Constitutional conventions

Constitutional conventions are non-legal rules of constitutional behaviour. They are not merely habits, courtesies, or political slogans. A convention is a rule accepted by constitutional actors as binding in constitutional practice, although not enforceable as law by the courts. The orthodox distinction is therefore not between important and unimportant rules, but between legal and non-legal modes of obligation. A convention may be more important to the working constitution than many statutory provisions; yet its breach will normally produce political, not legal, consequences.

The topic is central to Part IA Constitutional Law because the United Kingdom constitution cannot be understood by reading statutes and cases alone. The legal power of the Crown to appoint a Prime Minister, dissolve or prorogue Parliament, assent to Bills, command the armed forces, and conduct foreign affairs is intelligible only against a dense background of conventions. So too are collective Cabinet responsibility, individual ministerial responsibility, the expectation that ministers are accountable to Parliament, the confidence principle, the Salisbury-Addison convention, the Ponsonby convention before its statutory partial replacement, and the Sewel convention in relation to devolution.

The principal analytical tasks are threefold. First, one must identify whether a practice has become a convention. The classic Jennings test asks: what are the precedents; did the actors believe themselves bound; and is there a reason for the rule? Secondly, one must distinguish recognition from enforcement. Courts may take notice of conventions in order to understand the constitutional setting, but they do not enforce them as legal duties. That proposition is most clearly expressed in Miller (No 1), where the Supreme Court held that the Sewel convention, even though acknowledged in statute, remained outside judicial enforcement. Thirdly, one must evaluate whether this distinction is satisfactory. Conventions may secure democratic accountability, but their non-legal character means that compliance depends on political costs, institutional self-restraint, and constitutional culture.

The Cambridge examiner will expect more than a list of examples. The best answers treat conventions as a bridge between legal form and political reality. Parliamentary sovereignty, the prerogative, responsible government, and devolution all depend upon conventions. Yet conventions cannot be allowed to dissolve the distinction between law and politics. The question is not whether conventions matter; plainly they do. The harder question is what kind of constitutional obligation they create, and whether a constitution committed to legality can safely rely upon rules which courts will not enforce.

The modern doctrine of constitutional conventions is a product of the nineteenth-century attempt to describe the United Kingdom constitution as both legally monarchical and politically democratic. Dicey’s famous contrast between the law of the constitution and the conventions of the constitution remains the starting point. On the legal plane, the Crown possessed extensive prerogative powers: ministers were appointed by the monarch; Parliament was summoned and dissolved in the monarch’s name; royal assent was necessary for legislation; treaties were concluded by the Crown. On the political plane, those powers were exercised by ministers who depended on the confidence of the elected House of Commons and were answerable to Parliament. Conventions explained how an apparently monarchical constitution operated as a system of responsible parliamentary government.

This historical role matters. Conventions were not ornamental additions to the legal constitution. They were the devices by which legal powers were made compatible with democratic principle. The convention that the monarch appoints as Prime Minister the person best placed to command the confidence of the House of Commons prevents the legal power of appointment from becoming a personal monarchical discretion. The convention that royal assent is not refused prevents the Crown from blocking legislation enacted by the two Houses. The convention that ministers resign after losing confidence in the Commons, or seek an election where appropriate, connects executive authority to representative legitimacy.

The twentieth century added further layers. Party government and Cabinet discipline strengthened collective ministerial responsibility, though they also changed its substance. The growth of the administrative state made individual ministerial responsibility more complex: ministers remained answerable to Parliament, but the practical possibility that a minister would resign for every departmental error diminished. The House of Lords’ resistance to the Liberal Government’s People’s Budget and the subsequent Parliament Acts gave rise to understandings about the Lords’ restraint in relation to government legislation, including what later became known as the Salisbury-Addison convention: the Lords should not reject at second or third reading government Bills giving effect to manifesto commitments. That convention is politically significant but its content, application to coalition or minority governments, and continuing force remain contested.

Devolution gave conventions a new salience. The Sewel convention, announced during the passage of the Scotland Bill, stated that Westminster would not normally legislate with regard to devolved matters without the consent of the devolved legislature. This is a convention because Westminster legally retains power to legislate for Scotland, Wales and Northern Ireland, but politically accepts a restraint in ordinary circumstances. Its partial statutory recognition in the Scotland Act 1998 and the Wales legislation did not convert it into a legally enforceable rule. Miller (No 1) confirms that point.

Recent constitutional events have tested the resilience of conventions. The 2019 prorogation dispute, Brexit-related use of parliamentary procedure, disputes over ministerial standards, and tensions in intergovernmental relations all show that conventions presuppose actors willing to treat political morality as binding. A convention may survive occasional breach if the breach is acknowledged as wrongful. It may decay if actors deny that the rule binds them, or if political sanctions cease to operate. That historical lesson is crucial: conventions depend on constitutional culture, not judicial coercion.

1. Definition. A constitutional convention is a non-legal rule regulating the conduct of constitutional actors. The actors may include the monarch, ministers, Parliament, the House of Lords, devolved institutions, civil servants, and occasionally opposition parties. The rule is constitutional because it concerns the location, exercise, or accountability of public power. It is conventional because its normative force arises from acceptance within political practice rather than from statute, common law, or prerogative law.

1. The Jennings test. Sir Ivor Jennings’ formulation remains the standard analytical tool: first, what are the precedents; secondly, did the actors in those precedents believe that they were bound by a rule; thirdly, is there a reason for the rule? The first limb prevents a convention from being asserted on the basis of an isolated preference. The second distinguishes a binding convention from a repeated convenience. The third introduces constitutional purpose: a convention must be intelligible by reference to a principle such as democratic accountability, responsible government, parliamentary scrutiny, federal comity, or ministerial responsibility. In an essay, apply all three limbs. Do not merely say that a practice has often occurred.

1. Conventions are not enforceable law. This is the orthodox proposition. A court may recognise a convention as part of the constitutional background, but it will not grant a remedy simply because a convention has been breached. The distinction was classically expressed in Madzimbamuto and reaffirmed in Miller (No 1). The court may ask whether a legal rule exists, whether a legal power has been exceeded, or whether a statutory requirement has been met. It will not police political understandings as such. A convention may be obligatory in constitutional morality and yet legally unenforceable.

The case law on constitutional conventions is less extensive than students often suppose, because the central proposition is negative: courts do not enforce conventions. Yet the leading cases are important because they define the border between law and constitutional practice.

Madzimbamuto v Lardner-Burke remains the orthodox starting point. The dispute arose from Rhodesia’s unilateral declaration of independence. The Privy Council accepted that there were powerful constitutional understandings about the United Kingdom Parliament’s conduct towards Rhodesia, but insisted that such understandings could not limit Parliament’s legal power. Lord Reid’s statement is frequently cited because it captures the Diceyan distinction between political impropriety and legal invalidity. The case is useful whenever an essay asks whether conventions can limit parliamentary sovereignty.

Attorney General v Jonathan Cape is different. The court did not enforce a convention as such, but recognised the constitutional importance of collective Cabinet responsibility when considering whether Cabinet discussions were protected by confidentiality. The case demonstrates that conventions may have legal relevance indirectly. The court accepted that confidential Cabinet discussion serves the convention of collective responsibility, though on the facts publication of old ministerial diaries was not restrained. The lesson is subtle: conventions may inform the public interest analysis without themselves becoming directly enforceable legal rules.

Manuel v Attorney General concerned the Canada Act 1982 and objections that Westminster had legislated without sufficient provincial consent. The English court refused to invalidate the Act on the basis of alleged breach of Canadian constitutional conventions. The decision reinforces both the enrolled Act principle and the non-legal status of conventions. It also shows that courts will not use conventions to question the validity of primary legislation.

Miller (No 1) is the modern leading authority. The central issue concerned Article 50 and whether ministers could use prerogative power to initiate withdrawal from the European Union without statutory authority. Alongside that legal issue, the devolved claimants argued that the Sewel convention required devolved consent. The Supreme Court recognised the convention and its statutory acknowledgement but held that courts could not police its scope or operation. This is the clearest authority for the proposition that statutory recognition of a convention does not necessarily make it enforceable.

Miller (No 2) is not a case about enforcing conventions, but it is essential for avoiding a false inference. The Court held that the prorogation power was subject to legal limits because an unlawful prorogation could frustrate parliamentary sovereignty and accountability. That was a legal principle governing prerogative power, not judicial enforcement of a convention about responsible government. The case shows that courts may protect constitutional principles through law while leaving conventions to political enforcement.

Wheeler illustrates the non-justiciability of political commitments. A manifesto promise to hold a referendum did not generate an enforceable legal expectation. The case assists in distinguishing conventions and political undertakings from legitimate expectations grounded in clear legal representations within an administrative law framework.

FDA v Prime Minister is a useful recent caution. The Ministerial Code is generally a political instrument, but where a public authority adopts a legally relevant interpretation of its terms, the court may be able to review the legality of that interpretation. The case does not make ministerial responsibility justiciable. It shows that the legal frame chosen by government may matter.

See practice questions section below.

See further reading section below.