R (on the application of Majera (formerly SM (Rwanda)) v Secretary of State for the Home Department

This appeal raises a question of constitutional importance: whether the Government (or, indeed, anyone else) can lawfully act in a manner which is inconsistent with an order of a judge which is defective, without first applying for, and obtaining, the variation or setting aside of the order.

The appellant is a national of Rwanda who came to the United Kingdom as a child and was granted indefinite leave to remain. In 2006 he was convicted of serious offences, and an indeterminate sentence for public protection was imposed, with a minimum of seven years to be served. It was recommended that he should be deported. In November 2012 the Secretary of State made a deportation order. That order has never been implemented. An application to the Secretary of State for the revocation of the order, made in March 2015, remains outstanding.

During 2013 the appellant was transferred to open prison conditions, and undertook voluntary work in the community. During 2014 and 2015 he worked as a volunteer at a charity shop. On 30 March 2015 the Parole Board directed his release on licence. He was however immediately detained by the Secretary of State under paragraph 2 of Schedule 3 to the Immigration Act 1971 (“the 1971 Act”).

In July 2015 the appellant applied to the First-tier Tribunal for bail under paragraph 22 of Schedule 2 to the 1971 Act. The application was heard by First-tier Tribunal Judge Narayan on 30 July 2015. Having heard submissions on behalf of the appellant and the Secretary of State, the judge decided that bail should be granted, with the appellant’s mother acting as surety in respect of a specified sum of money. In relation to the conditions to be attached to bail, the judge decided that the appellant should report to his offender manager. Another issue concerned the appellant’s voluntary work. He wished to continue working as a volunteer at the charity shop. The Secretary of State, in her written submissions, sought a condition that the appellant must not enter employment, paid or unpaid. The judge decided not to impose a prohibition on unpaid work.

The judge’s decision was recorded in a notice of decision (“the bail order”) of the same date. Amongst other matters, it recorded the “primary condition of bail” as follows:

As I have explained, the bail order incorporated the terms of the appellant’s licence under the Criminal Justice Act 2003. The licence required him to report to his supervising officer (the offender manager to whom the bail conditions referred) at the offices of the National Probation Service in London before 3pm on 31 July 2015. It also required him to undertake only such work (including voluntary work) as was approved by his supervising officer. It also imposed a curfew between 8pm and 7am.

Later on 30 July 2015, shortly after the hearing before the judge, an immigration officer gave the appellant a notice notifying him that “the Secretary of State has decided that you should not continue to be detained at this time but, under paragraph 2(5) of Schedule 3 to the Immigration Act 1971, she now imposes the following restrictions on you”. The restrictions included, first, that “[y]ou may not enter employment, paid or unpaid”, and secondly, the imposition of a curfew between 8pm and 7am. The appellant was released.

The appellant’s solicitors subsequently wrote to the Secretary of State, maintaining that she had no power to impose the restrictions and asking that they be withdrawn. They contended that the appropriate course, if the Secretary of State was dissatisfied with the conditions imposed by the judge, was either to seek a variation of the conditions or to challenge the judge’s decision in proceedings for judicial review, there being no right of appeal. In response, the Secretary of State wrote on 19 October 2015 that “your client is no longer on immigration judge bail and is now on restrictions imposed by the Home Office”. In subsequent correspondence, the Secretary of State adopted the position that the bail order had ceased to have effect when the appellant reported to his supervising officer. Subsequently, it was argued on behalf of the Secretary of State that although the bail order remained in force, it was not inconsistent with the notice of restrictions. Neither of those positions is now maintained on behalf of the Secretary of State.

A further request for the prohibition of voluntary work to be withdrawn, supported by the appellant’s supervising officer, was refused by the Secretary of State on 3 December 2015. A request by the appellant’s supervising officer for the curfew restriction to be altered to 11pm to 6am, so as to correspond to the hours which were then considered appropriate for the appellant’s licence, was also refused on 4 January 2016. The appellant then applied for judicial review of those two decisions.

The powers of immigration detention relevant to these proceedings are contained in paragraph 2 of Schedule 3 to the 1971 Act. The provision has been subject to frequent amendment. As it stood at the relevant time, it provided, so far as material:

The appellant was a person to whom sub-paragraph (3) applied, by virtue of the deportation order made in November 2012. He was detained under that provision between 30 March and 30 July 2015.

The bail provisions relevant to these proceedings are contained in paragraph 22 of Schedule 2 to the 1971 Act. They applied to the appellant by virtue of paragraph 2(4A) of Schedule 3 (para 10 above). As has been explained, the appellant was, following his detention by the Secretary of State on 30 March 2015, a person who was detained under paragraph 2(3) of Schedule 3, and was therefore, by virtue of sub-paragraph (4A), a person to whom paragraph 22 of Schedule 2 of the 1971 Act applied.

Paragraph 22 of Schedule 2, as it stood at the relevant time, provided, so far as material:

Rule 43 of the Tribunal Procedure (First-tier Tribunal) (Immigration and Asylum Chamber) Rules 2014 (SI 2014/2604), provides:

The appellant sought judicial review of the Secretary of State’s decisions dated 3 December 2015 and 4 January 2016. When granting permission, the Upper Tribunal stayed the proceedings in order to permit the Secretary of State to consider whether to make an application to the First-tier Tribunal for variation of the bail order, but the Secretary of State declined to do so.

In the statement of the grounds for judicial review, it was argued on behalf of the appellant that the decisions in issue were ultra vires, since the Secretary of State had no power to attach additional conditions to bail which a bail judge had declined to order, and were in any event an irrational and disproportionate restriction of the appellant’s liberty. In response, the grounds of defence filed on behalf of the Secretary of State contended (in contrast to the position previously adopted, noted at para 8 above) that the bail order did not comply with paragraph 22(1A) of Schedule 2 to the 1971 Act (para 13 above), as it failed to require the appellant to appear before an immigration officer at a specified time and place. As has been explained, it required him instead to appear before his offender manager by a specified time and at a specified place. In those circumstances, it was argued, the purported grant of bail was void. On that basis, it was argued that it had been open to the Secretary of State to impose restrictions on the appellant under paragraph 2(5) of Schedule 3 (para 10 above). It was, however, conceded that the Secretary of State had no power to impose a curfew, having regard to the decision of the Court of Appeal in R (Gedi) v Secretary of State for the Home Department [2016] EWCA Civ 409; [2016] 4 WLR 93.

The Upper Tribunal (Upper Tribunal Judge Peter Lane) quashed the decisions of 3 December 2015 and 4 January 2016, and granted a declaration that the appellant remained on bail granted by the First-tier Tribunal on 30 July 2015: R (Majera) v Secretary of State for the Home Department (bail conditions: law and practice) [2017] UKUT 00163 (IAC). The tribunal observed at para 46 of the judgment:

The tribunal went on to note that a finding that a grant of bail was of no legal effect was liable to have serious consequences, and inferred that it could not have been Parliament’s intention that defects in the grant of bail should render it void. Rather, there was a valid but defective grant of bail. The defect could be corrected by the judge on its being drawn to his attention, as the Secretary of State should immediately have done. Since the grant of bail on 30 July 2015 was valid, albeit defective, it remained in force. Its defective nature could be remedied by the First-tier Tribunal. It followed that the restrictions purportedly imposed by the Secretary of State were of no effect.

The tribunal also observed, expressly obiter (para 64), that the judge could not properly impose a bail condition relating to employment, as it fell outside the purpose of paragraph 22(2) of Schedule 2 to the 1971 Act (para 13 above); that paragraph 2(5) of Schedule 3 (para 10 above) applied to the appellant following his release on bail; and that the Secretary of State had the power under that provision to restrict the appellant’s employment or occupation while on bail.

The Secretary of State appealed against that decision so far as it related to the decision dated 3 December 2015. The Court of Appeal (Underhill, Asplin and Haddon-Cave LJJ) allowed the Secretary of State’s appeal, quashed the declaration granted by the Upper Tribunal, and made a declaration that the purported grant of bail by the First-tier Tribunal on 30 July 2015 was void and a nullity: Secretary of State for the Home Department v SM (Rwanda) [2018] EWCA Civ 2770; [2019] Imm AR 714.

In the view of the Court of Appeal, the central issue in the proceedings was not the validity of the Secretary of State’s decision, which was the subject of the application for judicial review, but the validity of the bail order, as Haddon-Cave LJ made clear in the opening sentence of his judgment:

Following that approach, the bail order was considered to be a nullity, on the basis that (1) non-compliance with the provisions of paragraph 22(1A) of Schedule 2 to the 1971 Act rendered the grant of bail unlawful, and (2) the speeches of Lord Diplock in F Hoffmann-La Roche & Co AG v Secretary of State for Trade and Industry [1975] AC 295, 365 (“ Hoffmann-La Roche ”) and Lord Irvine of Lairg LC in Boddington v British Transport Police [1999] 2 AC 143, 158 (“ Boddington ”) established that, as Haddon-Cave LJ stated at para 47, “when an act or regulation has been pronounced by the court to be unlawful, it is then recognised as having had no legal effect at all”.

Proceeding on that basis, the appellant’s argument that orders of the court are binding until set aside or varied was regarded as “academic” (para 74):

Haddon-Cave LJ added that, when it became apparent that there was an issue as to the validity of the bail order, it would have been “good practice” for the Secretary of State to have brought the matter back before the First-tier Tribunal for appropriate resolution, rather than simply asserting her right to impose restrictions on the appellant in the face of the extant order (para 76).

The other members of the court agreed. Underhill LJ observed at para 79 that the implication of the court’s decision (issued on 11 December 2018) was that the appellant had been unlawfully at large since his release on 30 July 2015. He and Asplin LJ also agreed that the matter had not been “handled well” by the Secretary of State (paras 80 and 82).

The appellant now appeals against the decision of the Court of Appeal. Put shortly, counsel for the appellant argues that the First-tier Tribunal’s order had to be obeyed until it was varied or set aside, regardless of whether it was defective. A decision by the Secretary of State which was inconsistent with it was necessarily unlawful. The appellant’s arguments are supported by submissions on behalf of the intervener, Bail for Immigration Detainees, which the court has found of assistance. The Secretary of State supports the reasoning of the Court of Appeal (preferring that of Underhill LJ, in so far as it differed from that of Haddon-Cave LJ in relation to the correct approach to the construction of paragraph 22 of Schedule 2 to the 1971 Act; Asplin LJ expressed no view on that matter).

The Court of Appeal’s approach to the present case, based on the characterisation of invalid administrative acts and decisions as null and void, was as I shall explain inapposite to the order of a court or tribunal such as the First-tier Tribunal. But it is also worth explaining why, even in relation to administrative acts and subordinate legislation, Haddon-Cave LJ’s statement that “when an act or regulation has been pronounced by the court to be unlawful, it is then recognised as having had no legal effect at all” is, with great respect, an over-simplification of the position. Although judges have commonly used expressions such as “null” and “void” to describe unlawful administrative acts and decisions, it has nevertheless been recognised that the notion that such acts and decisions are utterly destitute of legal effect, as if they had never existed at all, is subject to important qualifications.

Although Haddon-Cave LJ’s dictum was confined to the situation where there has been a judicial pronouncement - which I take to mean an order, since it is orders, not the reasons given for them in judgments, which have legal effects - determining that an act or regulation is unlawful, it is illuminating to consider first the position before such a pronouncement is made. A significant point was made by Lord Radcliffe in Smith v East Elloe Rural District Council [1956] AC 736, 769-770, where he considered an argument that an ouster clause preventing a compulsory purchase order from being challenged after the expiry of a time limit must be construed as applying only to orders made in good faith, since an order made in bad faith was a nullity and therefore had no legal existence. Describing the argument as “in reality a play on the meaning of the word nullity”, Lord Radcliffe observed:

Accordingly, if an unlawful administrative act or decision is not challenged before a court of competent jurisdiction, or if permission to bring an application for judicial review is refused, the act or decision will remain in effect. Equally, even if an unlawful act or decision is challenged before a court of competent jurisdiction, the court may decline to grant relief in the exercise of its discretion, or for a reason unrelated to the validity of the act or decision, such as a lack of standing (as in Durayappah v Fernando [1967] 2 AC 337) or an ouster clause (as in Smith v East Elloe ). In that event, the act or decision will again remain in effect. An unlawful act or decision cannot therefore be described as void independently of, or prior to, the court’s intervention.

The courts have long been aware of this point. In Calvin v Carr [1980] AC 574, for example, Lord Wilberforce observed at pp 589-590, in relation to a contention that an appeal could not lie against a decision which was void, that until a decision was declared to be void by a competent body or court, “it may have some effect, or existence, in law”. He added that “[t]his condition might be better expressed by saying that the decision is invalid or vitiated”. In the context of a question as to whether an appeal lay, “the impugned decision cannot be considered as totally void, in the sense of being legally non-existent”. So to hold, he said, “would be wholly unreal”. The decision in question, to disqualify a racehorse owner, was a fact, and its existence had serious consequences for the plaintiff, because the Jockey Club complied with it by preventing him from entering his horses for races. As that example illustrates, the law has to address the fact that unlawful acts can have legal consequences.

Even where a court has decided that an act or decision was legally defective, that does not necessarily imply that it must be held to have had no legal effect. As the Court of Appeal correctly noted in the present case, it may be, in the first place, that to treat the decision as a nullity would be inconsistent with the legislation under which it was made (see, for example, R v Soneji [2005] UKHL 49; [2006] 1 AC 340). Or the result of treating the decision as legally non-existent may be inconsistent with legal certainty or with the public interest in orderly administration: it may, indeed, result in administrative chaos, or expose innocent third parties to legal liabilities (as where they have acted in reliance on the apparent validity of the unlawful decision). In some such circumstances, the act or decision may be held to have had some legal effects in accordance with principles of the common law (as, for example, where police officers are held to have acted lawfully in arresting and detaining individuals in pursuance of byelaws which are subsequently held to be invalid: see Percy v Hall [1997] QB 924). In other circumstances, the court may be able to secure an appropriate outcome through the exercise of statutory powers (as, for example, in Salvesen v Riddell [2013] UKSC 22; [2013] HRLR 23), or through the exercise of its discretion in granting relief (as, for example, in Fishermen and Friends of the Sea v Minister of Planning, Housing and the Environment [2017] UKPC 37). Lord Phillips, giving the judgment of the Privy Council, observed in Mossell (Jamaica) Ltd (t/a Digicel) v Office of Utilities Regulations [2010] UKPC 1, para 44:

These considerations have led judges to be critical of the description of unlawful administrative acts or decisions as “null” or “void”, and have sometimes led them to speak of voidness as a “relative” concept (see, for example, R (New London College Ltd) v Secretary of State for the Home Department (Migrants’ Rights Network intervening) [2013] UKSC 51; [2013] 1 WLR 2358, paras 45-46). The language of voidness and nullity, drawn from the law of contract, can be useful for some purposes in administrative law, but it depends upon an analogy between defective contracts and defective administrative acts which is inexact. The complexity and variability of the practical consequences of unlawful administrative acts necessitate a more flexible approach than is afforded by a binary distinction between what is valid and what is void. Judges have therefore expressed reservations not only about the use of words such as “void” and “null”, but more importantly about reasoning in the field of administrative law which allows the logic of those concepts to override important values underpinning the court’s supervisory jurisdiction, such as the public interest in legal certainty, orderly administration, and respect for the rule of law.

In that regard, the speech of Lord Hailsham of St Marylebone LC in London & Clydeside Estates Ltd v Aberdeen District Council [1980] 1 WLR 182, 189-190, has been particularly influential. The Lord Chancellor noted that in reported decisions in the field of administrative law “there is much language presupposing the existence of stark categories such as ‘mandatory’ and ‘directory’, ‘void’ and ‘voidable’, a ‘nullity’, and ‘purely regulatory’.” He accepted that such language was useful, but observed that “I am not at all clear that the language itself may not be misleading in so far as it may be supposed to present a court with the necessity of fitting a particular case into one or other of mutually exclusive and starkly contrasted compartments, compartments which in some cases (eg ‘void’ and ‘voidable’) are borrowed from the language of contract or status, and are not easily fitted to the requirements of administrative law”. He continued (ibid):

The authorities cited by Haddon-Cave LJ in the present case are not inconsistent with that approach. The case of Hoffmann-La Roche concerned an application by the Crown for an interim injunction to enforce the application of a statutory order pending the determination of proceedings in which the validity of the order was challenged. The sole issue arising was whether the Crown should be required to give a cross-undertaking in damages. It was held that it should not. Lord Reid observed at p 341 that “an order made under statutory authority is as much the law of the land as an Act of Parliament unless and until it has been found to be ultra vires”. He went on to state that “the order which the appellants seek to annul is the law at present and if an interim injunction is refused that means that the law is not to be enforced and the appellants are to be at liberty to disregard it” (p 342). Similar observations were made by the other members of the majority.

Lord Diplock, whose speech was cited by Haddon-Cave LJ, commented at p 366 that “it leads to confusion to use such terms as ‘voidable’, ‘voidable ab initio’, ‘void’ or ‘a nullity’ as descriptive of the legal status of subordinate legislation alleged to be ultra vires for patent or for latent defects, before its validity has been pronounced on by a court of competent jurisdiction”. These were, he said, “concepts developed in the private law of contract which are ill-adapted to the field of public law”. All that could usefully be said, in his view, was “that the presumption that subordinate legislation is intra vires prevails in the absence of rebuttal, and that it cannot be rebutted except by a party to legal proceedings in a court of competent jurisdiction who has locus standi to challenge the validity of the subordinate legislation in question”.

In an earlier passage in his speech, Lord Diplock also said that it would be “inconsistent with the doctrine of ultra vires … if the judgment of a court in proceedings properly constituted that a statutory instrument was ultra vires were to have any lesser consequence in law than to render the instrument incapable of ever having had any legal effect upon the rights or duties of the parties to the proceedings” (p 365). That was the passage relied upon by Haddon-Cave LJ in the present proceedings. It was, however, obiter, since the case did not concern the effects of an instrument which had been held to be ultra vires.

Haddon-Cave LJ also relied upon the speech of Lord Irvine in Boddington . The issue in that case was whether a defendant who was prosecuted for the breach of a byelaw could raise, in his defence before the criminal court, the question whether the byelaw was invalid. The House of Lords held that he could, following authorities going back to the classic decision in Kruse v Johnson [1898] 2 QB 91. In the course of his speech, Lord Irvine entered into a discussion of the question raised in some earlier cases of whether the unlawfulness of an administrative act rendered it void or voidable, and the impact upon that debate of the decision in the Anisminic case, as interpreted in R v Hull University Visitor, Ex p Page [1993] AC 682. He concluded by stating that, where subordinate legislation or an administrative act is impugned as unlawful, “the legislation or act which is impugned is presumed to be good until pronounced to be unlawful, but is then recognised as never having had any legal effect at all” (p 155).

A number of observations may be made about this dictum. First, it was unnecessary for the decision of the case, which was concerned solely, in this regard, with the question whether a person could be convicted in criminal proceedings of violating subordinate legislation which was held to be unlawful. Secondly, Lord Irvine relied, in support of his statement of the law, upon the passage in the speech of Lord Diplock in Hoffmann-La Roche which was considered in para 36 above. As was there explained, that passage was obiter. Thirdly, and most importantly, a majority of the appellate committee dissociated themselves from Lord Irvine’s statement of the law.

In that regard, Lord Browne-Wilkinson stated at p 164 that he agreed with the speech of Lord Irvine but for one point:

Lord Slynn of Hadley stated at p 165:

Lord Steyn stated that he accepted “the reality that an unlawful byelaw is a fact and that it may in certain circumstances have legal consequences” (p 172). The remaining member of the committee, Lord Hoffmann, agreed with both Lord Irvine and Lord Steyn.

It is unnecessary in this appeal to embark upon a detailed consideration of the legal consequences of administrative measures which have been held to be unlawful. It is necessary to focus only upon the question whether it is a defence to a challenge to the lawfulness of the Secretary of State’s decision, on the basis that it was inconsistent with the order of the First-tier Tribunal, to establish that the order was unlawful.

As I have indicated, the Court of Appeal were mistaken in treating cases such as Hoffmann-La Roche and Boddington as governing the present case. Those cases were concerned with subordinate legislation, and also considered administrative acts and decisions. The present case is concerned with the order of a court or tribunal. That aspect of the case gives rise to different issues, with the consequence that it is governed by different principles.

It is a well established principle of our constitutional law that a court order must be obeyed unless and until it has been set aside or varied by the court (or, conceivably, overruled by legislation). The principle was authoritatively stated in Chuck v Cremer (1846) 1 Coop temp Cott 338; 47 ER 884, in terms which have been repeated time and again in later authorities. The case was one where the plaintiff’s solicitor obtained an attachment against the defendant in default of a pleaded defence, disregarding a court order extending the period for filing the defence, which he considered to be a nullity. The order in question had been intended to give effect to an agreement between the parties, but had mistakenly allowed the defendant longer to file a defence than had been agreed. The Lord Chancellor, Lord Cottenham, set aside the attachment, and stated at pp 342-343:

Three important points can be taken from this passage. First, there is a legal duty to obey a court order which has not been set aside: “it must not be disobeyed”. As the mandatory language makes clear, this is a rule of law, not merely a matter of good practice. Secondly, the rationale of according such authority to court orders, as explained in the second and third sentences, is what would now be described as the rule of law. As was said in R (Evans) v Attorney General (Campaign for Freedom of Information intervening) [2015] UKSC 21; [2015] AC 1787, para 52, “subject to being overruled by a higher court or (given Parliamentary supremacy) a statute, it is a basic principle that a decision of a court is binding as between the parties, and cannot be ignored or set aside by anyone, including (indeed it may fairly be said, least of all) the executive”. This principle was described (ibid) as “fundamental to the rule of law”. Thirdly, as the Lord Chancellor made clear in Chuck v Cremer , the rule applies to orders which are “null”, as well as to orders which are merely irregular. Notwithstanding the paradox involved in this use of language, a court order which is “null” must be obeyed unless and until it is set aside.

This rule was applied by the Court of Appeal in Hadkinson v Hadkinson [1952] P 285. In a well-known passage, Romer LJ stated at p 288:

This rule was not engaged in the case of Boddington (nor, for that matter, in Hoffmann-La Roche ), since that case was not concerned with a court order. That was one of the grounds on which the Divisional Court correctly distinguished Boddington in Director of Public Prosecutions v T [2006] EWHC 728 (Admin); [2007] 1 WLR 209, where the defendant was charged with breaching an anti-social behaviour order which was alleged to be invalid. Richards LJ cited Boddington and observed, at para 27:

As Richards LJ pointed out in Director of Public Prosecutions v T at paras 30-31, although Romer LJ referred in Hadkinson v Hadkinson to “a court of competent jurisdiction” (para 46 above), and although that case, like Isaacs v Robertson and M v Home Office , was concerned with a court of unlimited jurisdiction, the rule has also been applied to courts of limited jurisdiction: see, for example, Johnson v Walton [1990] 1 FLR 350 and In re B (Court’s Jurisdiction) [2004] EWCA Civ 681; [2004] 2 FLR 741.

That is consistent with the rationale of the rule. As explained in para 45 above, it is based on the importance of the authority of court orders to the maintenance of the rule of law: a consideration which applies to orders made by courts of limited jurisdiction as well as to those made by courts possessing unlimited jurisdiction. In the present case, the First-tier Tribunal was in any event a court of competent jurisdiction: it possessed jurisdiction under paragraph 22 of Schedule 2 to the 1971 Act to hear and determine applications for bail.

It is relevant to note some other recent examples of the application of this approach, in contexts more closely analogous to that of the present case. First, the case of R v Central London County Court, Ex p London [1999] QB 1260 concerned the compulsory detention of a patient in hospital under mental health legislation. The application for his admission by the hospital managers was made by the local social services authority, which had been authorised to make the application by orders made by the county court. The patient subsequently applied for judicial review to quash the court orders and the compulsory admission based upon them. The Court of Appeal concluded that the orders were valid, but went on to consider what the position would be if the county court had no jurisdiction to make them. Stuart-Smith LJ, with whom Robert Walker and Henry LJJ agreed, cited Hadkinson v Hadkinson , Isaacs v Robertson , Boddington and Percy v Hall , and concluded at para 36 that even if the county court had no jurisdiction to make the orders in question, the decision of the hospital managers to admit the applicant was valid. As he explained at para 30, if the orders were made by the county court without jurisdiction, then the applicant was entitled to have them quashed, but he was not entitled to a declaration that the decision to admit him was unlawful: that decision could only be quashed if it was ultra vires the hospital managers at the time when it was made, and it was not.

Another recent example, which also illustrates the point that the rule set out in Chuck v Cremer is not confined to orders made by courts possessing unlimited jurisdiction, is the decision of the Court of Appeal (Simon Brown, Mummery and Dyson LJJ) in R (H) v Ashworth Special Hospital Authority [2002] EWCA Civ 923; [2003] 1 WLR 127. The case arose out of the decision of a hospital authority to re-detain a patient after a mental health tribunal had ordered his discharge from detention. The hospital authority then applied for judicial review of the tribunal’s order, on the ground that it was unreasonable and unsupported by adequate reasons, and the patient applied for judicial review of the authority’s decision, on the basis that it was incompatible with the tribunal’s order. Both applications succeeded: the tribunal’s order was held to be unlawful and was quashed, but the authority was also held to have acted unlawfully in making a decision which was inconsistent with the tribunal’s order at a time when that order had not been set aside. The mental health tribunal was, of course, a body exercising a limited jurisdiction.

Dyson LJ based his reasoning upon article 5(4) of the European Convention on Human Rights, but it was entirely consistent with the common law. He stated at para 56:

Reference might also be made to the judgment of the Court of Appeal in R (Lunn) v Governor of Moorland Prison [2006] EWCA Civ 700; [2006] 1 WLR 2870, which concerned an error in a warrant of imprisonment. Moore-Bick LJ, giving the judgment of the court, stated at para 22:

Another recent example is the case of Rochdale Metropolitan Borough Council v KW (No 2) [2015] EWCA Civ 1054; [2016] 1 WLR 198, where a judge of the Family Division took the view that a decision of the Court of Appeal was ultra vires. Lord Dyson MR, giving the judgment of the Court of Appeal, stated at para 22:

A further example is the decision of the Court of Appeal in R v Kirby (John Martin) [2019] EWCA Crim 321; [2019] 4 WLR 131, which concerned convictions for the breach of a non-molestation order that was subsequently set aside because of a procedural irregularity. The convictions were upheld. Singh LJ, giving the judgment of the court, based the decision on “a long-standing principle of our law that there is an obligation to obey an apparently valid order of a court unless and until that order is set aside. This is a crucial feature of a civilized society which has respect for the rule of law” (para 13). In that regard, Singh LJ cited Chuck v Cremer , Hadkinson v Hadkinson , Isaacs v Robertson and M v Home Office , among other authorities, and followed Director of Public Prosecutions v T in distinguishing the case of Boddington .

In the light of this consistent body of authority stretching back to 1846, it is apparent that the alleged invalidity of the order made by the First-tier Tribunal had no bearing on the challenge to the decision of the Secretary of State. Even assuming that the order was invalid, the Secretary of State was nevertheless obliged to comply with it, unless and until it was varied or set aside. The allegation that the order was invalid was not, therefore, a relevant defence to the application for judicial review of the Secretary of State’s decision. As there was no other basis on which the Court of Appeal reversed the Upper Tribunal, and the Secretary of State does not ask the court to dismiss the appeal on other grounds, it follows that the appeal should be allowed.

Before leaving this case, it is appropriate to make some observations about the procedure followed by the Court of Appeal in allowing a challenge by the Secretary of State to the decision of the First-tier Tribunal to be mounted in the form of a defence to an application for judicial review of a decision made by the Secretary of State. As I have explained, that should not have been allowed, because the challenge could not constitute a relevant defence in the light of the principle laid down in Chuck v Cremer . It was for that reason that challenges to court orders, by way of defence to criminal proceedings for their breach, were not permitted in Director of Public Prosecutions v T (para 47 above) and R v Kirby (John Martin) (para 55 above).

Even if the defence had been relevant, however, the procedure adopted raises a number of questions. One way of dealing with such a situation, which could not have given rise to any procedural difficulties or unfairness, would be to have adjourned the appeal, if that was considered appropriate, so as to enable the Secretary of State to apply for permission to apply for judicial review. In the Ashworth case, for example (para 51 above), there were two separate applications for judicial review of the two decisions in issue, which were considered together by the Court of Appeal. Allowing a challenge to an act or decision to be mounted instead in the form of a defence to an application for judicial review of another act or decision raises a number of issues which require consideration. As those issues do not arise for decision in this case, and were not fully addressed, this is not the occasion for a detailed consideration of the matter, but some relevant points can be noted:

For the reasons summarised at para 56 above, I would allow the appeal.