Disclosure Failure and Third-Party Bad Character: Later Rape Conviction Not “Substantially Probative” of Responsibility for a No-Body Murder (s.100 CJA 2003)

1. Introduction

Osment v R [2026] EWCA Crim 1165 is a Court of Appeal (Criminal Division) decision on the safety of a no-body murder conviction where the prosecution accepted a post-trial disclosure failure: it had not revealed that a pub regular present on the night of the victim’s disappearance (Franklin) was later convicted of rape and attempted rape of a lone intoxicated woman.

The appellant had been convicted (majority verdict) of the murder of Claire Holland, who disappeared after leaving Seamus O’Donnell’s public house in Bristol in June 2012. Her body was never found. The prosecution case depended heavily on repeated admissions/confessions allegedly made by the appellant over many years, including to an undercover police officer (“the UCO”).

The appeal focused on whether non-disclosure of Franklin’s later rape conviction (i) rendered the conviction unsafe and (ii) would have supported admissible third-party bad character evidence under section 100 of the Criminal Justice Act 2003. The appellant also renewed challenges (refused by the single judge) to the admissibility of undercover confession evidence and to the exclusion of proposed expert evidence.

2. Summary of the Judgment

  • Grounds 1–4 (renewed): Leave refused. The Court agreed the trial judge was entitled to admit the UCO evidence and to exclude the proposed additional expert evidence (including a “peer review” report by Professor Gudjonsson and trial evidence from Professors McManus/Oxburgh).
  • Ground 7 (disclosure failure re Franklin): The Court accepted the material ought to have been disclosed, but held it would have been inadmissible under s.100 CJA 2003 because it lacked “substantial probative value” on the central issue—whether the appellant murdered Ms Holland. The Court therefore declined to receive the fresh evidence under s.23(2)(b)–(c) Criminal Appeal Act 1968, and dismissed the appeal.

3. Analysis

3.1 Precedents Cited and Their Influence

  • R v H [2014] EWCA Crim 1555 and the Criminal Practice Directions, para 7.12: supported the structured approach to admissibility of expert evidence, including the need for expertise to assist the tribunal on matters beyond ordinary experience, and to avoid experts opining on the ultimate jury question.
  • Pora v New Zealand [2015] UKPC 9: relied on as authority that even eminent expertise on confessions/suggestibility (here, Professor Gudjonsson) may be excluded where the opinion lacks a sufficient evidential foundation (e.g., insufficient underlying material) or adds nothing of admissible assistance.
  • R v Stagg WL 17112471: referenced in the defence’s critique of risks in confession-gathering tactics (confirmation bias / extraction-focused methods), but the Court ultimately held the proposed academic evidence did not meet the threshold for admission at trial.
  • R v Lawson and Jefferson citing Christou: used in the undercover-operation analysis to mark the boundary between permissible covert “opportunity” to speak and impermissible “interviewing”/improper inducement likely to undermine admissibility under PACE.

3.2 Legal Reasoning

(A) Expert evidence on confessions: “peer review” and jury-domain concerns

The Court endorsed the trial judge’s exclusion of Professor Gudjonsson’s proposed evidence on both the voir dire and at trial. Although he was pre-eminent in the field, he had not examined the appellant, had not reviewed the underlying undercover recordings, and his report was essentially a commentary on other experts’ work and on the significance of a “shift” finding. The Court treated this as either duplicative of admissible clinical evidence already before the court (from clinicians who tested/examined the appellant), or as trespassing into the jury’s task of deciding reliability.

Similarly, the Court upheld the exclusion at trial of the academic evidence of Professors McManus/Oxburgh: the jury had the recordings/transcripts and admissible clinical opinion on vulnerabilities and suggestibility; whether admissions were reliable in this case remained a quintessential jury determination.

(B) Undercover confession evidence: PACE s.76 and s.78

The Court agreed there was no basis to overturn the trial judge’s conclusion that the UCO’s conduct was not oppressive (s.76) and did not render confessions unreliable, nor was there such unfairness as to require exclusion (s.78). Central to that conclusion were findings that:

  • the operation was lawful, authorised, and supervised;
  • all interactions were recorded, enabling jury evaluation of context and tone;
  • the appellant’s vulnerabilities (alcohol dependence; depression/anxiety) did not amount to the kind of fragility that made the operation improper;
  • the appellant had a demonstrated willingness to confess/boast to others outside any police pressure, weakening the claim that the covert context itself generated unreliable admissions.

(C) Disclosure failure and s.100 CJA 2003: why Franklin’s conviction did not qualify

The doctrinal heart of the judgment lies in its treatment of the conceded disclosure failure. The Court framed the safety question through two connected gateways: (i) whether the new material would have been admissible (s.100 CJA 2003), and (ii) whether it could properly found the reception of fresh evidence on appeal (s.23(2)(b)–(c) Criminal Appeal Act 1968).

The Court stressed that “substantial probative value” is context-dependent: the probative force of third-party misconduct cannot be assessed in isolation; it must be evaluated against the evidential picture on the disputed issue (here, identity of the killer). Applying s.100(3), the Court treated the similarities as insufficiently distinctive and the dissimilarities as significant:

  • Franklin’s later misconduct was rape/attempted rape, not murder; and it did not show a propensity to kill or dispose of bodies.
  • The victim in the later case was a stranger; Ms Holland was known in the pub community and had spoken of meeting the appellant.
  • There was no evidence Ms Holland was raped, and no independent evidence that Franklin interacted with her in a way pointing to homicide responsibility beyond being present and buying a drink.
  • The “similarity” reduced to an unfortunately common risk factor: a lone intoxicated woman at night, which the jury would already appreciate as a general possibility.

On that basis the Court held the Franklin material would not have met s.100(1)(b): it did not have substantial probative value on the matter in issue (whether the appellant killed Ms Holland), and thus could not properly be deployed to suggest Franklin as an alternative perpetrator in a way that crossed the statutory threshold.

Importantly, the Court drew a principled distinction between (a) material that merely makes an alternative-perpetrator hypothesis feel more imaginable and (b) material that is probatively strong enough to be admitted as third-party bad character. The Franklin conviction fell into the first category.

3.3 Impact

  • Disclosure litigation: The decision underscores that even a conceded disclosure failure does not, without more, establish unsafety; the appellate court will ask whether the withheld material would have been admissible and materially probative of a live issue.
  • s.100 CJA 2003 threshold (non-defendant bad character): The judgment reinforces a relatively strict approach where the defence seeks to use third-party convictions to imply responsibility for the charged offence—especially where the third-party misconduct is of a different character (rape vs murder) and the similarities are generic rather than distinctive.
  • Undercover confession operations: The ruling is a reminder that recorded covert interactions, absent oppressive conduct or improper “interviewing,” will often be left to the jury as to reliability, with clinical vulnerabilities addressed through appropriately qualified expert evidence rather than broad academic critique.

4. Complex Concepts Simplified

“No-body” murder
A murder prosecution where the victim’s body is not recovered. The prosecution must still prove death, that death was not accidental/suicidal/natural, and that the defendant caused it—often by inference and admissions.
PACE s.76 and s.78
s.76 can exclude confessions obtained by oppression or likely unreliability due to what was said/done. s.78 gives a broader discretion to exclude prosecution evidence if admitting it would adversely affect trial fairness.
Suggestibility: “yield” vs “shift”
“Yield” refers to giving in to leading questions; “shift” refers to changing answers under pressure/negative feedback. The Court accepted “shift” vulnerability existed but considered it of limited utility where the interactions were not a classic interrogative scenario.
s.100 Criminal Justice Act 2003 (non-defendant bad character)
Allows evidence of another person’s misconduct only in limited circumstances—most relevantly where it has substantial probative value on an important issue (e.g., identity). Generic similarity and mere opportunity are usually insufficient.
s.23 Criminal Appeal Act 1968 (fresh evidence)
The Court considers whether new material may afford a ground of appeal and whether it would have been admissible at trial; inadmissible material will ordinarily not be received.

5. Conclusion

Osment v R [2026] EWCA Crim 1165 affirms that a prosecution disclosure failure—though serious and conceded—will not render a conviction unsafe where the undisclosed material would not have been admissible at trial. The Court’s key contribution is its contextual, threshold-focused application of s.100 CJA 2003: a third party’s later conviction for a different kind of predatory offence, coupled with mere presence and temporal proximity, did not have “substantial probative value” as to responsibility for an earlier no-body murder. The judgment also reiterates the limits of expert evidence on confession reliability, particularly where it duplicates clinical testimony or intrudes on the jury’s role.