Latimer Is Not a Freestanding Basis to Quash Historic Confession Convictions: Safety Turns on Case-Specific Link to Reliability

Case: McDevitt, R. v
Neutral citation: [2026] NICA 37 (Court of Appeal in Northern Ireland)
Date: 12 September 2026
Court: Keegan LCJ, Colton LJ and Fowler J (Keegan LCJ delivering)

1. Introduction

This was a reference by the Criminal Cases Review Commission (the “CCRC”) under section 10 of the Criminal Appeal Act 1995 concerning a historic Belfast Crown Court conviction (December 1986, judge-alone trial) for murder and related firearms offences arising from a coordinated petrol bomb and firearms attack on an army patrol in Londonderry in April 1984.

The appellant accepted (then and now) his involvement in petrol bomb offences, but challenged the murder and firearms-related convictions. Those convictions depended entirely on admissions said to have been made during police interviews at Castlereagh—particularly a later statement in which he was recorded as admitting he transported men with a case and realised firearms were involved.

The CCRC’s single ground was that there was a “real possibility” the convictions would be held unsafe in light of subsequent findings in R v Latimer, Hegan, Bell and Allen [1992] 1 NIJB 89, where certain interviewing officers (also involved here) were found to have acted improperly in relation to interview note practices and to have given untruthful evidence.

The appeal therefore raised a recurring question in “Troubles-era” cases: when (and how) later-discovered misconduct findings about interviewing officers can undermine the safety of convictions founded on confession evidence.

2. Summary of the Judgment

3. Analysis

3.1 The legal framework applied: “unsafe” and fresh evidence

The court treated the controlling question as whether the verdict was “unsafe”. It adopted and applied the Northern Ireland approach in R v Pollock [2004] NICA 34, emphasising that an appellate court:

“should concentrate on the single and simple question ‘does it think that the verdict is unsafe.’” (Pollock, quoted at [57])

Where fresh evidence is involved, the court endorsed the well-known “might reasonably have affected” cross-check from R v Pendleton [2002] 1 WLR 72 (Lord Bingham), again quoted in terms:

“...it will usually be wise... to test their own provisional view by asking whether the evidence, if given at the trial, might reasonably have affected the decision... to convict. If it might, the conviction must be thought to be unsafe.” (Pendleton, quoted at [70])

For admissibility of fresh material, the court set out section 25 of the Criminal Appeal (Northern Ireland) Act 1980 and its factors (capable of belief; may afford a ground; admissible; reasonable explanation for omission). It also reaffirmed the “interests of justice” discretion, drawing from R v Pendleton (on the duty to run available cases at trial) and R v Erskine [2009] 2 Cr App R 29 (warning against subverting trial process absent a persuasive explanation).

The court also relied on its own articulation in The King v James Alexander Smith [2023] NICA 86, including the synthesis that, in practice, the appeal turns on whether the new material could have reasonably affected the conviction decision (again anchored in Pendleton and referenced to O'Doherty [2002] NILR 263).

3.2 Confession admissibility at the time: strict statutory exclusion and common law voluntariness

Because the trial was for a scheduled offence, the admissibility regime included section 6 of the Northern Ireland (Emergency Provisions) Act 1973. The court highlighted two layers:

• Statutory exclusion under section 6(2) where prima facie evidence of torture or inhuman or degrading treatment was adduced (placing a burden on the prosecution to satisfy the court the statement was not so obtained).
• Common law voluntariness discretion to exclude confessions not freely and voluntarily given.

This mattered because the appellant’s original case was that the incriminating admissions were induced by threats/inducements and improper interviewing practices. If accepted, the confession would be excluded; therefore, credibility findings about the interview circumstances were central.

3.3 The judgment’s core “precedent” point: Latimer-type findings are relevant but not determinative

The court accepted the general proposition that later-discovered misconduct by interviewing officers can be relevant to reliability and safety. It noted the approach endorsed in R v Michael Devine (endorsing R v O'Toole [2006] EWCA Crim 951): misconduct in other cases may bear on credibility in a confession case.

However, the court drew a critical boundary—building on, and aligning with, its more recent historic-conviction decisions (R v Thompson [2024] NICA 30 and R v Kirkpatrick, Cullen and Cullen [2025] NICA 5):

• Latimer can raise a serious question about officer credibility and interview-note integrity.
• But the “Latimer issue” is not a freestanding ground of appeal; outcomes depend on a “close and careful analysis of the facts of a particular case”.

The court explicitly cited and applied the caution expressed in R v Kirkpatrick, Cullen and Cullen [2025] NICA 5, including:

“...it does not follow that because some similarities arise with cases such as Latimer and Thompson that a conviction is automatically unsafe...” (Kirkpatrick, Cullen and Cullen, quoted at [68])

“...the fact that issues arise from Latimer... is not a freestanding ground of appeal.” (Kirkpatrick, Cullen and Cullen, quoted at [69])

3.4 How Latimer was distinguished on the facts

The court undertook a comparative exercise. In R v Latimer, Hegan, Bell and Allen, ESDA testing revealed numerous instances of rewritten interview notes and, crucially, untruthful police evidence about whether notes were contemporaneous. That created:

• an evidential foundation to suspect “reconstruction” or worse; and
• a direct attack on police credibility in cases turning on police-versus-accused conflicts.

In the present case, the court accepted that Latimer “damages the credibility” of the two officers criticised there. But it found the “automatic-unsafety” inference unjustified because:

• No fabrication allegation: the appellant accepted the statements were his words and he signed them; there was “therefore no claim” of concoction or doctoring (distinguishing a central driver in Latimer).
• Officer mix: only some interviewing officers were criticised in Latimer; two others involved here were not. The main alleged threats/inducements were attributed to an officer not subsequently discredited.
• Credibility landscape at trial: the appellant’s evidence contained extensive admitted lying on multiple topics. This framed the trial judge’s rational preference for police testimony as to interview conditions.
• Procedural context: the “critical” statement was not obtained in a continuous, escalating interview; questioning paused overnight and the appellant received legal advice before the final statement.
• No contemporaneous complaint: the trial judge relied on medical evidence (including rejection of the claim that a complaint had been made to the responsible medical officer), further undermining the oppression narrative.

The practical result was that Latimer did not create (on these facts) the “significant sense of unease” necessary to find unsafety under Pollock.

3.5 Thompson and the “constellation of factors” approach—why it did not translate

The court treated R v Thompson [2024] NICA 30 as fact-specific. It summarised the basis on which the conviction was quashed there: contemporaneous allegations, later-validated by misconduct/complaint material and Latimer, and fresh evidence capable of operating as “bad character” material undermining police credibility—together creating a real possibility that the confession would not have been admitted.

Here, by contrast, the court found:

• the Latimer linkage was weaker (no fabrication issue; admissions accepted as made);
• complaint-file material added no substantive value in this case; and
• there was no sustainable disclosure case capable of shifting safety.

This is a significant practical clarification: the court treated complaint/disciplinary material as potentially supportive (as in Thompson) but not generically transformative, and in any event dependent on the structure of the trial issues and the evidential foundation for reliability doubts.

3.6 Fresh evidence applications: disciplined use of section 25

3.6.1 Firearms present at Castlereagh

The appellant relied on material (including extracts from prior CCRC reasons) showing firearms seized in another case were taken from a forensic laboratory to Castlereagh and shown to suspects during interviews, contradicting the trial judge’s scepticism that a firearm would be brought into an interview room.

The court nonetheless rejected the point as incapable of affecting safety, substantially aligning with the earlier CCRC analysis: the appellant’s own trial evidence located the alleged gun-showing on a date when continuity evidence indicated the relevant firearm was not at Castlereagh; and, critically, the contested “gunmen transport” confession came later, after legal advice—weakening any claimed causal link.

3.6.2 Speculation about audio recording and missing original notes

The court characterised the “audio recording” point as speculative (no recordings identified). The missing original interview notes were treated as, at most, a lost forensic opportunity rather than affirmative evidence of unreliability—especially given the appellant’s acceptance that the statements recorded his words.

3.6.3 Collateral materials from other proceedings and complaint files

Materials from other cases (including R v Nash and references to McDaid & Nash) were regarded as general/collateral and not probative of safety on these facts. Complaint-file material—while potentially relevant in some cases—did not add anything substantive given the court’s assessment of the weakness of the Latimer point here.

3.6.4 Expert evidence on confessions and linguistics

The appellant sought to introduce expert evidence (Professors Shepherd and Coulthard) addressing coercion, false confessions, and the interview note process, met by a response from Professor Flowe.

The court refused to admit the expert evidence. Its key reasons were:

• the experts’ “fabrication/reverse engineering” thesis lacked a reliable evidential foundation on this record (given accepted, unchallenged admissions on petrol bombing, acceptance of signing/accuracy, and absence of contemporaneous complaint);
• methodological concerns were noted (including failure to record a pre-interview);
• the linguistics report was queried as potentially irrelevant because this was “not a fabrication case”; and
• most importantly, even if admitted, the evidence would not reasonably have affected the trial judge’s decision to convict.

3.7 The judgment’s practical impact

The decision consolidates a clear operational message for future CCRC references and historic appeals in Northern Ireland:

• “Latimer linkage” must be case-specific: later findings that particular officers lied about interview note practices are relevant, but unsafety depends on whether those findings materially intersect with the trial’s actual reliability disputes (e.g., fabrication allegations, ESDA-type vulnerabilities, credibility turning points, contemporaneous complaint evidence).
• Confession cases split into types: the court implicitly distinguishes (a) cases where the authenticity/creation of interview records is in issue (Latimer-type), from (b) cases where the accused accepts what was said but disputes voluntariness/pressure—where general misconduct findings may carry less weight absent supporting features.
• Fresh expert evidence must “touch safety” on the facts: generic critique of historic detention conditions will not, without an evidential anchor, satisfy section 25 or the Pendleton “might reasonably have affected” threshold.
• Complaint/disclosure arguments remain contextual: complaint files may matter (as in Thompson), but will not carry an appeal where the core credibility/reliability issues do not align.

4. Precedents Cited (and their role)

• R v Latimer, Hegan, Bell and Allen [1992] 1 NIJB 89 — the foundational misconduct case: ESDA-revealed rewriting and untruthful police evidence about contemporaneity; used here as the benchmark for when officer credibility issues can imperil confession convictions.
• R v Pollock [2004] NICA 34 — Northern Ireland statement of the “unsafe” test and the “significant sense of unease” formulation guiding appellate restraint.
• R v Pendleton [2002] 1 WLR 72 — the “might reasonably have affected” cross-check for fresh evidence; also emphasises not saving defences/evidence for appeal.
• R v Michael Devine and R v O'Toole [2006] EWCA Crim 951 — recognition that police misconduct evidence from other cases can, in appropriate circumstances, bear on credibility and confession reliability.
• R v Thompson [2024] NICA 30 — illustrates a successful “constellation of factors” model: complaint files + validated allegations + officer credibility collapse affecting admissibility; distinguished as fact-specific.
• R v Kirkpatrick, Cullen and Cullen [2025] NICA 5 — emphasises fact sensitivity and cautions against treating Latimer as automatically outcome-determinative; quoted to reject “freestanding Latimer ground”.
• R v Gallagher [2024] NICA 63 — cited for caution in historic cases at a remove of time.
• R v Brown & others [2012] NICA 14 (citing R v O'Halloran) — referenced in the context of defining/understanding ill-treatment concepts in historic confession litigation.
• The King v James Alexander Smith [2023] NICA 86 — guidance on section 25 fresh evidence assessment; includes a practical synthesis (with reference to O'Doherty [2002] NILR 263).
• R v Erskine [2009] 2 Cr App R 29 — stresses that absent persuasive explanation, attempting to run on appeal what could/should have been run at trial will rarely satisfy “interests of justice”.
• R v Nash — referenced as collateral material the appellant sought to deploy; found to have no relevance on these facts.

5. Complex Concepts Simplified

• “Unsafe” conviction: The Court of Appeal does not decide whether the appellant is guilty; it asks whether the conviction can safely stand given the trial evidence and any properly admitted new material.
• CCRC “reference”: The CCRC does not quash convictions; it refers cases where it considers there is a “real possibility” the appeal court will find the conviction unsafe.
• ESDA testing: A forensic technique that can reveal indented writing on paper, sometimes demonstrating that interview notes were rewritten rather than made contemporaneously (central in Latimer).
• Section 25 fresh evidence: The appeal court can receive evidence not adduced at trial if necessary/expedient in the interests of justice, paying close attention to credibility, relevance to a ground of appeal, admissibility, and why it was not used at trial.
• “Not a freestanding ground”: A later finding that an officer behaved improperly in another case does not automatically overturn every conviction involving that officer; the appellant must show how it matters to the particular trial issues and safety analysis.
• De bene esse: Material can be received provisionally “for what it is worth” without conceding it meets the formal standard for admission as fresh evidence.

6. Conclusion

[2026] NICA 37 reinforces a disciplined approach to historic confession appeals in Northern Ireland. R v Latimer, Hegan, Bell and Allen remains a powerful credibility touchstone, but it is not a universal solvent: the court will not quash a conviction merely because a discredited officer featured. Safety turns on whether the later misconduct findings meaningfully connect to the trial’s actual reliability contest—particularly where (as here) the appellant accepted he made and signed the statements, did not allege fabrication, had legal advice before the critical statement, and the trial judge’s credibility findings were supported by multiple, case-specific features.