Anthony Rae v United States of America

The appellant, Anthony Rae, is sought by the USA pursuant to a request dated 14 May 2020 and certified on 17 June 2020. He is wanted for trial on two indictments dating from 2003, each containing three counts alleging sexual abuse of a child and on two further indictments dating from 2019, each containing two counts of failing to surrender in relation to the sexual assault charges. The indictments were preferred by grand juries in Grayson County, Texas. The request is for Anthony Stevens, by which name the appellant used to be known.

On 21 June 2021, after a hearing at Westminster Magistrates Court, District Judge Tempia (“the judge”) decided to send the appellant’s case to the Secretary of State. On 6 August 2021, the Secretary of State ordered the appellant’s extradition.

Permission to appeal was refused on the papers by Jay J on 2 March 2022. After a hearing on 4 May 2022, however, I granted the appellant permission to appeal against the judge’s decision. The sole ground of appeal is that the judge was wrong to conclude that extradition would be compatible with the appellant’s rights under Article 3 of the European Convention on Human Rights (“ECHR”), given the prison conditions he may face if convicted in Texas.

This appeal was originally listed for hearing on 10 November 2022. In accordance with the usual practice where an appellant wishes to attend his hearing, HMP Wandsworth had been notified in advance that the appellant was to be produced in the video suite in time for the hearing to begin at 10.30am, so he could observe and listen to the proceedings remotely. A further reminder was sent on the morning of the hearing. No indication was given that there would be any difficulty until the hearing started, at 10.30am, and the appellant had not been produced. The court was informed that it would not be possible to produce him until 2pm because there were no free video suites. Since half a day was not sufficient for argument, it was necessary to adjourn the hearing. HMP Wandsworth’s failures (i) to ensure that the appellant was produced on time and (ii) to make any attempt to inform the court that it would not possible to produce him meant that a day of court time (which could have been used to hear another case) was wasted and substantial additional costs were incurred by the parties.

Since permission to appeal was granted, the respondent has produced a letter dated 24 May 2022 from Jason Clark, Chief of Staff of the Texas Department of Criminal Justice (“TDCJ”), providing what is variously described as an “assurance” or “information”. There is a further email dated 17 October 2022 providing further information. The respondent invites me to admit this fresh evidence under my inherent jurisdiction. The appellant does not object and indeed relies on the further information. I am satisfied that it is in the interests of justice to consider it, even though it was not before the judge. I therefore grant the application to adduce the letter and email.

The appellant’s main evidence on prison conditions in Texas took the form of an expert report from Michele Deitch, a Distinguished Senior Lecturer at the University of Texas in Austin and also an attorney. Ms Deitch had been a federal court-appointed monitor of prisons from 1984-1990 and continues to go into prisons every year. She served as the original reporter (draftsperson) for the American Bar Association’s Standards for the Treament of Prisoners (2010), a set of “aspirational” standards intended to guide courts, policy makers and correctional leaders.

Ms Deitch submitted two reports, the second a revised version of the first. The contents were based on her own personal knowledge of the Texas prison system, gained through first hand observation of prison conditions, her research team’s analysis of relevant data, her own research into correctional oversight, her continuous review of extensive media coverage, her reading of relevant court opinions, testimony at legislative hearings and letters and information received from incarcerated people and their families and from prison staff.

Ms Deitch noted that in Texas there is a distinction between “jails” and “prisons”. Remand prisoners and those sentenced to terms of less than a year for misdemeanour offences were typically held in county jails. Mr Rae would be held in the Grayson County Jail if remanded in custody. His stay there would likely be relatively short as most trials take place within a year. If convicted, he would be turned over to the Texas Department of Criminal Justice (“TDCJ”), which runs the prison system, which houses those convicted of felony offences. There are about 100 of these prisons, ranging in capacity from 1,000 to 3,000. They are generally in rural areas. A convicted person could be housed in any of these.

Assuming that he is convicted of the aggravated sexual assault of a child (a first degree felony in Texas), he could face a sentence of up to 99 years’ or life imprisonment. He would be required to serve a minimum of half his sentence before becoming eligible for parole, which Texas authorities are notoriously unwilling to grant to those convicted of serious sexual offences.

Ms Deitch’s report deals with many aspects of the treatment of prisoners in the Texas prison system. These include the lack of independent external oversight mechanisms, COVID risks and the lack of effective precautions against those risks, poor quality food, extreme heat, physical conditions, safety issues, the use of solitary confinement, work requirements and lack of pay and visitation and family contact. She also addressed risks particular to the appellant, given his health issues and age.

Given the way the case was presented before the judge and before me, it is not necessary to set out the parts of her report dealing with issues other than extreme heat and personal space.

As to extreme heat, Ms Deitch said this:

As to physical conditions, Ms Deitch said this:

The evidence for the respondent included comprehensive information about Grayson County Jail, where the appellant would be held if remanded in custody. It is not necessary to summarise that information, because the appellant’s contention that there is a real risk of a violation of Article 3 ECHR is based on conditions in the prisons under the responsibility of the TDCJ, in which he would serve any sentence imposed. As to those conditions, the respondent relied on a series of letters from Mr Clark on behalf of the TDCJ.

Again, given the focus of the appellant’s arguments before me and below, I concentrate here on the passages relevant to extreme heat and personal space. On extreme heat, Mr Clark said this in his letter of 14 December 2020:

Information about the sizes of cells was provided by Mr Clark in a letter of 19 January 2021:

Mr Clark provided further information in a letter of 17 February 2021, noting that the UK defines “unencumbered space” differently from the US and that the TDCJ has a mixture of dormitory and cellblock housing. He continued:

At [137] et seq., the judge set out Article 3 ECHR and the key cases interpreting it. At [142], she described the approach of the Grand Chamber of the Strasbourg Court in Muršić v Croatia (2017) 65 EHRR 1. The Court confirmed that in multi-occupancy cells, 3 sq. m. was the minimum floor space per detainee. Where it fell below that, there was a strong presumption of a violation of Article 3, which was normally capable of being rebutted only where (1) the reductions in the required minimum personal space are short, occasional and minor, (2) such reductions are accompanied by sufficient freedom of movement outside the cell and adequate out-of-cell activities and (3) the applicant is confined in what is, when viewed generally, an appropriate detention facility and there are no other aggravating aspects of the conditions of his or her detention.

The judge noted that the appellant accepted that the required personal space would be provided on remand, so the focus was on the facilities that would be provided by the TDCJ if he should be convicted. The judge undertook a careful, analysis of the written and oral evidence of Ms Deitch. At [148], she considered the issue of personal space. At [149], she noted:

By contrast the judge considered that the evidence adduced by the respondent, in the form of the letters from Mr. Clark, was more specific. The judge noted Mr Clark’s evidence that there was currently under-occupancy of the prisoner estate.

At [150]-[151], the judge noted that Mr Clark had set out the specific standards for spacing in cells as required by the ACA, through which the TDCJ was accredited. The judge found it clear that Mr Clark was referring to the whole prison estate, and not merely to some specific prisons. On this basis, she held that the appellant had not provided clear and cogent evidence of a real risk of a breach of the appellant’s Article 3 rights in respect of personal space if extradited.

At [152], the judge considered the question of extreme heat. She drew attention to Ms Deitch’s evidence on this topic. However, at [153], she found that these concerns had been “comprehensively addressed” by Mr Clark in his letter dated 14 December 2020. Therefore, there was no clear and cogent evidence of a real risk of an Article 3 violation on account of extreme heat or temperature.

The material part of Mr Clark’s letter of 24 May 2022 is as follows:

The email of 17 October 2022 shows that the US authorities were asked to provide “information identifying whether there was a less than 5% prospect of the appellant being detained in nonconforming housing taking into account based on: how many inmates are currently housed in cells where they receive less than 3 sq. m. of personal space; the current TCDJ prison population; the percentage of inmates in dormitory in cell block housing; the percentage of cell block housing designed for housing two inmates; and at present the percentage of housing areas allowing two individuals which are occupied by two inmates”. The email continues:

Ms Rebecca Hill for the appellant submitted that, on the respondent’s own evidence, the appellant faced a real risk of Article 3 ill-treatment. She focussed on two aspects: personal space and extreme heat.

Mr Clark’s letter of 17 February 2021 showed that 13% of cell accommodation did not meet the Muršić standard if there were two prisoners in a cell. Ms Deitch’s evidence was that cells designed for two people were used to house two people. Taken together with the email of 17 October 2022, the respondent’s own evidence showed that the proportion of prisoners in accommodation with less than 3 sq. m. of unencumbered space was somewhere between 5% and 13%. The Texas authorities were conspicuously silent about how many cells were occupied by only one person.

The judge had placed considerable emphasis on the occupancy figures, but there was no proper basis for deriving any comfort from that, because overall occupancy rates for the prison estate tell one nothing about occupancy rates in particular institutions. Mr Clark’s evidence suggests that the authorities respond to under-occupancy by closing or mothballing institutions or wings, rather than by accommodating prisoners more generously in the institutions or wings that remain open.

Ms Hill relied on the recent judgments of the Divisional Court (Jay J and Stuart-Smith LJ) in A v France, in which that Court formulated a detailed request for information from the French authorities about personal space and other conditions in relation to particular prisons ([2021] EWHC 2543 (Admin)) and decided, after consideration of the response, that there was no real risk of a breach of Article 3 ([2022] EWHC 841 (Admin)). She pointed out that the latter conclusion was reached on the basis that the proportion of cells with less than 3 sq. m. of space was less than 5% (see at [11]) and of these almost all had 2.96 sq. m. of personal space (see at [67]). This was far removed from the present case, where the statistical risk was higher, possibly much higher, and the non-confirming cells in many cases had considerably less than 3 sq. m. of personal space and in some cases as little as 1.86 sq. m.

As to the letter of 24 May 2022, Ms Hill submitted that it would have been easy for the Texas authorities to give an assurance that the appellant would be housed in a “conforming” cell. But the letter gives no such assurance, just an assurance that “a good faith effort” would be made to achieve that outcome. As the letter candidly made clear, this meant that there was no guarantee that he would not be placed in such a cell “due to operation or inmate needs”. This alone meant that there was a real risk of treatment contrary to Article 3.

The position was even clearer when the effect of extreme heat was taken into account. As to that, she referred to the decisions of the Strasbourg Court in Muršić v Croatia (2017) 65 EHRR 1, [139], Štrukelj v Slovenia (App. No. 6011/10), Judgment 24 February 2014, [10], Mathew v Netherlands (2006) 43 EHRR 23, [211], [214] and [217] and Peers v Greece (App. No. 28524/95, Judgment 19 April 2001), [64], [70], [72] and [75]. The temperatures found to give rise to a breach of Article 3 in some of those cases were well below those evidenced in this case. The evidence of Ms Deitch, taken with the conclusions of the US federal judge to which she referred, were such that it was not open to the judge to conclude that the measures relied upon by the Texas authorities adequately addressed the effects of extreme heat.

For the respondent, David Perry KC made six general submissions as to the proper approach to Article 3 ECHR.

First, ill-treatment must reach a certain level of severity in order to fall within the scope of Article 3 ECHR: Ireland v UK (1980) 2 EHRR 25, [162]. The level is relative and depends on all the circumstances of the case, including the age and characteristics of the prisoner: see Dougoz v Greece (2002) 23 EHRR 1480, at [44].

Second, the test to be applied when considering a violation of Article 3 in foreign cases is a high one: R (Ullah) v Special Adjudicator [2004] 2 AC 323, [24] (“it is necessary to show strong grounds for believing that the person, if returned, faces a real risk of being subjected to torture or to inhuman or degrading treatment or punishment”).

Third, while the prohibition on Article 3 ill-treatment remains absolute, the extradition context is important. In Sanchez-Sanchez v United Kingdom (Application No. 22854/20, Judgment 3 November 2022), the Grand Chamber of the Strasbourg Court had found that a prospect of treatment which would breach Article 3 in a contracting state would not necessarily prevent extradition to a state like the USA.

Fourth, it is not sufficient to demonstrate that Article 3 is generally violated in a requesting state; it is necessary to establish that the requested person will in particular be subject to such treatment: Miklis v Lithuania [2006] EWHC 1032 (Admin).

Fifth, lack of personal space does not necessarily amount to a violation of Article 3, even where there is less than 3 sq. m.: Muršić, [138].

Sixth, in Case C-404/15 PPU Aranyosi EU:C:2016:198 [2016] QB 921, it was established by the Court of Justice of the European Union that in European arrest warrant cases, where a prospective finding of violation of Article 3 is likely, there is an obligation on the court to request further information rather than to discharge the accused: [91] and [95]. In Larco v Ecuador [2020] EWHC 1797 (Admin), the Divisional Court accepted at [43]-[44] that a similar approach applies in relation to a Part 2 case: see, e.g., Dzgoev v Russia [2017] EWHC 735 (Admin), and Yilmaz and Yilmaz v Turkey [2019] EWHC 272 (Admin). (I adopted the same approach in Kaderli v Turkey [2021] EWHC 1096 (Admin), [76].)

Mr Perry emphasised that assurances were to be presumed to be given in good faith. He relied on the Divisional Court’s decision in Snowden v Ghana [2018] EWHC 198 (Admin), [24]-[25]. In that case, extradition was ordered even though the Ghanaian authorities could not promise that the requested person would be kept in a particular prison in all circumstances. The presumption of good faith was particularly strong in the case of assurances given by a state with a long-standing history of respect for democracy, human rights and the rule of law. More generally, Mr Perry submitted that the approach of the Divisional Court in A v France should not be followed. In that case, the very detailed list of questions which the court had formulated for the French authorities amounted to an inappropriate attempt to micromanage the conditions of detention.

As to personal space, Mr Perry submitted that the ACA standards comply with the requirement for a minimum of 3 sq. m. personal space, because the former do not include space occupied by furniture, whereas the latter does. The information as to the proportion of cells where the space falls below 3 sq. m. does not establish a 13% real risk that the appellant would be provided with inadequate personal space, given that the appellant might be housed in dormitory accommodation (where 3 sq. m. is provided), there is low prison occupancy and there is no basis for concluding that two inmates are being detained in cells where 3 sq. m. is not provided. Even if there were a 13% chance of being detained in a cell with less than 3 sq. m. of personal space, this would not establish a “real risk”.

The email of 17 October 2022 does not affect the analysis. It was understandable that the TDCJ declined to commit themselves to the assessment that the risk of the appellant being housed in non-conforming accommodation was lower than 5%. This reflects the caution which in Sanchez-Sanchez was recognised as an inherent part of the predictive exercise. The assurance that the TDCJ would make a good faith effort to place the appellant in a conforming cell was given in good faith; its cautious terms reinforce that.

Generally, the judge was entitled to conclude that other material conditions, including extreme heat, did not create a real risk of a violation of Article 3 ECHR.

In Elashmawy, at [49], Aikens LJ summarised in a set of propositions the correct approach in extradition cases where it is alleged that prison conditions in the requesting state give rise to a real risk of treatment that would breach Article 3 ECHR:

At [50], Aikens LJ set out another set of principles derived from Krolik v Poland [2012] EWHC 2357 (Admin), [2013] 1 WLR 490, which on their face apply when considering extradition to ECHR contracting states and EU Member States, but which Mr Perry submitted were also of relevance when considering extradition to other states with long-established traditions of democracy and adherence to the rule of law, such as the USA:

Neither party takes issue with [49] of Elashmawy as a summary of the law applicable at the time of the judgment, though it predated Muršić. The applicability of [50] of Elashmawy to a request for extradition to a non-ECHR state was not, however, common ground. In my judgment, the presumption described there does not apply to such requests.

The reason why ECHR states are presumed, absent cogent evidence to the contrary, to comply with ECHR standards is because they have assumed the obligation in international law to do so and taken steps to implement that obligation in their domestic law – and their good faith in that regard is to be presumed. There are other liberal democratic states with at least as firm a commitment to the rule of law, which are neither contracting parties to the ECHR nor members of the EU, but which have their own binding constitutional or human rights standards. These standards may be higher or lower in particular respects than those of the ECHR. In their case, good faith supplies no reason for them to conform to the standards set by the ECHR, to which they have never committed themselves.

To take an obvious example, the Eighth Amendment to the US Constitution contains a prohibition against “cruel and unusual punishment”. Many punishments which would contravene ECHR standards if carried out in a contracting state would also contravene the Eighth Amendment, but not all. The death penalty contravenes the minimum standards as elaborated in the case law of the Strasbourg Court and now codified in the 13^th Protocol to the ECHR, but has generally been held by the US Supreme Court to be consistent with the Eighth Amendment (save during a short period in the 1970s). This is a reflection of the fact that, even among mature democratic polities committed to the rule of law, there are differences of view about what substantive standards should be imposed.

Minimum standards of the kind found in the case law of the Strasbourg Court on personal space are even less likely to attract universal agreement. The selection of 3 sq. m. as the presumptive minimum draws a bright line which is helpful in achieving consistency among contracting states. But there is no a priori reason to expect or presume that a state which has not committed itself to comply with the ECHR should or will meet the 3 sq. m. standard. That raises the question whether it is appropriate to require non-contracting states to adhere to such a standard as a precondition for extradition.

In Wellington v USA [2008] UKHL 72, [2009] 1 AC 335, the House of Lords considered whether extradition to the USA was compatible with Article 3 ECHR where there was a real prospect of the imposition of a sentence of life imprisonment without parole. The leading opinion was given by Lord Hoffmann. Having analysed the Strasbourg jurisprudence, he held that the imposition by a contracting state of a whole life term did not ipso facto infringe Article 3 ECHR, provided that the sentence was “de jure and de facto reducible”, and that in deciding that question, it was relevant whether there was a system providing for consideration of the possibility of release: see at [9]-[10], citing Kafkaris v Cyrpus (2009) 49 EHRR 35, at [98]-[99].

Lord Hoffmann then considered how this applied in the extradition context. He drew a distinction between torture and other forms of inhuman and degrading treatment. So far as the latter was concerned, he noted that at [89] of the Strasbourg Court’s judgment in Soering v United Kingdom (1989) 1 EHRR 439, it had recognised the importance of extradition arrangements and of avoiding safe havens. This, he said at [24], meant that:

Lord Carswell and Lady Hale agreed with Lord Hoffmann on this point. Lord Brown and Lord Scott did not. At [86], Lord Brown said that there was:

In Ahmad v United Kingdom (2013) 56 EHRR 1, the Strasbourg Court had to consider the proper approach to Article 3 in extradition cases in the context of a request by the USA for extradition of individuals to face charges for terrorism offences. One of the complaints was that conditions of detention at a particular prison – ADX Florence – fell below Article 3 minimum standards. A key issue was whether Article 3 was “relativist” in the sense used by Lord Hoffmann in Wellington. As to that, the Court held at [171] that no distinction could be drawn between torture and other forms of inhuman and degrading treatment. At [172]-[175], it held that the question whether the treatment reached the minimum level of severity needed for an Article 3 violation could only be assessed independently of the reasons for removal or extradition. The Court then said this:

Language almost identical to this can also be found in the Strasbourg Court’s judgment in Harkins & Edwards v United Kingdom (2012) 55 EHRR 19, [124]-[131]. The catalogue of case-specific factors set out by the Strasbourg Court in Ahmad at [178] was cited with approval by the Supreme Court in R (AB) v Secretary of State for Justice [2021] UKSC 28, [2022] AC 487, [42].

Ahmad and Harkins & Edwards distinguish between two forms of “relativism” in the context of Article 3 ECHR. They disapprove the kind espoused by Lord Hoffmann, according to which the question whether treatment in another state is inhuman or degrading depends on the reasons for removal to that state. Such reasons might include case-specific ones (such as the seriousness of the offence and, thus, the weight to be given to the public interest in ensuring that the requested person is tried for it) or more general ones (such as the weight to be given to honouring extradition treaties and avoiding safe havens). The effect of Ahmad and Harkins & Edwards is that none of these can be relevant to the question whether extradition is compatible with Article 3 ECHR.

By contrast, Ahmad and Harkins & Edwards appear to endorse the different form of relativism referred to by Lord Brown (which might be better termed “contextualism”) in which the question whether treatment reaches the minimum level of severity could be sensitive to context, including the conditions in the country concerned. Thus, a practice like “slopping out” might be considered debasing or humiliating in a country like Scotland, but not so in a poorer country, for example if such a practice were common outside the prison context. A failure to provide particular forms of medical treatment might also reach the minimum level of severity in a contracting state but not in a country where such treatment was not generally available.

The passage at the end of [178] of Ahmad also has some significance here. It emphasises that assessing whether treatment contravenes Article 3 is an intensely fact-sensitive exercise and, for that reason, underlines the difficulty of establishing a breach prospectively in an extradition or expulsion context.

The significance of [179] is that, while not suggesting that states with a long history of respect of democracy, human rights and the rule of law should be presumed to comply with ECHR standards (as ECHR contracting states are), the Strasbourg Court has laid down a marker that it will be very rare to find a real risk of a treatment which would breach Article 3 ECHR in such states.

Are the minimum space requirements in Muršić, which apply in a domestic case, also applicable in the context of extradition to a non-ECHR state? In Serra v Paraguay [2017] EWHC 2300 (Admin), the Divisional Court (Burnett LJ and Sir Wyn Williams) described Muršić as “the latest incremental step” in driving up minimum standards in prison conditions in ECHR states: see at [13]. The Court then said this:

The question identified in Serra does not appear to have been answered, though the Divisional Court has on several occasions assumed that the Strasbourg Court’s minimum space requirements do apply without modification in the context of extradition to a non-ECHR state: see Cato v Peru [2016] EWHC 914 (Admin), [23(ii)] (Laws LJ and Nicol J); Chawla v India [2018] EWHC 1050 (Admin), [27] (Leggatt LJ and Dingemans J); Pimenta v Brazil [2017] EWHC 2588 (Admin), [21(7)] (Hamblen LJ and Dingemans J); Larco v Ecuador, [41] and [60] (Holroyde LJ and William Davis J); Sanchez v USA [2020] EWHC 508 (Admin), [10(vi)] and [50] (Fulford LJ and Elisabeth Laing J). The first two of these authorities applied the case law before Muršić, which also attached significance to the 3 sq. m. minimum standard, through a slightly weaker presumption.

But none of these authorities squarely addressed the question raised in Serra. Mr Perry invites me to grasp the nettle and hold that a real risk of being held in a cell with less than 3 sq. m. of personal space should not, in and of itself, be a bar to extradition, even if the state concerned is unable to satisfy the cumulative requirements in Muršić at [138]. The principal foundations for that submission are the very recent judgments of the Strasbourg Court in Sanchez-Sanchez and McCallum v Italy (App. No. 20863/21), both given by the same constitution of the Grand Chamber on 3 November 2022 and both concerning the compatibility of extradition to the USA with Article 3 in the context of the possible imposition of a whole life term of imprisonment.

In order to understand what the Grand Chamber decided in those cases, it is necessary to explain the background. In Vinter v UK (2016) 63 EHRR 1, the Grand Chamber held that Article 3 ECHR requires whole life sentences to be “reducible” by way of a review which allows the domestic authorities to consider whether any changes in the prisoner or progress towards rehabilitation are such that continued detention was no longer justified: [119]. What was required in the domestic context was “a dedicated mechanism guaranteeing a review no later than 25 years after the imposition of a life sentence, with further periodic reviews thereafter”: [120]. In Trabelsi v Belgium (2015) 60 EHRR 21, the Strasbourg Court applied these criteria in a case about extradition to the US, holding that the US mechanisms of review relevant to that case did not satisfy them: [137].

In Sanchez-Sanchez, the Grand Chamber of the Strasbourg Court accepted the UK Government’s invitation to depart from Trabelsi. It pointed out at [91] that Vinter was not an extradition case. At [92], it noted that, in the domestic context, the applicant will have been convicted and sentenced, so his position will be known. In the extradition context, by contrast, a complex risk assessment is called for and this justified caution in applying the Vinter principles. At [93], it distinguished between the “substantive” obligation to ensure that a life sentence does not over time become a penalty incompatible with Article 3 and “related procedural safeguards”, which were not ends in themselves, but served to avoid a breach of the substantive obligation by contracting states. To apply the procedural safeguards in the extradition context would amount to “an over-extensive interpretation of the responsibility of a Contracting State” and would be unduly difficult for domestic authorities deciding on extradition requests. At [94], the Court then gave this additional reason:

Thus, the procedural safeguards set out in Vinter were not a prerequisite for compliance by the sending Contracting State with Article 3 ECHR: [96]. However, at [99], the Court said this:

In McCallum , the Grand Chamber of the Strasbourg Court dismissed as manifestly inadmissible a complaint brought by a woman sought by the US for trial in Michigan. The applicant had argued that, even for second-degree murder, there was a real risk of a sentence that was irreducible in the Vinter sense, because of the role of the Governor of Michigan (a politician) in the decision whether to grant parole. At [53], the Court rejected that argument because it “relates to a matter that cannot be regarded as pertaining to the Vinter safeguard… but rather is more in the nature of a procedural guarantee”. The court “refers to the distinction between a substantive obligation and the related procedural safeguards that derive from Article 3 when it comes to the issue of life sentences in the extradition context”. The availability of procedural safeguards in the requesting state was not a prerequisite for compliance by the requested contracting state.

Thus, the position can be summarised as follows:

Other than Serra , where the point was not decided and the Divisional Court expressly indicated that it might in due course have to be, there is no authority directly addressing the applicability of the Muršić minimum space requirements to extra-territorial cases. I therefore consider the issue from first principles.

Sanchez-Sanchez represents a genuine development in the Strasbourg Court’s case law, but the development is not as fundamental as Mr Perry suggests. The judgment represents a reversal of one branch of the Article 3 case law concerning extra-territorial cases where there is a real prospect of a whole life prison sentence in the requesting state. Even in that narrow field, the Strasbourg Court took pains to distinguish between substantive requirements (which were applicable even in extra-territorial cases) and procedural requirements (which were not). This is consistent with the well-established line of authority on the application of the procedural guarantees of Articles 5 and 6 to extra-territorial cases. In that context, it is not enough to show a real risk that ECHR procedural standards will not be honoured; it is necessary to show a real risk of a “flagrant denial of justice”: see e.g. Othman v United Kingdom (2012) 55 EHRR 1, [259]; Elashmawy, [38].

It is easy to see why it would cause real difficulties if ECHR contracting states were required to insist, as a precondition for extradition to a non-ECHR state, on the same procedural safeguards as contracting states are required by Articles 3, 5 or 6 to confer on those within their own jurisdictions. The procedural safeguards offered by a non-ECHR state are likely to depend on the constitutional and legal traditions of that state or (in a federal state) of its component parts. For example, the constitution may reserve certain tasks to a member of the executive, or the law may prescribe a particular procedure for applications for parole or clemency. It may well be constitutionally, legally or politically impossible for a non-ECHR state to comply with ECHR procedural safeguards either generally or, following a request for an assurance, in particular cases. The practical consequence of imposing such standards as a precondition for extradition would be to turn the espace juridique of the ECHR into a safe haven for fugitives from the justice system of such states.

The same point seems to have underlain Lord Brown’s caution, endorsed in Harkins & Edwards and Ahmad , about assuming that treatment which would reach the Article 3 minimum level in a contracting state must necessarily do so in a non-ECHR state. The effect of transposing a substantive requirement for flushing toilets or particular types of medical intervention to the extra-territorial context might be to prevent extradition to states where such facilities or interventions were not available.

But it is important not to allow this point to be stretched further than justified by the policy concerns which animated it. Suppose, for example, that a state where flushing toilets are universally used by the general population decides for policy reasons to deny its prisoners access to them in circumstances which would give rise to a breach of Article 3 in a contracting state. In that case, it would be very difficult to say, when considering extradition to that state, that there was no real risk of a breach of Article 3 without infringing the principle in para. 64(a) above. This would be so whether the lack of flushing toilets was designed to be part of the punishment or merely reflected a desire not to spend public money on prisoners. Such a state might choose to provide the facilities required to avoid a breach of Article 3 minimum standards for prisoners it is seeking to extradite from ECHR contracting state, but not for others. If it decided not to do that, Article 3 would preclude extradition. If this made the espace juridique of the ECHR a safe haven for those sought by its justice system, that would be the result of its own policy choice.

In my judgment, the same approach applies to the two aspects of prison conditions relied upon by the appellant in this case: personal space and extreme temperature. In the case of personal space, Muršić creates a strong presumption that Article 3 will be breached in multi-occupancy accommodation if less than 3 sq. m. per prisoner is provided, unless the requirements in [138] of that decision are cumulatively met. There is no reason why it would be inappropriate to apply the same rule in a case concerning extradition to the US. No-one has suggested that the US in general, or Texas in particular, lacks the resources to provide accommodation meeting the Muršić minimum requirements: indeed, the evidence suggests that Texas does provide conforming accommodation for 87% of those housed in two-person cells. Even applying a contextual approach to the question whether treatment reaches the minimum level of severity necessary to engage Article 3, there is no convincing reason of principle why accommodation that falls below Article 3 standards because of inadequate personal space in a contracting state should be held not to breach such standards in a case concerning extradition to the US.

It may be objected that this approach involves refusing extradition to a non-ECHR state because of a real risk of treatment to which many prisoners are in fact subject in contracting states. In my judgment, this objection is not convincing. No doubt conditions which breach Article 3 standards have at times persisted, and may still do so, in some contracting states. If so, those states are required by Article 13 to provide domestic legal remedies to address them. If those remedies are not effective, complaint can be made to the Strasbourg Court, the implementation of whose judgments is, in turn, subject to supervision by the Council of Ministers. Cases where Article 3 is relied upon extra-territorially are different because the act of extradition, deportation or removal puts the individual beyond the protection of ECHR system. If the constitution or law of the requesting state imposes standards that are, in certain respects, lower than the minima insisted upon by Article 3, there will be nothing that can be done to enforce those minima. This is why, since Soering , the Strasbourg Court has held that a real risk of a breach of Article 3 by a non-ECHR state engages the responsibility of a contracting state considering whether to extradite or expel.

The question for the judge in this case, therefore, was whether there were serious grounds to believe that the appellant would run a real risk of being housed in accommodation with less than 3 sq. m. of personal space, other than in circumstances satisfying all three of the conditions in [138] of Muršić. In answering that question, the principles I have summarised in para. 64(c) and (d) above applied. Given the contextual nature of the Article 3 assessment, it will be a rare case where the “real risk” test is met prospectively, but each case depends on its own facts.

As to the number of cells of particular dimensions, the judge preferred Mr Clark’s evidence over Ms Deitch’s. Having heard Ms Deitch cross-examined, she was entitled to do so. In relation to personal space, the “real risk” question therefore had to be answered on the basis of Mr Clark’s evidence. At [149] of her judgment, she recorded his statement, in his letter of 17 February 2021, that 87% of the two person cells actually occupied by two persons and 100% of the two person cells occupied by one person conformed to the Muršić minimum space requirement. The judge then said this:

Thus, a key part of the judge’s reasoning depended on an inference from the underoccupancy of the TDCJ system to the unlikelihood that the appellant would be placed in a non-conforming cell. This inference, however, was unsound. The significance of under-occupancy depends on the authorities’ response to it. If the response is to close down, idle or mothball particular prisons, or parts of them, there would not necessarily be any impact on the likelihood of the appellant being placed in a non-conforming cell. In fact, there was specific evidence from Mr Clark that this was the TDCJ’s response to under-occupancy. In his letter, he had said:

In my judgment, there was a material error in the judge’s reasoning. On the evidence before her, she could not properly infer from the under-occupancy of the Texas prison system that there was no real risk of the appellant being placed in a non-conforming cell. On that evidence, the precise extent of the risk was unclear, but it might be as much as 13%.

At one point in the argument, Mr Perry submitted (in response to a hypothetical question from me) that, even if cells were allocated by drawing lots, a 13% risk of being placed in a non-conforming cell would not be a “real” risk. I have no hesitation in rejecting that submission. Assessing whether the “real risk” test is met involves looking at both the risk of being subjected to the treatment and the extent to which the treatment would fall below the Article 3 minimum: see e.g. A v France (No. 2), where Divisional Court regarded as relevant both the low risk of the treatment eventuating (under 5% on an arithmetic basis) and the fact that if it did eventuate the breach of the Muršić minimum space requirement would be likely to be marginal. In this case, however, the non-conforming cells had substantially less space than the Muršić minimum. There was no evidence that they meet any other international standard, as the Divisional Court in Serra speculated might be the case in a non-ECHR country. Moreover, there was no information about the length of time for which a prisoner might be held in such a cell. In that context, a risk which may be as high as one in eight is, on any view, “real”.

The only remaining question is whether the fresh evidence is such as to reduce the risk below the requisite level. In my judgment it is not. The email of 17 October 2022 supports the conclusion that, leaving the “assurance” of 24 May 2022 aside, the statistical risk of the appellant being placed in a non-conforming cell is somewhere between 5% and 13%, i.e. a “real” one.

Whether the letter of 24 May 2022 gives an “assurance” or is better characterised as providing “further information” does not matter. Either way, it is far removed from the Ghanaian assurance accepted as sufficient in Snowden. In that case, the assurance said that the Ghana Prison Service was prepared to accommodate the requested person in a particular named establishment and gave details of the cell accommodation in which the requested person “will be detained”: see at [14]-[15]. The limited caveat was that the Ghana Prison Service could “only guarantee its commitment to the extent allowable under its governing laws”. Here, by contrast, there is no commitment about where the appellant will be held, just an undertaking that the TDCJ will make a “good faith effort” to place him in a conforming housing area. But, as the letter makes clear, this is not a guarantee; and the reasons why he might be placed in non-conforming accommodation go well beyond the force majeure-type circumstances envisaged in Snowden and include “operation or inmate needs”.

I should emphasise that there can be no legitimate question about the bona fides of Mr Clark, or about the credit to be accorded as a matter of principle to any assurance from the US authorities: see, in this regard, Ahmad v USA [2006] EWHC 2927 (Admin), [2007] HRLR 8, [74]-[75] (Laws LJ). The question here is not whether the assurance will be honoured: one can properly assume that it will. The question is what will happen if it is. As to that, Mr Clark’s letter is both careful and frank. Even taking into account the genuine undertaking to make a “good faith effort”, a real risk remains that, for operational or inmate-related reasons, he will have to be placed in non-conforming accommodation in circumstances that would breach Article 3 standards.

I turn next to the question of extreme heat. As to this, the judge did not reject Ms Deitch’s evidence that temperatures can reach 110F (43C). She recorded this evidence at [152] of her judgment. Ms Deitch had in fact been more specific. She had said that temperatures can stay that high for weeks on end during the summer months and that indoor temperatures were “even higher”. This was not disputed. The sole basis for concluding that this state of affairs did not give rise to a real risk of treatment contrary to Article 3 was that these concerns had been “comprehensively addressed” by Mr Clark in this letter of 14 December 2020: [153].

The difficulty with this is that, aside from air conditioning (which Mr Clark said was limited to “some units” and Ms Deitch said only 1 in 5 prisoners had), the precautions listed were all said to be directed at reducing heat-related illnesses. But the case law of the Strasbourg Court suggests that the effects of extreme temperatures on prisoners confined in small, multi-occupancy cells (even ones that meet the Muršić minimum space requirements) may contribute to a finding of breach of Article 3 even if they do not result in heat-related illnesses.

In Štrukelj, average late afternoon indoor temperatures of 28C (exceeding 30C on seven days) during the second half of July and August contributed to a finding that Article 3 had been breached. Ms Deitch’s evidence suggests that the outdoor temperatures can be very considerably higher than this for long periods in the summer and that indoor temperatures are higher still.

In Mathew, iced water was provided to detainees in a cell whose situation exposed it to the heat of the sun. Despite this, the Strasbourg Court found it “unacceptable that anyone should be detained in conditions involving a lack of protection against… extreme temperatures”: [214]. This contributed to the finding that Article 3 had been breached: [217]. Here, the provision of water and ice is one of the measures said to have been introduced to reduce the number of heat-related illnesses. Ms Deitch’s evidence, recorded by the judge at [152] of her judgment, was that prisoners use this water to wet the floor and then lie on it in an attempt to cool down.

The opinion of US District Judge Ellison in Cole v Collier is also of some significance. It provides a detailed and vivid account of the effects of extreme temperatures on the lives of the plaintiff prisoners and an explanation of the ineffectiveness of measures other than air conditioning in addressing these effects. The case related to one particular prison housing older prisoners, but its significance cannot be confined to that prison, even though the remedy the court was empowered to grant was so confined.

In my judgment, the judge erred in confining her inquiry to the adequacy of measures taken to address heat-related illnesses. Given Ms Deitch’s apparently uncontroverted evidence as to the actual temperatures that could be endured, the evidence as to the effects of these temperatures in Cole v Collier and the significance attributed to much lower temperatures (without any suggestion of heat-related illness) in the Strasbourg authorities on Article 3, she was wrong to find that the concerns about extreme temperatures were “comprehensively addressed” by Mr Clark’s evidence.

The question whether treatment reaches the high level of severity necessary to engage Article 3 depends on a holistic assessment of the conditions of detention. As to personal space, unusually, Muršić creates a bright line rule giving rise to a strong presumption of breach. As to other conditions of detention, it will be rare that one element taken on its own will be sufficient to trigger the application of Article 3 in the domestic context and, a fortiori, in an extradition case: see para. 64(c) and (d) above.

However, in this case, the evidence about extreme temperatures was not presented on its own and out of context. The context was that these temperatures would have to be endured by persons most likely sharing two person cells, which – even if they met the Muršić minimum space requirements – would still be very small. The cells include a toilet with no privacy partition and, in older blocks, no window. Taken together, on the information presently before the court, the conditions described give rise to a real risk of treatment that would breach Article 3 standards applying the jurisprudence of the Strasbourg Court.

It may be that the US authorities are able to provide evidence which discounts this risk, for example by showing, contrary to Ms Deitch’s evidence, that indoor temperatures do not reach unacceptable levels for substantial periods, and/or that – despite the lack of air conditioning in most cells – the measures employed to mitigate the effects of those temperatures are such as to avoid treatment contrary to Article 3, as well as to reduce the incidence of heat-related illnesses. But such evidence has not yet been provided.

For these reasons, I have concluded that the judge was wrong to decide that the appellant had failed to establish serious grounds for believing that he would be subject to a real risk of treatment contrary to Article 3 ECHR if convicted and imprisoned in Texas. The fresh evidence does not displace that conclusion.

In the light of the Divisional Court’s decision in Larco (and the other case law cited at para. 37 above), it is common ground that I should follow the approach in cases to which Framework Decision 2002/584/JHA applies. This means that the court must “postpone its decision on the surrender of the individual concerned until it obtains the supplementary information that allows it to discount the existence of such a risk”: see Aranyosi, at [104]. I shall accordingly make no final order on the appeal at this stage.

The request for further information was the subject of agreement between the parties following the circulation of this judgment in draft. Subject to a minor modification from me, it appears in the Annex to this judgment.

ANNEX – REQUEST FOR SUPPLEMENTAL INFORMATION