reliance on an EU law rule, or an Israeli rule stating that marital disputes shall be settled based on Jewish Law. Conversely, we can ask why a certain legal norm belongs to a certain legal order; this way of putting the question emphasizes the type of connections, if any, between the norms comprising the system. As will be seen, positivist theories diverge on their answer to this question.

A second ambiguity arises because there are at least two ways of accounting for legal phenomena, generally, and for the unity of law, specifically. On the one hand, traditional positivist jurisprudence has long prided itself on being “realist and unromantic in nature,” seeing “the existence and the content of the law as a matter of social fact.” ³ On this view, legal phenomena are to be explained by recourse to natural facts alone; I shall thus refer to this tradition as naturalist positivism. On the other hand, some positivist thinkers believe that while law and legal phenomena surely are related to natural facts, they cannot fully be accounted for by recourse to such facts. In this vein, in the Introduction to the English version of the first edition of The Pure Theory of Law (PTL), Stanley Paulson, an expert on Kelsen, writes that the Pure Theory differs from naturalist positivism in light of its “normativist dimension.” ⁴ According to Kelsen, the Pure Theory’s purity is to be secured not merely against the infiltration of moral facts into our explanations—a commitment shared by naturalist positivism—but also against “the claims of a so-called ‘sociological’ point of view, which uses the methods of the causal sciences to appropriate the law as a part of nature.” ⁵ To wit, the strand that I shall term normative positivism holds that law and legal phenomena are not to be confused with morality or fact, as law has a specific meaning of its own.

The two positivist strands correspond to two ways of explaining unity. External explanations see the unity of law as rooted not in the interaction of the elements of a given system, e.g. rules or court decisions, but in the shared way of their coming into existence. The external conception accordingly highlights the pedigree of law rather than the relations between its different parts; as such, it is not necessarily systematic. Internal explanations, in contrast, designate the kind of unity which transpires based on the interrelations between legal elements themselves. Harkening back to the opening questions, while external explanations must provide a unifying ground for law that is not part of it, they are not committed to any kind of relations between legal elements. In contrast, internal explanations eschew the invocation of any foreign component to consolidate law, but they are committed to an account of the type of interrelations between legal elements that would culminate in a unified system.

It is a central claim of this essay that the correspondence between naturalism and external explanations, on the one hand, and between normativism and internal explanations, on the other hand, heralds two distinct models of legal authority. Take first Austin’s Command Theory. According to it, all the rules of a legal system S are bound by their provenance, i.e., the fact that they are all commands issued by the sovereign. This explanation echoes a model of legal authority that I designate criterial. Moving to the other end, according to the Pure Theory, a plurality of norms forms a unified order if the validity of all the norms can be traced to a single ultimate norm. Kelsen’s rejection of naturalism shifts the attention to the systematic relations between norms; it accordingly gives rise to an alternative model of legal authority, which I shall call coherence-based. Notably, the two models represent two distinct visions of law. While criterial models connote a centralized, vertical governance structure, coherence-based models imply a more decentralized structure of legal authority.

Yet this bivalence notwithstanding, contemporary positivist theory is almost exclusively naturalist, to the point of flatly ignoring normative positivism. What is more, on the rare occasion that Kelsen is discussed, his theory is taken to be a version of naturalist positivism. This characterization belies not only the methodological divergence between naturalist and normative positivism, but, more crucially, the divergence in the models of authority corresponding to each strand.

The reader might wonder, however, why we need to discuss Kelsen’s theory anyway. The traditional naturalist view, according to which the unity of law is secured by recourse to a (factually grounded) rule of recognition, seems perfectly plausible. Kelsen’s contributions—inasmuch as they are not confounded by the mystifications of his own Pure Theory—can be, and in fact have been, accommodated within the traditional model, particularly in the early work of Raz. Why do we need, then, an alternative explanation of the unity of law? Three reasons support the exposition.

First, while the traditional explanation of unity is usually taken as correct, it hardly follows that it gets everything right or that it cannot be improved. Indeed, Raz’s continuous engagement with the topic itself attests both to the importance of the question and to the difficulty of doing it full justice. While his contributions over the years, especially in The Concept and Practical Reason and Norms, ⁶ have generated what is undeniably the most comprehensive modern treatment of the question of unity, close examination of his theory shows that it fails to account for important legal phenomena. I will argue that it falters in explaining the complex processes of law implementation typifying modern legal systems: Particularly, it cannot account for the many instances of purposive interpretation by courts and administrative bodies. Kelsen’s theory, in contrast, boasts resources for accounting precisely for such interpretive activity, without succumbing to the Dworkinian temptation of arguing that, legally speaking, every dispute has one right answer. In expounding the Pure Theory, this article does not suggest that it is necessarily superior to the traditional theory. Rather, it aims to frame the controversy between the strands, and thereby contribute to a more rigorous conversation on the limits of prevailing positivist theory—a conversation that has already been sparked by recent takes on Kelsen, most notably by David Dyzenhaus and Lars Vinx. ⁷

However, and this is the second central contribution the article ventures, this alternative has effectively been gutted. Modern positivist theory, that is, takes Kelsen to be a ‘neo Austinian,’ and it thereby not only misunderstands the Pure Theory and its view of unity, but also obscures an alternative vision to the authority model undergirding naturalist positivism. Indeed, while traditional jurisprudence regularly treats Kelsen as a command theorist, this label can scarcely be further from the truth. In fact, Kelsen’s is a dynamic, coherence-based theory of the positivity of law that challenges the very foundations of the naturalist hegemony. In this vein, the discussion purports to demonstrate how the treatment that the Pure Theory has undergone, especially in Raz’s works, has all but mutilated it. This is seen in the common jurisprudential assumption that the only coherence theories are content-based ones, e.g., Ronald Dworkin’s Law as Integrity or Ernest Weinrib’s Formalism—both of which are obviously dismissed as natural law theories. Kelsen, however, is anything but a natural law theorist, and his effort shows that a coherence-based version of positivism may well be worth our attention. Recovering his theory from the distortions that characterize the Anglo-American engagement with it, then, seems warranted in and of itself.

Finally, articulating the Pure Theory’s unorthodox view allows us to present the trajectory of an influential version of modern positivism, reflected in Raz’s thought. Initially, in The Concept, Raz revolts against the Command Theorists’ purely external understanding of unity, and puts forward a novel explanation that combines external and internal elements. This reflects Raz’s ambivalence: On the one hand, Raz subscribes to the importance of internal explanations, yet on the other hand, he does away with any element that cannot be accommodated within a naturalist framework, such as the Grundnorm or Kelsen’s principle of coherence. He then revisits this early criterion, in the subsequent Practical Reason, and updates it to a predominantly internal explanation which is nonetheless still naturalist. Ultimately, in Between Authority and Interpretation, ⁸ he goes even further, now explicitly incorporating coherence as well as doubling down on the internal systematicity of law. However, while the story of Raz’s grappling with the unity conundrum brings naturalist positivism closer to Kelsen, the transformation is hampered by the very different views both writers have on the ability of law to predetermine normative content, viz., to guide behavior.

The article proceeds as follows. Section I spells out the connection between naturalist positivism and external explanations, and cashes it out in terms of the criterial model of legal authority. Section II delineates the main tenets of the Pure Theory, and substantiates the connection between normative positivism, internal unity, and coherence-based models of authority. Against this backdrop, there arises a problem for Raz, for his theory attempts to link a naturalist approach with an internal explanation. After setting up this dilemma, Section III purports to show how it undermines Raz’s early theory of unity, which turns out to oscillate between Kelsen and Austin. A conclusory Epilogue then argues that the problem persists in Raz’s two later works.

Section I—Naturalistic Positivism and the Criterial Model

1. External Explanations: Austin’s Command Theory of Law

The Command Theory explains the unity of law using the doctrine of sovereignty. Let us look a bit more closely at the theory and what this implies.

Austin’s theory holds that a legal system contains all and only the laws issued by one person (or body of persons). Hence, a given law belongs to that legal system which contains laws issued by the legislator of that law. ⁹ However, a legal system only exists if “the common legislator of its laws is a sovereign”—namely if, in point of fact, the person or group issuing the laws lives up to the conditions of sovereignty. ¹⁰ Following his intellectual predecessor, Jeremy Bentham, Austin explains:

If a determinate human superior, not in a habit of obedience to a like superior, receive habitual obedience from the bulk of a given society, that determinate superior is sovereign in that society, and the society (including the superior) is a society political and independent. ¹¹

We can complement this passage with Austin’s related discussion of the sources of law. In a central sense, Austin defines a source of law as “its direct or immediate author.” ¹² Yet he also makes clear that “either directly or remotely, the sovereign, or supreme legislator, is the author of all law; and all laws are derived from the same source.” ¹³ That is, Austin obviously was aware of the fact that not all the laws of a given legal system are commands issued by its (current) sovereign. But this would entail that some laws are at the same time (1) part of the legal system, yet (2) not the sovereign’s commands, thus defying Austin’s criterion. Hence the italicized part in the passage just quoted, alluding to the idea of indirect legislation or tacit command. Raz explains:

The sovereign … is the direct or indirect legislator of all the laws in the system. The important fact is that when we trace the source of the laws of a system we end with one person (or group) who is the ultimate source of every one of them. ¹⁴

In Raz’s reconstruction of Austin, a unified legal system is an assemblage of laws, all of which have been directly or indirectly issued by the same legislator—assuming that this legislator is in fact (still) sovereign and can enforce the laws. If they are, the unified assemblage of laws will be a valid legal system.

Thus we have the basic ingredients of Austin’s doctrine of unity: First, every law can be traced—directly or indirectly—to an act of legislation; ¹⁵ second, the ultimate origin of all the laws is the legislative activity of one and the same person or group, who is the ultimate source of legislation; and third, the laws thus enacted are in force so long as this ultimate source continues to be sovereign, thereby consolidating them to a valid system. ¹⁶

The unity of an Austinian system is thus fully accounted for based on the origin of its laws. For this reason, Raz ascribes to his theory the principle of origin. According to it, “the membership of laws in a system, and the identity of the system, are completely determined by the origin of the laws; the origin of a law being the set of facts which brings it into existence.” ¹⁷ Austin’s doctrine is therefore a paradigmatic case of external explanations: By relying solely on the origin of laws, it explains unity strictly on factual grounds. ¹⁸

Notably, the reliance on a purely external explanation allows the Command Theory to dispense with any necessary relations between laws. Hence Raz also attributes to Austin’s theory the principle of independence. According to it, a legal system can be unified as a mere heap, viz. even if there is no connection between its laws other than their shared source. ¹⁹ This entails that Austinian laws are self-standing, as it were: “The fact that every law is a command entails that every law can be an independent unit, the existence, meaning or application of which is not logically affected by other laws.” ²⁰ Raz does not spend too much time discussing the implications of this feature. But as will be now argued, it is in fact central for the model of authority undergirding Austin’s theory.

2. Criterial Models

If a legal system is unified purely externally, it is committed to the principle of independence. If so, its unity is not a function of internal relations between its laws, and, for the same reason, the validity of any law in the system does not stem from other laws. Thus, purely external explanations correspond to what I shall call ‘criterial models.’ ²¹

Under a criterial model, establishing the validity of a certain legal element means investigating if it conforms to extralegal criteria. The criteria may vary between different systems, and they may resort to all kinds of facts, be they the say-so of the elderly or the enactments of the Queen in Parliament. Yet whatever their content, it is always the case that the criteria logically precede—and accordingly are independent of—the individual elements that they validate. From the standpoint of accounting for the validity of a certain element, the criteria are taken as fixed: If a given law adheres to them, it is part of the system, and is therefore valid; otherwise, it is not. There is no reciprocal influence between the laws and the criteria. As the criteria are factually grounded, they are not a function of the validity of this or that particular legal element, and they remain intact even if a given law, or a bunch of laws, fails to conform to them. ²²

Austin’s theory is a case in point. A law belongs to an Austinian system, and is thus valid, only if it can be shown to live up to the criteria that that system presupposes. For Austin, these include the conditions defining sovereignty and those constituting sovereign acts: Any law that has been enacted by the sovereign is a valid part of the system, insofar as the sovereign remains habitually obeyed by the bulk of the relevant population. Accordingly, if for some reason the supreme legislator ceases to be sovereign, all the elements in the legal system instantly lose their validity. ²³

But we can easily come up with counterexamples to this analysis, already alluded to: Some laws clearly do not conform to the factual criteria set by Austin. How, then, can his theory explain their validity, as parts of the legal system? One possibility, which Austin adopts in the case of constitutional laws, is to revoke the validity of such laws altogether. Hence Austin concluded that some rules of constitutional law that cannot be shown to be enacted by the sovereign, such as those setting up law-making procedures, are not laws properly so-called at all; they are rather part of the community’s positive morality. ²⁴ Alternatively, as in the case of customary law, one has to conjure up explanations for the validity of apparently unenacted laws. Here Austin’s solution resorted to the doctrine of tacit adoption.

The doctrine sought to explain how deviant sources of law, such as custom and judicial legislation, can be reconciled with the Austinian criterion that all laws are commands of the sovereign. Simplifying, the idea is that the sovereign ‘acquiesces’ in the application of customs and laws that were enacted judicially, by accepting them as instances for the exertion of sanctions. Since the sovereign is aware of common-law adjudication and customary law—and backs both sources by exerting sanctions upon defiance—they must have tacitly embraced them. Hence, Austin concluded, they too are admitted as valid parts of the legal system. Yet, as several critics have noted, this solution in fact equivocates on the relation between the sovereign and the deviant sources, as it is not clear precisely what the conditions for ascribing to the sovereign such tacit recognition are. ²⁵

Raz, however, wants Austin to go further. First, in the spirit of this last misgiving, Raz argues that Austin misused the doctrine, because he (Austin) did not plausibly explain the circumstances under which hypothesizing a tacit command is acceptable. But second, and more importantly, Raz thought that the doctrine is unnecessary for Austin. As mentioned, in Raz’s reconstruction of Austin’s theory, the important thing is that, when one traces the source of every law in the system, one eventually ends up with an ultimate source that binds all of them. This allows Raz to supplant tacit adoption with the idea of indirect or “non-ultimate” legislation, relying on what he dubs “obedience laws.” ²⁶ Importantly, obedience laws are only hypothetically laws. ²⁷ They have not been enacted, but are assumed to exist by virtue of their function (had they been enacted, one would not have resorted to fictions). According to Raz, such rules grant legislative powers to actors which are thereby designated ‘non-ultimate’ legislators. A non-ultimate legislator is competent to legislate if, and only if, there is a law that confers upon them the power to make law, by instructing the relevant group of subjects to obey their commands. Since obedience laws are presumed to have been enacted by the sovereign, they amount to indirect commanding; hence any non-ultimate legislation would be valid via its indirect relation to the sovereign.

Obedience laws, let me reiterate, were not part of Austin’s own exposition. In fact, as Raz notes in passing, “Austin never explains how exactly delegated legislation takes place.” ²⁸ But contrary to Raz’s assumption, this is not necessarily an omission on Austin’s part. For, if one admits hypothesized laws into the Command Theory, laws can then be validated by recourse to other laws without any positing of this validating relation, i.e., absent any discernable act on the part of the sovereign. This drifts even further away from the logic of an Austinian system, which severs questions of validity from necessary relations between laws.

More schematically, we can say that Raz’s suggestion impairs the direct relation between the sovereign and all the system’s laws. It thus turns an immediate criterial model into a mediated one. An immediate model grounds each legal element directly in facts, and it is therefore necessarily an instance of a purely external explanation. Mediated criterial models, however, allow for the invocation of intermediate elements—like unenacted laws—that are themselves not factually grounded amidst accounting for the system’s unity, insofar as all legal elements are ultimately accounted for in factual terms. Such models thus correlate to explanations that are ultimately, yet not purely, external. To understand why Austin might not welcome such alteration, we need to flesh out the latent relation between criterial models and the more general aspirations of naturalist positivism.

3. The Political Underpinnings of Naturalist Positivism

It was Kelsen who noted that the naturalist concept of law is an upshot of the triumph of 19th century bourgeois liberalism over the then-reigning monarchical and religious orders. ²⁹ The victory marked the beginning of an outspoken reaction against the metaphysics of the time and, specifically, against natural law theory. With the progress of the empirical natural sciences and the breakdown of religious ideology, ‘bourgeois legal science’ shifted to legal positivism—the naturalist strand. As part of this shift, naturalist positivists, especially Bentham and Austin, molded their theories of legal authority with an eye to their political ambitions. As Jeremy Waldron explains, naturalist positivism “represents not merely a descriptive but also a normative impulse.” ³⁰ Its proponents accordingly were not entirely neutral when they identified parliamentary sovereignty as the prime source of law. Rather, “[t]hey valued the opportunity that legislation represented as a way by which the community … could take control of its law as something explicitly man-made.” ³¹

Put differently, naturalist positivists have long been committed to the idea that law is a mundane, social phenomenon, devoid of the mystifications and enchantments imparted upon it by natural law thinkers. This commitment might explain Austin’s reluctance toward admitting any unenacted laws into his theory—even at the cost of equivocation. Such an uncompromising stance may well reflect the insistence that no metaphysical entities, emerging from thin, unenacted air, should be allowed to infiltrate the positive law of the people by way of so-called conceptual derivation. Solutions such as Raz’s put this commitment at stake, as laws and relations between them might be deduced as valid parts of a legal system without any purposeful human action.

This analysis is corroborated when we unearth the linguistic assumptions of naturalist positivism. We saw that immediate criterial models operate based on the principle of independence, namely, they forego any necessary internal relations between laws. This was captured in Raz’s previous claim, that “[E]very law can be an independent unit, the existence, meaning or application of which is not logically affected by other laws.” ³² If the meaning and application of every law are not a function of other laws, it follows that so is the execution of laws. This claim rests on the assumption, also famous in Hart, that the core of a legal rule is generally speaking sufficiently clear and distinct from its penumbral areas, allowing for the application of laws as a matter of course. ³³ In the vast majority of cases, then, implementing a legal rule does not require interpretation. Accordingly, as Hart noted,

The picture that the command theory draws of life under law is essentially a simple relationship of the commander to the commanded, of superior to inferior, of top to bottom; the relationship is vertical between the commanders or authors of the law conceived of as essentially outside the law and those who are commanded and subject to the law. ³⁴

This centralized structure of legal authority is tightly connected to the character of law-implementation under the Command Theory. First, the immediacy of the relation between commander and commanded guarantees that no unenacted elements surreptitiously gain legal validity without a corresponding positing act. Second, for Bentham and Austin, it had to be the case that law is the unmediated sovereign’s command, for otherwise questionable mediators—read: Common law courts—might thwart the salutary implementation of Utilitarian ideals that they wanted to promote through enlightened legislation. ³⁵ The centralized verticality of criterial models therefore goes well with the claim that the laws posited by the sovereign are clear, logically independent commands. Issued immediately from ruler to ruled, they neither require nor admit interpretive mediation by courts (or other interpreters).

But whether or not Hart and Raz are taken to endorse similar motives, it must be understood that their model too relies on a similar linguistic fulcrum. According to Hart, legal rules are applied simply by “classifying particulars,” and only “penumbral” applications of laws—the rare situations where legal rules lack clarity—call for interpretation, under the famous positivist doctrine of extralegal discretion. ³⁶ Accordingly, the execution of a law is generally taken to be a simple task, necessitating little if any interpretative leeway. This assumption clearly undergirds Raz’s Concept, and it reinforces the criterial nature of his theory. ³⁷ So while one should be careful not to overlook the differences between mediated and immediate models, one should also bear in mind that both are naturalist, criterial, and ultimately resort to external explanations of unity.

To recapitulate, classic naturalist positivists reacted against the metaphysics of natural law theories by grounding law directly in facts. They have done so by devising a centralized, vertical model of legal authority that is criterial. The immediacy of their criterial model is premised, first, on the idea that for the most part, the sovereign commands subjects directly, and second, on the notion that the commands thus posited can be implemented to subjects using little if any interpretation. Laws are assumed to be fixed, clear rules, unified into a system based solely and directly on their provenance—save for specific exceptions, to be accounted for by the sovereign’s implicit recognition of customary law and common law adjudication. For the same reason, any discretion granted to official interpreters of the law is necessarily confined to the penumbra. In the normal case, valid laws are implemented to concrete cases based simply on applying the general terms of the law to the case at hand. The linguistic stipulation, concerning the core of settled meaning of legal rules; its corollary, the sidelining of interpretation to penumbral cases; and the principle of independence, corresponding to a purely external explanation of unity, thus go well under a criterial framework.

Raz’s restructuring of Austin, pushing the Command Theory from an immediate criterial model to a mediated one, is therefore far from trivial. For classic naturalist positivists such as Austin, legal unity was surely a feature pending explanation, but hardly did it epitomize the phenomenon named ‘law.’ Sufficed it to systematize law externally; insofar as laws are enacted by one sovereign, the legal system exhibits all the unity necessary. For Austin, the crucial point was that all the system’s laws are grounded in facts, barring the creation of unenacted elements, and narrowing the ability of official mediators to interfere with parliamentary legislation. Raz’s reconstruction of the Command Theory adds a mediated, internal layer to Austin’s immediate, purely external explanation. As will be seen, this apparently innocent revision heralds a profound change in the structure of legal authority envisaged by naturalistic positivism.

Section II—Normative Positivism and the Coherence-Based Model

1. The Pure Theory: A Scientific Theory of Law

To understand the internal explanation adopted by Kelsen, and its correlative model of authority, it must be understood that his theory differs in kind from naturalist positivism. Yet as the Pure Theory has been the target of much misconception and confusion—at least partly, I contend, due to its treatment by naturalist positivists—it is worthwhile to open up by briefly sketching its central tenets. The point is not so much to embark on a full-blown discussion of Kelsen’s intricate thought, as it is to provide the reader with the theoretical essentials that will allow the examination of his contrasting theory of unity.

The Pure Theory is a scientific theory of law. It aims at an accurate cognition of its object, which is a legal system, and attempts to demonstrate that such knowledge is possible. ³⁸ In exploring these conundrums, the Theory does not confine itself to a particular system. Rather, it is a general theory that takes as its object the phenomenon of law, as in a unified system or order of norms, wherever it takes place.

The scientific posture of the Theory informs its methodology: The basic principle is that law is to be understood strictly legally. Put differently, the Pure Theory sees law as an ideal reality, to be kept distinct from both morality and the natural reality. ³⁹ Law is the kind of object that can be understood purely in its own terms and, therefore, must be so understood, if true insight into its nature is to be gained. Hence the scientific purity of the Pure Theory rejects all external explanations of law, moral or natural in character.

This claim Kelsen sometimes puts as the idea of freeing the legal norm or ‘ought.’ Against the moral standpoint, the legal norm must be kept apart from the notion that the ultimate goal, the ideal, of legal norms is justice. ⁴⁰ Despite ostensible similarities, the legal norm differs in kind from its moral counterpart. First, legal norms exist, or are valid, only by virtue of their being validly posited; a legal norm’s validity is not assessed based on the strength of its content, i.e., on whether it realizes some external moral ideal. ⁴¹ This idea echoes one of Kelsen’s more famous claims, namely that law can have any content: No human behavior is excluded a priori from being the content of a legal norm. Second, legal norms must be separated from their moral equivalents because, unlike the latter, the former are not simply imperatives.

This last claim also comes out of the separation of the norm from natural reality. The legal norm, Kelsen contends, must be distinguished from the factual circumstances bringing it about, as an ‘ought’ can never be derived from an ‘is.’ Thus, the norm cannot be equated with the set of mental events or physical processes that bring it into existence. ⁴² In contrast to Raz’s description of the Pure Theory, then, ⁴³ Kelsen is quite explicit that norms are not imperatives. ⁴⁴ Rather, a norm is the elemental unit that facilitates the translation of natural reality to legal reality. This process is clarified using a second distinction, between a natural act and the legal meaning imparted upon the act. Kelsen explains:

There is an act perceptible to the senses, taking place in time and space, an external event, usually an instance of human behavior. And there is a specific meaning, a sense that is, so to speak, immanent in or attached to the act or event. People assemble in a hall, they give speeches, some rise, others remain seated—this is the external event. Its meaning: that a statute is enacted. ⁴⁵

Another way at the contrast follows the central principles undergirding the natural science and the legal one. If a material event, taking place in space and time, is to be known as an object of scientific cognition, it must be ordered within a systematic understanding. Here Kelsen follows Kant, to argue that such understanding unifies natural phenomena based on laws of nature. ⁴⁶ Laws of nature link a certain material fact as cause with another material fact as effect, e.g., when we say that A happened because B did. ⁴⁷ Thus, material facts and events are part of nature inasmuch as they are governed by the principle of causality. Analogously, legal norms connect a certain fact—most often, human behavior—with a sanction. ⁴⁸ If the principle tying the occurrence of a natural fact to that of another is causal, the relation between a legal condition and a legal consequence relies on a normative principle. What renders an act legal, therefore, is not its physical being, but its interpretation as legal:

The specifically legal meaning of … [an] act is derived from a norm whose content refers to the act; this norm confers legal meaning to the act, so that it may be interpreted according to this norm. The norm functions as a scheme of interpretation. ⁴⁹

Thus, our conclusion that a given natural act is the execution, say, of the death penalty, rather than simply murder, is the result of the process of interpreting this act based on the criminal code. ⁵⁰ In this sense,

Just as the chaos of sensual perceptions becomes a cosmos, that is, “nature” as a unified system, through the cognition of natural science, so the multitude of general and individual norms, created by the legal organs, becomes a unitary system, a legal “order,” through the science of law. ⁵¹

It follows that while the natural and normative principles are similar in form, the type of connection they establish is different. While the causal principle says: “If A is, then B is (or will be),” the normative principle says: “If A is, then B ought to be.” ⁵² Unlike laws of nature, then, legal norms are hypothetical judgments: A natural law stipulates that under certain conditions, a certain effect is to happen, while a legal norm stipulates that under certain conditions, a certain effect ought to happen. ⁵³

The corollary is that just like we can comprehend nature, as a unified system of actual judgments we can comprehend law, which is a unified system of hypothetical judgments. ⁵⁴ The unit facilitating this legal understanding is the norm, which links an abstractly defined legal condition—a subject’s behavior—with an abstractly defined legal consequence—a sanction. We can thus make sense of law and legal phenomena based on the unification of legal norms.

The exposition conditions the unique understanding of unity envisaged by the Pure Theory. Legal unity is not the unity of a set of commands to be accounted for externally, based on this or that factual condition. Rather, it is a unity of hypothetical judgments, and it is thoroughly cognitive in nature. We interpret the relations between the manifold general and particular legal norms, to constitute them as an interrelated order that transforms natural acts and events to legal ones. The Pure Theory thus relies on an explanation of unity that is qualitatively different from the external, factual one.

2. Internal Unity and the Source of Law

The Pure Theory calls us to explain the unity of law in purely legal, i.e. internal, terms. This task Kelsen frames as an inquiry concerning the dynamics of the norms that culminate in a unified legal system. It is closely related to the question: “Why is a norm valid?” ⁵⁵ and, as mentioned, this question is related to the question: ‘Why does a legal norm belong to a legal system?’ According to the Pure Theory, the reason for the validity of a norm is that it is part of a valid order. A valid order is one all the norms of which can be traced to the same basic norm. If this basic norm is taken to be valid, then so are all the norms that can be traced back to it. Accordingly, a legal norm is part of a certain order if its validity can be traced to the basic norm unifying that order.

Kelsen’s answer to the unity question, one might worry, seems quite similar to Austin’s, unifying laws if their creation can be traced back to the ultimate law-creating authority. What is more, Kelsen is not always careful to clearly distinguish the terminology and mechanics of his internal explanation from those of an external one. Particularly worrying is his tendency to refer to the Grundnorm as a “source” of norms, for example in the following passage:

[The basic norm] qua common source constitutes the unity in the plurality of all norms forming a system. That a norm belongs to a certain system follows simply from the fact that the validity of the norm can be traced back to the basic norm constituting the system. ⁵⁶

In light of such paragraphs, it is perhaps unsurprising that at least in Anglo-American jurisprudence, The Pure Theory is usually read as a (misguided) version of naturalist positivism. ⁵⁷ Can one really blame commentators like R.K. Gooch who, in equating the analytical projects of Kelsen and Austin, argue that “it seems fair to say that Herr Kelsen is essentially a neo-Austinian”? ⁵⁸

Especially problematic in this connection has been Raz’s engagement with Kelsen, which effectively assimilated his theory into the naturalist strand. ⁵⁹ First, as noted earlier, by designating the Pure Theory an “imperative theory,” despite Kelsen’s insistence that norms are not imperatives but hypothetical judgments. But second, and even more importantly, by recasting it as a “theory of a common source.” ⁶⁰ In Section I, we saw that Raz took Austin’s theory to be a variant of such theory: ‘The important fact’ about Austin’s unity criterion, Raz stressed, ‘is that when we trace the source of the laws of a system we end with one person (or group) who is the ultimate source of every one of them.’ Similarly, by presenting the Grundnorm as the common source of all the system’s norms, Raz equates Kelsen with Austin:

Austin thought of a legal system as the set of all the laws enacted, directly or indirectly, by one sovereign. Kelsen substitutes the basic norm for Austin’s sovereign, and leaves the rest of the definition unaltered: A legal system is the set of all the laws enacted by the exercise of powers conferred, directly or indirectly, by one basic norm. ⁶¹

But this claimed similarity is doubly misleading. First, as we saw, Raz’s recasting of the Command Theory as a mediated criterial model might not be welcomed by Austin. Second, Kelsen emphatically rejects the concept of a ‘source’ of law as a fixed set of criteria from which legal norms derive, since the Pure Theory’s basic principle forbids any derivation of norms from facts. ⁶²

Yet this seems to fly in the face of passages such as those quoted: What is Kelsen talking about, then, in discussing the Grundnorm “qua common source”? The answer is to be found in the epistemic character of the Pure Theory. For Kelsen, the unity of a plurality of norms is a function of our ability to cognize the relation of each norm to its validating counterpart, all the way to the Grundnorm. But it does not follow that, similarly to the criterial model, the Grundnorm emanates, and thus logically precedes, the norms of the system. In fact, the order is reversed: The Grundnorm is presupposed not prior to the validity of the system’s norms, but because they must be understood as valid to make sense of the legal activity already taking place, viz. if they are to facilitate the interpretation of (actual) acts and events as objectively legal. Putting it somewhat cryptically for the moment, the efficacy of a legal order logically precedes the presupposition of its basic norm. This slogan relies on Kelsen’s explanation of law as a hypothetical order, comprising two elements: A chain of validation, and a Grundnorm. A close examination of these will clarify that the Grundnorm is the ‘source’ of the system’s norms in a sense very different than Raz’s.

3. Validity Chains and the Grundnorm

a. Two Types of Normative Validity

A chain of validation, or validity chain, connects valid norms to each other. The idea is that validity is an ‘inheritable’ quality: A norm can validate another, rendering it part of the same order. Kelsen distinguishes two modes of validation in normative systems, corresponding to two types of relation between norms: Static and dynamic.

Static validation characterizes systems of moral norms, where the upper-level norm maintains a substantive relation, as a general to particulars, to the lower-level one. Hence, moral norms are validated based on their content: “The human behaviour specified by these norms is … obligatory because the content of the norms has a directly evident quality that confers validity on it.” ⁶³ Static derivation is therefore an act of intellect. It consists in analyzing the premise so as to infer the subsidiary norm, similarly to the logical relations between factual propositions, where the premise already contains all the content that can be derived from it. If an upper-level norm mandates truth-telling, extracting it to derive a more concrete norm—e.g., forbidding fraud in business transactions—relies on static validation. ⁶⁴ Accordingly, the lower-level norm is deduced, rather than created, by the applying organ. An upper-level norm therefore supplies both (1) the reason for validity, and (2) the content of the norms derived from it.

Dynamic relations, on the other hand, are formal. A formal process does not assume an upper-level norm that encompasses every norm derived from it. Rather, dynamic relations consist in the authorization to create lower-level norms. An upper-level norm contains not the content of the norm derived from it, but the manner of its creation. Likewise, the basic norm in a dynamic order contains only “the authorization of a norm-creating authority or … a rule that stipulates how the general and individual norms of the order based on the basic norm ought to be created.” ⁶⁵ Such a basic norm supplies only the reason for the validity of subsidiary norms, but not their content.

For Kelsen, a legal order is a dynamic order. A legal norm is not valid because it has a certain content, deducible from a validating norm. Rather, it is valid because it was created in a certain way—as prescribed by its validating norm and, ultimately, in the way set by the basic norm: “For this reason alone does the legal norm belong to the legal order whose norms are created according to this basic norm.” ⁶⁶ Here is, then, the reason that any content can be law: Legal norms are created by a specific process, and they are valid solely because thus posited, namely, as elements of a positive order.

We can flesh out the dynamic relations typifying such an order by revisiting a previous example. If we ask: ‘Why is this hanging a legal act and not murder?’, the answer is that the act can be interpreted as legal if it was prescribed by an individual legal norm, viz.—if it ought to have been performed, say based on a judicial decision. And if we ask further, ‘What allows this judicial decision, mandating the execution of this death penalty?’, the answer is that the decision, creating the individual norm authorizing the execution, was issued in accordance with the general criminal law norm, which prescribes that the death penalty ought to be inflicted given such and such conditions. And the criminal law itself is valid because it was created by the legislature, in line with the constitutional authorization, and so on. ⁶⁷ This tracing back of norms goes all the way to the basic norm, which is the reason for the validity of all norms traceable to it.

b. The Grundnorm: A Hypothetical Basic Norm

If we attempt to understand an act of hanging as the execution of the death penalty—rather than as murder—we need to trace back the act to the valid legal norm authorizing it. Since the legal order is hierarchical, this process can be thought of as taking place gradually, in the manner just described. But the question may be raised as to the validity of the upmost echelon: What is the reason, one might ask, for the constitution’s own validity? It is of course scarcely helpful to reply that the constitution has been authorized by a prior constitution, e.g., a national convention or a constituent assembly. Immediately the question could be raised again, as to the validity of this historically first constitution.

Kelsen’s answer to this endless regress is to be found in the hypothetical nature of the legal order. The validity of the historically first constitution must be presupposed if we want to interpret the acts performed according to it as the creation of valid general legal norms, and the acts performed in application of these general norms as the creation of valid individual legal norms. ⁶⁸ This necessity, we should recall, is epistemic. According to the Pure Theory, the raw data of ostensibly legal acts and processes is given, and the theory assumes that systematic sense can be made of these seemingly chaotic circumstances. ⁶⁹ For this reason, the basic norm always refers directly to a specific constitution that is both actually established and which is, by and large, efficacious. ⁷⁰ Put differently, only once we have the raw data that could potentially be unified as a legal order can we postulate, at its apex and as its ultimate reason for validity, a valid Grundnorm. Thus, the basic norm is the “transcendental-logical condition,” allowing the interpretation of the subjective acts of law-creation as objective legal acts. ⁷¹ Just like we interpret the natural facts perceived by our senses using the laws of nature, so it is possible to interpret the subjective meaning of certain facts as a system of objectively valid legal norms, described in rules of law. ⁷²

This conditional character of the Grundnorm, unifying norms as hypothetical judgments, is fleshed out by Kelsen using the limiting case of revolution. According to the principle of legitimacy, a legal norm is valid until its validity is either terminated in a legally determined way or replaced by the validity of another norm of the order. This principle, however, does not apply in the case of revolution, defined as a change of the constitution or its replacement by another “in a manner not prescribed by the constitution valid until then.” ⁷³ Often in these situations, we say that most of the statutes created under the old constitution remain valid. But this, Kelsen notes, is inaccurate: If the statutes are to be regarded as valid under the new constitution, they must have been validated by it. The content of the statutes remains unchanged, but the reason for their validity has changed: “As the new constitution becomes valid, so simultaneously changes the basic norm,” viz., the presupposition allowing us to interpret the act of constitution-creation and its derivatives as objectively valid, or legal. ⁷⁴ Suppose, tells us Kelsen, that an old constitution had the character of an absolute monarchy, and that the new, post-revolution constitution is one of a parliamentary democracy. The new basic norm “does not make it possible—like the old one—to regard a certain individual as the absolute monarch,” but it does make it possible to regard the democratically elected parliament as a legal authority. ⁷⁵ Thus, “The change of the basic norm follows the change of the facts that are interpreted as creating and applying legal norms.” ⁷⁶

The Grundnorm, then, is in no way a ‘source’ of law in a Razian sense. It is the necessary normative fulcrum without which no sense could be made, no cognizing would be possible, of the multitude of acts and processes typifying a complex of rule-based coercion as a legal system. It is accordingly structured hypothetically: If we are to be able to interpret natural facts and events as legal, it behooves us to presuppose an ultimate normative reason for the validity of the order. Therefore, no Grundnorm would be presupposed unless it could correspond to an at least partly efficacious order of norms. If no one obeys the orders of officials, if norms are meaningless within a certain community, then no minimal efficacy exists, and no unified sense could be made of the quasi-legal acts—which turn out to be nothing but acts of force, rather than instances of legal enforcement.

This analysis undercuts one of Raz’s central arguments against the Grundnorm. In contrast to his assumption, it is not the case that the Grundnorm is designed to identify the (already existing) norms of an order. ⁷⁷ Rather, the Grundnorm is the logical condition for the possibility of rendering quasi-legal acts as objectively legal. One may, for example following Paulson, criticize Kelsen for presupposing that we indeed have sufficient knowledge of the data to be interpreted as legal. ⁷⁸ And whether the Pure Theory can accommodate this difficulty may be open to debate. ⁷⁹ But one cannot, with Raz, blame the Grundnorm for failing to identify pre-existing norms because, on Kelsen’s understanding, legal norms pre-existing the Grundnorm is confusion in terms. ⁸⁰

Raz is correct, however, to note that Kelsen does not tell us exactly what efficacy means. ⁸¹ Arguably, one way at it denotes the grip that the normative system has on a population. We know that this grip requires enforcement, but it cannot be reduced to it. ⁸² It is also clear enough that for norms to be valid—i.e., to denote the objective sense of ‘ought’ interlinking an act and its legal consequence—it must be the case that they are actually applied; namely, that the consequences indeed take place following the acts on which the consequences are conditioned. But it must also be the case that we can sensibly interpret the law-implementing acts as legal. In our modern cognition, for this to happen there must be in place at least preliminarily a recognizable system of officials who interpret and execute the lower-level norms based on the authorization set in upper-level norms. If a person presenting themselves as an ‘official’ enters my house without uniform, a warrant, or any explanation, and demands that I pay them what they call a ‘fine,’ I could hardly make legal sense of the demand. The demand does not seem to follow from a binding norm, nor is it performed in a manner generally perceived as legal. I can trace the act to no recognizable authorizing norm, and, therefore, there is no justification for presupposing a basic norm that could validate it. Hence, the act would be taken not as an objective application of law, but as a subjective instance of power.

The crucial point is that the interpretive act of tracing norms’ validity back to the Grundnorm assumes an already ongoing (quasi-) legal conversation. Only once a (purported) legal order is at least minimally effective, in the sense explained, can we interpretively trace norms to their validating counterparts and to a Grundnorm. Hence, it is utterly misguided to view the Grundnorm as the ‘source’ of the system’s norms, if by ‘source’ one has in mind authoritative criteria from which law emanates down to subjects. For the same reason, it is hardly an argument against the Grundnorm that it fails to identify already existing legal norms. Such an argument assumes a model of authority alien to the one undergirding the Pure Theory, in which legal norms can be known as such only after a Grundnorm is presupposed.

4. The Coherence-Based Model and the Dynamics of Legal Interpretation

We now delineate the contours of Kelsen’s coherence-based model. Unlike criterial models, it does not see validity as wafting down from the external authority at the apex of the system. Rather, validity is a function of normative relations: If a norm coheres with other valid norms, it is part of a valid unity, and is itself valid. Valid law is therefore the set of those norms that cohere with one another and are unified by a valid Grundnorm. The relevant benchmark for assessing a norm’s validity, then, is always to be found in another norm. As positivism assumes that legal norms can be created, amended, or repealed on demand, the standard for norms’ validity is not fixed. ⁸³

Since for Kelsen legal relations are dynamic, that a norm ‘coheres’ with another means that the validated, lower-level norm lives up to the dictates of its validating, higher-level counterpart. But as the foregoing discussion makes clear, this precept should immediately be qualified as epistemic: We must be able to interpret the totality of norms as comprising a hierarchical order of formal validation. Since the basic norm of a legal order is thoroughly dynamic, our cognizing of the relations between norms cannot be static: Lower-level legal norms cannot logically be deduced from the basic norm. This brings Kelsen to one of his more dramatic claims: The issuance of a legal norm relies on (the interpreter’s) will, not intellect. ⁸⁴

Consider how the Pure Theory characterizes adjudication as the implementation of general norms, e.g., statutes. As a hypothetical judgment, a general norm attaches an abstractly determined consequence, viz. a sanction, to an abstractly defined material fact, viz. human behavior. Implementing a general norm means fleshing out its normative meaning in the concrete case. ⁸⁵ This requires that the fact which the general norm determines abstractly be ascertained as present in the actual dispute, and that the abstractly defined punishment shall too be concretely determined. ⁸⁶ Accordingly, any general legal norm requires individualization, which is the task of the judicial decision. ⁸⁷ Notably, the judicial function is not declaratory:

The act of the court is not simply a matter of pronouncing or discovering the law already complete in the statute, the general norm. Rather, the function of adjudication is constitutive through and through; it is law creation in the literal sense of the word. ⁸⁸

The court must create the individual norm according to which this defendant must be punished with this specific sanction; and this individual legal norm did not exist prior to the court’s determination. ⁸⁹ The determination becomes a link in the chain of dynamic law creation, concretizing the legal norm as it journeys down from the upper, abstract echelons of the system. ⁹⁰

This brings us to interpretation, which plays a dual role in Kelsen’s theory, corresponding to two modes of, or points of view on, the legal order. ⁹¹ He distinguishes between authentic and non-authentic interpretation. ⁹² The second kind of interpretation is the province of the legal scientist trying to make sense of rules of law, describing the norms of the system as seen in a snapshot. This static view is important for inquiring into positive law, as well as for the legal subject, say when they are trying to understand the system ‘in rest.’ Our discussion focuses on the first kind, which connotes the operation of legal authorities within the order dynamically, i.e., from the standpoint of law as an order constantly in the making. Authentic interpretation is the creative interpretation exercised by the organs actually applying the law, e.g., a court’s decision applying a statute. The court’s decision creates a lower-level norm, based on the authorization set in the statute. This individual norm is the implementation of the abstract, general norm in the concrete case, its meaning in the actual dispute before the court.

Here we reach the essence of the dynamism of a legal order. On the one hand, the relation between a higher- and lower-level norm is one of determining or binding. The higher-level norm regulates the procedure for creating the lower-level norm, and also, possibly, its content. But on the other hand, and this is the crucial point, this determination is never complete. ⁹³ There must always remain less or more room for discretion on the part of the implementing organ, as even the most detailed command must leave some discretion to the executing party. Kelsen gives the example of a norm decreeing the arrest of a suspect:

If official A orders official B to arrest suspect C, B must use his own discretion to decide when, where, and how he will carry out the warrant to arrest C; and these decisions depend upon external circumstances that A has not foreseen and, for the most part, cannot foresee. ⁹⁴

In contrast to the linguistic clarity stipulated by criterial models, the inherent indefiniteness of norms under the Pure Theory entails that interpreting a norm can only mean the ascertainment of a ‘frame.’ Several possibilities are open for applying the norm, and, so long as the applying act remains within the frame, it is validly authorized, i.e., legal. ⁹⁵ Interpretation does not lead to a single decision as the legally correct result, as if the content of the validating norm need only be teased out, intellectually, by the interpreter. Rather, interpretation entails several options of equal value, one of which—the one chosen by the applying organ—will become positive law. Hence, that a judicial decision is based on a statute means that it keeps inside the frame; not that it is the individual norm, but only that it is one of the individual norms that may be created.

It follows that law implementation is necessarily law creation. The means of creation is legal interpretation, which exists at every stage of implementation. ⁹⁶ Creative interpretation, then, is not confined to the ‘penumbra’: Even the so-called ‘core’ of a law requires interpretation to be implemented in a concrete case. The corollary, to quote Dyzenhaus, is that “law made at the higher levels of the legal hierarchy … [is] progressively concretized as it journeys down to the moment of actual application to a particular legal subject.” ⁹⁷ Yet the Pure Theory’s dynamism runs even deeper. We conclude this Section by demonstrating this rather radical feature, using Kelsen’s discussion of the (im)possibility of conflict between norms. ⁹⁸

5. Coherence and the Possibility of Norm Conflicts

A legal order represents a hierarchy of upper- and lower- level norms. A norm belongs to such an order only because and insofar as it accords with the higher norm that determines its creation. But what is the law if a lower-level norm does not conform to the norm that prescribes its creation and, especially, its content?

Surprisingly, Kelsen believes that such conflicts are merely apparent: If a norm truly were in conflict with a norm that determines its creation, it could not have been regarded as valid. Such a “norm” would be null, which means that it would be inexistent, and hence—no norm at all. ⁹⁹ The positive law, however, does not take such norms to be null, precisely because it normally includes procedures to annul them: “If a legal order authorizes the annulment of a norm it must first recognize this norm as an objectively valid, that is, lawful, legal norm.” ¹⁰⁰ Accordingly, even if a court’s decision, creating an individual legal norm, is deemed ‘unlawful’ by a higher court authorized to rescind it, this only means that the norm was annullable, i.e. that it could have been invalidated by a legally regulated procedure. ¹⁰¹

What is more, if the procedure for annulment has been exhausted, and the individual legal norm can no longer legally be questioned, the deviant norm would nonetheless remain valid. The same holds for all final decisions in a legal order:

[E]ven if a general norm, to be applied by the court, is valid, which predetermines the content of the individual norm to be created by the court, an individual norm, created by the court of last instance, can become valid whose content does not conform to this general norm. ¹⁰²

Alongside the general norm that predetermines the content of the judicial decision, then, there also exists an implicit norm, according to which the court may itself determine the content of the individual norm that it is to create. The court of last instance is therefore authorized to create either (1) an individual norm the content of which is predetermined by the general norm, or (2) an individual norm whose content is not so predetermined but is to be decided by the court itself.

In the same vein, the fact that the decision of a court of first instance is annullable means that courts are authorized to create individual norms, whether their content lives up to the content predetermined by the general norm or not. The only qualification is that the decisions of courts other than the court of last instance are provisionally valid—if a legal process allows that they may be rescinded by a higher instance—while the decisions of the court of last instance are definitively valid. Put simply, so long as a decision has been issued by a court (or any other authorized organ), it cannot be unlawful. ¹⁰³ It is valid until rescinded, and, if it cannot be rescinded, until its validity is terminated by another legal norm. Kelsen concludes that there can be no genuine conflict between the individual norm, created by the court based on the general statutory norm, and this latter general norm.

The notable corollary is that the possibility of predetermining the creation of individual norms by courts is “very limited,” ¹⁰⁴ particularly with regards to the norms’ content. The same applies to any apparent conflict between upper- and lower-level norms throughout the order:

[N]o conflict is possible between statute and judicial decision, constitution and statute, constitution and ordinance, statute and ordinance, or, formulated generally: no conflict is possible between a higher and lower norm of a legal order, which would destroy the unity of this system of norms by making it impossible to describe it in noncontradictory rules of law. ¹⁰⁵

Because presupposing the validity of a legal order means presupposing a Grundnorm that coherently unites it, it is impossible for genuine conflicts between norms to plague such an order. Had such conflicts been real, either the order could not have been formed, or the particular conflict could not have been cognized as part of it. Unfortunately, the cognitive part of Kelsen’s theory is completely sidelined by Raz who, in a rather laconic fashion, dismisses both coherence as a necessary requirement for the unity of normative systems and its corollary, the radical indeterminacy of content relations under an epistemic view of unity. ¹⁰⁶ This dismissal, it will now be seen, has dramatic consequences.

Section III—Raz’s Predicament

1. Situating the Dilemma

Raz’s theory of unity has at least two distinct stages, which I shall dub an early and an updated one. These will be discussed in turn. This Section examines the first iteration, extracted from Raz’s Concept. It propounds two overarching claims. First, that Raz’s early theory struggles to account for legal unity. The reason is that his unity criterion attempts to reconcile two antagonistic tendencies: A naturalist outlook and an internal explanation of unity. As we saw, Raz sets the stage for this attempt by juxtaposing Austin and Kelsen’s theories as similar in the relevant respects. This leads to the second overarching claim, which is that this juxtaposition seriously distorts the Pure Theory.

In light of the preceding discussion, it is no coincidence that Austin’s theory adopts a purely external explanation of unity whereas Kelsen adopts a purely internal one. External explanations go well with a criterial model, allowing little if any intermediation between the commands of the authority and the implementation to subjects. Kelsen’s internalism, however, is emphatically non-naturalist, seeing legal unity as the upshot of a cognitive endeavor of interpretively interrelating lower- and upper-level norms into a single coherent order. It accordingly relies on the vast discretion of official mediators, not on the correspondence of norms to fixed, factual criteria bringing law about.

Raz’s Concept is interesting because it is the first prominent work that attempts to reconcile the tendencies. On the one hand, Raz’s early theory of unity is, of course, naturalist. Throughout his works, Raz is emphatic that law and legal phenomena are to be accounted for by social facts. This comes out both from his adoption of Hart’s rule of recognition (albeit with several modifications that needn’t detain us), ¹⁰⁷ as well as from his meticulous analysis of the structure of a law, following Bentham. ¹⁰⁸ It is also manifest in his adoption of the ‘basic power’ doctrine, discussed below. Yet on the other hand, Raz openly revolts against the apparent corollary of naturalism—external explanations of unity—for his theory is also clearly internalist. This is seen in the repeated comments throughout The Concept in which he disparages Austin precisely for neglecting internal relations between laws. “[I]t is impossible to grasp the nature of law,” Raz insists, “if we limit our attention to the single isolated rule. In a sense almost all the rest of this study [The Concept] is designed to convince the reader of the truth of this quotation.” ¹⁰⁹ The systematic nature of law, Raz explains, embodies at least part of its unified nature, and it also serves to distinguish it from adjacent systems of social ordering. ¹¹⁰ Focusing exclusively on external unity thus obscures law’s nature, as well as undermines our explanation of important legal phenomena. ¹¹¹ Finally, he commends Kelsen’s insistence on the validation of norms based on other norms as “a great improvement upon Bentham’s and Austin’s ideas,” and he incorporates validity chains—a clearly internal element—in his unity criterion. ¹¹²

The Concept, then, is both a proud link in the naturalist legacy, and the forerunner of a critique of one of its central commitments—the immediate relationship between laws and the factual criteria unifying them into a system. This duality is anything but trivial. Recall that, for Austin, sufficed it that law was unified externally; insofar as laws are enacted by one sovereign, the legal system exhibits all the unity necessary. The next phase saw Hart’s seminal account, rejecting the Command Theory for its inability to account for the continuity of legal systems, and stressing that ‘the key’ to jurisprudence lies in the unity between primary and secondary rules. ¹¹³ Yet Hart’s focus was the diachronic unity of law, i.e., its continued existence. Raz’s endorsement of Kelsen’s notion of internal relations should be understood as the continuation of Hart’s legacy, by way of highlighting the synchronic unity of law. Yet this undercuts a most fundamental tenet of classic naturalism, namely that law is a social, man-made phenomenon. For recall that a central aim of classic naturalists was to dispel the mysteries and enchantments of natural law, which served to hijack law from the people as well as to obscure its nature. Insofar as the internal unity of norms is put on the pedestal—authorizing the derivation of laws independent, at least partly, from the earthly enactments which are supposed to bring law about—law’s positive nature, at least as conceived by classic naturalists, might be slipping away. The rise of internal at the expense of external unity thus marks the culmination of a tectonic shift in modern positivist thinking.

We now turn to see how the conflicting tendencies interact in Raz’s early theory of unity. I argue, first, that the naturalist commitment colors Raz’s adoption of two of Kelsen’s most central doctrines; second, that this recasting of Kelsen’s theory distorts it; and, finally, that the same tendency prevents Raz from accounting for law implementation.

2. Between Austin and Kelsen

We start by presenting Raz’s early criterion of unity, the strands of which run through different parts of The Concept. It can be reconstructed as follows:

1. (1) A legal system consists of the first constitution and all the laws created, directly or indirectly, by the exercise of powers conferred by the first constitution;

2. (2) The basic power, directly authorizing this constitution, (a) is identified by not being created by any law, i.e., it is not conferred by another norm; and (b) is recognized as such by the courts; ¹¹⁴

3. (3) A law belongs to a legal system if and only if it is either part of the first constitution, or has been enacted by the exercise of powers directly or indirectly conferred by it, based on the doctrine of internal relations; ¹¹⁵ and

4. (4) An internal relation exists between laws if and only if one of them (a) is (part of) a condition for the creation of the other; or (b) affects the meaning or application of the other. ¹¹⁶

Raz’s is a mediated criterial model, combining internal and external explanations of unity. Externally, all the system’s laws have to be validated by the conferral of powers that ultimately depend on the basic power. Internally, all the laws that are not part of the first constitution are validated in recourse to it, through internal relations to other laws.

The criterion reflects Raz’s oscillation. On the one hand, following Kelsen, it puts a premium on the internal relations of laws in explaining the boundaries and the conceptual pattern of law. But on the other hand, it both curbs Kelsen’s understanding of internal relations and retains a strong Austinian flavor. This is seen in Raz’s commitment to naturalism which, at important points, entails centralized implications that fly in the face of Kelsen’s theory. The question, then, is whether Raz’s integration is tenable. We first examine the external part, then shift to the internal one.

a. The Basic Power and the Grundnorm

According to Raz’s early theory, the first constitution is authorized by the basic power (2), and this power is defined as a power not conferred by law (2)(a). A law belongs to a legal system if it was part of the first constitution, or has been enacted by the exercise of powers conferred by it (3). The conjunction of the claims reflects Raz’s naturalist commitments. Since a law belongs to the first constitution only if its creation was not authorized by any other law, the criterion secures a factual basis for the system. ¹¹⁷ The basic power is an external element, then, which raises the question: How does it come to be? Raz’s answer demonstrates his attempt simultaneously to draw on Austin and Kelsen.

According to Raz, the solution to the question of the apex of law resembles Austin’s, in that it identifies a legislator whose powers were not conferred by law. ¹¹⁸ Moreover, similarly to the sovereign, this basic authority is unlimitable. ¹¹⁹ At the same time—and here Raz explicitly turns to the Grundnorm—the basic power need not be “ever-present.” ¹²⁰ While the sovereign must remain habitually obeyed by the bulk of the population, “[a] legal system [based on the basic power] continues to exist even after the holder … of its basic power ceased to exist.” ¹²¹ This feature of the basic power is celebrated by Raz as Kelsen’s “main improvement” over Austin. ¹²² And it is this feature which, later on, will buttress Raz’s rejection of the Grundnorm, in favor of a group of self-referring laws. ¹²³

Yet Raz acknowledges that the Grundnorm is not identical to the basic power; ¹²⁴ nor does Kelsen mention this concept in his Pure Theory, let alone ground it in facts. Raz overcomes the hurdle by transmuting the Grundnorm to a static ‘source of law.’ This gambit is facilitated by the reconstruction of the sovereign and the Grundnorm as equivalent, viz. their conceptual juxtaposition. Austin’s solution to the unity question, Raz writes, “rests on the combination of two concepts: validity chains and sovereignty. Kelsen accepts the first, and thereby also the principle of origin, and rejects the second, substituting his own concept of a basic norm.” ¹²⁵ Hence the Grundnorm is to be understood the same way that Austin thought of the sovereign; it is precisely for this reason that Raz characterized both the Pure Theory and the Command Theory as theories of “common source.” ¹²⁶ That this argument distorts the Pure Theory should by now be clear enough. Raz converts the Grundnorm from an interpretative, normative postulate hypothesized in legal cognition to a (social-) fact-of-the-matter source of law. But the Grundnorm is not the ‘source’ of the system’s norms any more than the sovereign is a normative postulate hypothesized in legal cognition. For the same reason, it is not the ‘origin’ of the system’s norms, and, contrary to Raz’s stipulation, Kelsen hardly would have accepted the principle of origin.

If the basic power is genuinely to be the equivalent of the Grundnorm, then, it cannot be factually grounded. But if it is not, it can neither comport with Raz’s naturalist commitments, nor be modeled after Austin’s sovereign—and Raz is explicit that it is both. By drawing on two incompatible doctrines, then, the basic power reads as a placeholder, an equivocation in lieu of a clear grounding of the legal system.

But assume that the Grundnorm can indeed be equated with the sovereign. Still the difficulty persists because it remains unclear what it is that buttresses the basic power, when the sovereign—and, with it, the habitual obedience of the population—drops out of the picture. If Raz jettisons the sovereign, and declines Kelsen’s presupposition of a basic norm, what is it that does elicit the basic power? Raz’s direct answer is not altogether clear: “The improvement [that the basic power represents over the sovereign] is made possible by positing obedience to the laws, instead of obedience to the sovereign, as a condition for the existence of a legal system.” ¹²⁷ My own guess is that Raz is not yet sure how to explain the basic power, so he resorts to the recognition by courts (2)(b). After all, a court is an institution and recognition a material act, hence the naturalistic assumption is not flouted. This solution seems to resemble Hart’s, viz. the convergence of official behavior constituting the rule of recognition. But if that is the case, it becomes unclear why one needs a ‘basic power’ in the first place.

One possibility is that Raz believed that the basic power more naturally comports with Kelsen’s validity chains. This suggestion is supported by Raz’s further claim, that the shift from a lawmaker to a (factually grounded) power at the apex of the system underlines the importance of validity chains in explaining unity. ¹²⁸ If so, then the function of the basic power is to retain an external, factual fulcrum, while keeping alive the possibility of connecting this fulcrum to the rest of the laws through chains of validity. Put differently, the purpose of the basic power is to moor Raz’s early theory, which will momentarily be seen to rely heavily on internal relations, in a factual foundation that nonetheless secures its naturalist character.

Naturalism, however, points to a model directly at odds with Kelsen’s; his coherence-based model is interpretive, dynamic, and decentralized. Not only does this fact entail the equivocations just discussed, but it also leads Raz to distort the Grundnorm. He dismisses its logical-cognitive character, and the reader is left with the impression that the Pure Theory relies on a basic norm in a sense virtually identical to that in which the sovereign unites an Austinian system. For the same reason, Raz’s critiques of the Grundnorm miss their mark, grounded as they are in a naturalist environment alien to Kelsen’s epistemic normativism.

b. Internal Relations and Validity Chains

Let us now consider the internal element in the early criterion (4). It relies on a distinction between two kinds of relation between laws. According to it,

There are innumerable possible relations among laws, many of them of interest. In searching for the structure of legal systems we shall be concerned only with one kind of relation which we call an internal relation. An internal relation exists between laws if, and only if, one of them is (part of) a condition for the existence of the other or affects its meaning or application. ¹²⁹

Hence, interpretation laws are internally related to the laws they help interpret, as they affect their meaning. At the same time,

The relations between laws according to their social consequences are a very important kind of external relations. A local by-law, demanding certain sanitary conditions in pubs, complements, in different ways, both the by-law of the same local authority demanding similar conditions in restaurants, and the by-law of the neighboring authority which demands the same conditions in pubs. ¹³⁰

The first of these quotes indicates Raz’s adoption of validity chains, as part of his discussion of Kelsen’s view of the existence conditions of norms. ¹³¹ The discussion is presented as a rather innocuous reconstruction of the Pure Theory. ¹³² Raz follows J.B. Bryce’s study, ¹³³ and remarks that a chain of validity can also apply to Austin’s theory. The theory could then be represented in a validity-based tree diagram, but, instead of a normative postulation at the top (or bottom) of that tree, one replaces the Grundnorm with the sovereign.

The first thing to note is that this move may undermine Austin’s theory, because it turns his model from an immediate criterial model to a mediated one. ¹³⁴ Moreover, it is quite clear that the reconstruction distorts the Pure Theory. For Kelsen, legal validation denotes the interpretive tracing of each of the norms in a legal order back to its upper-level counterpart, and ultimately up to the Grundnorm. Internal unity, we saw, is a cognitive endeavor: The interpreter hypothesizes validity chains to make sense of the hierarchical order as a coherent whole. Put differently, validity chains are the result of our ability coherently to interpret the relations between upper- and lower-level norms. Accordingly, they reflect the dynamism of legal interaction, where the standards for norms’ validity are in flux.

Raz’s employing of validity chains presupposes a diametrically opposite logic. Under his mediated criterial model, the creation-conditions of norms are logically prior to the creation of norms themselves. Reversing the order is, of course, out of the question. The criteria do not change based on the creation of this or that lower-level norm, or, what amounts to the same thing, a norm could never come to be valid if its issuance did not adhere to the fixed mandate in its upper-level counterpart. ¹³⁵ Likewise, Raz does not even mention the epistemic dimension of the interrelation between norms. Just like the Grundnorm is warped to fit the naturalist disposition, so the doctrine of validity chains is transformed from an interpretive-based, cognitive element into a fixed, criterial conduit, transmitting authority from the basic power down to the lower norms.

It might be argued that this analysis mistakenly equates Raz’s theory with Austin’s. Even if Raz was wrong to conclude that Austin’s theory could incorporate validity chains, his early criterion can nonetheless accommodate them. But this claim overlooks the naturalist posture of Raz’s theory. This conceptual mismatch can be substantiated by visiting the very different views of Raz and Kelsen on the relations between upper- and lower-level norms.

Take first the case of subordinate legislation. In Raz’s reconstruction of Austin, a subordinate legislator exists only if there is a law conferring upon them the power to create law by certain acts. ¹³⁶ Hence, if an organ is authorized to create lower-level norms, it must be the case that the created norm does not exceed the limits of the delegated power. If a legislative body issues a law that it was not empowered to issue, “its unconstitutional command would not be legally binding, and disobedience to that command would therefore not be illegal.” ¹³⁷ Likewise, if we want to find out what legislative powers a subordinate legislator has, we must consult the laws conferring these powers upon them. ¹³⁸

Next, take Raz’s assumption that the content of the lower-level norm is determined by the intention of the creator of the upper-level norm. Again Raz argues that this position is shared by both Kelsen and the classic naturalists:

It is important to understand that it is the manifested intention that determines the content of the norm. If the act is performed with the intention of making certain people, x, behave in a certain manner, a, then the norm is that x ought to do a. This [purportedly Kelsenian] doctrine is strongly reminiscent of those of Bentham and Austin. ¹³⁹

More schematically, it is the intent of the creator of norm n ₁ , which authorizes an organ O to create n ₂ , which determines n ₂ ’s content. Similarly, when Raz discusses the kinds of act that can be seen as creating norms, he notes in passing that

A norm-creating act must, of course, determine at least in part the content of the norm created by it.… The norm investing an act with the character of a norm-creating act will indicate the exact way in which to interpret which norm is created by that act. ¹⁴⁰

Connecting these examples is the implicit assumption that the content of lower-level norms is on the whole determined by their upper-level counterparts. Yet this logic flatly contradicts Kelsen’s analysis, both generally and on the specific point.

c. Kelsen Transmuted

On this point, Raz himself admits that the intention Kelsen discusses is not an intention to affect the norm-subject’s behavior in the way the norm prescribes. Many times, Kelsen noted, the legislator does not even know the content of the law for which they voted. ¹⁴¹ How, then, can it be argued that the legislator intended to determine the behavior of the norm subject, i.e., that he predetermined the lower-level norm’s content? The simple answer is that Kelsen does not advocate for such a position, nor would he accept the claim that sub-legislation which did not conform to the authorizing law would be unconstitutional and invalid. ¹⁴²

This conclusion is vindicated by his more general views on the question of content (in)determinacy. Recall that dynamic validation is intimately related to the implementation of law. This process, we saw, relies on the creative function of official interpreters. When a court implements a general statute, it creates a lower-level norm based on the statute. But while the general statute normally determines the organ authorized to create the concrete norm, and, sometimes—though by no means necessarily—also the content of the norm to be created, the determination is never complete, and the court retains a significant degree of discretion. Accordingly, such authentic interpretation cannot elicit a single option to be applied as the lower-level norm. Implementing norms leads to the recognition of several possibilities each of which can legally be created in the concrete case, i.e., the mentioned frame. Insofar as the court chooses one of the options—insofar as it keeps within the frame—the lower-level norm would be valid.

Already at this point it is evident that Kelsen’s model contravenes Raz’s early theory. Yet we saw that even this does not exhaust the radical dynamism of the Pure Theory. For Kelsen also insists that lower-level norms can contradict their upper-level counterparts and, to the extent that the positive order does not rescind the created norm, it would remain valid. So much follows from Kelsen’s insistence that no conflicts between norms are possible, as the unity of the system is the necessary assumption regulating the legal order: If a norm has been issued by the authorized organ, it is valid—whether it appears to conflict with its upper-level counterpart or not.

To reiterate, this peculiarity relies precisely on the epistemic nature of Kelsen’s theory. It is because the relation between higher- and lower-level norms is essentially cognitive that the leeway granted to interpreters is considerably wider than that allowed to them in the centralized criterial model. For the same reason, unlike a Razian chain, a Kelsenian chain of validity would hold even if the lower-level norm created deviates significantly from the determination set in its upper-level counterpart.

d. Raz’s Early Criterion and Purposive Interpretation

Let us now revisit Raz’s internal relations distinction. Suppose that a case comes before a court which has to implement a statute to resolve the dispute. For Kelsen, implementing the norm relies on creative interpretation. The general norm is analyzed and a frame retrieved, from the various options within which the court has to choose one that would become the individual legal norm that settles the dispute (e.g., A owes B $100 in damages). Since the assumption is that every norm requires implementation, the court may well rely on adjacent rules that might seem relevant to the decision (as well as on extralegal materials).

Not so for Raz. If a dispute comes before a court, one of two possibilities obtains: We are either in the core of the governing norm, or in the penumbra. If the governing norm is clear, resolving the dispute amounts to applying the terms of the norm to the case, by means of “classify[ing] particulars.” ¹⁴³ Notably, Raz’s distinction entails that the purposes of adjacent rules—e.g., the ‘social consequences’ of by-laws—are considered external, and are excluded from the application process. If the court relies on inspiration from adjacent materials, then, it has exceeded the limits of its authority. Only if we are dealing with a penumbral case does interpretation enter the picture, allowing the judge to rely on whatever materials they think pertinent to the case, purposive relations included.

Raz’s early criterion, then, admits purposive relations between laws under one of two conditions: Either the case is a penumbral one, or a specific interpretation law has been posited. In all other situations—namely, in most situations—purposive relations between laws are defined as external, and one would not be able to argue that the shared purpose of two by-laws renders a suggested interpretation of a third by-law the preferred one. Similarly, finding that a given rule or directive is part of a legal body which is seen to promote a certain aim would not, in itself, provide sufficient grounds for reaching legal conclusions. And there can likewise be no instance where the meaning of a clear norm is updated based on changes in its legal surroundings, be they legislative or otherwise. Accordingly, the intuitive ‘hanging together’ of legal materials is considered external, while purposive interpretation, its natural companion, renders the court’s decision ultra vires. ¹⁴⁴

Yet we know well that judges and administrative agencies regularly apply law based on the assumption that a legal rule ‘lives’ in its habitat; it is a legal triviality that implementation relies, inter alia, on purposive relations. Yet for Raz such activity can only take place in the penumbra, and so his early criterion cannot explain the interpretive dynamics that make up the everyday operation of law.

Epilogue—Raz’s Updated Theory of Unity

1. Practical Reason and Norms

These problems, I believe, did not escape Raz. Nineteen years later, at the outset of Practical Reason, he maintains that the central task is—still—to explain what unifies norms into a system. ¹⁴⁵ He now zeroes in on the unique internal mechanics of normative systems, to put forward an impressive institutional theory of law. But a naturalist impulse still animates the analysis, reproducing difficulties similar to those discussed.

Legal systems, Raz argues, are the most important kind of institutionalized normative systems. These latter systems are characterized by three central features, which provide the first leg of Raz’s updated unity theory. First, they have a criterion of existence which does not require that all of their norms would be practiced, and which assigns “considerable weight” to the activities of officials and institutions. ¹⁴⁶ Second, all the norms of an institutionalized system are internally related to the norms setting up the norm-creating or norm-enforcing institutions of the system; accordingly, the rules of such systems are identified by their relation to the institutions which characterize them. ¹⁴⁷ Finally, the validity of norms belonging to such systems is conditioned by the system actually being practiced. ¹⁴⁸ The updated criterion thereby links the unity and the existence of the system, so that “[r]oughly speaking a norm belongs to a certain institutionalized system only if it was enacted by the organs of that system or is applied by them.” ¹⁴⁹

Based on this setup, Raz revisits the old problem of the primacy of either norm-creating or norm-applying institutions: To which of the two must a norm be related, to be part of an institutionalized normative system? Raz argues for the primacy of norm-applying institutions, particularly courts, which he dubs “primary” norm-applying institutions. ¹⁵⁰ Of interest here is his reasoning, which compares his theory with Austin and Kelsen’s. Again, ambivalence is present. On the one hand, Practical Reason represents an important development, towards a more decentralized model of authority. But on the other hand, since the updated criterion retains criterial commitments, it reinforces the distorted view of Kelsen.

Raz frames the question of the primacy of norm-creating or norm-enforcing organs as part of his discussion of “[s]ystems of a common origin.” ¹⁵¹ These have two versions: Hobbes and Austin’s version, and Kelsen’s version. The former defines a legal system as the set of all norms issued, directly or indirectly, by one legislator; the latter regards it as a set of all the norms deriving their legal validity, directly or indirectly, from one norm. ¹⁵² In keeping with his naturalist reading of Kelsen, Raz argues that despite several differences, both versions see every legal system as having a shared source: For Austin it is the Sovereign, for Kelsen—the Grundnorm. Such theories, Raz claims, cannot account for two major features of complex institutionalized systems: They fail to explain both their unity and their existence.

Focusing on unity, Raz claims that legal systems often include more than one ultimate source of legislative authority. ¹⁵³ No single norm can be said to confer powers on all legislators in a given system, because the scope and manner of exercising the different powers differ too much; there have to be at least two norms which do that. He gives the example of Britain, where the authority of Parliament is not derived from the common law, yet neither does the common law’s authority derive from Parliament. Essentially the critique is that theories of common origin assume too centralized an allocation of powers, an assumption which cannot hold for modern law. Legal systems cannot be explained by their common origin, then, and so both Austin and Kelsen’s theories must be rejected. Along with them, and without mention, the basic power doctrine, undergirding the early criterion, vanishes as well.

Two points should be noted. First, Raz replaces the basic power with the relation between primary institutions, the norms setting them up, and the norms applied by them. He relies solely on the rule of recognition as the fundamental apex of the system; though, unlike Hart, he allows that a system could have more than one such rule. In addition, he essentially retains his internal relations doctrine, and so while his updated criterion is still naturalist, relying on the coordinated action of primary institutions, it is much less centralized. ¹⁵⁴ The second point, however, is that it is ironic that Kelsen’s theory is rejected on grounds of its centralized character. Firstly, because, as was repeated time and again, the Pure Theory is simply not a ‘common origin’ theory. But secondly, even after foregoing the basic power, Kelsen’s theory is still far less centralized than Raz’s. It would thus be quite absurd to reject the Pure Theory on the pretext that it takes legal authority as wafting down from a single source, as if it were a criterial model. Moreover, the very creativity of law implementation under the Pure Theory means that for Kelsen, legislation and adjudication differ only quantitatively, not qualitatively; accordingly, the Pure Theory has no problem (epistemically) unifying all law-making authority under one Grundnorm. ¹⁵⁵

The first leg of the updated criterion grounded unity in the norms setting up norm-applying institutions, these institutions, and the norms they apply. To this claim Raz adds a second leg, by contrasting legal systems with “systems of absolute discretion.” ¹⁵⁶ In an institutionalized system, primary institutions combine norm-making and norm-application in a special way: They authoritatively determine normative situations in accordance with pre-existing norms. ¹⁵⁷ Notably, ‘authoritative’ here does not entail that a court can never err in determining a normative situation, but that its decision is binding even if mistaken. Yet whether or not an institution’s decision is mistaken, its powers must have been exercised by applying existing norms to the case. The presence of primary institutions with power to settle normative situations based on pre-existing norms indicates that the system provides for an institutionalized and authoritative way of resolving disputes. Systems of absolute discretion, on the other hand, include norm-applying institutions with unfettered discretion, which authoritatively decide disputes without recourse to any pre-existing standard; they are not even bound by their own decisions. In contrast, “[l]egal and similar systems … provide guidance for individuals,” based on the preexisting norms directed to norm-subjects. ¹⁵⁸ These norms guide behavior by equipping subjects with “rights and duties in litigation before the primary organs.” ¹⁵⁹ The corollary is that legal systems are coordinated systems of guidance and evaluation.

Does the complete updated criterion fare better in explaining law implementation? Unfortunately, the short answer is “No.” It is true that in the early parts of Practical Reason, the reader gets an impressive account of the mechanics of law based on the exclusionary nature of legal reasons. And the second leg provides important insight into the role of courts in Raz’s theory. Yet missing still are recognition of the necessity of interpretation to law implementation, and a discussion of the interpretative dynamics. These remain obscured in Practical Reason, and the work cannot in this sense be seen as an advancement over The Concept.

2. Between Authority and Interpretation

Surprisingly, things take a quite radical turn in Between Authority and Interpretation, Raz’s last book on the subject. I cannot here embark on a full discussion of the book. Instead, I shall confine myself to two remarks, venturing a somewhat inconclusive conclusion.

First, the main tenets of the updated criterion—relying on the internal relations between norms setting up primary institutions, the institutions themselves, and the norms they apply to subjects—remain intact. However, in the first sentence of Between Authority, we learn that “[g]eneral theories of law struggle to do justice to the multiple dualities of law. The law combines power and morality, stability and change, systemic or doctrinal coherence and equitable sensitivity to individual cases, among others.” ¹⁶⁰

Not only is this the first time Raz explicitly endorses coherence as part of his theory, but the quote also seems to suggest that interpretation now occupies center stage. And indeed, the third and last part of the book is dedicated to interpretation. Raz explains that interpretive action is central to law, and he also mentions the coherence of doctrine, though he does not explain if or how, in light of his internal relations distinction, legal interpreters can rely on coherence to derive legal conclusions in concrete cases. ¹⁶¹ He further refers to the fact that the pre-emptive nature of legal reasons is an upshot not of the operation of a single isolated rule, but rather

[I]t is the law, the legal system as a whole, which pre-empts those background considerations, not any of the legal rules taken singly. What the law requires, what rights it grants, what its conditions of responsibility etc [sic] are, depend on the combined effect of all its rules, though in each case only some of them are relevant. ¹⁶²

Should we infer that a judge deciding a case based on a clear rule may be inspired by the legal materials surrounding the rule? The answer is not altogether clear.

On the one hand, Raz stresses the necessarily interpretive nature of law as a coherence-based complex: “[M]any participants and close observers perceive the law as a complex web held together by subtle logic.” ¹⁶³ Accordingly,

[A]ny changes to the law should, other things being equal, be continuous with existing law. Typically we would expect any changes to fit well with existing legal defences, remedies, procedures, doctrines of legal competence and much else, besides having to cohere with existing law on the matter they deal with … or else they are likely not to achieve their aims, and are likely to undermine the working of other parts of the law. ¹⁶⁴

What is more, the interpretive action facilitating such coherence has been understudied: “There are aspects of legal practice in many countries which have often been neglected by philosophers. … Prominent among those is the way the law is interpreted by the courts.” ¹⁶⁵

Does it not follow that greater leeway is accorded to interpreters? Not necessarily, because while Raz understands well that the aspiration to coherence goes hand in hand with interpretive leeway, he is reluctant to take the extra step of allowing such unrestrained interpretive dynamics. He claims that the difficulty is that in interpretive reasoning, the distinction between law application and law creation is obscured, and this undermines theories about the nature of law. This problem, he insists, does not plague only positivist theories:

In recent times some so-called ‘legal positivist’ theories were sometimes accused … of inability to deal with interpretation because of the way they draw the boundaries of the law. This, I believe, misidentifies the difficulty. It afflicts any theory of law. ¹⁶⁶

Now Raz is correct that the difficulty has been misidentified to apply solely to positivist theories. But he is wrong to conclude that it afflicts any theory.

The problems that interpretation creates for legal theory—that of dissolving the dichotomies of making law and applying it, of what is part of law and what is not, and so on—primarily afflict theories that put a premium on content determination, viz., those that connect the authority structure of law to its content. The problem, in other words, is that despite highlighting the interpretive nature of law and the coherence of legal materials, Raz is unwilling to forego the guidance function. But if the law is to guide behavior it must do so uniformly, based on the moral considerations that it pre-empts. Interpretation must therefore be at least minimally standardized, and here Raz faces a problem: For it is very difficult, if possible, to tame the unruly interpretive horse.

Hence, Raz rightly notes that the difficulty also applies to Dworkin, and that accordingly his (Dworkin’s) theory cannot account for important legal phenomena. But unsurprisingly, he is silent about the Pure Theory. For if one, with Kelsen, relinquishes the presupposition that law guides behavior by predetermining content, or at least mitigates one’s expectations as to the possibility of such complete determination, many problems disappear. We then have no worry whether something is or is not part of law, for interpreters may draw on whatever content they deem relevant; and we care little if their activity creates law or merely applies it. We simply say that they do both, and that, in any case, the distinction between the two is exaggerated. The difficulties are mitigated, in other words, because the ‘subtle logic’ holding the web of law together turns out to be formal, not substantive.

The real issue, then, is whether we are ready to subscribe to a purely posited view of law, which takes it as a formal construct of interrelated authorizations, rather than ties law in through content. The question for theories such as Raz’s, then, is whether, at every instance of its (re)making, any content truly can be law. Clearly, Dworkin has no problem answering in the negative. ¹⁶⁷ But can a self-proclaimed positivist resort to a similar answer?

The suggestion might strike many as unpalatable. If law cannot provide standardized guidance, if it really is unified only by virtue of its manner of issuance, why would we even want law? What is the point of having, and theoretically expounding, a social institution which, so far as governing social relations and solving coordination problems go, is all but random? These are difficult questions, but I think that they overstate the difficulty. For Kelsen argues not that content determinacy is completely unattainable, but that we must be more realistic about its prospects. Accordingly, his unity criterion does not provide a carte blanche to any implementation dynamics whatsoever. It is quite obvious that there is a limit on one’s ability to epistemically interrelate norms, and if a certain assemblage of norms shows no connection whatsoever, Kelsen would hardly claim that it can be unified. Rather, he seems to have held that complete predetermination—in the sense of controlling the most intimate details—of any normative situation in advance is probably the result of wishful legal thinking. So while the law can provide some guidance, amongst other functions that it may secure, it cannot be reduced to its guiding function; it cannot be construed as necessarily and invariably achieving this aim, come what may.

Let me stress that the point is not to argue that determining legal content is never possible, nor is it to argue that Kelsen can iron out all the quirks in his own theory. Rather my purpose is to flag the difficulty that content predetermination poses, and to insist that, in contrast to this naturalist assumption, Kelsen has at least attempted to provide an alternative view which frees legal theory from the predicament of accounting for predetermining content. This may not be a neat view of law. It may undermine many ideals we would like to ascribe to this institution, democratic ones included. But it may well be the only coherent way to think about its unity.

This leads to the second point, which harkens back to an admittedly cryptic claim in the Introduction, and which we can introduce as our final question: What is the general trajectory of Raz’s thought? While the later developments evidently drive Raz further away from Austin, would it be accurate to describe the updated criterion as bridging the gap between Raz and Kelsen? The answer, it seems, is yes and no. On the one hand, Raz is willing to admit heretofore unheard-of elements such as doctrinal coherence, as well as to rely more heavily on the notion of a “legal point of view”—both of which seem congenial to Kelsen’s epistemic, normative understanding of law. ¹⁶⁸ But on the other hand, Raz’s updated thought retains its external tendency; only this time, it is closer to morality. Raz now argues that law is part of morality; ¹⁶⁹ that, when legitimate, it provides reasons because it “instantiates certain background moral reasons;” and that, in such cases, “[law] is an applied moral system, always to be developed along moral lines.” ¹⁷⁰ True, the law retains independence to the extent that its pre-emptive force relegates moral (or otherwise normative) discourse from the application of legal rules. But he also makes clear that the separation is in no way hermetic; legal discourse, Raz claims, is moral discourse. It goes without saying that such claims are anathema to the Pure Theory, which vehemently separates the legal from the moral ‘ought,’ and which repeatedly warns against the assimilation of the ideal of law into that of morality, namely, justice. ¹⁷¹

It follows that Raz’s theory remains externally grounded—albeit morally no less than factually; rather centralized—though significantly less than Austin or Hart’s theories; and hesitantly interpretive—though it does allude toward interpretation and doctrinal coherence as conceptual parts of law. The only truly unequivocal commitment, beyond these dualities, is to the claim that the law, conceptually, guides the behavior of subjects.

In the final analysis, then, the true divide between positivist theories turns on the nature of law as an institution that must be able to guide behavior. For this reason, Raz’s updated theory comes closer to Dworkin’s than to Kelsen’s. And so the latter’s exhortations, that the modeling of law after morality collapses positivism into natural law may, after all, be vindicated. ¹⁷²