Re Spectrum Plus Ltd  UKHL 41;  4 All ER 209
LORD NICHOLLS OF BIRKENHEAD:
 The starting point is to note some basic, indeed elementary, features of this country's judicial system. The first concerns the essential role of courts of law. In the ordinary course the function of a court is adjudicative. Courts decide the legal consequences of past happenings. Courts make findings on disputed questions of fact, identify and apply the relevant law to the facts agreed by the parties or found by the court, and award appropriate remedies.
 The second feature concerns the wider effect of a court decision on a point of law. To promote a desirable degree of consistency and certainty about the present state of 'the law', courts in this country have long adopted the practice of treating decisions on a point of law as precedents for the future. If the same point of law arises in another case at a later date a court will treat a previous decision as binding or persuasive, depending upon the well-known hierarchical principles of 'stare decisis'.
 The third feature is that from time to time court decisions on points of law represent a change in what until then the law in question was generally thought to be. This happens most obviously when a court departs from, or an appellate court overrules, a previous decision on the same point of law. The point of law may concern the interpretation of a statute or it may relate to a principle of 'judge-made' law, that is, the common law (which for this purpose includes equity). A change of this nature does not always involve departing from or overruling a previous court decision. Sometimes a court may give a statute, until then free from judicial interpretation, a different meaning from that commonly held.
 The fourth feature is a consequence of the second and third features. A court ruling which changes the law from what it was previously thought to be operates retrospectively as well as prospectively. The ruling will have a retrospective effect so far as the parties to the particular dispute are concerned, as occurred with the manufacturer of the ginger beer in Donoghue v Stevenson  AC 562. When Mr Stevenson manufactured and bottled and sold his ginger beer the law on manufacturers' liability as generally understood may have been as stated by the majority of the Second Division of the Court of Session and the minority of their Lordships in that case. But in the claim Ms Donoghue brought against Mr Stevenson his legal obligations fell to be decided in accordance with
Lord Atkin's famous statements. Further, because of the doctrine of precedent the same would be true of everyone else whose case thereafter came before a court. Their rights and obligations would be decided according to the law as enunciated by the majority of the House of Lords in that case even though the relevant events occurred before that decision was given.
 People generally conduct their affairs on the basis of what they understand the law to be. This 'retrospective' effect of a change in the law of this nature can have disruptive and seemingly unfair consequences. 'Prospective overruling', sometimes described as 'non-retroactive overruling', is a judicial tool fashioned to mitigate these adverse consequences. It is a shorthand description for court rulings on points of law which, to greater or lesser extent, are designed not to have the normal retrospective effect of judicial decisions.
 Prospective overruling takes several different forms. In its simplest form prospective overruling involves a court giving a ruling of the character sought by the bank in the present case. Overruling of this simple or 'pure' type has the effect that the court ruling has an exclusively prospective effect. The ruling applies only to transactions or happenings occurring after the date of the court decision. All transactions entered into, or events occurring, before that date continue to be governed by the law as it was conceived to be before the court gave its ruling.
 Other forms of prospective overruling are more limited and 'selective' in their departure from the normal effect of court decisions. The ruling in its operation may be prospective and, additionally, retrospective in its effect as between the parties to the case in which the ruling is given. Or the ruling may be prospective and, additionally, retrospective as between the parties in the case in which the ruling was given and also as between the parties in any other cases already pending before the courts. There are other variations on the same theme.
United Kingdom practice
 Prospective overruling has not yet been adopted as a practice in this country. The traditional approach was stated crisply by Lord Reid in Birmingham City Corp v West Midland Baptist (Trust) Association (Inc)  3 All ER 172 at 180,  AC 874 at 898-899, a case concerning compulsory acquisition;
'We cannot say that the law was one thing yesterday but is to be something different tomorrow. If we decide that [the existing rule] is wrong we must decide that it always has been wrong, and that would mean that in many completed transactions owners have received too little compensation. But that often happens when an existing decision is reversed.'
 In Morgans v Launchbury  2 All ER 606 at 611,  AC 127 at 137, Lord Wilberforce noted: 'We cannot, without yet further innovation, change the law prospectively only.' More recently, in Kleinwort Benson Ltd v Lincoln City Council  4 All ER 513 at 536,  2 AC 349 at 379, Lord Goff of Chieveley said the system of prospective overruling 'has no place in our legal system'.
 The possibility of a change in this practice has been raised from time to time. In Hudson v Secretary of State for Social Services, Jones v Secretary of State for Social Services  1 All ER 145 at 189,  AC 944 at 1015, Lord Diplock said this topic deserved consideration. Lord Simon of Glaisdale said the possibility of prospective overruling should be seriously considered (see  1 All ER 145 at 198,  AC 944 at 1026). He expressed a preference for legislation, saying that 'informed professional opinion' was probably to the effect that the House had no power to overrule decisions with prospective effect only. Lord Simon repeated his plea in Milliangos v George Frank (Textiles) Ltd  3 All ER 801 at 832,  AC 443 at 490. In the Court of Appeal in R v Governor of Brockhill Prison, ex p Evans (No 2)  4 All ER 993 at 1003,  QB 1043 at 1058, Lord Woolf MR expressed the view that prospective overruling has much to commend it. In your Lordships' House this issue was left open (see  4 All ER 15,  2 AC 19). Lord Slynn of Hadley, with his Luxembourg experience in mind, considered there may be situations in which it would be desirable, and in no way unjust, that the effect of judicial rulings should be prospective or limited to certain claimants (see  4 All ER 15 at 19,  2 AC 19 at 26). Lord Hobhouse of Woodborough was hostile to prospective overruling, describing it as a denial of the constitutional role of the courts (see  4 All ER 15 at 39,  2 AC 19 at 48). In the advocates' immunity case of Arthur JS Hall & Co (a firm) v Simons, Barratt v Ansell (t/a Woolf Seddon (a firm)), Harris v Scholfield Roberts & Hill (a firm)  3 All ER 673,  1 AC 615 the House departed from the earlier decision of the House in Rondel v Worsley  3 All ER 993,  1 AC 191. The decision on the immunity point in Arthur JS Hall & Co (a firm) v Simons did not affect the actual outcome in that case. In that context my noble and learned friend Lord Hope of Craighead expressed the view that the change in the law made by Arthur JS Hall & Co (a firm) v Simons should take effect only from the date of the judgment in that case (see  3 All ER 673 at 725,  1 AC 615 at 726). He said ( 3 All ER 673 at 709,  1 AC 615 at 710):
'I consider it to be a legitimate exercise of your Lordships' judicial function to declare prospectively whether or not the immunity-which is a judge-made rule-is to be available in the future and, if so, in what circumstances.'
 Perhaps the nearest the House has come to giving non-retroactive rulings was in two decisions on the law of undue influence. The decisions concerned cases where, typically, a wife claims her consent to a mortgage of her share in a jointly-owned home was procured by her husband exercising undue influence over her (see Barclays Bank plc v O'Brien  4 All ER 417,  1 AC 180 and Royal Bank of Scotland v Etridge (No 2), Barclays Bank plc v Coleman, Bank of Scotland v Bennett, Kenyon-Brown v Desmond Banks & Co (a firm)  UKHL 44,  4 All ER 449,  2 AC 773). In both cases the House said that, in order to avoid being fixed with constructive notice of the wife's rights, a bank could reasonably be expected to bring home to the wife the risks she was running. But in both cases the House sought to give guidance by being more specific on what that test meant in practice. It was in this limited respect that in both cases the House, having regard to realities, drew a distinction between past and future transactions. In Barclays Bank plc v O'Brien, Lord Browne-Wilkinson said ( 4 All ER 417 at 429-430,  1 AC 180 at 196-197) that whether the steps taken by the creditor satisfied the prescribed test would, for past transactions, depend on the facts of each case. As to the future, an appropriately worded warning given at a private meeting between the creditor and the wife would generally suffice to satisfy the test. Despite this admonition the banks did not adopt the course of holding such a meeting. In Royal Bank of Scotland v Etridge (No 2) the House decided that holding a private meeting was not the only way a bank could discharge its obligation to bring home to the wife the risks she was running. I set out other steps ( 4 All ER 447 at -,  2 AC 773) which would generally be regarded as discharging this obligation as to future transactions and, separately, as to past transactions.
 These two decisions illustrate the flexibility inherent in this country's legal system. In passing, another instance of this flexibility can be noted. This illustrates how the House has been prepared to depart from a strict and narrow interpretation of the judiciary's adjudicative role. From time to time situations occur where a point of law of general importance is raised by court proceedings but the outcome will have no practical effect in the particular case. The general principle is that the court will not entertain such proceedings. Nevertheless, when there is good reason for doing so, the House, in the cautious exercise of its discretion, may proceed to decide the point of law. A recent example occurred in the Judicial Committee of the Privy Council in A-G for Jersey v Holley  UKPC 23,  3 All ER 371,  3 WLR 29. There an enlarged Board resolved a conflict between previous decisions of the Board and the House of Lords on an important issue concerning the defence of provocation to a charge of murder. The Board decided this point even though the outcome, either way, would have no effect on the conviction or sentence of the defendant in that case.
 One further matter may be noted regarding the present position on prospective overruling. In the devolution legislation of 1998 Parliament made express provision for courts to have power to limit the temporal effect of a particular class of decisions. The Scotland Act 1998, s 102, provides that where a court decides a provision in an Act of the Scottish Parliament is not within the legislative competence of the Parliament the court may make an order removing or limiting any retrospective effect of the decision or suspending the effect of the decision to enable the defect to be corrected. Comparable provisions appear in the Government of Wales Act 1998, s 110, and the Northern Ireland Act 1998, s 81. These provisions show that Parliament does not perceive non-retroactive rulings by courts as being of their nature inconsistent with the judiciary's proper function.
 In other common law countries prospective overruling has taken root as such only in the United States of America and India. In the United States the fortunes of prospective overruling, sometimes known colloquially as 'Sunbursting', have waxed and waned. Prospective overruling, although without that label, occurred as long ago as the mid-19th century in the Ohio case of Bingham v Miller (1848) 17 Ohio 445. In 1932 the Supreme Court, in a famous judgment of Cardozo J, held that the Constitution neither prohibits nor requires prospective overruling. The Federal Court, he said, 'has no voice upon the subject' (see Great Northern Railway Co v Sunburst Oil & Refining Co (1932) 287 US 358). Prospective overruling by the Supreme Court itself reached its apogee in the 1960s and 1970s when the court decided that in both criminal and civil cases 'the accepted rule today is that in appropriate cases the Court may in the interest of justice make the rule prospective' (see Linkletter v Walker (1965) 381 US 618 at 628). In 1971 in the leading case of Chevron Oil Co v Huson (1971) 404 US 97 at 106-107, the court summarised three factors taken into account when considering if a decision should be applied non-retroactively: whether the decision established a new principle of law, whether retrospective operation would advance or retard the operation of the new rule, and whether the decision could produce substantial inequitable results if applied retrospectively.
 Since then the Supreme Court has retreated. In Griffith v Kentucky (1987) 479 US 314 the court abandoned prospective overruling when directly reviewing criminal cases. Selective overruling has been abandoned in civil cases (see James B Beam Distilling Co v Georgia (1991) 501 US 529 and Harper v Virginia Dept of Taxation (1992) 509 US 86). Whether the court has abandoned 'pure' prospective overruling in civil cases remains to be resolved (see Glazner v Glazner (2003) 347 F 3d 1212, a decision of the Court of Appeals of the Eleventh Circuit).
 Taking its lead from the United States jurisprudence the Supreme Court of India has made prospective overrulings but only in constitutional cases. The first case was IC Golak Nath v State of Punjab (1967) 2 SCJ 762. In that case the court reversed two earlier decisions of its own in circumstances where meanwhile constitutional amendments had been made, and state laws enacted, on the basis of the court's earlier two decisions. The jurisdiction is not confined to cases where an earlier decision is overruled. Non-retroactive effect may be given to a ruling which decides an issue for the first time (see India Cement Ltd v State of Tamil Nadu (1990) 1 SCC 12). The Supreme Court founds its jurisdiction to make rulings of this character on art 142 of the Indian Constitution. This article empowers the Supreme Court to 'make such order as is necessary for doing complete justice in any cause or matter pending before it'. In exercise of this power it is a 'well settled proposition that it is open to the court to grant, mould or restrict relief in a manner most appropriate to the situation before it, in such a way as to advance the interests of justice' (see Orissa Cement Ltd v State of Orissa 1991 Supp (1) SCC 430).
 In Ireland in Murphy v A-G  IR 241 the Supreme Court held that certain taxation provisions were unconstitutional and void. The court rejected an argument that it was for the courts to say whether these statutory provisions should be held to be invalid prospectively or with only limited retrospective effect. The provisions were invalid from the date on which they were enacted. However, the court also held that the plaintiffs' restitutionary right to recover amounts paid by way of taxes unconstitutionally imposed began with the first year in which they raised their objections. Further, unless other taxpayers had already made tax recovery claims, only the plaintiffs could maintain a claim pursuant to the court's decision.
 In Canada prospective overruling has not found favour. In Re Edward and Edward (1987) 39 DLR (4th) 654 the Saskatchewan Court of Appeal said prospective overruling would be a 'dramatic deviation from the norm in both Canada and England'. Bayda CJS said (at 664) 'the most cogent reason for rejecting this technique is the necessity for our courts to maintain their independent, neutral and non-legislative role'. He approved comments that prospective overruling 'would distort our expectations of the judicial role' and that 'confidence may recede at the point where the courts are not seen as adjudicative agencies but as legislators' (see Lord Lloyd of Hampstead, Introduction to Jurisprudence (4th edn, 1979) pp 858-859). But in the extreme circumstances of a Reference re Manitoba Language Rights under the Manitoba Act 1870 (1985) 19 DLR (4th) 1 the Supreme Court of Canada declined to give retroactive effect to its decision on the constitutional invalidity of all statutes and regulations of the Province of Manitoba not printed and published in both English and French. A declaration that the unilingual laws of Manitoba were of no effect would have created a legal vacuum with consequent legal chaos. Refusing to take a narrow and literal approach to constitutional interpretation, the court held it could have regard to unwritten postulates such as the principle of the rule of law. Faced with the task of recognising the unconstitutionality of Manitoba's unilingual laws while avoiding a legal vacuum and ensuring the continuity of the rule of law, the court made a ruling which gave deemed temporary validity to all laws rendered invalid by reason of their unilingual defect.
 As with all controversial subjects prospective overruling attracts arguments both ways. The arguments against prospective overruling are both principled and practical. It will be convenient to note first the major practical difficulties attendant upon prospective overruling. The retrospective nature of a court ruling on a point of law means that the ruling applies in all cases, past as well as future. This is subject only to defences of general application, such as limitation, laches, and res judicata. Whatever its faults the retrospective application of court rulings is straightforward. Prospective overruling creates problems of discrimination. Born out of a laudable wish to mitigate the seeming unfairness of a retrospective change in the law, prospective overruling can beget unfairness of its own.
 This is most marked in criminal cases, where 'pure' prospective overruling would leave a successful defendant languishing in prison. 'Selective' prospective overruling avoids this consequence but it could see a successful defendant freed while others in like case stayed in prison. In civil cases 'pure' prospective overruling would hinder the development of the law by discouraging claimants from challenging a prevailing view of the law. 'Selective' overruling, if only the successful claimant benefits from the change, is likely to mean that persons in like case are treated differently. Further, it would introduce an arbitrary element into the law. The ability to obtain an effective remedy could depend upon which of several challenges reaches the House of Lords first. Even if everyone who had already commenced proceedings was given the benefit of the court ruling there would still be scope for discrimination: there would be discrimination between those who knew they might have a claim and started proceeding post-haste and those, lacking proper advice, who were unaware they might have a claim.
Objections in principle
 The essence of the principled argument against prospective overruling is that in this country prospective overruling is outside the constitutional limits of the judicial function. It would amount to the judicial usurpation of the legislative function. Power to make rulings having only prospective effect, it is said, is not inherent in the judicial role. A ruling having only prospective effect cannot be characterised as merely a less extensive form of overruling than overruling with both retrospective and prospective effect. Prospective overruling robs a ruling of its essential authenticity as a judicial act. Courts exist to decide the legal consequences of past events. A court decision which takes the form of a 'pure' prospective overruling does not decide the dispute between the parties according to what the court declares is the present state of the law. With a ruling of this character the court gives a binding ruling on a point of law but then does not apply the law as thus declared to the parties to the dispute before the court. The effect of a prospective overruling of this character is that, on the disputed point of law, the court determines the rights and wrongs of the parties in accordance with an answer which it declares is no longer a correct statement of the law. Making such a ruling would not be a proper exercise of judicial power in this country. Making new law in this fashion gives a judge too much the appearance of a legislator. Legislation is a matter for Parliament, not judges.
 In short, this argument raises this issue: would a decision by this House on a point of law having only prospective effect be so substantial a departure from established judicial procedure that it should be regarded as outside the function discharged by the judiciary under this country's constitution?
 In answering this question the Appellate Jurisdiction Act 1876 provides no assistance. The appellate jurisdiction of the House is formally regulated by this Act. Section 4 provides that appeals shall be brought by petition to the House praying that the order appealed may be reviewed 'before Her Majesty the Queen in her Court of Parliament' in order that this court 'may determine what of right, and according to the law and custom of this realm, ought to be done in the subject-matter of such appeal'. When Pt 3 of the Constitutional Reform Act 2005 comes into force the new Supreme Court will have power 'to determine any question necessary to be determined for the purposes of doing justice in an appeal to it' (s 40(5)). These general statements do not point either way on the issue now under consideration.
 The next point to note is that, broadly stated, the constitutional separation of power between the legislature and the judiciary in this country is that the legislature makes the law, the courts administer the law. Parliament makes new law, by enacting statutes having prospective and varying degrees of retrospective effect (see Wilson v First County Trust Ltd  UKHL 40 at ,  4 All ER 97 at ,  1 AC 816). When disputes arise, whether between citizens or between a citizen and the government, they are to be resolved in accordance with the law, and that is a matter for the judicial arm of the state. In this regard it is for the judiciary to decide what is the law, not the legislature or the executive.
 This broad generalisation is a good starting point but, like most constitutional generalisations in this country, it calls for qualification. The boundary between making and administering the law is not in all respects quite so clear-cut as this general statement suggests. Contrary to the broad generalisation and within strict bounds, judges themselves have a legitimate law-making function. It is a function they have long exercised. In common law countries much of the basic law is still the common law. The common law is judge-made law. For centuries judges have been charged with the responsibility of keeping this law abreast of current social conditions and expectations. That is still the position. Continuing but limited development of the common law in this fashion is an integral part of the constitutional function of the judiciary. Had the judges not discharged this responsibility the common law would be the same now as it was in the reign of King Henry II. It is because of this that-
'the common law is a living system of law, reacting to new events and new ideas, and so capable of providing the citizens of this country with a system of practical justice relevant to the times in which they live.' (See Lord Goff in Kleinwort Benson Ltd v Lincoln City Council  4 All ER 513 at 534,  2 AC 349 at 377.)
 Changes in the common law made by judges are usually described as 'development' of the common law. This is a helpful description, not a misleading euphemism. Judges do not have a free hand to change the common law. Judicial development of the common law comprises the reasoned application of established common law principles, of greater or less generality, in current social conditions. Development of the common law by the judges in any one case is usually marginal. Occasionally it is more far-reaching, as in Donoghue v Stevenson. In all cases development of the common law, as a response to changed conditions, does not come like a bolt out of a clear sky. Invariably the clouds gather first, often from different quarters, indicating with increasing obviousness what is coming. Cardozo J's colourful summary, in his Nature of the Judicial Process (1921), p 141, merits repetition:
'The judge, even when he is free, is still not wholly free. He is not to innovate at pleasure. He is not a knight-errant, roaming at will in pursuit of his own ideal of beauty or of goodness. He is to draw his inspiration from consecrated principles. He is not to yield to spasmodic sentiment, to vague and unregulated benevolence. He is to exercise a discretion informed by tradition, methodized by analogy, disciplined by system, and subordinated to "the primordial necessity of order in the social life"'.
 At one time the judicial function of overruling previous common law decisions was sought to be rationalised by the 'declaratory' theory. Sir William Blackstone said that 'if it be found that the former decision is manifestly absurd or unjust, it is declared, not that such a sentence was bad law, but that it was not law' (see Commentaries on the Laws of England (1st edn, 1765) vol 1, p 70). If 'law' is given one of its several possible meanings, this theory is still valid when applied to cases where a previous decision is overruled as wrong when given. Most overruling occurs on this basis. These cases are to be contrasted with those where the later decision represents a response to changes in social conditions and expectations. Then, on any view, the declaratory approach is inapt. In this context this approach has long been discarded. It is at odds with reality.
 For present purposes the distinction between these two types of cases is not important. Nor is the declaratory theory. For present purposes what matters is the practical impact overruling may have in both types of cases. The question now under consideration is whether, having regard to this practical impact, it is necessarily and always beyond the competence of the House as the supreme court in this country ever to limit the temporal effect of its ruling....
 [T]he interpretation the court gives an Act of Parliament is the meaning which, in legal concept, the statute has borne from the very day it went onto the statute book. So, it is said, when your Lordships' House rules that a previous decision on the interpretation of a statutory provision was wrong, there is no question of the House changing the law. The House is doing no more than correct an error of interpretation. Thus, there should be no question of the House overruling the previous decision with prospective effect only. If the House were to take that course it would be sanctioning the continuing misapplication of the statute so far as existing transactions or past events are concerned. The House, it is said, has no power to do this. Statutes express the intention of Parliament. The courts must give effect to that intention from the date the legislation came into force. The House, acting in its judicial capacity, must give effect to the statute and it must do so in accordance with what it considers is the proper interpretation of the statute. The House has no suspensive power in this regard. In Ha v State of New South Wales, Walter Hammond & Associates v State of New South Wales (1997) 146 ALR 355 at 372, (1997) 189 CLR 465 at 503-504, the High Court of Australia unanimously considered 'it would be a perversion of judicial power to maintain in force that which is acknowledged not to be the law'. This would especially be so where 'non-compliance with a properly impugned statute exposes a person to criminal prosecution'.
 The objections in principle and difficulties in practice mentioned above have substance, particularly in respect of the traditional interpretation of statutes. These objections are compelling pointers to what should be the normal reach of the judicial process. But, even in respect of statute law, they do not lead to the conclusion that prospective overruling can never be justified as a proper exercise of judicial power. In this country the established practice of judicial precedent derives from the common law. Constitutionally the judges have power to modify this practice.
 Instances where this power has been used in courts elsewhere suggest there could be circumstances in this country where prospective overruling would be necessary to serve the underlying objective of the courts of this country: to administer justice fairly and in accordance with the law. There could be cases where a decision on an issue of law, whether common law or statute law, was unavoidable but the decision would have such gravely unfair and disruptive consequences for past transactions or happenings that this House would be compelled to depart from the normal principles relating to the retrospective and prospective effect of court decisions.
 If, altogether exceptionally, the House as the country's supreme court were to follow this course I would not regard it as trespassing outside the functions properly to be discharged by the judiciary under this country's constitution. Rigidity in the operation of a legal system is a sign of weakness, not strength. It deprives a legal system of necessary elasticity. Far from achieving a constitutionally exemplary result, it can produce a legal system unable to function effectively in changing times. 'Never say never' is a wise judicial precept, in the interest of all citizens of the country.