Almost every panic about regulation comes from one of four confusions: mistaking a proposal for a rule, mistaking guidance for law, mistaking one country's rule for a global one, or mistaking an obligation on companies for an obligation on people.
All four dissolve once you understand how the system is built. That understanding does not expire. It applies across a wide range of jurisdictions, subjects and decades. It is, in our view, among the most useful things a non-lawyer can learn about law, and almost nobody teaches it outside a law faculty.
A course like this necessarily moves between several different kinds of claim, and they do not carry the same weight. Where it matters, the text says which one it is in; where it does not, assume the second.
- Structural. True of essentially any developed legal system, such as the existence of a hierarchy of norms.
- Common but not universal. The usual arrangement, with real exceptions. Most of this page.
- Tradition-specific. True of civil law or of common law systems, and labelled as such.
- Jurisdiction-specific. An example from one country, used to make a structure concrete.
Nothing here tells you the rule where you live. It tells you what kind of rule to look for, and where.
Contents
Thirty-three parts, grouped into eight sections. Each is self-contained, so you can start anywhere. Every technical term is defined in the companion glossary.
The state and its institutions
How law is made and organised
Private law building blocks
How law is applied
Rules about rules
The great principles, by branch
Reference
Begin with the question a first lecture always opens on, because the answer determines everything that follows.
Law is a system of general rules, produced by a recognised authority, binding on those they address, and backed by organised enforcement.
Each element of that sentence does work. Remove any one and you get something that is not law: a moral principle, an order, a suggestion, or a threat.
The four elements
A law addresses a category, not a person. "Drivers must stop at red lights" is a rule. "Marie must stop at this light" is an order.
Generality is what makes law predictable. You can know in advance what applies to you, because the rule was written without knowing who you are. It is also what makes law fair in a minimal sense: the drafter cannot target an enemy, because they must write a rule that catches everyone in the category, including their friends.
This is why constitutions frequently prohibit legislation aimed at a named individual. A rule that applies to exactly one person, drafted knowing who that person is, has the form of a law and the substance of a punishment.
Not every general prescription is law. A religious commandment, a company handbook, a philosophical principle and the rules of a sport can all be general and compelling without being law.
What makes a rule legal is that it was produced through a procedure the legal system itself recognises as producing law. A statute is law because parliament voted on it in the manner the constitution prescribes. A decree is law because a statute authorised the minister to make it.
This is circular, and the circularity is real: legal systems define their own criteria of validity. Part two examines what philosophers have made of that, because it turns out to be the deepest question in the subject.
Law claims obligation, not advice. It does not say what would be prudent; it says what is required.
This is the distinction between law and soft law, and it is the one most often blurred in public discussion. A regulator saying "firms should consider doing X" has not created an obligation, however firmly it was said and however many firms comply.
Law is associated with an apparatus that can compel: courts that can order, officials that can seize, penalties that can be imposed.
This does not mean every rule carries a penalty. Many rules are enabling rather than prohibitive, and breach of a formal requirement often means the act is simply void rather than punished. Nor does it mean enforcement is always effective. It means the legal system as a whole claims a monopoly on legitimate coercion, exercised through defined procedures rather than at will.
Apply the four elements to four rules that all feel binding.
- "Do not exceed 50 km/h in built-up areas." General, enacted by a competent authority, obligatory, enforced by fines. Law.
- "Employees must wear a badge at all times." General within the company and enforced by discipline, but the authority is the employer, not the state. Not law, though it may have legal consequences through the employment contract.
- "You should keep your promises." General and obligatory in a moral sense, but no recognised legal authority produced it and no organised enforcement attaches. Not law, though large parts of contract law grew out of it.
- "Firms are expected to review their client files annually." Published by the supervisor, addressed generally, and universally complied with. Still not law: it uses expected, no statute imposes it, and no penalty attaches to it directly. Soft law.
The fourth is the one people get wrong, and it is the one that matters most in regulated industries.
Law compared with its neighbours
| System | Source of authority | Sanction | Who enforces | Can you opt out? |
|---|---|---|---|---|
| Law | Recognised legal procedure | Organised, defined in advance | State institutions | No |
| Morality | Conscience, reason, tradition | Guilt, internal | The individual | In practice, yes |
| Social custom | Repeated practice | Disapproval, exclusion | The group, informally | By leaving the group |
| Religious precept | Doctrine, revelation | Spiritual, communal | Religious community | By leaving the faith |
| Private rules | Agreement or ownership | Exclusion, contractual | The private body | By not joining or not contracting |
The overlaps matter more than the distinctions.
Most of criminal law tracks widely shared moral intuitions, which is why it feels natural rather than imposed. Much of commercial law codified pre-existing merchant custom rather than inventing rules: the law of bills of exchange grew out of what traders were already doing. Several legal systems incorporate religious law directly for defined subjects. And private rules acquire legal force whenever the law chooses to enforce them, which is exactly what a contract is: a private rule the state will back with its courts.
You and a neighbour agree that she may cross your garden to reach the road. That is a private arrangement.
If you write it down and sign it, contract law makes it enforceable between the two of you: if you block the path, she can sue.
If you register it as a servitude on the property title, property law makes it enforceable against the world: it binds whoever buys your house next, even though they never agreed to anything.
Same underlying arrangement, three different legal strengths, depending entirely on which legal machinery you used. This is the enabling function of law in its purest form, and it is why the choice of legal instrument matters as much as the substance of the deal.
What law is for
Legal theory identifies several functions. Any real rule usually serves more than one, and the functions frequently conflict.
- Ordering conduct. Telling people what they must, may and must not do, so behaviour becomes predictable and coordination becomes possible. Traffic rules are the pure case: the content hardly matters, the coordination is everything.
- Resolving disputes. Providing a procedure producing an authoritative answer, so conflict does not escalate indefinitely or resolve by force. Note that the value here is finality as much as correctness. A system that produced perfect answers after forty years would fail at this.
- Allocating and constraining power. Deciding who may decide what, and setting limits. This is the whole of constitutional and administrative law.
- Creating facilities. Providing tools people can use to arrange their own affairs: contracts, companies, wills, marriages, mortgages, property titles. Often forgotten because it involves no prohibition, yet most of private law is this. Nobody is ordered to make a will; the law simply provides one and gives it effect.
- Distributing. Allocating burdens and benefits through taxation, social provision and regulation.
- Expressing values. Declaring what a society treats as unacceptable, sometimes with more symbolic than practical effect. Rules that are never enforced can still do this work, which is why campaigns to repeal dead-letter provisions are contested.
Some rules prohibit and some rules enable.
A rule against theft prohibits: it removes an option you previously had. The law of contract enables: it does not order you to make contracts, it provides machinery that turns your promises into enforceable obligations if you choose to use it. Company law does not require you to form a company; it makes one available.
Reading any legal text, ask which kind you are looking at. Enabling rules are constantly misread as prohibitions. "Digital assets may be held in custody by authorised institutions" creates an option for institutions; it does not prohibit anyone from holding anything themselves. That sentence, and sentences like it, generate a great deal of unnecessary alarm every year.
Why must you obey a rule? Three families of answer have dominated legal thought, and the disagreement is not academic. It determines what you say when a legal system produces a monstrous rule through impeccable procedure, which is a situation the twentieth century supplied repeatedly.
There exist principles of justice independent of human enactment, discoverable by reason or given by nature, and human law derives its authority from conformity to them. A sufficiently unjust rule is not merely bad law; it is not law at all. The classical formulation is that an unjust law is a corruption of law rather than an instance of it.
The tradition runs from Greek and Roman philosophy through medieval scholasticism to modern human rights theory. Its influence is visible wherever a legal text refers to inherent dignity, inalienable rights, or principles recognised by civilised nations.
Strength: it explains why we condemn the laws of tyrannies rather than merely disliking them, and it gives a foothold for resisting them from within law.
Weakness: reasonable people disagree profoundly about what nature or reason requires, and the tradition has historically been used to justify very different arrangements with equal confidence.
Law is a social fact. A rule is legally valid because it was produced according to the system's own criteria, not because it is just. Whether a law is good and whether it is law are separate questions, and conflating them muddles both.
Positivists are not indifferent to justice, and the charge that they are is the most persistent misunderstanding in the field. Their claim is that clarity requires separating the description of what the law is from the evaluation of what it should be. You can then say, coherently, "this is the law, and it is indefensible, and it should be changed."
This is the dominant working assumption of modern legal practice. It is why a lawyer can tell you accurately what the law requires while thinking it monstrous, and why that is a service rather than a betrayal.
Focus on what officials actually do. Rules underdetermine outcomes; what decides cases is the behaviour of judges and administrators, shaped by facts, institutions, incentives and social context. To know the law, study decisions rather than texts.
A useful corrective rather than a complete theory. Anyone who has read a statute and then discovered how it is actually applied has encountered the realist point directly. Its modern descendants include empirical legal studies and the economic analysis of law.
A statute, validly enacted by a parliament under a valid constitution, requires officials to confiscate property from members of a named ethnic group. A judge is asked to apply it.
Natural law: the statute violates fundamental principles of justice so gravely that it is not law. The judge should refuse to apply it, and in doing so is applying law, not defying it.
Positivism: the statute is law, because it satisfies the system's criteria of validity. It is also monstrous. The judge faces a genuine conflict between legal duty and moral duty, and should probably resign or refuse and accept the consequences. Calling it "not law" obscures the choice rather than resolving it.
Realism: ask what judges in this system actually do with such statutes, what pressures they face, and what happens to those who refuse. The answer to "what is the law here" is a prediction about official behaviour.
Notice that all three can lead the judge to refuse. They differ on how to describe the refusal, and that description matters enormously afterwards, when the system is being rebuilt and someone must decide whether acts done under those statutes were ever valid.
Two ideas that recur constantly
If every rule draws validity from a higher rule, the chain must terminate somewhere. Two influential answers:
The basic norm. Kelsen posited a presupposed foundational norm at the summit, not itself enacted, which grounds the validity of the constitution. It is presupposed rather than proved, a logical necessity rather than a discovered fact.
The rule of recognition. Hart's answer: the chain ends in a social practice. Officials of the system in fact accept certain criteria for identifying valid law, and that shared acceptance is the ultimate foundation. It is not a norm but a fact about behaviour.
The practical payoff: this is why revolutions are legally interesting. A successful coup produces a new foundation, and the question of when the old order ceased to be law and the new one began has no clean answer from inside either system.
A legal order needs both. A rule can be formally valid yet universally ignored, in which case its validity is real but hollow. A regime can be effective yet regarded as illegitimate, in which case obedience rests on force alone and is expensive to maintain.
Stable systems have both, and most interesting constitutional crises are about one of them failing. A widely disobeyed law is a symptom, not merely a nuisance: it usually indicates that the rule outran the consent it depended on.
Because the positivist distinction is the one you need in daily life. "Is this the law?" and "is this right?" are different questions with different answers, and running them together produces two opposite errors.
The first: concluding that because something is legal, it is acceptable. The second: concluding that because a rule is unjust, it can be safely ignored. Both are expensive, and the second is expensive quickly.
Law is not a single body of text. It is a set of sources, produced by different institutions, carrying different weight, interacting according to rules that are themselves law.
When someone says "the law says X", the useful reply is always: which source?
The founding text of a legal order. It creates the institutions, allocates power among them, sets the procedure for making law, and usually guarantees rights. It is the source that validates all others: a rule contradicting it is in principle void.
Most states have a single written constitution. A few, notably the United Kingdom and New Zealand, have no single document and instead treat certain statutes, judicial decisions, conventions and principles as constitutional in nature. This does not mean they lack a constitution; it means the constitution must be assembled from several places, and that identifying its content is itself a contested legal question.
Rules adopted by a parliament or legislature. Statutes, acts, lois, Gesetze. This is what people mean by "a law was passed". Its authority rests on democratic legitimacy, and it normally ranks immediately below the constitution.
A crucial and underappreciated point: legislation both creates rules and creates powers. Much of what a statute does is not to regulate citizens directly but to authorise the executive to regulate them. A financial services act may contain very few obligations aimed at you and a great many provisions saying "the authority may make rules concerning X".
This means reading a statute to find out what you must do frequently produces nothing, because the obligations are two levels down in instruments the statute merely authorised.
Rules made by the executive under authority conferred by a statute. Decrees, orders, regulations, statutory instruments. Generally binding and enforceable, but valid only within the boundaries of the enabling power. Delegated legislation is not simply equivalent to primary legislation: its constitutional status, the grounds on which it can be challenged, the procedure for making it and the degree of parliamentary control over it commonly differ.
In volume this dwarfs primary legislation, typically by an order of magnitude. Most rules that affect daily life are of this kind, and this is the source most often confused with the two above.
Decisions of courts. In common law systems, a formal source that binds later courts through the doctrine of precedent. In civil law systems the classical position is that decisions are not a formal source, but this is a simplification that no longer describes many of them well: constitutional court rulings are frequently binding by statute or constitution, supreme court jurisprudence can be binding or near-binding, and a settled line of decisions from a supreme court carries decisive practical weight.
Even where case law is not a source, it is indispensable, because a statutory provision means what the courts have held it to mean. Reading a code article without reading the decisions interpreting it produces confident errors.
A civil code provides: "Any act of a person which causes damage to another obliges the person by whose fault it occurred to repair it."
Twenty-one words. From them, courts across two centuries have built the entire law of liability: what counts as fault, whether omissions count, how to establish causation where several causes combine, whether purely economic loss is recoverable, when liability attaches without fault at all, how to value non-material harm.
Someone reading only the article would answer almost none of those questions, and would be wrong about most of them. The provision is the starting point of the law, not its content.
Treaties between states, plus customary international law formed by consistent state practice accepted as obligatory.
Its place in domestic law varies fundamentally:
Monist systems treat a ratified treaty as forming part of national law without separate implementing legislation. It may be invocable before a national judge, and in some such systems it ranks above ordinary statutes.
Dualist systems treat international and domestic law as separate orders. A treaty binds the state internationally but has no domestic effect until a statute transposes it. Until then, a national court will not apply it, however clearly the state has breached its obligations.
The two labels are a useful starting distinction and a poor finishing one. Whether a particular treaty provision is self-executing, directly invocable by an individual, judicially enforceable, dependent on publication, or superior to a later statute varies within monist systems as much as between monist and dualist ones. Establish the position for the specific treaty and the specific country rather than reasoning from the label.
This determines whether you can rely on a treaty in front of a national judge, and the answer differs from country to country. It is one of the first things to establish about any legal system you are unfamiliar with.
A practice treated as binding through long observance. It requires two elements, and both are necessary:
The material element: consistent, general, repeated practice over time.
The psychological element: a conviction among those following it that it is obligatory rather than merely convenient. Traditionally called opinio juris.
Without the second, a practice is a habit, not a custom. Everyone in a market may pay in cash out of convenience without anyone thinking they are legally required to.
Custom is marginal in most modern domestic systems, dominant in international law, and still significant in commercial law, where trade usages fill gaps in contracts and define what terms mean in a particular trade.
Principles a legal order treats as inherent even without enactment: good faith, the right to be heard, nobody may profit from their own wrong, proportionality, legal certainty, nobody may be judge in their own cause.
Courts rely on them to fill gaps and to interpret texts, and constitutional and administrative law is largely built from them. In several systems, entire bodies of administrative law were constructed by courts from general principles before any statute existed.
The writing of legal scholars. Not a source of law anywhere in the strict sense, but genuinely influential in civil law systems, where a leading treatise can shape how an entire field is understood and is routinely cited in argument and sometimes in judgments.
Its influence is greatest where the law is unsettled and where courts give short reasons, because someone has to do the systematic thinking and it will not be the judgment.
And the thing that is not a source
Soft law. Guidance, circulars, recommendations, opinions, FAQs, supervisory expectations, industry codes, speeches, consultation papers, letters to firms. None of it binds by itself. All of it shapes behaviour, often more than binding rules do.
Being able to distinguish soft law from law is the single most valuable skill on this page. A regulator's guidance tells you how that regulator currently reads a rule. That is important information. It is not the rule, it can change without procedure, and a court is not bound by it.
A financial supervisor publishes, in the same month:
- A regulation requiring firms to verify customer identity before opening an account, made under an express power in the anti-money-laundering statute. Binding. Breach is an offence and grounds for licence withdrawal.
- Guidance stating that firms "should generally re-verify identity every three years." Not binding. No statute requires it. A firm that does not do so has broken no rule, though it may face awkward questions at inspection.
- A speech by the chair saying the authority "will be paying close attention to onboarding practices." No legal weight whatsoever. Real predictive value about enforcement priorities.
All three will be reported as "the regulator has told firms to...". Only the first creates an obligation. And note the practical trap: most firms will comply with all three identically, which makes the guidance feel like law and lets it function as law without ever having gone through a law-making procedure.
It is fast, flexible, and requires no democratic procedure. Regulators use it to signal expectations without the cost and delay of rule-making, and without the risk that a court will strike the rule down.
Firms comply because the supervisor controls their licence, their inspections, their approvals and their reputation. The practical effect can equal that of binding law while none of the procedural safeguards apply: no consultation requirement, no publication requirement, no possibility of annulment, no clear standard to litigate against.
This is a live and legitimate criticism of modern administrative governance, and it is worth understanding structurally rather than simply resenting. The counter-argument is real too: soft law lets regulators give useful clarity quickly, and firms generally prefer knowing the supervisor's view to guessing it.
Sources are ranked. The idea is usually associated with Kelsen, who described a legal order as a pyramid in which each norm draws validity from the one above.
| Rank | Source | Made by | Changed by | Typical time to change |
|---|---|---|---|---|
| 1 | Constitution | Constituent power | Special amendment procedure | Years, often never |
| 2 | Ratified treaties (rank varies, see below) | States, approved domestically | Renegotiation or denunciation | Years |
| 3 | Legislation | Parliament | Ordinary majority | Months to years |
| 4 | Delegated legislation | Executive, under statutory authority | The executive, within the power | Weeks to months |
| 5 | Individual decisions | Ministries, agencies, authorities | The issuing authority | Days |
| — | Soft law | Anyone | Anyone, instantly | Hours. Binds no one. |
The pyramid above is the classical continental teaching model, and rank two is accurate for systems whose constitution expressly places ratified treaties above ordinary legislation. France is the standard example. It is not a general rule, and assuming it across borders produces confident errors.
In the United States the position is different. The Supremacy Clause names the Constitution, federal statutes and treaties together as the supreme law of the land, without ranking statutes and treaties against each other. The Supreme Court has resolved conflicts between them through the last-in-time rule: whichever was enacted later controls, as the latest expression of the sovereign will. So a treaty does not outrank an act of Congress there. A later statute can and does override an earlier treaty, and a later treaty can override an earlier statute.
Two refinements worth carrying. The rule applies only to self-executing treaties; a non-self-executing treaty yields to a federal statute whatever the dates. And no treaty overrides the Constitution itself.
The general lesson is larger than the example: establish where treaties sit in the specific system you are dealing with, because monist and dualist systems, and constitutions that rank treaties expressly and those that do not, all give different answers.
Three consequences follow, and together they explain most of what confuses people about regulatory news.
One: a lower norm cannot lawfully contradict a higher one
A decree exceeding its parent statute can be annulled. A statute violating the constitution can be struck down. So "the government announced a rule" is never the end of the analysis: the rule may exceed the power the government was given.
The technical term for acting beyond your powers, used across many systems, is ultra vires. It is the most common successful ground of challenge in administrative law, and it exists precisely because the pyramid is real.
A ministry publishes a decree requiring anyone transferring more than a certain sum to file a declaration. You want to know whether it binds you.
- What kind of instrument is it? A decree, so delegated legislation. It must rest on a statute.
- Which statute? The decree's opening lines will cite its legal basis. Find that provision and read it.
- Does the statute authorise this? Suppose it empowers the ministry to "set reporting thresholds for transfers by financial institutions." The decree, however, applies to any person. That is a gap: the power covers institutions, the decree covers everyone.
- Consequence. As applied to individuals, the decree exceeds its enabling power and is vulnerable to annulment. It remains in force until challenged, and someone must actually bring that challenge, in the right court, within the deadline.
This is one of the most useful analytical moves in the subject, and it takes about fifteen minutes. It is easy to skip, because it is natural to assume that a published official document is necessarily valid. Many are not.
Two: each level constrains the one below
The statute sets the frame. The decree fills it in. The individual decision applies it to you. To test whether a requirement is lawful, climb the pyramid until you find the provision authorising it. Frequently there is none, and what you are looking at is an institution's preference dressed in the language of obligation.
Three: speed is inversely proportional to rank
Look at the last column of the table. Constitutional change is deliberately hard, requiring supermajorities, referendums or multiple readings. Statutes need a parliamentary majority and scarce floor time. Delegated legislation can be made in weeks. Soft law can be published on a Tuesday afternoon.
This creates a structural pressure that shapes modern government: because the higher levels are slow and the lower ones are fast, policy migrates downward. More is done by delegated legislation and guidance, less by statute. Parliaments increasingly pass framework statutes that decide little and delegate much.
That migration is one of the central problems of contemporary public law, because each step down the pyramid trades democratic legitimacy and procedural safeguards for speed and flexibility. It is worth noticing when it happens, because it is rarely announced.
When an institution tells you "we are required to do this", the correct follow-up is: required by what?
Sometimes the answer is a statute. Often it is a regulator's guidance, which is not a requirement. Just as often it is the institution's own internal policy, which is not even soft law. All three are communicated to customers in identical language, and distinguishing them is frequently worth real money.
If the answer is the third, the request is negotiable. If it is the second, it is sometimes negotiable. If it is the first, it is not.
Law in the modern world is overwhelmingly state law. Understanding the state is therefore a precondition for understanding law.
The classical elements
- A territory. A defined space over which authority is claimed. Borders can be disputed without destroying statehood; what matters is that some core territory is effectively controlled.
- A population. People linked to the state, usually by nationality. Size is irrelevant: microstates are states.
- An effective government. An organised authority actually exercising control. This is the element that fails in collapsed states, and its absence is why some entities have territory and population without functioning as states.
- The capacity to enter relations with other states. Added by international practice. In substance this is where recognition by other states enters, and recognition is politically decisive even though theory holds that statehood exists independently of it.
This is why the legal status of contested territories is genuinely difficult rather than merely disputed: an entity can satisfy the first three elements and still be treated as a non-state by most of the world, or satisfy them poorly and be treated as a state by nearly everyone.
Sovereignty
Supreme authority within the territory. No competing authority may override the state's law there. This is what makes the state's rules law rather than one set of demands among many, and it is why a legal system can claim to be comprehensive: every question within the territory has, in principle, an answer under its law.
Independence from other states. No superior authority exists above the state internationally, only obligations it has consented to. This is the formal basis of the international order: states are equal, and none can legislate for another.
Both are heavily qualified in practice. States bind themselves by treaty, submit to international courts, join organisations whose decisions bind them, and in the case of supranational organisations transfer competences to institutions whose acts prevail over their own law.
Sovereignty in the twenty-first century is better understood as a bundle of powers that can be pooled, delegated and reclaimed than as an indivisible absolute. A state that joins a customs union has not stopped being sovereign; it has exercised sovereignty by committing itself, and it usually retains the legal right to leave.
The state as a legal person
A crucial abstraction, and one people rarely notice.
The state is not the people currently running it. It is a legal person that persists across governments, owns property, makes contracts, incurs debts, sues and is sued. When a government changes, obligations survive: the new administration inherits the treaties, the contracts and the debts, whether it likes them or not.
A ministry awards a construction contract. Two years later the government changes and the new minister considers the contract a mistake.
The contract was not with the previous minister personally. It was with the state, which still exists. The state remains bound. The new minister may terminate under whatever termination clause exists, or may terminate in the public interest where administrative law allows it, but in both cases compensation is generally due.
Without the abstraction of the state as a legal person, every election would void every public contract, and nobody would ever contract with a government.
The same abstraction is what lets you sue "the state" for the act of an official whose name you never learned, and why the judgment is paid from the treasury rather than from that official's salary.
The rule of law as a constraint on the state
The defining idea of the modern constitutional state is that the state is itself subject to law. Officials act only with legal authority, their acts can be tested before independent courts, and the same law binds the governor and the governed.
This is a historical achievement rather than a natural condition. For most of history the sovereign was the source of law and therefore above it, and the proposition that a subject could sue the crown was close to incoherent. The machinery that makes it possible, administrative courts and judicial review, is largely a nineteenth and twentieth century construction.
It produces a fundamental asymmetry, worth memorising because it governs every encounter with authority:
The private person
May do anything not prohibited. Freedom is the default; restriction requires a legal basis.
You do not need permission to act. Someone who wants to stop you needs a rule.
You need not justify your choices. You may act for good reasons, bad reasons, or no reason.
The public authority
May do only what it is empowered to do. Power is the exception; it requires a legal basis.
An authority acting without a legal basis acts unlawfully, even with excellent intentions and a beneficial result.
It must act for the purpose the power was given, on relevant considerations, and be able to explain why.
A bank closes your account without explanation. A tax authority freezes it without explanation. These feel identical and are legally opposite.
The bank is a private person. It may generally decline to contract with anyone, subject to its contract terms, anti-discrimination law and sector-specific rules. It usually need not give reasons, because it needed no authority to act in the first place. Your remedy, if any, lies in the contract or in specific consumer protections.
The tax authority is a public authority. It needed a legal power to freeze the account, must have exercised it for the purpose the power was granted, must generally give reasons, and must inform you how to challenge the decision. Absent any of those, the decision is likely annullable.
This is why "show me the power" is always a fair question to a public authority and rarely a useful one to a company. Knowing which you are dealing with determines the whole strategy.
A constitution does three things: it constitutes institutions, it allocates and limits their powers, and it usually guarantees rights.
Types of constitution
| Distinction | One side | The other | Why it matters |
|---|---|---|---|
| Form | Written: a single codified document | Unwritten: dispersed across statutes, cases and conventions | Determines whether you can point to a text |
| Amendment | Rigid: special procedure required | Flexible: amended like ordinary legislation | Determines how protected rights actually are |
| Enforcement | Justiciable: courts may strike down conflicting law | Political: enforced by institutions and convention | Determines whether you can litigate a breach |
| Origin | Granted: conferred by a ruler | Popular: adopted by assembly or referendum | Affects perceived legitimacy |
Note that rigidity and justiciability are more important than whether the constitution is written. A written but flexible constitution that any majority can amend protects less than an unwritten one backed by strong conventions and independent courts.
Constituent power
The authority to make a constitution differs in kind from the authority to legislate under one.
Original constituent power creates a new constitutional order and by definition operates outside any prior legal framework. It cannot be legally authorised, because the law it would need authorisation from does not yet exist. This is a founding or revolutionary act.
Derived constituent power amends an existing constitution according to the procedure that constitution itself lays down. This is a legal act, exercised under and limited by the existing order.
Most constitutional crises are arguments about which one is occurring. A government claiming to exercise derived power while ignoring the amendment procedure is in substance claiming original power, which is a different thing entirely.
Some constitutions declare certain provisions beyond amendment: the republican form of government, human dignity, federal structure, territorial integrity. These are sometimes called eternity clauses.
They pose a genuine puzzle. If constituent power is supreme, how can one generation bind the next? The usual answer is that they bind derived constituent power, which operates under the constitution, but not original constituent power, which could replace the whole order. In practice constitutional courts have sometimes struck down amendments for violating them, which is about as strong a form of judicial review as exists.
What constitutions contain
- A preamble, stating founding values and historical context. Sometimes treated as non-binding, sometimes given full legal force by courts, and the difference is decided by case law rather than by the text.
- Fundamental principles, defining the nature of the state: democratic, secular, federal, social, indivisible, subject to the rule of law.
- Rights and freedoms. The bill of rights, however styled.
- The organisation of powers. Composition, election, term and competences of each institution.
- The law-making procedure, including what must be done by statute and what may be left to the executive.
- Territorial organisation. Federal or unitary, and the division of competences.
- The judiciary, its independence and the mechanism of constitutional review.
- Emergency provisions. What may be suspended, by whom, for how long, and with what oversight. Historically the most abused part of any constitution, which is why modern texts hedge them with time limits and parliamentary control.
- Amendment procedure, sometimes with unamendable provisions.
- Transitional provisions governing the passage from the old order to the new.
Generations of rights
A standard classification. It is a simplification, criticised by many scholars, and it remains useful because it explains why some rights are enforced by courts and others largely are not.
Liberty, security, property, expression, assembly, association, religion, privacy, fair trial, due process, the vote, equality before the law.
Primarily negative: they require the state to abstain. Individually enforceable, immediately applicable, and the historical core of constitutional adjudication.
The negative framing is imperfect. A fair trial requires courts, judges and legal aid, all of which cost money, so even classical rights have a positive dimension.
Work, fair conditions, education, health, housing, social security, participation in cultural life.
Primarily positive: they require the state to provide. Often drafted as objectives to be progressively realised within available resources rather than as immediately enforceable entitlements.
Courts have generally been more cautious here, on the ground that allocating budgets is a political rather than judicial function. Some constitutional courts have nonetheless enforced a minimum core: whatever the resource constraints, the state may not fall below a basic threshold, and it must not act regressively without strong justification.
Self-determination, development, a healthy environment, peace, shared heritage.
Held by groups as much as individuals, which makes them hard to enforce through ordinary litigation, since procedure is built around individual claimants with personal interests. Environmental claims have nonetheless succeeded increasingly, often by being reframed as first-generation rights: the right to life and to private life, applied to pollution.
Rights are almost never absolute
Nearly every right can be limited. What matters is the test a limitation must satisfy. Across many constitutional and human rights systems the test has converged on a common structure, usually called proportionality. It is the single most useful analytical tool in modern rights law, and it is worth learning properly.
Legal basis
The limitation must be provided for by law: accessible, published, and sufficiently precise that people can foresee its consequences and regulate their conduct.
A restriction imposed by unpublished internal practice fails at this first step, however reasonable it is. So does one framed so vaguely that nobody can tell what it prohibits.
Legitimate aim
It must pursue an objective the constitution or treaty recognises: public safety, public health, national security, prevention of crime, protection of the rights of others, economic wellbeing.
This step is rarely decisive, because states can nearly always articulate a recognised aim. Its function is to force them to state one, which then constrains the following steps.
Suitability
The measure must be capable of achieving that aim. A measure that cannot work fails regardless of how important the aim is.
Also rarely decisive, but it does catch measures adopted for display rather than effect, where the government cannot explain the causal mechanism.
Necessity
No less restrictive measure would achieve the aim as effectively.
This is where most measures fail, and it is the step most worth invoking. The question is comparative: was there an alternative that intrudes less and works as well? A blanket rule where a targeted one would do, universal collection where sampling would do, permanent retention where a time limit would do.
Proportionality in the narrow sense
Even a necessary measure must not impose a burden excessive relative to the benefit gained. A balancing exercise, and the most openly evaluative step.
Courts vary in how much weight they give here, and this is where the accusation of judicial policy-making bites hardest.
A government requires every person entering a public library to present identity documents, which are recorded and retained for ten years, to combat terrorism.
Legal basis: assume it is set out in a published statute. Passes.
Legitimate aim: national security. Passes.
Suitability: arguable. Does recording library visitors actually contribute to preventing attacks? The government must be able to explain the mechanism. Weak, but probably passes.
Necessity: fails badly. Targeted measures based on individual suspicion would serve the aim without collecting data on the entire reading public. A blanket measure where a targeted one exists is the classic failure at this step.
Proportionality: would also fail. Ten-year retention of everyone's reading habits imposes a severe burden on privacy and on freedom of expression, since people read differently when watched, against a benefit that is speculative.
Note that the analysis never needed to say the aim was illegitimate or the government insincere. Proportionality lets you defeat a measure without impugning anyone's motives, which is precisely why it is the tool courts prefer.
Absolute rights and derogation
A small number of rights admit no limitation at all. The prohibition of torture and of slavery are the standard examples. These are termed absolute or non-derogable, and they remain in force even during declared emergencies.
Most other rights may be derogated from in a genuine emergency threatening the life of the nation, but only to the extent strictly required, only temporarily, and usually with a requirement to notify and to justify. Emergency powers are the historic weak point of every constitutional system, which is why modern instruments hedge them so heavily.
Horizontal effect
A question people rarely think about: do constitutional rights apply between private parties, or only against the state?
The classical answer is only against the state, since the constitution constitutes and constrains public power. But most legal systems have developed some form of horizontal effect, either directly, where certain rights bind private parties too, or indirectly, where courts interpret private law in conformity with constitutional values.
This matters enormously in practice. It is the difference between a constitutional right to privacy that only restrains the police and one that shapes what an employer may do with employee data.
The organising principle of the modern state, given its canonical form by Montesquieu in the eighteenth century, building on Locke: state power is divided among functions exercised by distinct institutions, so that no single body both makes and enforces the rules.
The underlying insight is not about efficiency. Concentrated power would be far more efficient. It is that concentrated power is dangerous regardless of who holds it, and that the reliable protection is structural rather than personal. You do not rely on the good character of the officeholder; you arrange matters so that a bad officeholder cannot do very much alone.
The three functions
Makes general rules for the future. Acts prospectively and abstractly, without knowing to whom the rule will apply. Deliberates publicly, decides by vote, and represents.
Its characteristic limitation: it cannot apply its own rules to cases, and it should not legislate for named individuals.
Implements and enforces. Acts concretely and continuously, on particular situations, controls the resources, the personnel and the information, and can act quickly.
Its characteristic danger: it holds the coercive apparatus and the practical initiative, which is why constraining it is the main business of administrative law.
Resolves disputes by applying rules to established facts. Acts retrospectively and passively: on the initiative of a party, only on what is brought before it, and only on the past.
In many systems a court will not act on its own motion to change the world, though the degree of judicial initiative varies considerably: some civil law and administrative procedures give judges substantially more power to act ex officio than the adversarial model suggests. It decides the case in front of it. This passivity is a defining feature and a real limitation: courts cannot fix a systemic problem nobody litigates.
Why it is never clean
- The executive makes rules constantly, through delegated legislation, in far greater volume than parliament. On any quantitative measure the executive is the principal legislator in every modern state.
- Courts make rules too, definitively in common law systems through precedent, and effectively everywhere, because applying a general text to unforeseen facts necessarily creates something new.
- Regulatory agencies combine all three functions. They write rules, investigate breaches, and adjudicate them, sometimes within the same building. This is the most significant structural challenge to the doctrine in modern government, and systems respond with internal separation, procedural safeguards and appeal to a court rather than by abolishing the arrangement.
- In parliamentary systems the executive is drawn from the legislature and normally commands its majority, so the separation between them is more theoretical than real. The government usually controls what parliament debates and how it votes.
The doctrine is therefore better read as a standard against which to assess arrangements than as a description of any actual state. The useful question is never "are the powers separated?" but "what stops this institution from doing whatever it wants?"
Three families of arrangement
The executive is separately elected and does not sit in the legislature. Neither can dismiss the other during a term, absent exceptional procedures.
Consequences: separation is sharp. Deadlock is a normal feature rather than a crisis. Policy consequently migrates toward executive rule-making and toward judicial resolution, because those are the routes that remain open when legislation is blocked. Appointments become intensely contested, since they are one of the few levers that still move.
The government emerges from the legislature and survives only while it retains that body's confidence. Fusion rather than separation.
Consequences: legislation moves quickly when a majority is stable and stalls entirely when it is not. The real check is the possibility of losing a confidence vote, which is a blunt instrument used rarely. Day-to-day constraint comes from within the governing party, from the second chamber, and from the courts.
A directly elected president alongside a prime minister answerable to parliament.
Consequences: the distribution of real power shifts depending on whether the president's party controls the assembly. When it does, the system behaves almost presidentially; when it does not, the president's domestic role can shrink sharply. This makes such systems the hardest to describe in the abstract, because the constitution reads the same while the practice differs completely.
A government wants to impose a new reporting obligation on an industry. Where will the rule appear?
If it has a comfortable parliamentary majority: possibly in a statute, especially if it wants the political credit and the durability. Statutes are harder for successors to undo quietly.
If its majority is thin or the subject is unpopular: in delegated legislation under an existing enabling power, avoiding a floor vote entirely. Look for an old framework statute with a broad rule-making clause.
If no enabling power exists and legislation is blocked: in regulator guidance, which requires nothing from anyone and is complied with anyway. Slower to bite, impossible to challenge directly, and reversible without notice.
Knowing the political configuration lets you predict the legal instrument, and knowing the instrument tells you what procedural rights you have. These are not separate subjects.
Separating powers is not enough. Each branch also needs means of restraining the others. Checks and balances means not that the branches are independent, but that they are mutually constrained.
Note the tension with separation: a pure separation would give each branch its sphere and no purchase on the others. Checks and balances deliberately violates that, giving each branch a partial role in the others' functions precisely so that it can obstruct. The veto is a legislative power held by the executive; confirmation of judges is a judicial power held by the legislature. The overlap is the point.
Judicial review of legislation
Diffuse review
Any court may decline to apply an unconstitutional statute in a case before it. Developed in the United States through its own case law in the early nineteenth century.
Review is concrete: it arises from a real dispute between real parties. And a posteriori: after the law is in force.
The formal effect is limited to the case, though precedent makes it general in practice. Advantage: accessible, grounded in facts. Disadvantage: uncertainty until a high court settles the point, and inconsistency in the meantime.
Concentrated review
A single specialised constitutional court holds a monopoly on the question. The Kelsenian model, adopted across most of continental Europe.
Review may be abstract, with no concrete dispute, and a priori, before entry into force, or both. Referral is often restricted to defined actors: the head of state, the government, a number of parliamentarians, or ordinary courts sending up a question.
The effect is normally to annul the provision for everyone. Advantage: certainty and uniformity. Disadvantage: narrower access, and abstract review decides without facts.
Many systems now mix the two: concentrated review plus a mechanism allowing ordinary courts, or litigants through them, to send a constitutional question up during a live case. That combination captures most of the advantages of both.
Judicial review of administrative action
Distinct from the above, far more common, and the mechanism that actually constrains regulators day to day. You are far more likely to encounter this than constitutional review.
The classical grounds, recurring across systems in slightly different vocabulary:
| Ground | The question | Typical example |
|---|---|---|
| Lack of competence | Did this body have the power at all? | A local authority regulating a matter reserved to the state |
| Procedural defect | Was a required step taken? | No consultation where consultation was mandatory; no hearing before a sanction |
| Defect of form | Does the act meet formal requirements? | No statement of reasons where reasons are required |
| Error of law | Did the authority misunderstand the rule? | Applying a definition wider than the statute's |
| Error of fact | Was the factual basis wrong or unsupported? | Refusing a licence on the basis of a conviction the applicant does not have |
| Improper purpose | Was the power used for its intended aim? | Using a safety power to punish a critic |
| Manifest error of assessment | Is the decision outside the range a reasonable authority could adopt? | A conclusion no reasonable body could reach on the evidence |
| Disproportionality | Does the burden exceed what the objective justifies? | Permanent exclusion for a minor first breach |
Note what is absent from that list: whether the decision was wise. Review asks whether a decision was lawful, not whether it was correct. Courts are generally careful not to substitute their judgment for the authority's on the merits, particularly where the subject is technical or involves resource allocation.
This deference is a persistent source of frustration for people who win the argument and lose the case, and it is deliberate: an unelected court second-guessing every policy choice would itself be a separation of powers problem.
A regulator refuses your licence application. You think the decision is wrong. Which ground do you argue?
Weakest: "the decision is wrong on the merits." Courts rarely engage, because assessing your suitability is what the regulator is for.
Stronger: "the decision relied on a factual error." If they refused because they believed you had a sanction record and you do not, that is reviewable, because it is a matter of fact rather than judgment.
Stronger still: "no reasons were given." Where reasons are required, their absence is a self-contained defect. You need not prove the decision was substantively wrong.
Strongest: "the regulator had no power to impose this condition at all." If the enabling statute does not authorise the condition, nothing else matters.
The lesson generalises: procedural and competence grounds are easier to win than substantive ones. Non-lawyers instinctively argue the merits, which is the hardest route. Lawyers look for the procedural defect first, not out of cynicism but because that is where courts have the clearest role.
The other checks
- Legislative control of the executive: budget approval, confirmation of appointments, committee inquiries, power to annul delegated legislation, questions, confidence votes, and in some systems impeachment.
- Executive checks on the legislature: veto, dissolution, control of the timetable, referral of laws to a constitutional court.
- Federalism, where a second level of government holds constitutionally guaranteed powers the centre cannot touch. A vertical check rather than a horizontal one.
- Independent authorities: central banks, electoral commissions, data protection regulators, audit offices, ombudsmen, statistical agencies. Deliberately insulated from the political cycle, and increasingly the subject of argument about whether they are insulated too much.
- Supranational courts, where jurisdiction has been accepted. A check that operates from outside the system entirely.
- A free press and civil society. No legal power, but they generate the information every other check depends on. A parliamentary committee cannot question a minister about something nobody has reported.
- Time itself. Fixed terms, mandatory elections and staggered appointments prevent any configuration of power from becoming permanent.
When a government announces something dramatic, the useful questions are procedural rather than political:
- Does it need legislation, or can it act by decree?
- If by decree, under which statute, and does that statute stretch that far?
- Who has standing to challenge it, in which court, on which ground, within what deadline?
- Is there a suspensive effect, or does the measure apply while being challenged?
- How long would the whole process take?
The answers usually reveal that the announcement is a proposal, that several institutions must agree, and that the timetable is measured in years. This is not complacency. It is the difference between what has happened and what someone has said.
Structure
Unicameral: one chamber. Simpler, faster, cheaper. Common in smaller and more homogeneous states.
Bicameral: two chambers, which must usually both approve a text. The second chamber typically represents a different principle: territorial units in federations, or seniority, indirect election, or professional and regional interests elsewhere.
Bicameralism deliberately slows legislation. That is the point. It forces reconsideration, gives a second body the chance to catch drafting errors and unintended consequences, and prevents a transient majority in one chamber from acting alone.
Where the chambers disagree, systems provide one of three solutions: a shuttle, where the text passes back and forth until agreement; a conciliation committee drawn from both; or a last word rule giving the directly elected chamber the final say after a defined number of readings.
The legislative procedure
Initiative
Who may propose a text. Usually the government and individual members, sometimes citizens through popular initiative, sometimes regional assemblies.
In practice the overwhelming majority of successful legislation originates with the government, which controls both the drafting resources and the parliamentary timetable. Private members' bills succeed rarely and usually only with government acquiescence. This is worth knowing because it means "parliament passed a law" almost always means "the government proposed a law and its majority approved it."
Committee examination
A specialised committee studies the text, hears evidence from officials, industry and civil society, and proposes amendments.
This is where most substantive change occurs and where lobbying concentrates. It is far less visible than floor debate and far more consequential. Anyone seriously trying to influence a text works at this stage, and anyone reading about a text only when it reaches the floor is reading about a decision already largely made.
Debate and amendment
The chamber discusses and votes on amendments article by article, then on the text as a whole.
Procedural rules matter enormously here and are frequently the real battleground: rules on the admissibility of amendments, on whether an amendment must relate to the subject of the bill, on time limits for debate, and on procedures allowing a government to force a single vote on a text of its own choosing.
Passage to the other chamber
In bicameral systems the process repeats. Disagreement triggers the reconciliation mechanism.
Final adoption
The definitive vote. The text is now fixed but not necessarily in force, and this is the distinction most often lost.
Constitutional review, where provided
In some systems a law can be referred to a constitutional court before promulgation, typically by the head of state, the government, or a defined number of parliamentarians.
This is a genuine check: an opposition too small to defeat a bill may still be large enough to refer it, which changes what governments dare to draft.
Promulgation and publication
Formal signature by the head of state and publication in the official journal.
In most systems an unpublished rule cannot bind anyone, on the principle that people cannot be held to rules they had no means of knowing. Publication is therefore a condition of validity, not an administrative formality.
The domain of the law
Some constitutions specify which subjects must be regulated by statute rather than left to the executive. The typical reserved list:
- Crimes and penalties
- Taxation
- Fundamental rights and their limitation
- Nationality and civil status
- Property and its expropriation
- The organisation of the judiciary
- The basic organisation of local government
The logic is that these subjects touch liberty and property so directly that only a democratically elected body should decide them.
Where such a division exists, it is legally significant: a decree intruding into the reserved domain is unlawful regardless of its content. Where it does not exist, the boundary is set by whatever each individual statute chooses to delegate, which means it varies subject by subject.
A ministry issues a decree creating a new offence punishable by a fine for failing to register a certain activity.
In a system where crimes and penalties are constitutionally reserved to statute, this is unlawful on its face, and no analysis of the merits is needed. The ministry may not create offences, whatever the enabling statute purported to allow, because the constitution places the subject beyond delegation.
The remedy is not to argue that registration is burdensome. It is to say: this instrument creates a criminal penalty and only a statute may do that. That argument wins or loses in one paragraph.
Contrast a decree setting the form of the registration document. That is administrative detail, properly delegable, and no such objection arises.
What legislatures do besides legislating
- Approve the budget. Historically the origin of parliamentary power: the executive needed money, and granting it was the lever for everything else. Still the most concrete control.
- Scrutinise. Questions, committee inquiries, summoning officials, requiring reports.
- Confirm appointments, in some systems.
- Authorise ratification of treaties.
- Represent. The function that justifies all the others and the one least visible in any legal analysis.
The branch that touches people most, and the one least understood.
Two layers
The political executive: head of state, head of government, ministers. Changes with elections. Sets direction, takes the major decisions, answers politically.
The permanent administration: civil servants, agencies, inspectorates, regulators, local authorities. Persists across governments. Does the work, and produces the overwhelming majority of decisions that affect individuals.
The relationship between the two is one of the recurring problems of governance. A permanent administration provides continuity, expertise and resistance to arbitrary direction. It can also frustrate elected policy, and it is accountable to voters only through a long and weak chain.
What the executive actually does
- Makes rules, through delegated legislation, in vastly greater volume than the legislature.
- Takes individual decisions: licences, permits, refusals, benefits, penalties, authorisations, assessments. Millions per year in any sizeable state.
- Enforces: inspection, investigation, sanction, prosecution.
- Delivers services and manages public property.
- Contracts, as a purchaser on an enormous scale, subject to public procurement rules designed to prevent favouritism.
- Negotiates internationally, subject to the legislature's role in ratification.
Independent regulatory authorities
A defining feature of modern governance and a genuine puzzle for classical theory. Bodies deliberately insulated from ministerial direction, typically with fixed-term appointments, protection from dismissal except for cause, and their own budget and staff.
The justification: certain functions require technical expertise and freedom from the electoral cycle. A central bank that raised rates only when politically convenient would not work. Nor would a competition authority that spared national champions, an electoral commission answering to the incumbent, or a data protection regulator dependent on the ministry it supervises.
The difficulty: accountability. These bodies frequently write rules, investigate breaches and impose penalties, combining all three classical functions while being answerable to voters only indirectly. They are powerful, technical, and hard to challenge politically.
The responses systems have developed:
- Internal separation between investigation and decision, so the same officials do not both build and judge the case
- Mandatory reasoned decisions
- Rights of defence before any sanction
- Appeal to an ordinary or specialised court
- Annual reporting to parliament and appearance before committees
- Audit and, in some systems, review of the authority's own procedures
Discretion and its limits
Where a rule leaves an authority a choice, that is discretion. It exists on a spectrum: from "the authority shall grant the licence if conditions A, B and C are met", which leaves almost none, to "the authority may take such measures as it considers appropriate", which leaves a great deal.
Discretion is necessary. No drafter can foresee every situation, and rigid rules produce absurd results at the edges. But it is never unlimited.
Discretion must be exercised:
- For the purpose the power was granted. Using a licensing power to punish criticism is an abuse regardless of how the decision is dressed.
- On relevant considerations only, and taking all relevant considerations into account. Ignoring something you were obliged to weigh is as much an error as weighing something irrelevant.
- Genuinely, case by case, without a preordained outcome.
- Proportionately, where the system applies that principle to administrative action.
- Consistently, or with an explanation for departing from previous practice.
Fettering discretion: an authority adopts a rigid internal policy and applies it mechanically, refusing to consider whether an individual case warrants departure. Policies are legitimate and promote consistency; refusing ever to look at the individual case is not. The formula courts use is that an authority may have a policy but must be willing to listen to an argument for an exception.
Abuse of power: a power granted for one purpose used for another. Often hard to prove, since authorities rarely state the real reason, so it is usually established by inference from the sequence of events.
A statute says an authority "may grant an extension of the filing deadline where circumstances justify it."
The authority publishes an internal policy: extensions are granted only for documented hospitalisation of the filer.
You apply, citing a fire that destroyed your records. The authority refuses, stating that its policy permits extensions only for hospitalisation.
The defect is not the policy. Having a policy is sensible and promotes equal treatment. The defect is that the authority treated its policy as exhaustive, when the statute gave it a discretion to consider any circumstances that justify an extension. It never asked whether a fire justified one; it asked only whether the case fitted its list.
The argument to make is therefore not "a fire is as bad as hospitalisation." It is: you did not exercise the discretion parliament gave you; you applied a rule of your own invention.
Procedural rights against the administration
These recur across systems and are frequently your only real protection. They matter because they are far easier to establish than substantive error.
- The right to be heard before an adverse decision. Sometimes called the rights of defence or the audi alteram partem principle. Usually requires that you be told what is proposed, on what basis, and given a genuine opportunity to respond before rather than after.
- The right to reasons. A decision must state why, in terms specific enough to allow it to be challenged. A boilerplate paragraph reciting the statutory test without engaging with your case often fails this. An unreasoned decision is frequently annullable on that ground alone.
- Access to the file on which the decision is based, subject to defined exceptions for confidentiality and investigations.
- Impartiality. The decision-maker must have no personal interest and no prior involvement that would compromise them.
- Notification of remedies. Many systems require the decision itself to state how and by when it can be challenged. Where this is required and omitted, the deadline may not start running.
- Decision within a reasonable time, with some systems attaching consequences to silence: either deemed acceptance, or deemed refusal which at least opens the right to appeal.
- Legitimate expectations. Where an authority has clearly induced someone to rely on a position, abruptly reversing it may itself be unlawful, or may require a transition period or compensation.
Court structure
Nearly all systems are hierarchical, with three levels performing genuinely different functions.
First instance
The trial court. Establishes the facts, hears evidence and witnesses, applies the law, and decides.
This is the only level at which facts are genuinely determined. That single sentence explains why the first-instance stage matters far more than people assume, and why weak preparation there is rarely recoverable later.
Appeal
Re-examines the case. In many systems the appeal court may review both facts and law, effectively retrying the matter on the existing record and sometimes with new evidence. In others it is confined to specified grounds.
Whether an appeal is available as of right or only by permission is a matter of the specific procedural code, and turns on the court, the subject matter, whether the order is final or interlocutory, and often the amount at stake. Do not assume a right of appeal exists; establish it.
Supreme or cassation court
In cassation systems, typically reviews questions of law rather than findings of fact, its role being to secure uniform interpretation of the law rather than to correct individual injustice.
This is not a universal description of a supreme court. Some review factual findings under defined standards, some combine cassation with constitutional jurisdiction, and constitutional adjudication routinely involves a factual record. Treat the three-level model as the common shape, not as a rule.
Depending on the system it either substitutes its own decision or quashes the judgment and sends the case back to a different lower court to be reheard, which can add years.
This is why "appealing to the supreme court" rarely means arguing the facts again. If the trial court believed the wrong witness, that is generally final.
The distinction decides what a supreme court will look at, and it is genuinely difficult at the margins.
Fact: what happened. Did he sign the document? Was the vehicle moving? What did the parties say?
Law: what the rule means. Does "vehicle" include a bicycle? What must be proved to establish fault?
Qualification, the operation of deciding whether established facts fall within a legal category, sits between them and is usually treated as reviewable. Whether the agreed conduct amounts to "negligence" is reviewable; whether the conduct occurred is not.
Many civil law systems run parallel court orders: ordinary courts for civil and criminal matters, and a separate administrative court system for disputes with public authorities, each with its own supreme court, plus sometimes a further body to resolve conflicts of jurisdiction between them. Common law systems typically use a single hierarchy for everything, with specialised tribunals feeding into it.
Jurisdiction
Three questions decide which court hears a case:
- Subject-matter jurisdiction: which type of court handles this kind of dispute. Commercial, labour, administrative, family, criminal.
- Territorial jurisdiction: which geographical court. Usually the defendant's domicile, sometimes the place of performance, the place of the harm, or the location of the property.
- Hierarchical jurisdiction: which level, often determined by the value or gravity of the matter.
Getting this wrong wastes months and sometimes destroys the claim, if the limitation period expires while you are in the wrong court. It is also why contracts routinely contain a clause designating the competent court, and why those clauses matter far more than their placement at the end of the document suggests.
Judicial independence
Not a privilege for judges but a condition of the whole system, because a court controlled by one of the parties cannot decide anything. It is secured structurally rather than by good intentions:
- Security of tenure. Judges cannot be dismissed at will, typically only for defined misconduct through a defined procedure before a defined body. Appointment for life or until a retirement age is common precisely because it removes the incentive to please whoever controls reappointment.
- Financial security. Remuneration cannot be reduced as a means of pressure.
- Independent appointment and discipline, often through a judicial council rather than a minister.
- Immunity for judicial acts. A judge cannot be sued personally for deciding a case, however wrongly. The remedy is appeal, not litigation against the judge.
- Random or rule-based case allocation, so cases cannot be steered to a chosen judge. Less discussed than the others and just as important.
- Budgetary autonomy, so the judiciary cannot be starved into compliance.
Independence is structural: the judiciary as an institution is free from external control.
Impartiality is individual: this judge, in this case, has no interest and no bias.
A perfectly independent judiciary can contain a judge who must withdraw from a particular case. The classical formulation of impartiality includes appearances: justice must not only be done but be seen to be done, so a connection that would make a reasonable observer doubt the judge is enough, without proof of actual bias.
Two roles judges have that surprise people
Judges make law. Formally in common law systems through binding precedent. Substantively everywhere, because rules are general and reality is not: applying a text to circumstances its drafters never imagined necessarily creates something new. The pretence that judges merely discover pre-existing law has been abandoned by nearly everyone who studies the question.
Judges cannot refuse to decide. Many civil law systems expressly prohibit a judge from declining to rule on the ground that the law is silent, obscure or insufficient, and doing so is itself an offence.
The judge must find an answer, using interpretation, analogy and general principles. This is why gaps in legislation do not produce legal vacuums, only judicial creativity, and it is a large part of how case law grows in systems that formally deny case law is a source.
A code written long before electricity provides that theft is the fraudulent taking of a thing belonging to another. Someone diverts electricity from a neighbour's supply.
Is electricity a thing? It cannot be picked up, it has no substance in the ordinary sense, and the drafters plainly never considered it.
The judge cannot say "the code does not cover this, come back when parliament legislates." A decision is required.
Route one: interpret "thing" purposively. The provision protects economic value that can be appropriated. Electricity is such value. It is a thing for these purposes.
Route two: refuse to extend, because criminal provisions are construed strictly and analogy is prohibited in criminal law. Acquit, and let the legislature act.
Both are defensible, and legal systems have gone both ways, with several ultimately legislating a specific offence to settle the matter. The example is worth carrying because it shows the two competing pressures in their purest form: the need to decide, and the principle that criminal liability must be foreseeable.
The practical heart of the page. Vocabulary differs between countries in genuinely confusing ways, so learn the categories rather than the words.
The four questions that classify any instrument
- Who made it? Legislature, executive, agency, court, or a private body.
- Under what authority? Directly from the constitution, or delegated by a statute. If delegated, how far does the delegation reach?
- Does it bind, and whom? Everyone, a defined class, one named person, or nobody.
- What follows from breach? Criminal penalty, administrative fine, civil liability, licence withdrawal, invalidity of the act, or nothing.
Answer those four and you know what you are looking at, whatever it is called.
The categories
Adopted by the legislature. Statute, act, loi, Gesetz. Highest ordinary source, slow to make and slow to change.
Creates the framework, and creates the powers for everything below it. Increasingly, framework statutes decide little themselves and delegate extensively, which is why finding the statute often does not answer your question.
Made by the executive under a power conferred by statute. Decree, order, regulation, statutory instrument, arrêté, Verordnung.
Binding and enforceable exactly like a statute, but valid only within the enabling power. Two tests: was the power actually delegated, and did the executive stay inside it?
Courts examine the enabling provision closely, and the broader the delegation the more suspicious some systems become, since an unlimited delegation would amount to parliament abdicating its function.
In some systems the executive holds a regulatory power directly from the constitution for matters outside the legislature's reserved domain, requiring no enabling statute at all.
Where this exists it changes the analysis considerably: the government can legislate in its own right on those subjects, and the question is not "which statute authorised this?" but "is this subject inside or outside the reserved domain?"
A subset of delegated legislation made by specialised regulators. In many systems these follow a mandatory procedure: publication of a draft, a public comment period, a reasoned response to the comments received, then final publication with a statement of basis and purpose.
That procedure is itself legally binding. Failing to follow it can invalidate the rule regardless of its content. This is why consultation periods matter and why filing a comment is genuine participation: an authority that ignores substantial comments without reasoned response is exposed on review.
Not rules but applications of rules to a named person: a licence, a refusal, a penalty, an authorisation, a tax assessment, an expulsion.
Binding on that person only. Normally challengeable before an administrative court, usually within a short and strictly enforced deadline. Missing that deadline generally ends the matter permanently, which is the harshest and most common trap in administrative law.
Binding on the parties through res judicata, subject to appeal and to the scope that doctrine takes in the system concerned, which treats interlocutory orders, procedural rulings and jurisdictional dismissals differently from final judgments on the merits. Binding on future courts in common law systems through the doctrine of precedent, which operates vertically from higher courts to lower ones in the same hierarchy. Civil law systems are not uniformly the opposite: many give constitutional court rulings binding force by statute or constitution, and supreme court jurisprudence can be binding or close to it. The reliable question is not which tradition a country belongs to but what force that particular court's decisions have there.
Guidance, circulars, recommendations, opinions, interpretive notices, supervisory expectations, FAQs, codes of conduct, letters to industry.
Binds nobody. Enormously influential anyway.
The tell is usually in the verbs. Binding instruments say shall and must. Soft law says should, may, the authority expects, good practice would be. Reporting collapses this distinction constantly.
One nuance: soft law can acquire indirect legal effect. An authority that publishes guidance and then departs from it without explanation may breach equal treatment or legitimate expectations. So guidance can bind the authority even while it does not bind you.
The same idea, different words
| Category | France | United Kingdom | United States | European Union |
|---|---|---|---|---|
| Primary legislation | Loi | Act of Parliament | Statute, Act of Congress | Regulation, Directive |
| Delegated legislation | Décret, arrêté | Statutory instrument | Agency regulation | Delegated and implementing acts |
| Individual decision | Décision | Administrative decision | Order, licence, adjudication | Decision |
| Soft law | Circulaire, recommandation | Guidance, code of practice | Guidance, no-action letter, bulletin | Recommendation, opinion, guidelines |
"Regulation" means opposite things. In the United States it usually means delegated legislation made by an agency, ranking below a statute. In the European Union it means a directly applicable act of the Union legislature, functioning as primary legislation. Same word, opposite rank. An article saying "a new regulation requires X" tells you almost nothing until you know which system it is describing.
A directive is not a rule that applies to you. In the EU, a regulation applies directly in every member state without national action. A directive obliges states to achieve a result, and each must legislate to implement it, usually within a transposition deadline of a year or two.
A directive therefore produces national laws that differ from one another and take effect at different times. "The EU has banned X" describes a regulation if X is immediately binding, and something considerably more complicated if it is a directive: one national law per member state, each with its own date, and each with its own variation within the margin the directive leaves.
"Brussels has banned anonymous crypto accounts from 2027."
Six questions:
- Which instrument? Find the name and number. Regulation or directive? The answer changes everything downstream.
- Adopted or proposed? "Brussels has" suggests adopted, but proposals are reported the same way. Check whether it is published in the official journal.
- Who is the obliged party? Read the scope article. If it says obligations apply to "credit institutions, financial institutions and crypto-asset service providers," it binds firms, not individuals.
- What does "anonymous account" mean here? Read the definitions. It will have a precise meaning, probably narrower than the ordinary sense.
- What is the date of application? Not the date of adoption. Look at the final article.
- What do the transitional provisions say about accounts that already exist?
Fifteen minutes with the instrument gives you an accurate answer to a question that a thousand articles will get wrong. The instrument is free and searchable. This is the whole method.
The second structural question, alongside separation of powers. Not who exercises which function, but at which level.
One source of sovereign authority. Sub-national bodies hold only the powers the centre grants, and the centre can alter or withdraw those powers by ordinary legislation.
Unitary does not mean centralised. Most unitary states are administratively decentralised, delegating substantial responsibility to regions and municipalities with their own elected assemblies and budgets. The distinguishing point is that the delegation is revocable by ordinary law.
A related distinction worth knowing: deconcentration means the centre posts its own officials to the regions, who remain answerable to the capital. Decentralisation means locally elected bodies with their own competences and their own democratic mandate. Many states do both simultaneously, which is why a region can contain both an elected council and a centrally appointed prefect with overlapping remits.
Two orders of government, each drawing authority directly from the constitution rather than from the other.
The essential feature is not decentralisation but constitutional guarantee. A federal unit has powers the centre cannot take away without amending the constitution, and it usually participates in that amendment, so it can block its own abolition.
Federations typically also feature: a second parliamentary chamber representing the units, a constitutional court to arbitrate competence disputes, the units having their own constitutions, legislatures and courts, and a fiscal arrangement determining who taxes what.
The intermediate case, and the one that causes most confusion. A unitary state transfers substantial powers to a region by statute. The region legislates genuinely and independently, but the power originates in an ordinary law that the centre could in principle repeal.
Politically this can look identical to federalism, and conventions may make repeal unthinkable. Legally it is not the same, and the difference surfaces precisely when centre and region disagree.
Devolution is also frequently asymmetric: different regions receive different powers, reflecting different histories and demands. That never happens in a classical federation, where the units are formally equal.
An association of states that remain sovereign, with a joint body exercising only conferred powers which the members retain the right to reclaim, and often the right to leave.
Historically unstable: confederations either dissolve or become federations, usually because the joint body cannot act effectively without powers the members will not surrender.
A state's name is not evidence of its structure. Switzerland is styled a confederation and is in substance a federation.
Neither federation nor ordinary treaty organisation. Member states transfer specific competences to common institutions whose acts can bind directly and take precedence over conflicting national law within those competences.
Distinguishing features: acts that apply without national re-enactment, a court whose rulings bind national courts, and a body of law that national judges must apply and give priority.
The European Union is the developed example and fits none of the classical categories, which is why analogies to federal states mislead in both directions.
How competences are divided
Three techniques, usually combined:
- Enumerated powers. The constitution lists what one level may do.
- Residual powers. Everything not enumerated goes to a specified level, usually the units in a federation. Whoever holds the residue has a structural advantage over time, as new subjects arise that nobody listed.
- Concurrent powers. Both levels may legislate, with a supremacy rule determining which prevails on conflict, and sometimes a rule that the units may act only until the centre does.
Two further principles appear in multi-level systems:
Conferral: the higher level holds only the competences expressly conferred on it. Anything not conferred remains with the members. This is the founding principle of supranational organisations and the basis of most competence challenges.
Subsidiarity: even within a shared competence, decisions should be taken at the lowest level capable of achieving the objective adequately. The higher level acts only where the lower cannot. Frequently invoked, hard to enforce judicially, because whether a lower level "could" act adequately is a political judgment dressed as a legal test.
A federal government wants to regulate a new technology. Before anything else, one question: does it have the power?
- Is the subject enumerated? If the constitution grants the centre power over, say, interstate commerce, currency or telecommunications, does this technology fall inside one of those heads?
- If not enumerated, who holds the residue? In most federations the units. So absent an enumerated head, the centre may have no power at all, and the competence belongs to each state or province separately.
- Is it concurrent? If so, the centre may act, and its rule prevails within its sphere, but only within it.
- How broadly is the enumerated head read? This is where the litigation happens. The scope of general commerce or economic powers has been the single most contested question in most federations, because a wide reading effectively gives the centre general legislative power.
The pattern to notice: in a federation, arguments that look like they are about a policy are frequently about competence, and the competence argument is usually the stronger one.
In a federal state, "is this legal in that country?" is often not a well-formed question. Something may be lawful under federal law and prohibited by a constituent unit, or the reverse. It may require authorisation from one level, the other, or both. Two neighbouring units may reach opposite answers on identical facts, and both may be correct.
Anyone giving you a single answer for a large federal state is either summarising heavily or wrong.
The two great Western legal traditions. Roughly: continental Europe, Latin America and much of Asia and Africa follow the civil law tradition; England and its former colonies, including the United States, follow the common law.
Note the vocabulary trap immediately. "Civil law" has two unrelated meanings: the tradition discussed here, and the branch of private law dealing with persons, property, family and obligations. A common law country has civil law in the second sense. Context always disambiguates, but the collision confuses everyone at first.
Where they come from
Civil law descends from Roman law, transmitted through the sixth-century compilation ordered by the emperor Justinian, rediscovered and taught in medieval Italian universities from the eleventh century, and spread across Europe by university-trained jurists who carried a shared learned law across political boundaries. It was eventually expressed in the great national codifications, of which the French Civil Code of 1804 and the German Civil Code that entered force in 1900 are the two most influential models. Most civil law systems descend from one or the other, and the difference between those two models is itself significant: the French tradition favours concise general principles, the German one favours precise technical structure.
Its characteristic move is codification: stating a whole field of law systematically in a single coherent text from which particular answers are deduced. A code is not merely a compilation of existing rules. It aspires to completeness and internal logical order, so that any question can in principle be answered from within it.
Common law descends from the practice of the English royal courts after the Norman conquest, which built a body of law common to the whole realm through accumulated decisions rather than legislation. Its characteristic move is precedent: the rule emerges from decided cases and is refined case by case.
A distinctive feature of the English tradition is the historical split between common law in the narrow sense and equity, a parallel body of principles developed in a separate court to soften the rigidity of the ordinary courts, which had become bound by narrow forms of action. Though the two jurisdictions were merged procedurally in the nineteenth century, the distinction survives substantively. It is the origin of the trust, one of the common law's most consequential inventions and one with no exact civil law equivalent, which is why estate and asset-holding structures differ so sharply between the traditions.
The differences that actually matter
Civil law tradition
- Primary source: codes and statutes. Case law classically not a formal source, though constitutional and supreme court decisions often carry binding or near-binding weight.
- Reasoning: deductive. From the general rule in the text to the particular case.
- Precedent: not binding. A settled line of supreme court decisions nonetheless carries great authority.
- The judge: a career official, typically entering the judiciary young after specialised training rather than after practice.
- Judgments: traditionally short and impersonal, sometimes a single sentence of reasoning. Dissents often not published.
- Procedure: tends inquisitorial. The court directs investigation and evidence-gathering.
- Public and private law: a fundamental division, often with separate court systems.
- Scholarship: influential and routinely cited in argument.
- Gaps: filled by analogy and general principles drawn from the code's own logic.
- Legal education: starts from concepts and systems, then applies them.
Common law tradition
- Primary source: statutes and judicial decisions together. Case law is a genuine source.
- Reasoning: analogical. From decided cases to the present facts by resemblance.
- Precedent: binding. A lower court must follow a higher court in its hierarchy.
- The judge: usually appointed from experienced practitioners, often late in a career. Makes law as well as applying it.
- Judgments: long, discursive, personally authored. Dissents published and sometimes eventually prevail.
- Procedure: tends adversarial. Parties drive the case; the judge umpires.
- Public and private law: a less rigid division. Ordinary courts hear administrative cases.
- Scholarship: historically less cited, though this has softened.
- Gaps: filled by extending existing case law to new facts.
- Legal education: starts from cases, and induces the principle from them.
Concepts worth knowing by name
The common law doctrine of standing by decided matters. It binds vertically, from higher courts to lower ones in the same hierarchy, and to varying degrees horizontally, meaning a court's own past decisions.
Horizontal binding force differs between systems and has generally weakened: most supreme courts now accept they may depart from their own earlier decisions where those were wrong and departure will not cause serious disruption.
Within a judgment, the ratio is the reasoning essential to the outcome, and it is what binds. Obiter dicta are observations made in passing, which persuade but do not bind.
Separating the two is a core skill of common law practice and much of what a first-year student spends the year learning. It is harder than it sounds: judgments do not label their own ratio, several judges may give different reasons for the same result, and later courts effectively determine what an earlier case decided.
Two techniques for escaping an inconvenient precedent.
Distinguishing: the earlier case turned on materially different facts and therefore does not apply. Available to any court, including a lower one bound by the precedent.
Overruling: the earlier decision was wrong and ceases to be law. Only a court of sufficient rank can do it.
Distinguishing is how the common law evolves without admitting error, and a precedent can be distinguished so often and so finely that it survives formally while governing nothing.
The civil law counterpart. No single decision binds, but a consistent series on the same point acquires authority such that departure requires serious justification.
The practical result converges with precedent, reached by a different route and with different formal commitments. A civil law advocate cites a line of decisions to show what the code means; a common law advocate cites a case because it is the law. The citations look similar and the theory behind them is not.
A refinement worth knowing. Where changing a settled rule would unfairly upset arrangements made in reliance on it, some courts announce the new rule while applying the old one to the case before them, or set a future date from which the change takes effect.
It sits awkwardly with the theory that judges declare what the law always was, which is one more reason that theory is not widely believed.
A supplier delivers goods that are defective in a way no one could have detected at delivery. The buyer sues two years later. Is the claim in time?
Civil law approach. Find the code article on limitation for sales claims. It says the action must be brought within a period running from a defined starting point. Then find the article defining that starting point: delivery, or discovery of the defect. Then read the supreme court decisions on what "discovery" means for latent defects. Reason from the text outward, using the decisions to fix its meaning.
Common law approach. Find the limitation statute, then find the cases on when the cause of action accrued for latent defects. Compare their facts to yours. Identify the closest precedent, argue it is analogous, and be ready to distinguish the cases going the other way on their facts.
Notice: both consult text and both consult decisions. The difference is the direction of travel and what carries the argument. The civil lawyer says "the code provides X, as the court confirmed." The common lawyer says "the court held X in a case like this one." Same materials, opposite emphasis, and frequently the same answer.
Mixed systems and convergence
Many systems are genuinely mixed, usually through layered colonial history: Scotland, Louisiana, Quebec, South Africa, Israel, the Philippines. Several Asian systems adopted a European code and subsequently absorbed substantial common law influence through commercial practice and constitutional design.
More importantly, the traditions have converged for a century:
- Common law jurisdictions now legislate massively. Most of what a practitioner works with is statute, and statutory interpretation has become the dominant skill.
- Civil law jurisdictions have developed constitutional courts whose decisions function as binding precedent in all but name, and supreme courts increasingly publish longer, more reasoned judgments.
- Procedural reform has run in both directions: common law systems have adopted stronger judicial case management, civil law systems have strengthened oral hearings and party rights.
- Supranational and international courts blend both traditions by necessity, since their judges come from both.
The distinction remains real, particularly in reasoning style and procedure. The caricature of a code-bound continental judge and a case-spinning English one describes the nineteenth century better than the present.
In a common law jurisdiction, an important question may be settled entirely by a court decision with no statute in sight. Searching the statute book will find nothing and you will wrongly conclude the point is unregulated.
In a civil law jurisdiction, look for the code provision first, then for how the supreme court has read it. Searching case databases without finding the underlying article will give you fragments without the structure.
Searching for the wrong kind of source is the most common way to conclude, incorrectly, that a country has no law on something.
Comparative law recognises more than two families. A short account, because assuming the world divides into civil and common law produces serious errors, particularly about the majority of the world's population.
Systems in which religious law is a formal source of state law. Several states apply religious law to personal status matters such as marriage, divorce, custody and succession, sometimes alongside a civil or common law framework for commercial and public matters, and sometimes with different rules for different religious communities within the same country.
The practical consequence is legal pluralism: which rules apply to a person may depend on their community, not only on their territory. Two neighbours can be subject to different succession rules in the same city.
This creates genuine difficulties in private international law, since the usual connecting factors assume one law per territory.
In many states, particularly across Africa and the Pacific, customary law formally coexists with a received European system. Constitutions frequently recognise it, subject to consistency with fundamental rights, and customary courts may have jurisdiction over defined matters, especially land and family.
The recurring tension is between recognising customary law as an authentic legal order and subjecting it to constitutional rights standards, particularly on gender equality and inheritance. Most modern constitutions attempt both.
States that operated under socialist legal orders, with distinctive doctrines of property, a subordinate judiciary and a leading role for the party, and have since transitioned toward civil law models.
The formal structures converged quickly, often by importing a foreign code wholesale. Institutional practice, particularly judicial independence and administrative culture, changed more slowly, which is a reminder that legal systems are institutions rather than texts.
The commonest situation globally: a colonial-era code, subsequent common law influence, a modern constitution drawing on international human rights instruments, and surviving customary or religious rules in defined areas.
Most of the world lives under something like this rather than under a pure type. When researching a country's law, expect layers and expect them to interact unpredictably.
Public law and private law
Public law governs the relationship between the state and everyone else, and the internal organisation of the state. Its defining feature is inequality of the parties: the state acts with public authority, and the law simultaneously empowers and constrains it.
Private law governs relationships between persons treated as formally equal. Its defining feature is party autonomy: within limits, people arrange their own affairs and the law enforces the arrangement.
The consequences are concrete and worth listing, because they determine strategy:
| Public law | Private law | |
|---|---|---|
| Court | Administrative court, in systems with a separate order | Ordinary civil court |
| Typical remedy | Annulment of the decision | Damages or specific performance |
| Default position | Authority may do only what is permitted | Person may do anything not prohibited |
| Deadline to act | Often very short, weeks or months | Usually years |
| Who initiates | The affected person, against the authority | Either party |
| Can the rules be varied by agreement? | Generally no | Largely yes, subject to mandatory rules |
The public law branches
- Constitutional law. The organisation of the state, allocation of powers, fundamental rights.
- Administrative law. The action of public authorities, its legality, and remedies against it. In practice the branch citizens encounter most, and the one least taught outside law schools.
- Criminal law. Which conduct is punishable and with what penalty. Classified as public law because the state prosecutes on behalf of society, even where a private victim exists. This is why a victim generally cannot "drop charges": the case belongs to the state.
- Criminal procedure. Investigation, arrest, detention, charge, trial. Often treated as a separate discipline because it is where fundamental rights bite hardest.
- Tax law. The levying of public revenue. Dominated by the principle of legality: no tax without a statute.
- Public international law. Relations between states and international organisations.
The private law branches
- Law of persons. Legal personality, capacity, name, domicile, protection of vulnerable adults.
- Family law. Marriage, partnership, parentage, divorce, parental responsibility, matrimonial property.
- Property law. Ownership, possession, real rights, security interests over assets, registration.
- Law of obligations. The engine room of private law: contract, extra-contractual liability, unjust enrichment.
- Succession law. What happens to a person's assets and liabilities on death, including forced heirship rules where they exist.
- Company and commercial law. Business entities, insolvency, negotiable instruments, sale of goods, agency.
- Intellectual property. Copyright, patents, trade marks, designs, trade secrets.
- Private international law. Which country's law applies to a cross-border relationship, and which court decides.
- Civil procedure. How private claims are brought and tried.
The in-between branches
Several fields resist the division, mixing mandatory public rules with private relationships. They are where most modern regulatory activity happens.
- Labour law. An individual contract, but with protective rules the parties cannot waive, plus collective bargaining and public inspection.
- Consumer law. Contract, but with mandatory protections, public enforcement, and rules that reverse the usual assumption that parties bargain as equals.
- Competition law. Private conduct, public investigation and sanctions, and private damages actions on top.
- Financial regulation. Private firms, public supervisors, licensing as the mechanism, and a great deal of soft law.
- Environmental law. Public authorisation, private liability, collective interests, and a persistent standing problem since the harm is diffuse.
- Data protection. Individual rights, private processors, public regulators, and extraterritorial reach.
Substance and procedure
A cross-cutting distinction. Substantive law defines rights and obligations: what you may do, what you owe, what you may claim. Procedural law defines how they are asserted: which court, within what time, in what form, with what evidence.
Non-lawyers underestimate procedure severely. A perfect substantive claim brought in the wrong court, after the deadline, or without admissible evidence, loses. Completely.
The commonest way to lose a good case is procedural, and it is almost always avoidable. This is not a defect of the system so much as the price of finality: rules about time and form exist so that disputes end and so that defendants are not exposed indefinitely.
The building blocks of private law. This part is the densest on the page and the most useful, because these categories are the vocabulary every other legal document assumes you already have.
Legal personality
Human beings. Legal personality generally begins at birth and ends at death.
Two refinements. Some systems protect the unborn conditionally for defined purposes, chiefly inheritance, on the principle that a child later born alive is treated as having existed from conception where that is to its advantage. And death raises the question of what happens to personality rights afterwards, which systems answer differently: the right to reputation may or may not survive.
Entities the law treats as persons: companies, associations, foundations, cooperatives, public bodies, and the state itself.
They can own, contract, sue, be sued, employ, and incur liability, distinctly from the individuals behind them. They are created by a legal act, have their own name, domicile and nationality, and end by a legal act.
This separation is the foundation of company law and of limited liability. It is why a company's debts are not automatically its shareholders' debts. It is also why the exceptional cases in which a court disregards the separation are called piercing or lifting the veil and are treated as genuinely exceptional, typically requiring fraud or the use of the entity as a mere façade.
Capacity
Two distinct concepts, frequently confused, and the confusion causes real errors.
Capacity to hold rights
The ability to be a subject of legal relations at all. To own things, to be owed things, to inherit.
Every person has it. A newborn can own a house.
Capacity to exercise rights
The ability to act validly on one's own behalf: to sell, to contract, to litigate.
Restricted for minors and for adults under protective measures. The newborn who owns the house cannot sell it.
Where exercise capacity is lacking, the law supplies a mechanism: representation by a parent or guardian, assistance requiring a co-signature, or authorisation by a court for significant acts.
Acts performed without the required capacity are typically voidable rather than void: they can be annulled at the request of the protected person, but they are not automatic nullities and the other party generally cannot invoke the defect. The point is protection of one party, not invalidity in the abstract.
Subjective rights
Civil law systems distinguish objective law, the body of rules, from subjective rights, the entitlements individuals hold under those rules. English uses "law" and "a right" and lacks a clean pairing, which causes recurring translation difficulty.
| Type | Against whom | Examples | Survives transfer of the asset? |
|---|---|---|---|
| Real rights | Everyone. Good against the world. | Ownership, usufruct, servitude, mortgage, pledge | Yes |
| Personal rights | One identified person | A contractual claim, a debt | No |
| Intellectual rights | Everyone, over intangibles | Copyright, patent, trade mark | Yes |
| Personality rights | Everyone, over one's own person | Life, bodily integrity, privacy, image, reputation, name | Not transferable |
The distinction between real and personal rights is one of the most consequential in private law, and the last column is why.
You lend money to a friend who owns a building. Two ways to secure it:
Option one: a personal promise. He signs a document promising to repay, and promising not to sell the building until he has. That is a personal right. If he sells anyway, you can sue him for breach. You cannot touch the building, and the buyer takes it free of your claim. If he is insolvent, you rank alongside every other unsecured creditor and probably recover little.
Option two: a registered mortgage. That is a real right over the building. It follows the asset. If he sells, the buyer takes the property subject to your mortgage. If he becomes insolvent, you are paid from the building's value ahead of unsecured creditors.
Same loan, same amount, same debtor. The difference between recovering everything and recovering nothing is which category of right you took, and that difference costs a registration fee and an afternoon.
This is why lawyers care about the taxonomy. It is not classification for its own sake; it is the difference between security and hope.
Ownership
Classically analysed as three components, a Roman division still used across civil law systems:
- Use (usus): the right to use the thing.
- Enjoyment of fruits (fructus): the right to its income and produce, whether natural, such as a harvest, or civil, such as rent.
- Disposal (abusus): the right to alienate, transform or destroy it. The defining attribute, because only the owner has it.
These can be separated. Where use and fruits are held by one person and disposal by another, the arrangement is a usufruct, and it is heavily used in estate planning across civil law countries: a surviving spouse takes the usufruct of the home, the children take the bare ownership, and full ownership reunites on the spouse's death.
Ownership is nowhere absolute. It is limited by planning and zoning rules, neighbour law, expropriation for public purposes against compensation, environmental constraints, cultural heritage protection, and the general prohibition of abusive exercise: using your right for no purpose other than to harm someone.
Ownership is the legal title. Possession is the factual control of a thing with the intention of holding it as one's own.
They usually coincide and can separate: a thief possesses without owning; a landlord owns without possessing.
The law protects possession independently of ownership, which surprises people. The reason is public order: if you could seize your property back by force from whoever holds it, disputes would resolve by strength. Instead, the possessor is protected until a court decides title, and even a wrongful possessor may not be dispossessed by force.
Obligations
A legal bond between two persons by which one, the debtor, owes a performance to the other, the creditor.
Every obligation has three elements: two parties and an object. Remove any and there is no obligation. This is why an obligation "to the world at large" is not an obligation but a duty, and why the person entitled to enforce is always identifiable.
The object of an obligation is classically one of three things: to give (transfer something), to do (perform an act), or not to do (abstain).
Obligation of result: the debtor promises a specific outcome. Failure to achieve it is breach, without more. The creditor proves only that the result was not achieved.
Obligation of means: the debtor promises diligent effort toward an outcome, not the outcome itself. Breach requires proving the effort fell below the required standard.
A carrier who promises delivery owes a result. A doctor owes means: not a cure, but competent care. A lawyer owes means: not victory, but competent representation.
Which category applies determines who must prove what, and it decides a large proportion of professional liability cases. It is often the whole argument.
Sources of obligations
- Contract. Agreement between parties.
- Unilateral undertaking. A promise binding without acceptance, recognised to varying degrees.
- Extra-contractual liability. Tort or delict: the duty to repair harm wrongfully caused.
- Unjust enrichment. The duty to restore a benefit received without legal cause at another's expense. The residual remedy when nothing else fits.
- Negotiorum gestio. Where someone voluntarily manages another's affairs without a mandate, in the other's interest. Recognised in civil law systems, largely absent from the common law, which is generally unsympathetic to uninvited helpers.
- Statute. Obligations imposed directly by law: maintenance duties between relatives, tax, restoration obligations.
Contract: the classical requirements
Terminology varies but the substance recurs:
- Consent, freely given and genuine. Vitiated by error (a mistake about an essential quality), fraud (deception inducing the agreement), or duress (illegitimate pressure). Each renders the contract voidable at the request of the affected party.
- Capacity of the parties.
- A determinate object, possible and lawful.
- A lawful cause or consideration. Civil law asks whether the obligation has a lawful reason; common law asks whether something of value was exchanged. Different tests, overlapping function, and the source of much confusion in cross-border drafting.
Some systems add form for specified contracts: writing, notarisation, or registration for sales of land, mortgages, marriage contracts and wills. Where form is required for validity, its absence voids the contract, not merely the proof of it.
The single most useful distinction for reading any contract or statute governing contracts.
Default rules apply unless the parties agree otherwise. Most of contract law is default: it supplies terms the parties did not think to write, and they can displace it.
Mandatory rules apply regardless of agreement. Consumer protections, employment protections, and rules protecting third parties are typically mandatory. A clause purporting to exclude them is simply void, and the rest of the contract usually survives.
When someone tells you "you signed it, so you are bound", the first question is whether the clause offends a mandatory rule. Frequently it does.
Extra-contractual liability
Across most systems, a claim requires three elements:
- A wrongful act or, in strict liability regimes, simply a defined event.
- Harm. Material (financial or physical), bodily, or moral (suffering, distress, reputational).
- Causation between the two.
The usual starting test is whether the harm would have occurred anyway without the defendant's act. That works for simple cases and fails for several common ones: multiple sufficient causes, where either would have caused the harm alone; long causal chains where an intervening act breaks the link; and loss of a chance, where the defendant's fault deprived the claimant of an opportunity without certainly causing the loss.
Systems have developed doctrines for each, and causation is frequently the real battleground in a case that looks like it is about fault.
Fault-based liability requires proving the defendant fell below the required standard: what a reasonably prudent person would have done in the circumstances.
Strict liability dispenses with fault, attaching liability to a risk created: defective products, dangerous activities, things in one's custody, animals, employees' acts within their functions.
The trend across the twentieth century was steadily toward strict regimes, for a practical reason: proving fault inside a complex organisation or an industrial process is often impossible for a victim, and the party creating the risk is better placed to prevent and insure against it.
A distinction taught early in civil law faculties and genuinely clarifying once you have it.
A manifestation of will intended to produce legal effects. A contract, a will, a resignation, a company resolution, an administrative decision.
The effects follow because someone wanted them. That is the defining feature, and it explains everything about how legal acts can be defective: since the will is what matters, anything corrupting the will corrupts the act. Hence vitiated consent, lack of capacity, and defect of form.
Legal acts are subdivided by how many wills are needed: unilateral (a will, a resignation), bilateral (a contract), or collective (a company resolution).
An event to which the law attaches consequences regardless of anyone's intention. Birth, death, the passage of time, an accident, a natural disaster, reaching majority.
Nobody intends to become liable by causing an accident. Liability attaches anyway, because the law attaches it to the event rather than to a choice.
You claim someone owes you money.
If the debt arises from a loan, that is a legal act. Many systems require proof in writing for loans above a threshold, precisely because the parties could easily have documented it at the time. Testimony alone may be inadmissible. No document, no claim.
If the debt arises from damage to your car, that is a legal fact. Nobody prepares evidence before an accident. Proof by any means is therefore admitted: photographs, witnesses, an expert report, a police record.
The rule is not arbitrary. It tracks what it was reasonable to expect someone to document in advance. Which is also why the practical advice follows immediately: for anything that is a legal act, document it at the time, because the law assumes you could have.
Validity, nullity and enforceability
Three separate questions, constantly conflated:
- Existence: did the act occur at all? An agreement never concluded is not an invalid contract; it is no contract.
- Validity: does it satisfy the legal requirements?
- Enforceability against third parties: can it be invoked against people who were not parties? This often requires publicity: registration of a mortgage, filing of a company charge, recording of a sale of land. A perfectly valid act may bind the parties and be worthless against everyone else.
And two grades of invalidity:
Absolute nullity
Protects the public interest. Unlawful object, missing essential element.
Void from the outset. Anyone with an interest may invoke it. A court may raise it on its own motion. Generally cannot be confirmed by the parties. Longer or no limitation period.
Relative nullity
Protects a particular party. Vitiated consent, incapacity.
Valid until annulled. Only the protected party may seek annulment, and may instead confirm the act. Shorter limitation period, often running from discovery of the defect.
Time
Time creates and destroys rights. Two mechanisms, both called prescription in civil law systems:
Extinctive prescription, or limitation. A claim not exercised within a defined period can no longer be enforced. The rationale is legal certainty and the practical impossibility of defending against very old allegations, when documents are gone and witnesses have died or forgotten.
Periods vary enormously by subject. Shorter periods are common in administrative and commercial matters, longer in property, and some claims, such as those for the gravest crimes, may not prescribe at all.
Acquisitive prescription. Prolonged possession in defined conditions, typically peaceful, public, continuous and as owner, can produce ownership. The rationale is to align legal title with long-settled factual situations rather than leave them permanently uncertain.
Two mechanisms that alter the clock, and confusing them is a classic error.
Suspension pauses the clock, which resumes where it stopped. Typical causes: the claimant is a minor, or is legally unable to act.
Interruption resets the clock, which starts again from zero. Typical causes: commencing proceedings, or the debtor acknowledging the debt.
This is why a debtor's casual written acknowledgement can revive a claim that was days from expiry, and why creditors chase for acknowledgements while debtors are advised not to give them.
How a lawyer gets from a rule and a set of facts to a conclusion. This is the skill a law degree actually teaches, and it is more mechanical than outsiders expect and harder than insiders admit.
The legal syllogism
Major premise: the rule
Identify the applicable rule and state precisely what it requires. This involves finding the rule, checking it is in force, and interpreting it.
Most of the difficulty in this step is that rules do not arrive labelled. Working out which of several plausible provisions governs is itself an argument.
Minor premise: the facts, qualified
Establish what happened, and then characterise it in the rule's terms. This second operation, called qualification or characterisation, is the hidden heart of legal work.
Deciding whether these facts count as a sale or a lease, as negligence or misfortune, as an employee or a contractor, as a security or a commodity. The facts are agreed; the label is fought over, because the label determines the rule.
Most legal disputes are not about what the rule says or what happened. They are about whether what happened falls within the rule's categories.
Conclusion
Apply the rule to the qualified facts and state the consequence.
A rule provides: "An employer must provide paid annual leave to every employee."
A courier delivers for a platform. She works when she chooses, uses her own vehicle, may refuse jobs, and is paid per delivery. She claims paid leave.
Major premise: not in dispute. Employers owe employees paid leave.
Facts: not in dispute either. Both sides agree on how the arrangement worked.
Qualification: the entire case. Is she an employee? That depends on how the legal system defines employment, which typically turns on subordination: control over how work is done, integration into an organisation, economic dependence, who bears the risk.
Notice that the parties will marshal the same agreed facts toward opposite qualifications. The platform emphasises freedom to refuse and ownership of the vehicle. She emphasises algorithmic direction, ratings, and dependence on one source of income.
This is what legal argument almost always looks like. Not "the rule is different" or "that did not happen", but "these facts belong in a different box".
Forms of argument
This case resembles that one in the respects the rule cares about, so it should be treated the same way. The engine of common law reasoning, and used in civil law systems to fill gaps.
The whole force lies in relevant resemblance. Two situations always resemble each other in some respects. The argument works only if the shared feature is one the rule was concerned with.
Its limit: analogy is generally prohibited in criminal law, where creating an offence by resemblance would violate the principle that crimes must be defined in advance.
The rule names a category; therefore it does not apply outside that category. If the text grants a right to spouses, it does not grant it to cohabitants.
Dangerous, because silence may signal deliberate exclusion or mere oversight, and the argument assumes the first without proof. Note that it is the exact opposite of analogy: given a gap, analogy fills it and a contrario leaves it open. Choosing between them is a judgment about what the silence means.
If the rule applies to the lesser case, it applies with greater force to the stronger one. If a bicycle is banned from the park, so is a lorry.
Requires that the cases sit on a genuine scale of the same quality. It fails where the difference is one of kind rather than degree.
The opposing interpretation produces a result the legislator cannot have intended, therefore it is wrong.
Powerful, and only as strong as the shared sense of what counts as absurd. Its cousin is the argument from coherence: an interpretation that makes the instrument contradict itself must be wrong, because texts are presumed internally consistent.
The rule exists to achieve something; the interpretation that best achieves it should prevail.
Dominant in modern courts and in supranational jurisdictions. The most flexible of these tools, which is both its strength and the standing objection to it: if purpose can override text, the drafter's words stop constraining and the interpreter's view of the purpose takes over.
Rules and principles
A distinction worth carrying, associated with Dworkin and widely adopted since.
Rules apply in an all-or-nothing way: either the conditions are met or they are not. Two conflicting rules cannot both be valid; one must yield or be invalid.
Principles have weight rather than triggers. They incline toward an outcome and can be outweighed by competing principles without being invalidated. A principle that loses in one case survives intact and may win in the next.
Good faith, proportionality, legal certainty, equality and the protection of legitimate expectations operate as principles. When two collide, the resolution is a balancing exercise rather than a determination that one is void, which is exactly what the proportionality test in part six is doing.
An authority has for years accepted a particular filing method. It announces that from next month only a new electronic format will be accepted, effective immediately, with no transition.
Principle one, legality: the authority may set the form of filings and is exercising a power it has.
Principle two, legitimate expectations: people arranged their affairs relying on the accepted practice, and abrupt reversal defeats reasonable reliance.
Principle three, proportionality: the objective, modernisation, could be achieved with a transition period at almost no cost to the authority.
Neither principle is invalid, and the authority is not acting unlawfully in the crude sense. The resolution is typically not annulment but a requirement of a reasonable transition, which is what balancing produces: not one side winning outright, but an adjustment.
This is why arguments framed as "you have no power to do this" often fail, while arguments framed as "not like this, not this fast" succeed.
Texts do not apply themselves. Words are ambiguous, drafters are imperfect, language shifts, and circumstances arise that nobody foresaw. Interpretation is unavoidable, and every legal system has developed methods for it.
The classical methods
Start with what the words say, in their ordinary meaning, unless the text defines them otherwise. Always the first step. Where the text is clear, most systems hold there is nothing to interpret, expressed in the maxim that a clear text is not to be interpreted.
The difficulty is that "clear" is itself contested: a text is clear until someone with an interest finds an ambiguity in it, and they usually can.
Read the provision in the context of the instrument as a whole and of the wider legal order.
Three working assumptions follow: a term should normally mean the same throughout a text; provisions should be read so as not to contradict one another; and no provision should be rendered redundant, since the drafter is presumed not to have written pointless words.
What did the drafters intend? Established from preparatory works, explanatory memoranda, committee reports and parliamentary debates.
Its weight varies enormously between systems, and there is a genuine debate about whether the intention of long-dead legislators should govern present cases, and indeed whether a collective body can be said to have had a single intention at all.
What is the rule for? Interpret so as to achieve its objective.
Dominant in modern practice, and particularly strong where instruments contain express statements of purpose in their recitals. It is also the method that best handles new technology, since the drafters' specific expectations are useless while their objective may still apply.
Where a provision admits several readings, prefer the one consistent with a higher norm: the constitution, a treaty, or supranational law.
This allows a court to save a provision rather than strike it down, and it is used constantly. It has a limit: a court cannot interpret a text against its clear wording, so conforming interpretation runs out where the conflict is genuine, at which point the higher norm prevails outright.
A statute drafted decades ago prohibits "vehicles" in the public park. A child's electric scooter appears. Is it prohibited?
Literal: a vehicle is a means of conveyance. A scooter conveys. Prohibited.
Systematic: the same statute elsewhere refers to "vehicles" alongside registration and insurance obligations that plainly contemplate cars. Read consistently, "vehicle" may mean motor vehicles. Not prohibited.
Historical: the debates show the concern was noise, fumes and danger from cars and motorcycles. Nobody contemplated scooters, which did not exist. Weak either way, but leans against prohibition.
Teleological: the purpose is the safety and tranquillity of pedestrians. A fast electric scooter threatens both. Prohibited. A slow one used by a child, arguably not.
Conforming: if a constitutional right to freedom of movement is engaged, prefer the narrower reading. Not prohibited.
Two methods point one way, two another, and one is genuinely ambivalent. This is normal. Methods are not an algorithm producing a single answer; they are arguments a court weighs. Which is why two competent lawyers reading the same provision reach opposite conclusions in good faith, and why the answer is only settled when a court gives one.
Maxims that recur across systems
| Maxim | What it means | Watch out for |
|---|---|---|
| Where the law does not distinguish, we should not distinguish | Do not read in exceptions the text does not contain | Sometimes an exception is genuinely implied by context |
| Exceptions are construed narrowly | A derogation from a general rule is read restrictively | Requires first identifying which is rule and which is exception |
| The expression of one thing excludes others | A list is presumed exhaustive | Only where the list was meant to be complete |
| General words following specific ones take their colour from them | A catch-all at the end of a list is limited to the same class | Needs an identifiable common class |
| Effectiveness | Prefer the reading that gives the provision practical effect | Can be stretched to rewrite the text |
| Penal provisions are construed strictly | Criminal liability is not extended by interpretation | Does not mean absurdly narrow readings |
| Contra proferentem | Ambiguity in a contract is read against the party who drafted it | Mostly for standard-form and consumer contracts |
Presumptions
Courts approach texts with default assumptions, which the drafter must displace expressly:
- Legislation is presumed not retroactive.
- Legislation is presumed not to abolish fundamental rights by implication. Doing so requires clear words.
- Legislation is presumed consistent with international obligations.
- Legislation is presumed territorial unless extraterritorial reach is stated.
- Legislation is presumed not to intend an absurd result.
- Legislation is presumed not to alter settled law by side wind: fundamental changes require clear language, not inference.
These presumptions do real work. Many disputes turn entirely on whether a text was clear enough to displace one, and drafters who know this write express words to avoid the fight.
Having the law on your side is worth nothing if you cannot prove the facts. This is the gap between being right and winning, and it is wider than non-lawyers imagine.
Who must prove what
The general rule, expressed in every system: the person asserting a fact must prove it. The claimant proves the facts founding the claim; the defendant proves the facts founding a defence.
The burden of proof is therefore not a technicality but a decision rule: if a fact remains unproved, the party who had to prove it loses on that point. Cases are routinely decided not because the court found the opposite but because it found nothing established.
Two burdens are usually distinguished. The legal burden is who must ultimately establish the fact, and it does not move. The evidential burden is who must produce enough material to put an issue in play, and it shifts back and forth during a case.
Presumptions shift the burden
A presumption tells the court to treat a fact as established from another fact, without direct proof of it.
Rebuttable presumptions shift the burden to the other side, who may disprove them. The presumption of innocence is the archetype: the accused proves nothing; the prosecution proves everything.
Irrebuttable presumptions admit no contrary proof. Rare, and effectively substantive rules disguised as evidential ones: a rule that a child below a certain age cannot commit a crime is not really about proof at all.
A tenant's flat floods and damages the flat below. The neighbour sues.
Without a presumption: the neighbour must prove the tenant was at fault. She cannot get into the flat, does not know the state of the pipes, and cannot show what happened. She probably loses despite plainly having suffered harm caused from above.
With a presumption, as many systems provide for things in one's custody: harm caused by a thing under a person's control is presumed attributable to that person. The neighbour proves only that water came from the flat above and caused damage. The burden shifts to the tenant to establish an external cause, such as a building-wide failure.
Nothing about the substantive standard changed. Only who must prove what changed, and that alone reverses the outcome. Presumptions are where a great deal of the real policy in a legal system is hidden.
Standards of proof
| Context | Typical standard | Rationale |
|---|---|---|
| Criminal | Very high. Any reasonable doubt benefits the accused. | Liberty and stigma are at stake; wrongly convicting is worse than wrongly acquitting |
| Civil | More likely than not, or the judge's reasoned inner conviction | Both parties are private; no reason to favour either |
| Administrative | Varies; often intermediate where the measure is punitive | Depends how closely the sanction resembles a criminal penalty |
| Interim measures | A serious case and urgency, not full proof | Decided quickly on incomplete material, and reversible |
Means of proof
- Documentary evidence. The strongest in most systems. Some documents drawn up by public officials carry heightened force and can be challenged only through a special procedure alleging forgery.
- Witness testimony. Universally available, universally distrusted by researchers on memory, and still decisive in many cases. Its weaknesses are well documented: confident witnesses are not more accurate, memory reconstructs rather than replays, and questioning shapes recall.
- Expert evidence. A qualified person's opinion on a technical question. In inquisitorial systems typically appointed by the court, producing one neutral report. In adversarial systems typically retained by the parties, producing two opposed reports, with predictable consequences for divergence and cost.
- Admissions. A party's acknowledgement of a fact against their own interest. Powerful, which is why advice to say nothing is advice about evidence, not about guilt.
- Presumptions of fact. Inferences the court draws from established circumstances. Not a formal category so much as ordinary reasoning.
- Physical and electronic evidence. Objects, records, logs, metadata. Authentication and chain of custody have become central: proving what a record is and that it has not been altered is now often harder than obtaining it.
- Inspection. The court examining the site or object itself.
Admissibility
Not all evidence may be used. Systems restrict it for several reasons:
- Unlawfully obtained. Rules vary sharply. Some systems exclude strictly; others weigh the seriousness of the breach against the interest in establishing the truth. One of the areas where the difference between traditions has the most concrete consequences.
- Privileged. Communications between a lawyer and client are protected in essentially every system, and the protection belongs to the client. Other privileges may cover doctors, journalists' sources, and settlement negotiations.
- Prejudicial. Where the risk of improper influence outweighs the value, particularly with juries.
- Formal requirements. Where a system requires certain acts to be proved only in writing.
Evidence is created before the dispute, not during it. By the time you need it, you either have it or you do not.
This is why keeping records, confirming oral agreements in writing, dating documents, and preserving originals matters far more than knowing the substantive law. The best legal position in the world is worth nothing if the facts cannot be established.
A corollary: once a dispute is foreseeable, destroying relevant material is a separate and serious problem, and in many systems it permits the court to draw adverse inferences against you.
Two models
Inquisitorial
- The court directs the proceedings and the search for evidence.
- A judge may investigate, order expert reports, summon witnesses on the court's own motion.
- The file, built progressively, is central; hearings are shorter.
- Proceedings are heavily written.
- The stated objective is establishing the truth.
- Cheaper, since the court rather than the parties bears the investigative effort.
- Risk: an investigating judge who forms an early view may shape the file around it.
Adversarial
- The parties drive the case and gather their own evidence.
- The judge umpires, ruling on admissibility and outcome.
- The oral hearing is central, including cross-examination.
- Extensive pre-trial disclosure between parties.
- The stated objective is resolving the parties' dispute fairly.
- Expensive, since each side investigates independently.
- Risk: outcomes track the resources each side can deploy.
No system is purely one or the other, and both have converged: adversarial systems have increased judicial case management to control cost, and inquisitorial systems have strengthened party rights and oral hearings.
The stages of a case
Standing and admissibility
Before anything else: does this person have the right to bring this claim, in this court, within the time limit, in the correct form?
Many claims die here without the merits ever being examined. Standing generally requires a personal, direct and current interest. Some systems permit associations, public bodies or any citizen to act in a collective interest, and the scope of that is a live question everywhere, particularly in environmental litigation where the harm is shared by everyone and personal to no one.
Commencement and pleadings
The claim is filed and served on the defendant. Service matters: a defendant who was never properly notified can usually have any resulting judgment set aside, sometimes years later.
The defendant responds, and the issues genuinely in dispute are progressively narrowed.
Interim measures
Urgent relief before the final decision: freezing assets, suspending an administrative decision, preserving evidence, ordering interim payment.
Typically requires urgency and a serious case rather than full proof. Often where the practical outcome is really determined, because a party that cannot obtain interim relief may have nothing left to win by the time judgment arrives.
Evidence gathering
Disclosure between parties, expert appointment, witness statements, document production orders.
In adversarial systems this phase is long, expensive and frequently decisive, and it is the main driver of the cost that makes litigation inaccessible.
Hearing
Oral argument, examination of witnesses, submissions. Public in principle, with defined exceptions for privacy, minors, trade secrets and security.
Judgment
The decision, with reasons. Reasoning is generally required: it allows the parties to understand the outcome, permits meaningful appeal, and disciplines the judge, who must be able to justify the result in writing.
Appeal
Within a strict deadline, often short. Some systems require permission. The scope depends on the level, as set out in part eleven.
Finality
Once appeals are exhausted the judgment becomes final and acquires res judicata: the same dispute between the same parties on the same subject cannot be relitigated.
This is what allows disputes to end, and it is why a judgment you consider wrong is nonetheless the answer.
Enforcement
A separate stage with its own procedures: seizure of goods, attachment of bank accounts or earnings, forced sale, registration of charges.
A judgment is not money. Converting one into the other is a distinct exercise that frequently fails, because the defendant has nothing, has moved assets, or is abroad. Checking whether a defendant can actually pay is a step to take before suing, not after winning.
The right to a fair trial
A cluster of guarantees recognised in nearly every constitutional order and every human rights instrument:
- Access to a court. A right that is theoretical if unaffordable, which is why legal aid is treated as part of the guarantee rather than as charity.
- An independent and impartial tribunal established by law. The last phrase matters: ad hoc bodies created for a particular case do not qualify.
- Equality of arms. Neither party placed at a substantial disadvantage relative to the other.
- Adversarial process. Each party sees and can respond to everything the court will rely on. A court deciding on material one side never saw breaches this even if the decision was correct.
- A public hearing and public judgment, subject to defined exceptions.
- Decision within a reasonable time.
- Reasons.
- In criminal matters additionally: presumption of innocence, prompt notice of the charge in a language understood, adequate time and facilities to prepare, legal assistance including free assistance where necessary, interpretation, the right to examine prosecution witnesses and call one's own, and the right not to incriminate oneself.
Alternatives to litigation
- Negotiation. Most disputes end here, invisibly, and this is a feature rather than a failure.
- Mediation. A neutral facilitates agreement without deciding. Increasingly mandatory as a first step in many systems. Confidential, and what is said cannot usually be used later.
- Conciliation. A neutral proposes terms rather than merely facilitating.
- Arbitration. The parties agree that a private tribunal will decide bindingly. Widely used in commercial and international matters. Faster, confidential, with the parties choosing the arbitrators, and generally with very limited appeal, which is either the main attraction or the main risk depending on the outcome.
- Expert determination. A technical question referred to a specialist whose decision the parties accept.
- Ombudsmen and sector schemes. Free or cheap, often binding on the firm but not the consumer, and normally requiring the firm's internal complaints process to be exhausted first.
A supplier in one country contracts with a buyer in another. A dispute arises. Two routes:
Courts. Which country's courts? Each side prefers its own. Whoever loses that argument litigates abroad, in a foreign language, under unfamiliar procedure. Then the winner must have the judgment recognised in the loser's country, which depends on whether a treaty exists between them, and often none does.
Arbitration. The parties chose the seat, the language, the procedure and the arbitrators in advance, in a clause drafted before any dispute existed. And the resulting award benefits from a long-established, very widely adopted framework for recognition and enforcement across borders that court judgments generally lack.
That last point is the decisive one, and it is a procedural fact rather than a substantive one. An arbitral award is easier to enforce abroad than a court judgment, which is why sophisticated cross-border contracts almost always contain an arbitration clause.
What actually follows from a breach. The category matters because different sanctions carry different procedures, different standards of proof and different protections.
Criminal sanctions
Imposed by a criminal court, on behalf of society, following a finding of guilt. Imprisonment, fines, community measures, disqualification, confiscation.
They carry the full set of criminal procedural protections and, importantly, a stigma the law recognises as a distinct harm, separate from the penalty itself.
No offence and no penalty without a prior law defining them. Usually expressed as nullum crimen, nulla poena sine lege.
Four consequences follow, and they are among the most protected principles in any legal system:
- Criminal law is not retroactive, with the single exception that a more favourable later law generally applies.
- Offences must be defined with sufficient precision that conduct can be foreseen. Vague offences are struck down.
- Criminal provisions are construed strictly and not extended by analogy.
- Only a statute, not an executive act, may create serious offences in systems with a reserved domain.
Administrative sanctions
Imposed by a public authority rather than a court: fines, licence suspension or withdrawal, exclusion from a market or from public contracts, publication of the decision.
Procedurally lighter than criminal proceedings, which is precisely why they have grown so much. They are faster, cheaper for the state, and do not require a prosecutor or a jury.
A significant development across many jurisdictions is the recognition that a sanction which is punitive in substance attracts criminal-type protections regardless of its formal label. Severity and purpose, not the name, determine which safeguards apply. A very large "administrative" fine imposed for deterrence may in substance be a criminal penalty, and the person facing it may be entitled to the presumption of innocence, the right not to self-incriminate, and protection against double jeopardy.
This is a genuinely important doctrine and worth knowing about, because it is the main defence against the migration of punishment out of the courts and into administrative bodies.
Civil remedies
- Damages. Money compensating loss. The classical principle is full reparation: restore the victim to the position they would have occupied, no more and no less. This means damages compensate rather than punish, and it is why punitive damages exceeding actual loss are available in some common law systems and viewed with suspicion in many civil law ones.
- Specific performance. An order to do what was promised. More readily available in civil law systems, where it is the primary remedy; in common law systems damages are primary and specific performance more exceptional, granted where damages would be inadequate.
- Injunction. An order to do or stop doing something. Often the only useful remedy where harm is ongoing, and available on an interim basis before trial.
- Termination or rescission. Undoing the contract and returning what was exchanged.
- Price reduction. Keeping a defective performance while paying less for it.
- Restitution. Giving back a benefit unjustly received.
- Declaration. An authoritative statement of the legal position, without any order. Useful where the parties need certainty rather than compulsion.
Administrative remedies
- Annulment. The decision is treated as never having existed, generally with retroactive effect. The characteristic remedy of administrative law, and a powerful one.
- Suspension. Interim halting of a decision pending judgment. Often the remedy that actually matters, since annulment three years later may come too late.
- Injunction against the authority. An order to act, or to retake a decision properly. Available in some systems and not others; where unavailable, an annulled decision can simply be retaken with better reasoning and the same outcome.
- State liability. Compensation for damage caused by unlawful administrative action, and in some systems for lawful action imposing an abnormal and special burden on one person for the common good.
The same conduct can attract several sanctions simultaneously, from different bodies, each with its own standard of proof and its own timetable, and each capable of reaching a different conclusion.
A single incident might produce a criminal prosecution, a regulatory fine, a professional disciplinary measure, and a private damages claim.
An acquittal in criminal proceedings does not preclude civil liability, because the standards differ: not proved beyond reasonable doubt is not the same as shown not to have happened. This surprises people constantly and is entirely coherent once you see that the two proceedings ask different questions.
The limit is double jeopardy: being punished twice for the same facts in proceedings that are both criminal in substance. That is prohibited in most systems, and it is precisely why the classification of administrative sanctions as punitive or not matters so much.
Every rule passes through stages. Reporting routinely describes a rule at stage two as though it were at stage six. Learning the stages inoculates you against most regulatory panic.
Agenda and announcement
A government, party or agency states an intention. A minister gives a speech, a manifesto makes a commitment, a strategy paper sets a direction.
No legal effect whatsoever. Many announcements never become anything, and this stage generates a disproportionate share of headlines because it is cheap to produce and easy to report.
Draft and consultation
A text is published for comment. Where consultation is legally required, this stage has procedural significance: skipping it, or running it as a formality without genuinely considering responses, can invalidate the eventual rule.
It is also the only stage at which an outsider can realistically influence content. After adoption you can only challenge; here you can persuade. Almost nobody outside the affected industry files a response, which is why consultations are dominated by the regulated.
Deliberation and amendment
Committees, readings, negotiation between chambers or institutions, trilogues in supranational systems.
Texts change substantially here. A provision reported alarmingly at stage two frequently does not survive to adoption, and provisions nobody noticed frequently appear.
Adoption
The competent body votes and approves. The text is now fixed.
It is still not necessarily in force. This is the distinction most often lost, and the next two stages explain why.
Promulgation and publication
Formal signature and publication in an official journal or register.
In most systems an unpublished rule cannot bind anyone. Publication is a condition of validity, grounded in the principle that people cannot be held to rules they had no means of knowing.
Entry into force
The date from which the rule actually applies. Frequently later than adoption, sometimes by years. Complex instruments often phase in different provisions on different dates, so the same instrument can be partly in force and partly not.
This is the stage that matters, and the one most often omitted from reporting. A rule adopted today and applying in three years is not a rule you are subject to today. It may also be amended or repealed before it ever applies.
Implementation
Where the instrument requires further action: national transposition, implementing decrees, technical standards, forms, registers, agency guidance.
Until this is done, the framework may exist while the operative detail does not, and firms genuinely cannot comply because nobody has yet specified what compliance looks like.
Enforcement, interpretation and challenge
Authorities apply the rule; courts interpret it; parties challenge it.
What the rule means in practice is often settled here, years after adoption, and can differ substantially from what the text appeared to say. The first enforcement action against a major firm typically does more to define a rule than the text ever did.
Amendment or repeal
Rules are modified and withdrawn constantly. A text read three years ago may no longer say what you remember.
This is why consolidated versions matter and why quoting an instrument from memory is unsafe even for people who work with it daily.
"New law will require identity verification for all transfers."
Six questions, in order:
- Which stage? "Will require" suggests it is not yet in force. Is it adopted, or a proposal? Search for the instrument by name and number.
- If a proposal: at what stage of deliberation? Proposals die routinely and change beyond recognition.
- If adopted: what is the date of application? Look at the final article, not the press release.
- Does it need implementation? If it is a directive or a framework instrument, national measures must follow, on their own timetable.
- What are the transitional provisions? Existing arrangements are frequently grandfathered.
- Who is the obliged party? "All transfers" almost certainly means all transfers by regulated firms.
In a large majority of cases this exercise turns an alarming headline into: a proposal, applying to firms, taking effect in two to four years, with transitional relief. That is a completely different fact.
A general principle in most systems is that rules apply prospectively.
This is strongest in criminal law, where it is usually constitutionally guaranteed, with one universal exception running the other way: where a later law is more favourable to the accused, it generally applies retroactively, because there is no reason to punish under a standard society has abandoned.
Outside criminal law the principle is weaker. Tax and administrative measures with retroactive effect exist, though they normally require particular justification and may be reviewable, especially where they defeat settled expectations.
Separate and often more important is transitional law: how a new rule treats situations already under way. Grandfathering of existing positions, phase-in periods, and rules on contracts concluded before the change are frequently the provisions that actually determine the impact on anyone who already exists. They are also the provisions nobody reports.
Legal systems produce contradictions constantly, because thousands of rules are made by different bodies at different times without anyone reading all of them. Three classical maxims resolve most conflicts.
The higher rule prevails over the lower. A statute beats a decree; a constitution beats a statute.
The hierarchy of norms doing its work. Note the consequence: the lower rule is not merely displaced but is typically invalid, and can be annulled with effect for everyone.
The specific rule prevails over the general, without repealing it for anything else.
A rule for a particular sector displaces the general rule within that sector only. Both remain valid; they simply occupy different territory. This is the most frequently applied of the three and the least noticed.
The later rule prevails over the earlier. Applies between rules of equal rank; a later statute does not override a constitution.
They interact, and the interaction matters. A later general rule does not usually displace an earlier specific one, because specificity is treated as the stronger signal of legislative intent: the drafter of the general rule probably was not thinking about the special case at all.
Three provisions appear to conflict on whether a particular filing is required.
- A is a constitutional provision guaranteeing privacy.
- B is a statute from 2015 requiring filings from all businesses.
- C is a statute from 2010 exempting small enterprises in one sector.
Step one, lex superior. A outranks both. But a constitutional guarantee rarely resolves a filing question directly; it constrains how far B and C may go. Set aside unless the filing is genuinely disproportionate.
Step two, lex specialis. C is specific: one sector, one size category. B is general: all businesses. C prevails for the businesses it covers.
Step three, lex posterior. B is later. Does that displace C? Not automatically, because C is more specific. It would only displace C if B expressly repealed it, or if B is so plainly incompatible with C that both cannot stand.
Result: small enterprises in that sector remain exempt. Everyone else files.
The lesson: apply the maxims in that order, and remember that lex specialis usually beats lex posterior. Non-lawyers instinctively assume the newest rule wins, and that instinct is wrong more often than it is right.
Repeal
Express repeal: the new text states which provisions it abolishes. Clean, and good drafting practice.
Implied repeal: the new text is incompatible with the old, so the old ceases to apply to the extent of the inconsistency. Messy, and a frequent source of litigation. Courts are reluctant to find implied repeal where the texts can be reconciled, precisely because it produces uncertainty about what is in force.
Some systems protect constitutional or fundamental statutes from implied repeal altogether: they can only be repealed by express words, so that a legislature cannot dismantle them by accident.
Conflicts between levels
In federal and supranational systems a fourth rule operates: within the sphere where the higher level is competent, its law prevails over conflicting law of the lower level.
The whole of the argument then shifts to whether the higher level was competent in the first place. This is why competence disputes dominate multi-level litigation: whoever wins on competence has usually won the case.
A rule is not a fact about the world. It is an instruction issued by a particular authority to a particular set of people. Identifying that set is often the whole question.
Bases of jurisdiction
| Basis | What connects the rule to the person | Typical use |
|---|---|---|
| Territoriality | Conduct occurred in the territory | The default for nearly everything |
| Nationality | The person is a citizen | Tax in some systems; certain crimes committed abroad |
| Passive personality | The victim is a citizen | Crimes against nationals abroad |
| Protective | The conduct threatens the state itself | Counterfeiting currency, espionage |
| Effects | Conduct abroad produces consequences inside | Competition law, securities regulation, sanctions |
| Establishment and targeting | The firm is established there, or directs activity at people there | Online services, data protection, consumer law |
| Universality | Nothing. The offence itself justifies jurisdiction | A small set of grave international crimes |
The effects basis is the contested one, because it has no natural limit: almost any economic activity anywhere has some consequence somewhere else. States accept it when they use it and object when others do, which is why extraterritorial regulation generates persistent friction.
The targeting basis is the modern workhorse for anything online, and it is why the location of servers has become almost irrelevant. What matters is whether the service is offered to people in the territory: language, currency, marketing and delivery arrangements are the usual indicators.
Private international law
A distinct discipline addressing cross-border private relationships. It asks three separate questions that are routinely confused, and confusing them wastes enormous effort.
Which court has jurisdiction?
Determined by connecting factors: the defendant's domicile, the place of performance of the obligation, the place where the harm occurred, the location of property, or a choice-of-court clause agreed in advance.
Note that more than one court may have jurisdiction, which produces a race to file, and doctrines to manage it: rules giving priority to the court first seised, or discretion to decline in favour of a more appropriate forum.
Which country's law applies?
A completely separate question. A court routinely applies foreign law. A judge in one country may decide a case entirely under another country's rules, hearing expert evidence on what those rules say.
Conflict rules select the applicable law through connecting factors that vary by subject: contracts often by the parties' choice, failing which by closest connection; property by the location of the asset; succession by the deceased's habitual residence or nationality; torts by the place of the damage; company matters by the place of incorporation or of the real seat.
Two limits on applying foreign law: public policy, allowing a court to refuse where the foreign rule is incompatible with its own fundamental values, and overriding mandatory rules, which apply regardless of the law otherwise selected because the forum considers them essential.
Will the judgment be recognised elsewhere?
A judgment has effect where it was given. Enforcing it abroad requires recognition, governed by treaties or national rules.
Recognition is typically refused where the original court lacked jurisdiction by the recognising state's standards, the defendant was not properly notified, the judgment conflicts with an existing one, or recognition would offend public policy.
A consumer in country A buys online from a seller established in country B. The contract says country B's law applies and country B's courts have jurisdiction. The goods are defective.
Jurisdiction. The clause points to B. But many systems have protective rules for consumers that override choice-of-court clauses, allowing the consumer to sue in their own country where the seller directed activity there. So probably A, despite the clause.
Applicable law. The clause points to B's law. But protective rules commonly provide that a choice of law cannot deprive a consumer of the mandatory protections of their home country. So B's law applies, subject to A's mandatory consumer protections, which is a genuinely hybrid answer.
Recognition. If the consumer wins in A and the seller has no assets there, the judgment must be enforced in B. Whether that works depends on the arrangements between A and B.
Three questions, three different answers, and none of them is what the contract said. This is why the clause at the end of the terms of service is not the end of the analysis.
When you read that a country has adopted a rule, ask: does it reach me, and how?
Usually the answer is that it binds firms established or operating there, and reaches you only indirectly, through the firms you deal with. That indirect reach is real and can be substantial: if every regulated platform must ask you something, you will be asked, and the practical effect is close to being regulated yourself.
It is nonetheless not the same as the rule applying to you personally, and the difference determines who can be penalised for what, who can challenge the rule, and what happens if you deal with someone outside the system.
The idea that power is exercised through law rather than through will. Not a single rule but a cluster of requirements, and the standard against which legal systems are judged.
The formal requirements
A widely used formulation, associated with Fuller, identifies eight ways a legal system can fail. Each is stated here as the requirement rather than the failure.
There must be rules at all, rather than case-by-case decisions. A system that decides each situation on its own with no general standards is not a legal system, whatever else it is.
Rules are published and accessible. Nobody can be bound by a rule they had no means of knowing. This is why publication in an official journal is a condition of validity rather than a courtesy.
Rules govern the future. People cannot comply with a rule that did not exist when they acted, and punishing them for failing to is punishing them for something other than a choice.
Rules are intelligible enough that people can determine what is required. Provisions so vague that conduct cannot be planned around them fail this requirement and are sometimes struck down on that basis alone.
Rules do not contradict one another, and where they appear to, resolution rules exist. A system that requires and forbids the same act makes compliance impossible.
Rules do not demand the impossible. A duty nobody could discharge is not a duty.
Rules do not change so frequently that planning becomes impossible. Related to legitimate expectations: where an authority has induced someone to rely on a position, abrupt reversal may itself be unlawful, or may require a transition period.
Officials actually apply the rules as announced.
A system with excellent published law and arbitrary administration does not satisfy the rule of law. This is the requirement most often absent in practice, and the hardest to fix, because it depends on institutional culture rather than on drafting.
The institutional requirements
- Independent courts able to review official action.
- Effective access to those courts, which requires affordability and reasonable time limits. A right that costs more to vindicate than it is worth is not a right.
- Equality before the law. The same rules for governor and governed, and no immunity from ordinary law by virtue of office.
- Due process before adverse decisions.
- Enforceable judgments, including against the state itself. A state that ignores judgments against it has defeated the whole arrangement.
- Legal certainty, including limitation periods and res judicata, so that matters can be settled and stay settled.
Thin and thick conceptions
A genuine and unresolved debate.
The thin conception holds that the rule of law concerns the form of legal governance only: generality, publicity, clarity, non-retroactivity, independent adjudication. A system meeting these requirements satisfies the rule of law regardless of the content of its rules.
The thick conception holds that it also requires substantive commitments: fundamental rights, democratic legitimacy, human dignity. On this view a system can be procedurally impeccable and still fail.
The disagreement matters because it determines whether "rule of law" is a technical description or a moral endorsement.
- The thin conception is analytically cleaner, but it can seem to launder oppressive systems: a regime that publishes its unjust rules clearly and applies them consistently would satisfy it.
- The thick conception captures what people usually mean, but it risks collapsing into a general theory of justice, at which point the phrase stops doing any distinct work.
A common middle position: the formal requirements are genuine constraints even on bad regimes, because generality, publicity and consistency make certain kinds of arbitrariness harder. They are not sufficient for a just system, and they are not nothing.
These criteria are a diagnostic. When something feels wrong about how a rule is being applied, one of them is usually being violated, and naming it converts an inchoate objection into an argument.
- Was there a legal basis?
- Was the rule published before it was applied?
- Is it clear enough to comply with?
- Is it being applied to conduct that predates it?
- Are officials following it consistently, or is my case being treated differently?
- Can it be tested before an independent court, in time, at a cost I can bear?
Those six questions are more useful than any amount of indignation, and they are the questions a lawyer would actually ask.
Legal texts follow a convention. Once you know it they become far less intimidating, and reading the source directly is almost always faster than reading commentary about it.
The usual anatomy
- Title and identifying number. Tells you the type of instrument and its date, which tells you its rank.
- Citation of legal basis. Which provision authorised this instrument. If the enabling power does not stretch far enough, the instrument is vulnerable. This is where a serious reading starts.
- Preamble or recitals. Purpose and context. Not usually binding, but used to interpret the operative provisions. In some instruments this section is long and genuinely informative. Read it first.
- Subject matter and scope. What the instrument covers, and to whom it applies. Read second, before anything alarms you.
- Definitions. Read third, and read them properly. Defined terms mean exactly what the instrument says, which is frequently not what they mean in ordinary speech. Most misreadings happen here.
- General principles, in longer instruments, which colour everything that follows.
- Operative provisions. The actual obligations.
- Exemptions and derogations. Who escapes and on what conditions. Often where the real content is, and frequently where you find that the instrument does not apply to you at all.
- Supervision and enforcement. Which authority acts, with what powers.
- Penalties. What follows from breach.
- Transitional provisions. How existing situations are treated during phase-in.
- Amendments to other instruments, often buried near the end and frequently more significant than anything in the main text.
- Entry into force and application. Usually at the very end, and frequently the single most important provision in the text.
- Annexes. Technical detail, lists, thresholds, forms. Often where the substance actually lives.
Language that carries weight
| Wording | What it signals |
|---|---|
| shall, must, is required to | Obligation. Binding. |
| shall not, may not, is prohibited | Prohibition. |
| may, is entitled to | Permission or discretion. Not an obligation. |
| should, is expected to, is encouraged to | Almost always soft law. Not binding. |
| without prejudice to | This provision does not displace the rule referred to. |
| notwithstanding | This provision overrides the rule referred to. |
| subject to | This provision yields to the rule referred to. |
| mutatis mutandis | Apply the other provision with the necessary changes. |
| for the purposes of this Article | The definition applies here only, not throughout. |
| in particular, including, such as | The list is illustrative, not exhaustive. |
| exclusively, only where, solely | The list is exhaustive. |
| by way of derogation from | An exception to the rule named. Construed narrowly. |
| where appropriate, as far as possible | A qualified obligation. Considerably weaker than it looks. |
| Member States shall ensure that | An obligation on the state, not on you. Requires national implementation. |
You open a thirty-article regulation and Article 12 says every operator must appoint a compliance officer. Alarming.
Wrong approach: read from Article 1 forward, panic at Article 12, stop.
Right approach, in this order:
- The final article. Application date. Suppose it is three years away. The urgency evaporates immediately.
- The scope article, usually Article 1 or 2. Suppose it applies to "operators providing services on a professional basis." Are you one?
- The definitions article. "Operator" will be defined. Suppose it means an entity providing services to third parties for remuneration. If you act only for yourself, you are outside the instrument entirely and Article 12 never applied to you.
- The exemptions article. Even if you are inside, there may be a threshold below which the obligation does not bite.
- Only then, Article 12, read with everything above in mind.
- The transitional provisions, for what happens to existing arrangements.
Four articles out of thirty answer the question, and none of them is the one that caused the alarm. This ordering is the single most useful habit on this page.
Three habits worth building
Always read a consolidated version. The original text is a historical document. Amendments accumulate, and the version you want incorporates all of them. Official databases usually offer both, and reading the wrong one is a common and entirely avoidable error.
Follow every cross-reference at least once. Legal instruments are networks, not documents. A provision that seems clear frequently depends on a definition, an exemption or a threshold located somewhere else, sometimes in a different instrument entirely.
Read scope and definitions before the operative provisions. Most people read a legal text in order, hit an alarming obligation, and stop. Half the time the scope provision three articles earlier excluded them entirely.
The great principles of public law
The rules that govern how the state may act. Every encounter with an authority runs on these, and almost nobody outside law knows them by name.
Public law has a small number of organising principles that recur in every system, whatever the local vocabulary. Learning them by name is worth more than learning any particular country's administrative code, because they are the arguments that actually work.
Every act of a public authority must have a legal basis. The authority may do only what it is empowered to do, and only in the manner and for the purpose the empowering rule provides.
This is the foundational asymmetry: a private person may do anything not forbidden; an authority may do only what is permitted. An authority acting without a basis acts unlawfully even where the act is beneficial, popular and well intentioned.
The practical form: "under what provision are you doing this?" That question is always legitimate, and the answer is frequently unsatisfying.
An administrative act must conform to every norm above it: the constitution, treaties, statutes, and higher administrative rules. An act conforming to its parent decree but violating the statute above that decree is unlawful.
This is why testing a requirement means climbing the whole pyramid, not stopping at the first document that mentions it.
Like situations must be treated alike, and unlike situations differently in proportion to their difference. Distinctions are permitted where they rest on a relevant difference and pursue a legitimate purpose; otherwise they are discrimination.
Two useful corollaries. An authority departing from its own established practice must explain why. And a general rule applied to a genuinely exceptional case without consideration may breach equality precisely by treating unlike cases alike.
A measure must be suitable to its aim, necessary in the sense that no less restrictive alternative would work as well, and not impose a burden excessive relative to the benefit.
Applies both to limitations on rights and, in many systems, to administrative measures generally. The necessity limb is where most measures fail, and it is the argument to reach for first.
People must be able to know the rules and plan around them. Rules must be published, clear, and reasonably stable.
Where an authority has induced someone to rely on a position, by a decision, an assurance, or a consistent practice, reversing it abruptly may itself be unlawful, or may require a transition period or compensation. The expectation must be legitimate: induced by the authority, reasonable, and not contrary to law.
An administrative decision takes effect for the future. Retroactive effect requires an express legal basis and is exceptional.
A separate rule governs the withdrawal of favourable decisions: many systems allow an authority to withdraw an unlawful favourable decision only within a short window, after which the beneficiary's reliance prevails even though the decision was wrong.
Before any decision adversely affecting a person, that person must be informed of what is proposed and on what basis, and given a genuine opportunity to respond. Often expressed as audi alteram partem.
Genuine means before the decision, not after; with enough information to respond meaningfully; and with enough time. A consultation conducted after the decision has been taken satisfies nothing.
An adverse decision must state the factual and legal grounds on which it rests, specifically enough that the person can understand it and challenge it.
Boilerplate reciting the statutory test without engaging with the case frequently fails. Where reasons are required and absent, annulment often follows without any examination of the merits, which makes this one of the most efficient grounds available.
The decision-maker must have no personal interest in the outcome and no prior involvement compromising them. Includes appearances: the test is usually whether a reasonable observer would doubt impartiality, not whether bias is proved.
Essential public services must operate without interruption. The classical justification for restrictions on strikes in some sectors, for requisition powers, and for rules ensuring that a service continues when a contractor fails.
The administration may modify how a service operates to meet changing needs. Users of a public service have no vested right in its continuation in a particular form, which distinguishes the relationship sharply from a contract.
A power must be exercised for the public purpose it was granted. Using it for another purpose, however legitimate that other purpose may be in itself, is an abuse and a ground of annulment.
Using a fire safety power to close a business you dislike is unlawful even if the business is genuinely unpleasant, and even if some fire risk exists.
Administrative documents are in principle accessible, subject to defined exceptions for privacy, commercial secrets, security and ongoing investigations. The principle is access with exceptions, not secrecy with concessions, and that ordering matters when an authority refuses.
Every administrative decision must be challengeable before an independent court, within a real opportunity to do so: a knowable deadline, an affordable procedure, and a remedy capable of repairing the harm.
A right of appeal that is theoretical, because it is unaffordable or because the decision will have taken irreversible effect before it is heard, does not satisfy the principle. This is why interim suspension matters so much.
A regulator withdraws a licence you have held for six years, by letter, effective immediately, citing "concerns regarding the applicant's suitability." You were not contacted beforehand.
Rather than arguing you are suitable, which is the hardest ground and the one the regulator is best placed to judge, run through the principles:
- Rights of defence: you were never told what the concerns were or given a chance to answer. Self-contained defect.
- Duty to give reasons: "concerns regarding suitability" is not reasons. It identifies no fact and no provision.
- Proportionality: withdrawal is the most severe measure available. Was a condition, a warning or a suspension considered?
- Legitimate expectations: six years of holding the licence, presumably with inspections passed, created a position you relied on.
- Effective judicial protection: immediate effect means the business is destroyed before any appeal is heard. Seek interim suspension first, before anything else.
Five arguments, none of which requires you to prove you are suitable. This is how administrative litigation is actually conducted, and it is why procedure is not a technicality.
The great principles of private law
The rules governing relations between equals: contracts, property, liability and family. Older than any modern state and remarkably stable across systems.
Private law changes far more slowly than public law. Many of these principles are Roman, and a lawyer from two thousand years ago would recognise most of them. That stability is exactly why they are worth learning.
Foundational principles
Persons may arrange their own affairs by agreement, and the law will enforce the arrangement. You choose whether to contract, with whom, and on what terms.
Three limits, everywhere: mandatory rules that cannot be excluded, public policy and morality, and the rights of third parties.
Agreements are to be kept. A contract validly formed binds the parties as if it were law between them, and neither may unilaterally withdraw.
The classical exception is hardship or imprévision: where circumstances change so fundamentally that performance becomes ruinous rather than merely unprofitable, some systems permit renegotiation or judicial adaptation. Common law systems are more reluctant, recognising a narrower doctrine of frustration that discharges the contract rather than adjusting it.
Parties must behave honestly and loyally in negotiating, forming and performing obligations. It generates duties nobody wrote down: to inform, to cooperate, to not exercise a right purely to harm.
A general principle in civil law systems. The common law traditionally resisted a general duty, preferring specific doctrines, though it has moved considerably toward the same results by other routes.
A contract binds the parties and neither benefits nor burdens third parties. You cannot impose an obligation on someone who did not agree.
Two qualifications. Contracts can confer benefits on third parties where the system allows a stipulation for another. And third parties must respect the existence of a contract, so deliberately inducing its breach can itself be a wrong.
The default is that agreement alone forms a contract: no writing, no ceremony, no registration.
Formality is the exception, imposed where the stakes justify slowing people down: transfers of land, mortgages, wills, matrimonial arrangements, guarantees. Where form is required for validity, its absence voids the act rather than merely complicating proof.
Legal facts may be proved by any means. Legal acts frequently require written proof above a threshold, on the reasoning that the parties could easily have documented what they deliberately did.
Principles of liability
Harm nobody. The general duty not to cause wrongful damage, from which the whole of extra-contractual liability descends.
Damages restore the victim to the position they would have occupied: the whole loss, and nothing beyond it.
Two consequences. Compensation is measured by the victim's loss, not the wrongdoer's gain or fault, so a very careless act causing small harm produces small damages. And punitive awards exceeding the loss sit uneasily with the principle, which is why they are contested in many systems.
A person suffering harm must take reasonable steps to limit it. Loss that reasonable action would have avoided is generally not recoverable.
Where the victim's own conduct contributed to the harm, compensation is reduced proportionately, and in extreme cases extinguished.
A person who creates or profits from a risk answers for its realisation, without proof of fault. Products, dangerous activities, things in one's custody, employees.
The justification is practical rather than moral: the risk-creator can prevent, insure and price the harm; the victim usually cannot prove what went wrong inside someone else's operation.
Principles of property
Parties may invent any contract they like, but they may not invent new categories of real right. The list is fixed by law: ownership, usufruct, servitude, mortgage, pledge and a few others.
The reason is that real rights bind the whole world, including future buyers who never agreed to anything. Allowing private parties to create novel world-binding rights would make property unknowable.
Rights that bind third parties must be discoverable by them, through registers of land, companies, security interests and matrimonial regimes.
The corollary: an unregistered right may be perfectly valid between the parties and worthless against anyone else. This single point decides an enormous number of insolvency disputes.
The possessor is protected against disturbance independently of who owns the thing, and even a wrongful possessor may not be dispossessed by force.
The purpose is public order. If self-help were permitted, disputes would resolve by strength rather than by title.
Nobody can transfer more rights than they have. A seller who does not own cannot make you owner.
Every system tempers this to protect commerce, typically protecting a good faith purchaser who buys movables from someone in apparent possession, or who relies on a public register. The tension between protecting true owners and protecting honest buyers is one of the oldest arguments in private law.
Principles of succession
This is one of the sharpest divides between the traditions, and getting it the wrong way round is expensive.
Civil law: universal succession. Heirs succeed to the whole legal position of the deceased, assets and debts together, and the estate vests in them directly at the moment of death. Unqualified acceptance can therefore expose an heir personally to debts exceeding the assets. That is exactly why these systems build in safety valves: acceptance limited to the value of the net assets, or outright renunciation, usually within a deadline.
Common law: administration by a personal representative. The estate vests first in an executor or administrator, who collects the assets, pays the debts, and distributes only what is left. A properly acting personal representative is liable only to the extent of the assets that passed through their hands, and beneficiaries are not personally liable for the deceased's debts at all. There is no acceptance step carrying that risk.
The practical consequence: advice about whether to accept or renounce an inheritance, and warnings about inheriting debt, are civil law advice. They do not translate to England, the United States or other common law jurisdictions, where the question does not arise in that form.
The two great models. Common law systems generally permit a testator to leave their estate to whomever they wish, subject to family provision claims. Civil law systems typically reserve a fixed share for close relatives, which the testator cannot defeat.
This is one of the sharpest divergences in comparative private law, and it means an estate plan that is perfectly effective in one system can be partly void in another.
Wills are subject to strict formal requirements because the person who could confirm their authenticity is dead. Formality substitutes for testimony.
You agree with a buyer that they will purchase your building on condition that a room stays available to your elderly aunt for life.
As a contractual promise: binds the buyer personally. If they sell to someone else, the aunt has nothing against the new owner, only a damages claim against the buyer, who may have disappeared.
As a real right: you must use a category the law already recognises, such as a usufruct or a right of habitation. You cannot invent "a right for my aunt to use the blue room". That is numerus clausus.
And it must be registered. An unregistered usufruct may bind the buyer who agreed to it and not the person who buys from them. That is publicity.
Two principles, one arrangement, and the difference between a right that lasts a lifetime and a promise that evaporates on the next sale.
Regulation changes every year. The law of obligations changes every century.
If you learn one body of law properly, learn this one. It underlies every contract you will ever sign, every asset you will ever hold, and every estate you will ever inherit, and it will still be true in thirty years.
The great principles of criminal law
Where the state's power is at its sharpest, and where the constraints on it are correspondingly strictest.
Criminal law is the branch where the state does the most severe things to individuals. Every principle here exists to constrain that power, and most were won at some cost.
The principle of legality and its four consequences
No offence and no penalty without a prior law defining them. The foundational principle, from which four separate rules follow.
Offences must come from a written legal source, and in most systems from a statute rather than an executive act. Custom cannot create a crime, and neither, in principle, can a court.
Conduct is judged by the law in force when it occurred. You cannot be punished for something that was lawful when you did it.
The one exception runs in the defendant's favour: where a later law decriminalises conduct or reduces a penalty, it generally applies retroactively, because there is no justification for punishing under a standard society has abandoned. This asymmetry is deliberate and universal.
Offences must be defined precisely enough that a person can foresee what is prohibited. A provision so vague that nobody can tell what it catches fails, and can be struck down for that reason alone.
This is the principle behind challenges to offences framed in terms like "conduct contrary to the public interest".
Criminal provisions are read narrowly, and analogy cannot be used to extend an offence to conduct the text does not cover.
Where a gap exists, the answer is acquittal and legislation, not creative interpretation. This is why the electricity example in part eleven, on the judiciary, is genuinely hard: extending "thing" to cover electricity is arguably exactly what this principle forbids.
Principles of responsibility
A person answers for their own acts, not for another's. Collective punishment and liability for family members are prohibited.
The corollary, sometimes forgotten: criminal liability generally does not pass to heirs. A fine imposed on a person who then dies is commonly extinguished rather than enforced against the estate, though the position varies and confiscation of proceeds frequently survives. Check the rule where you are rather than assuming either outcome.
A crime classically requires both an act and a state of mind. The act, sometimes called actus reus, is the prohibited conduct or omission. The mental element, mens rea, is intention, knowledge, recklessness or negligence as the offence requires.
Where the required mental element is absent, there is no offence, however harmful the outcome. Some minor regulatory offences are strict, requiring no mental element, and their expansion is a persistent subject of criticism.
Thoughts are not punishable. Status is not punishable. There must be conduct, or an omission where a duty to act existed.
The penalty must be proportionate to the gravity of the offence and the culpability of the offender. Grossly disproportionate penalties can be unconstitutional independently of whether the offence itself is valid.
Sentencing considers the particular offender and the particular circumstances, within the range the law sets. Mandatory fixed penalties sit in tension with this and are frequently litigated.
Procedural principles
The accused proves nothing. The prosecution proves everything, to a high standard. Where doubt remains, the accused benefits, expressed as in dubio pro reo.
It extends beyond the trial: it constrains how officials and, in many systems, the press may speak about a person before conviction.
Nobody may be compelled to contribute to their own conviction. Includes the right to silence, and the rule that silence should not by itself be treated as evidence of guilt.
Its boundaries are genuinely contested where compulsion is used to obtain documents, passwords or keys, since these are arguably pre-existing material rather than testimony. Systems differ, and this is an active area.
Prompt notice of the charge in a language understood, adequate time and facilities to prepare, access to the file, legal assistance including free assistance where necessary, an interpreter, and the right to examine prosecution witnesses and call one's own.
No one may be tried or punished twice for the same facts. The protection extends to proceedings that are criminal in substance whatever their label, which is why very severe administrative penalties raise the question.
Evidence must be obtained lawfully. Systems differ on the consequence: strict exclusion in some, a balancing exercise in others weighing the seriousness of the breach against the interest in establishing the truth.
Pre-trial detention is exceptional, must be justified by specific risks such as flight, interference with evidence or repetition, must be reviewed periodically, and must not become an anticipated sentence.
Criminal trials are public, so that justice is done visibly. Exceptions exist for minors, privacy and security, and they are construed narrowly.
Proceedings must conclude within a reasonable period. Excessive delay is itself a violation and may lead to reduction of sentence or, in extreme cases, termination.
A statute criminalises "operating a financial service without authorisation". A regulator prosecutes someone for running software that other people used to transact with each other.
Rather than arguing the software was harmless, run the four limbs of legality:
- Lex scripta: is the definition of "financial service" in the statute, or in guidance? If the conduct is caught only by the regulator's interpretation and not by the text, that is fatal.
- Lex praevia: did the interpretation catching this conduct exist when the conduct occurred, or was it announced afterwards?
- Lex certa: could a person reading the statute have foreseen that writing and publishing software fell within "operating a service"? If not, the provision is too vague as applied.
- Lex stricta: is the prosecution extending "operating" by analogy from businesses that hold customer funds to a person who holds nothing?
None of these arguments requires denying the facts or defending the outcome. They ask whether the state gave fair warning, which is the question criminal legality exists to force.
The great principles of tax law
Not rates, not thresholds, not this year's rules. The structural principles that determine how any tax system works and how to read any tax question.
Tax is the branch people most want a shortcut on and where a shortcut is most dangerous. Rates and thresholds change annually and are worthless to memorise. The principles below have been stable for a century and answer most questions before you reach the rates.
Constitutional principles
A tax may be imposed only by statute, adopted by the elected legislature. The executive cannot create a tax, widen a tax base, or raise a rate by decree.
Historically the origin of parliamentary power itself: legislatures acquired authority by controlling the purse. It remains the strongest structural protection any taxpayer has, and it is why "the tax authority has decided that..." is always worth testing against the statute.
Tax is authorised for defined periods, typically annually through the budget. The consent of the represented is renewed rather than granted once.
Taxpayers in comparable situations must be taxed comparably. Distinctions require a relevant difference and a legitimate objective.
This grounds two familiar concepts. Horizontal equity: equal treatment of those with equal capacity. Vertical equity: appropriately different treatment of those with different capacity, which is the justification for progressive rates.
Tax should track economic capacity. It underlies progressivity, the deduction of costs incurred in earning income, allowances for dependants, and relief where a taxable event produces no cash.
The recurring practical problem this principle raises is liquidity: a tax charge on an unrealised or non-cash gain can exceed the taxpayer's available cash, which is why systems build in deferrals, instalments and rollovers.
Tax rules should apply to future periods. Retroactive tax legislation is not always prohibited outright, but it requires serious justification and is reviewable, particularly where it defeats arrangements made in reliance on the previous position.
A taxpayer must be able to determine their liability in advance. This is the principle behind advance rulings, published administrative positions, and the protection of legitimate expectations where an authority has given a clear assurance.
Tax being a burden imposed by the state, provisions creating liability are generally construed strictly, and doubt is not resolved in the treasury's favour by default.
The mirror rule: exemptions and reliefs are also construed strictly, since they are exceptions to the general charge.
Structural concepts
Every tax attaches to a defined event: earning income, realising a gain, making a supply, transferring property, holding an asset on a given date, dying.
Identifying the taxable event is the first move in any tax question, and usually the decisive one. Most confusion about the taxation of anything dissolves once you ask precisely which event the statute attaches to.
Most systems tax gains when they are realised through a disposal, not as they accrue. A rising asset generates no charge until something happens to it.
Two consequences. Timing becomes a genuine variable a taxpayer controls, which is the foundation of most legitimate planning. And the definition of "disposal" carries enormous weight: whether an exchange, a conversion, a transfer between one's own accounts, or a use as payment counts as a disposal decides everything downstream.
A gain is proceeds minus basis. The basis is what the asset cost, adjusted for defined items.
Where identical units were acquired at different times and prices, the system needs an identification convention to decide which unit was sold: first in first out, average cost, or specific identification. The convention can change the taxable amount dramatically without changing anything about the transaction, which is why knowing which one applies matters more than knowing the rate.
Two models. Territorial: tax income arising in the territory. Worldwide: tax residents on income wherever it arises. Most systems combine both, taxing residents on worldwide income and non-residents on domestic-source income.
The consequence is overlap, which is what double taxation treaties exist to resolve, through residence tie-breakers, allocation of taxing rights by category of income, and relief by exemption or credit.
The two connecting factors. Residence attaches tax to a person, determined by physical presence, home, centre of vital interests, or place of management for entities. Source attaches tax to income, determined by where the activity, asset or payer is located.
Both are legal constructs rather than facts, which is why a person can be resident in two places at once and why treaties need tie-breaker rules.
Tax follows the economic reality of a transaction, not merely its label. Calling a payment a loan does not make it one if it functions as remuneration.
The counterweight is legal certainty: taxpayers are entitled to rely on the legal form of what they actually did. The tension between these is the central argument in most sophisticated tax disputes.
The distinction that matters most, and the one most often blurred.
Evasion is concealing or misrepresenting facts to reduce tax. Unlawful in every system that levies the tax concerned, and treated as a criminal offence in most, though whether a given instance is prosecuted criminally or dealt with through civil penalties depends on the jurisdiction, the amount and the degree of intent that can be established.
Avoidance is arranging genuine affairs to attract less tax. Lawful in principle, on the classical proposition that nobody is obliged to arrange their affairs to maximise the revenue.
Between them sits abusive avoidance: arrangements that are legally genuine but wholly artificial and designed only to obtain a tax advantage contrary to the purpose of the rule relied on. Most systems now have a general anti-abuse rule allowing such arrangements to be disregarded. The line between planning and abuse is drawn by purpose and artificiality, and it is genuinely difficult.
Tax should not distort economic choices without reason. Similar activities should bear similar burdens, so that decisions are driven by economics rather than by tax.
Frequently sacrificed deliberately, when tax is used to encourage or discourage behaviour, but it remains the default from which departures must be justified.
If gains are taxable, losses should be relievable. Systems routinely limit this by restricting loss relief to the same category of income, or by capping or deferring it, and each restriction is a departure from the principle that requires justification.
Administrative and procedural principles
Most modern systems require the taxpayer to compute and declare their own liability, with the authority verifying afterwards. This shifts the initial burden onto the taxpayer and makes accurate record-keeping a legal obligation rather than good practice.
This allocation differs fundamentally between systems, and assuming the wrong one is one of the more expensive mistakes available.
In some systems the burden is shared and asymmetric. The authority bears the burden of establishing that income exists or that an assessment is justified; the taxpayer bears the burden of establishing entitlement to a deduction, relief or exemption.
In the United States the default runs the other way, and it matters enormously. A notice of deficiency carries a presumption of correctness, and the burden sits on the taxpayer to prove it wrong. That has been the rule since Welch v. Helvering in 1933, and it is written into the Tax Court's own procedural rules. It covers not only deductions and reliefs but whether the income existed at all. You do not begin from a position where the authority must prove your gain occurred; you begin from a position where you must disprove their figure.
Two genuine exceptions cut against that default, and both have conditions:
- The naked assessment doctrine. The presumption does not attach to an assessment that is arbitrary and without foundation. In unreported-income cases courts require the authority to show some minimal evidentiary link before the presumption applies at all.
- Statutory burden-shifting. Where an individual taxpayer introduces credible evidence on a factual issue, and has complied with substantiation and record-keeping requirements and cooperated with reasonable requests, the burden on that issue shifts to the authority. The prerequisites do the real work, and the taxpayer must establish that they are satisfied.
Notice what both exceptions have in common: they are unlocked by records. Which leads to the same practical rule under either system, for a different reason under each.
You must be able to evidence everything you claim, and in the United States, everything you deny as well. Deductions without documentation fail regardless of being genuine, and a disputed figure you cannot contradict with records stands.
The authority may reassess only within a defined period, extended in cases of non-declaration or fraud. Records must be retained for at least that period, which is why retention obligations are set by reference to it.
Before an adjustment, the taxpayer must be informed of the proposed change and its basis, and given a genuine opportunity to respond. Assessments issued without this can be annulled on procedure alone.
Tax penalties must be proportionate to the breach, and where they are punitive in substance they attract criminal-type protections whatever they are called: presumption of innocence, rights of defence, and protection against double punishment.
Modern tax administration rests on automatic international exchange between authorities and on reporting obligations imposed on intermediaries such as banks and platforms.
The structural consequence: information asymmetry has largely reversed. The default assumption should be that cross-border holdings held through regulated intermediaries are visible to the residence country's authority.
Categories of tax
| Category | Taxable event | Base |
|---|---|---|
| Income tax | Earning income during a period | Net income after deductible costs |
| Capital gains tax | Disposal of an asset | Proceeds minus basis |
| Wealth tax | Holding assets at a date | Net value of assets held |
| Consumption tax | Supply of goods or services | Value of the supply |
| Transfer tax | Transferring an asset | Value transferred |
| Inheritance or estate tax | Death, or receipt by an heir | Value of the estate or the share received |
| Gift tax | Gratuitous transfer during life | Value given |
| Corporate tax | Profit of an entity | Accounting profit as adjusted |
Two structural observations. Inheritance systems divide into estate taxes, charged on the estate as a whole before distribution, and inheritance taxes, charged on each heir according to what they receive and their relationship to the deceased. The difference changes planning entirely.
And gift tax exists almost everywhere inheritance tax does, for an obvious reason: without it, the inheritance tax would be avoided by giving everything away the week before death.
How systems close that gap divides into two models, and the difference is decisive for anyone planning around it.
The lookback window. Gifts made within a defined period before death are pulled back into the estate; gifts made before that window are outside it. The United Kingdom and France use versions of this, with different periods. Under this model, surviving the window is the objective.
The cumulative lifetime model. Gift and estate tax share a single unified exemption, and every taxable gift you have ever made is tracked and counted against it, with no time boundary at all. The United States works this way: taxable gifts are reported as they are made, each one permanently reduces the exemption available at death, and there is no period after which a gift falls out of the calculation. Treasury regulations have also confirmed that gifts made while the exemption was higher are not clawed back if it later falls.
Giving assets away during life is frequently suggested as a way to keep them out of a probate file or an estate return. Whether that achieves anything depends entirely on which model applies.
Under a lookback window, outliving the period works. Under the cumulative lifetime model there is no period to outlive. A reader planning around "gifts made more than X years before death" would be planning around a window that does not exist for them. Establish which model your jurisdiction uses before building anything on it.
Someone asks how a particular transaction is taxed. Work through the structure rather than searching for the answer.
- Is there a taxable event? Identify precisely which event the statute attaches to. If none of them occurred, there is no charge, and the analysis stops.
- Which category? Income, gain, consumption, transfer. The categorisation determines the rules, and this is where the real argument usually is.
- Who is the taxpayer? The person on whom the charge falls is not always the one who bears the economic cost.
- Which jurisdiction? Apply residence and source. Check whether more than one country has a claim.
- What is the base? Gross or net. What is deductible. Which cost basis convention applies.
- When? Which period, and on what timing rule: accrual, realisation, or receipt.
- Any relief? Exemptions, allowances, rollovers, deferrals, treaty relief.
- What must be declared and retained? The obligation exists independently of whether tax is due.
Notice that the rate appears nowhere in that list. The rate is the last and easiest step, and it is the only part that changes every year.
Keep contemporaneous records of every acquisition and disposal: what, when, how much, at what price, and from whom.
Tax disputes are almost never lost on the law. They are lost because the taxpayer cannot evidence the cost basis, cannot show when an asset was acquired, or cannot document a claimed relief.
The rules will change several times over your holding period. The records you keep today will still be the records you need under whatever rules apply then.
Everything above, reduced to questions you can ask about any legal claim you encounter.
Identifying what you are looking at
- What kind of instrument is it? Constitution, statute, delegated legislation, individual decision, judicial decision, or soft law.
- Who made it, and under what authority? If delegated, does the enabling power actually stretch this far?
- Is it binding, or does it just say should?
- Am I reading the instrument, or a description of it? The source is usually free and usually faster.
Establishing whether it is real yet
- Has it been adopted, or is it a proposal? These are reported identically and are entirely different things.
- Has it been published in an official journal?
- Is it in force? Adoption and application are separate dates, sometimes years apart.
- Does it require implementation to have effect? A framework without implementing measures may bind nobody yet.
- What do the transitional provisions say? Existing situations are often treated differently.
Establishing whether it reaches you
- Which jurisdiction, and at which level? In a federal state, national and sub-national answers can differ.
- Whom does it bind? Firms, individuals, or public authorities. Most financial regulation binds firms.
- On what basis would it reach me? Territory, nationality, effects, or targeting.
- Am I inside the scope? Read the scope article and the definitions before anything else.
- Does an exemption apply? Thresholds and carve-outs are common and rarely reported.
Establishing what follows
- What is the consequence of breach? Criminal, administrative, civil, invalidity, or none.
- Which authority enforces it, and with what powers?
- Can it be challenged, by whom, on what ground, and within what deadline?
- Has it already been challenged? A rule under challenge may not survive.
- Is there anything higher in the hierarchy it might conflict with?
Checking yourself
- Am I reading the current consolidated version?
- Have I followed the cross-references?
- Have I confused what the law is with what I think it should be? Both questions are worth asking, separately.
If you remember nothing else, remember these four:
- Is it in force? Not adopted. In force.
- Does it bind firms or people? Almost always firms.
- Am I inside the scope? Read the definitions.
- What is the actual instrument? Go and read it.
Those four questions, asked before reacting, will spare you the large majority of unnecessary alarm about regulation, and will make you better informed than almost everyone discussing it.
Where to go next
Every technical term used above, and a great many you will meet elsewhere, is defined in the companion legal glossary. It is a separate page, arranged alphabetically, and it is designed to be kept open beside anything you are reading.
Everything here is written to stay true. It explains how the machinery works rather than what today's numbers are, because thresholds, rates and deadlines change every year and a stale legal page is worse than no page at all. Where a current figure matters to your decision, this page tells you how to find it rather than guessing on your behalf.
General information, not legal advice. This site does not provide legal advice, and no professional or advisory relationship is created. This is an introductory account of general concepts in comparative law and legal theory, necessarily simplified. Every legal system has its own structure, vocabulary and exceptions, and the categories described here are analytical tools rather than precise descriptions of any particular country's arrangements. Terminology varies significantly between systems, and a word meaning one thing in one legal order frequently means something else in another. Institutions, procedures and doctrines described in general terms may operate differently, or not exist at all, in any given jurisdiction. Examples are illustrative and simplified, and are not descriptions of any real case. For any actual question, consult the law of the relevant jurisdiction and a qualified lawyer licensed there.