No question in jurisprudence is older, or more contested, than the deceptively simple one: what is law? Salmond called jurisprudence "the science of the first principles of the civil law", and the first of all first principles is the definition of law itself. Naturalists answer that law is the dictate of reason; Austin that it is the command of a sovereign backed by sanction; the realists that it is what courts in fact do; and H.L.A. Hart that it is a structured union of primary and secondary rules. Each definition captures a real feature of legal order and misses another, which is why a judiciary or CLAT-PG aspirant must hold all four in view rather than memorise one. This article maps the four dominant conceptions, tests each against its standard criticisms, and anchors the abstract debate in Indian constitutional authority where natural-law reasoning has visibly shaped the law.
Why the definition of law matters
Jurisprudence does not state rules of contract or tort; it asks the prior question of what makes any rule a legal rule at all. As our introduction to jurisprudence explains, the discipline investigates "the nature of law, its sources, and an analysis of legal concepts" — and every one of those inquiries presupposes a working definition of law. Get the definition wrong and the whole edifice tilts: if law is merely a sovereign's command, then international law and customary law cease to be law; if law is only what courts do, then a statute is not law until a judge applies it. The definitional debate is therefore not academic hair-splitting but the load-bearing foundation of the subject.
Three features recur in every serious attempt to define law and supply a useful checklist. First, law is normative — it tells us what ought to be done, not merely what is. Second, law is institutional — it issues from, and is administered by, recognised authorities rather than private opinion. Third, law is generally coercive — non-compliance attracts an organised response. The four theories examined below each privilege one or two of these features. Austin foregrounds coercion, the naturalists foreground normativity rooted in morality, the realists foreground the institution of the court, and Hart attempts a synthesis in which rules of different kinds supply all three. The art of answering a definition question is to show why no single theory exhausts the concept.
Law as the dictate of reason: the natural-law conception
The oldest definition treats law as the dictate of reason. On this view law consists of principles of justice and morality deduced from the objective moral order of nature, discoverable by natural reason and common sense. The Stoics gave the idea its classical form — "man should live according to nature", since man is by nature endowed with reason, so that true law equals right reasoning. Cicero distilled it into the formula that law is "right reason in agreement with nature", universal, unchanging and everlasting. Medieval natural law, above all in Aquinas, located human law within the eternal law of God knowable by reason, and the modern, secularised version in Grotius held that natural law would bind "even if we were to concede that God does not exist". The deep treatment of this tradition lives in our dedicated note on the natural law school across the Greek, Roman, medieval and modern periods.
Two merits explain the theory's staying power. It supplies a superior standard against which positive law can be judged: when an enacted law falls short of an ideal, people appeal to a higher law, and the rallying cry becomes lex injusta non est lex — "an unjust law is no law at all". And it grounds natural rights: equality, the naturalists say, rests on nature itself, for "a dwarf is as much a man as a giant is". The standard criticisms are equally familiar. Natural law tells us how people ought to behave but is frequently not followed in practice; it lacks universality, since slavery was lawful in Greece and Rome and moral contents vary across states (monogamy here, polygamy there); and courts find moral standards difficult to apply to concrete disputes. Yet, as the next section shows, the natural-law instinct has never died — it resurfaces wherever judges insist that law must be just.
Natural law in Indian constitutional reasoning
The natural-law conception is not a museum piece in India; it is woven into the most consequential decisions of the Supreme Court. In A.K. Gopalan v. State of Madras, AIR 1950 SC 27, the Court initially read Article 21 narrowly, holding that "procedure established by law" required only a procedure enacted by a competent legislature and refused to import the American "due process" standard — a positivist reading in which the legislature's command sufficed. That austere positivism was decisively rejected in Maneka Gandhi v. Union of India, AIR 1978 SC 597, where the Court held that any procedure depriving a person of life or personal liberty must be "right, just and fair" and not arbitrary, fanciful or oppressive. The injection of fairness and reasonableness into the bare words of Article 21 is natural-law reasoning in constitutional dress: an enacted procedure that is unjust is treated as no law at all for the purposes of Article 21.
The same instinct animates the basic-structure doctrine. In Kesavananda Bharati v. State of Kerala, AIR 1973 SC 1461; (1973) 4 SCC 225, a thirteen-judge bench held by 7-6 that Parliament's amending power under Article 368 cannot be exercised to damage or destroy the "essential features" or basic structure of the Constitution. The notion that there exist inviolable principles which no sovereign legislative act can touch is, in jurisprudential terms, a natural-law limit on positive law — and it is precisely the kind of restriction Austin's command theory cannot comfortably explain, as we will see. For the aspirant, the lesson is that the natural-law/positivism debate is live Indian doctrine, not foreign abstraction.
Law as command: Austin's imperative theory
The sharpest rival to natural law is the imperative or command theory, whose classic statement belongs to John Austin and which forms the core of the analytical, imperative school. Austin defined positive law as the command of a sovereign backed by a sanction, identifying three characteristic features. First, law is a species of command: a wish, issued by a determinate superior, that another behave in a certain way, coupled with the power and intention to inflict an evil if the wish is disregarded. Crucially, to qualify as law a command must be general, directed at a course of conduct rather than a single act — Austin's own example is that a ruler's order to a servant to shut a door is a mere desire, not law. Second, law emanates from a political sovereign — a determinate person or body that receives habitual obedience from the bulk of society but is not itself in the habit of obeying anyone else. Third, law is enforced by a sanction: for Austin it is the fear of the threatened evil that gives a command its obligatory force, so a command without a sanction is no law.
Austin's achievement was to separate the question "what is law?" from "what ought law to be?" — the hallmark of legal positivism. By insisting that the existence of law is one thing and its merit another, he gave jurisprudence a clear analytical method and a workable criterion for identifying the law of a system: trace the chain of commands back to the sovereign. This "law as it is" approach made the theory rigorous and teachable, and it captures an undeniable truth — that law characteristically emanates from an authority and is visited with a penalty.
The collapse of the command model
Powerful as it is, Austin's theory has been heavily attacked, and the criticisms repay close study because examiners love them. Customary and international law fit badly: custom is law though no sovereign commanded it, and on Austin's logic international law — which issues from no determinate superior — would not be law at all, which contradicts reality. Power-conferring rules are not commands: the rules that confer the right to vote, to make a will or to contract enable people to do things rather than ordering them about, and threatening no sanction they escape Austin's mould. Persistence of law embarrasses the theory too: laws continue in force long after the death of the law-giver who supposedly commanded them, and some constitutional provisions cannot be changed by the present sovereign at all.
The Indian material sharpens this last point. The basic-structure limitation in Kesavananda Bharati means that even Parliament — the apparent sovereign — cannot validly command an amendment that destroys the Constitution's essential features; and under judicial review courts routinely declare legislative "commands" void, which is impossible if the sovereign's word is by definition law. Hart would later show that Austin's "gunman writ large" mistakes being obliged (coerced by a threat) for being under an obligation (bound by a rule). And yet, as the notes fairly record, "this theory contains a lot of truth": law genuinely does emanate from, and is enforced by, an authority, and no rival explains that coercive dimension better. The honest conclusion is that command is a necessary but insufficient ingredient of the concept of law.
Law as the practice of courts: Salmond and legal realism
A third conception shifts the focus from the sovereign to the court. Sir John Salmond defined law as "the body of principles recognised and applied by the State in the administration of justice" — the rules recognised and acted on by courts of justice. Whereas the positivist locates law in the will of the legislature, the realist locates it in the medium through which that will is actually expressed and enforced: the judicial process. Salmond's point is practical. All law, however made — by legislation, custom or precedent — is in the end recognised and administered by the courts, and rules the courts do not recognise are simply not followed. Therefore, to know the true law, look to what the courts do, not merely to what the statute book says.
The American realists pressed the same insight harder. Oliver Wendell Holmes Jr. declared that "the life of the law has not been logic: it has been experience", and reduced law to prediction: "the prophecies of what the courts will do in fact, and nothing more pretentious, are what I mean by the law." On this "bad man" account, a person who cares only for consequences looks to the courts to learn what will actually happen to him, and it is the judges, exercising discretion as they interpret and apply texts, who "put life into the dead words of the statute". Legal realism thus pictures the judge not as a mechanical mouthpiece of the legislature but as a genuine, if constrained, maker of law — an emphasis that won a major victory in the United States and that resonates with the candid acknowledgement in Indian and English practice that case-law is judge-made.
Testing realism: certainty, statutes and tribunals
Legal realism's court-centred definition has its own weaknesses. It risks affecting the definiteness of law: if law is only a prediction of what judges will do, then strictly speaking no one can ever say what the law is, only guess what a court might do — yet in daily life the law is treated as known and settled. It also struggles with statute law: a statute is operative the moment it is enacted and does not wait for judicial recognition; courts apply statutes because they are law, not the other way round. Salmond's answer is that so long as courts and legislature work in harmony it scarcely matters whether we say the statute is law because the courts recognise it or the courts apply it because it is a statute — though the problem bites when a court strikes a statute down as void.
Two further objections complete the picture. There are, in the modern state, agencies other than courts — administrative tribunals, quasi-judicial bodies, and (in England) the House of Commons enforcing its own privileges — that recognise and apply law, so a purely court-focused definition is too narrow. And realism gives only a partial perspective: to know the law fully we must consult statutes, rules and regulations, not case-law alone. The enduring value of the theory, however, is that it exposes the reality behind the words — that statutes, rules and regulations come alive only through the courts that interpret and enforce them. Realism is best read not as a complete definition but as a corrective to the legislative formalism of Austin.
Law as a system of rules: Hart's synthesis
The most influential twentieth-century answer is H.L.A. Hart's, in The Concept of Law (1961). Hart agreed with Austin that law is a matter of rules and that law and morality are conceptually distinct, but he rejected the command model as crude. His central move is the union of two kinds of rule. Primary rules impose duties — they tell people what they must or must not do, like the rules forbidding theft or violence. Secondary rules are "rules about rules": they confer powers, public and private, to create, modify, adjudicate and identify the primary rules. A society with only primary rules would suffer from three defects — uncertainty about what the rules are, staticness because there is no way to change them, and inefficiency because there is no authoritative way to settle breaches.
Hart's three secondary rules cure exactly these defects. The rule of recognition supplies an authoritative test of validity, telling officials which primary rules count as law and so curing uncertainty — it is, in effect, the master rule of a legal system. Rules of change empower a legislature or other authority to enact and repeal rules, curing staticness and making the system dynamic and responsive to social change. Rules of adjudication empower courts to determine authoritatively whether a primary rule has been broken, curing inefficiency. A legal system, for Hart, simply is this union of primary and secondary rules. The notes capture the essence: primary rules impose duties, secondary rules confer the power of rule-making and vest it in an authority such as Parliament, producing a uniform yet adaptable system.
Hart's internal point of view and its critics
Hart added a crucial psychological dimension that distinguishes a legal rule from a mere habit or a gunman's threat: the internal point of view. Rules have two aspects — an external aspect of regular behaviour, observable from outside, and an internal aspect, the reflective critical attitude in which members of the group treat the rule as a common standard, use it to guide and criticise their own and others' conduct, and speak in terms of "ought". It is this internal attitude that explains why people generally obey law out of a sense of obligation rather than mere coercion — answering Austin's failure to distinguish being obliged by a threat from being bound by a rule. Even those who dislike a particular rule recognise it as a standard they are obligated to obey.
The theory is convincing but not unassailable. Critics question whether the neat division into primary and secondary rules is satisfactory, especially when the legislative sovereign changes. Some argue that not every part of a legal system reduces to rules: every developed system also rests on fundamental principles that are not rules in Hart's sense — a line of attack pressed most famously by Ronald Dworkin, who argued that legal reasoning draws on principles and policies that the rule of recognition cannot capture. Others doubt that the internal/external account of rule-following is true to life. Even so, Hart's framework remains the standard modern map of the concept of law, precisely because it absorbs the strengths of its rivals: it preserves the positivist separation of law and morals while explaining the normativity that Austin missed and the institutional reality that realism stressed.
How the theories map onto the schools of jurisprudence
The four definitions are not free-floating; each grows out of a school with its own method, surveyed in our overview of the schools of jurisprudence. The natural-law definition belongs to the ethical or philosophical tradition, which asks what law ought to be and sets forth ideals for the future rather than describing present rules. Austin's command definition is the flagship of the analytical school, which dissects "law as it is" through rigorous analysis of concepts such as sovereignty, command and sanction. Salmond's and the realists' court-centred definition is a refinement within and reaction against analytical positivism, relocating the centre of legal gravity from legislature to bench. Hart's rule-based definition is analytical jurisprudence brought to maturity, repairing Austin from within.
A complete account must also leave room for the schools this topic only brushes. The historical school of Savigny and Maine defines law not as command or reason but as the organic expression of a people's Volksgeist or spirit, evolving with the nation — an answer to "what law is, from what it was". The sociological school of Ihering, Pound and Ehrlich defines law functionally, as an instrument of "social engineering" for balancing competing interests, locating the reality of law in society rather than in statute or court. Seen together, the definitions are less rival contestants than complementary lenses, which is exactly why Salmond insisted that "a study of all the schools is essential, because the three schools are closely related and interwoven".
Comparing the four conceptions
It helps to line the four definitions up against the three features of law identified at the outset — normativity, institutional source and coercion. Natural law excels at normativity: it explains why law claims moral authority and why we feel entitled to condemn wicked statutes, but it is weak on institutional source (it cannot easily tell us which of competing moral views is the law) and indifferent to coercion. Austin's command theory excels at coercion and institutional source — law plainly issues from an authority and is enforced — but it cannot explain power-conferring rules, customary and international law, or the persistence of law beyond its maker, and it confuses obligation with mere force. Realism brilliantly captures the institutional reality that courts, not statute books, deliver law in practice, but threatens the certainty of law and underplays the binding force of un-litigated statutes.
Hart's system of rules scores best across all three: the internal point of view supplies normativity, the rule of recognition supplies an institutional source, and primary rules backed by rules of adjudication supply organised coercion — all without collapsing law into morality. That is why, for examination purposes, the safest position is a Hartian one tempered by an acknowledgement that natural-law reasoning supplies the moral floor (visible in Maneka Gandhi and Kesavananda Bharati) and that realism supplies a healthy reminder that the law in books is not always the law in action. No single definition is complete; the mature jurist treats them as overlapping answers to a genuinely hard question.
Answering 'what is law?' in the exam
For judiciary and CLAT-PG papers, definition questions reward structure over recitation. Open by stating why the question is hard — that law is normative, institutional and coercive, and that each theory privileges different features. Then take the four conceptions in historical order: natural law (Cicero, Aquinas, Grotius; lex injusta non est lex), Austin's command theory (command, sovereign, sanction), Salmond and the realists (law as the practice of courts; Holmes's "experience" and "prophecies"), and Hart (the union of primary and secondary rules, the rule of recognition, the internal point of view). For each, give one or two merits and one or two crisp criticisms — examiners award the criticism, not just the statement.
Wherever possible, anchor the abstraction in Indian authority. Cite A.K. Gopalan and Maneka Gandhi to show positivism giving way to a natural-law-flavoured insistence on just procedure under Article 21, and Kesavananda Bharati to show natural-law limits on sovereign legislative power that Austin cannot explain. Close with a reasoned conclusion: that the definitions are complementary, that Hart offers the most complete modern synthesis, and that natural-law and realist insights remain indispensable correctives. For the wider context, return to the jurisprudence notes hub and read this topic alongside the natural, analytical, historical and sociological schools, since the concept of law is the thread that runs through all of them.
Frequently asked questions
What is the simplest way to define 'law' in jurisprudence?
There is no single accepted definition, which is itself the point. The four dominant answers are: law as the dictate of reason (natural law), law as the command of a sovereign backed by sanction (Austin), law as the practice of courts (Salmond and the legal realists), and law as a union of primary and secondary rules (Hart). Each captures a real feature of legal order, so a sound answer presents them as complementary rather than choosing only one.
How does Austin's command theory differ from Hart's theory of rules?
Austin defines law as a sovereign's command backed by the threat of a sanction, treating obedience essentially as coerced. Hart rejects this "gunman writ large" picture: he distinguishes being obliged by a threat from being under an obligation created by a rule, and explains law as the union of primary (duty-imposing) and secondary (power-conferring) rules, accepted from an internal point of view. Hart thereby accounts for power-conferring rules, customary and international law, and the persistence of law, all of which embarrass Austin.
Where does natural law appear in Indian constitutional law?
Most clearly in Maneka Gandhi v. Union of India, AIR 1978 SC 597, where the Supreme Court held that any procedure under Article 21 must be just, fair and reasonable, overruling the narrow positivism of A.K. Gopalan v. State of Madras, AIR 1950 SC 27. The basic-structure doctrine in Kesavananda Bharati v. State of Kerala, AIR 1973 SC 1461, also reflects natural-law thinking, since it places inviolable limits on even Parliament's amending power.
What is the 'rule of recognition' in Hart's theory?
It is the master secondary rule that supplies an authoritative test for which primary rules count as valid law, curing the uncertainty that would afflict a society governed by primary rules alone. Alongside rules of change (which cure staticness by empowering the enactment and repeal of rules) and rules of adjudication (which cure inefficiency by empowering courts to determine breaches), the rule of recognition is what turns a mere set of rules into a functioning legal system.
What did Holmes and the legal realists mean by saying law is the practice of courts?
They shifted the focus from the legislature's will to what courts actually do. Salmond defined law as "the body of principles recognised and applied by the State in the administration of justice". Holmes went further, calling law "the prophecies of what the courts will do in fact" and insisting that "the life of the law has not been logic; it has been experience". The insight is that statutes and rules come alive only through judicial interpretation and enforcement, though critics note this threatens the certainty of law and underplays statutes that are never litigated.
Why can't one theory simply be declared correct?
Because each illuminates a different feature of law and obscures another. Natural law explains law's moral authority but not which institution declares it; Austin explains coercion and authority but not power-conferring rules or the persistence of law; realism explains the centrality of courts but threatens certainty; Hart offers the fullest synthesis yet is criticised for reducing everything to rules and ignoring underlying principles. The mature position treats them as overlapping lenses on a genuinely hard concept.