Scots criminal law has long prided itself on a distinctive safeguard: no one should be convicted on the testimony of a single witness alone. Rooted in the writings of Baron Hume, corroboration was intended to protect against the fallibility of human testimony and the potentially devastating consequences of wrongful conviction.
Recent decisions of the High Court of Justiciary have fundamentally changed how that safeguard operates. Following Lord Advocate’s References Nos 1, 2 and 3 of 2023, and most recently HM Advocate v PK[2026] HCJAC 42, the question is whether corroboration remains a meaningful safeguard against wrongful conviction or has become a matter of form rather than substance—concerned only with the number of sources rather than with what they actually prove.
What did the courts decide?
In Lord Advocate’s Reference No 1 of 2023 [2023] HCJAC 40, the court rejected the requirement that each essential element of an offence must be corroborated separately. It overruled Smith v Lees 1997 JC 73 and Cinci v HM Advocate 2004 JC 103, holding that the evidence should be considered as a whole.
The court also confirmed that evidence of a complainer’s distress, together with statements made in the immediate aftermath of an alleged offence, could corroborate the complainer’s account of the crime, rather than being restricted to corroborating a particular element such as absence of consent.
Lord Advocate’s References Nos 2 and 3 of 2023 [2024] HCJAC 43 extended that approach. The court held that a de recentistatement could provide corroboration even where there was no accompanying evidence of distress. Such a statement could, in appropriate circumstances, corroborate both the commission of the crime and the identification of the accused.
The most recent decision, HM Advocate v PK [2026] HCJAC 42, went further in clarifying the consequences of the new approach. The court held that forensic evidence establishing penetration could corroborate an allegation of rape even where the accused accepted intercourse but maintained that it was consensual.
The reasoning was that corroborative evidence need not independently establish the criminality of the conduct. It can be equally consistent with innocence when considered in isolation, provided that it supports the complainer’s account when considered alongside the evidence as a whole.
The formal rule remains: two sources of evidence are required to establish the crime and the accused’s responsibility. But what those sources must actually establish has changed considerably.
Two sources of evidence, but how much independent support?
This is where I believe the real difficulty lies.
If we consider a de recenti statement in a case involving an allegation of rape. Shortly after the alleged incident the complainer tells a friend that she has been raped. The friend gives evidence of that conversation.
There are two witnesses, but only one original source of the allegation that rape occurred. The friend did not witness the intercourse and cannot independently establish whether it was consensual. The statement may strengthen the complainer’s credibility, but does it provide independent confirmation that the crime was committed?
The same difficulty arises with DNA evidence.
Where an accused accepts sexual intercourse but disputes consent, DNA may reliably establish penetration and identity. It does not establish whether the intercourse was consensual. Yet, following PK, that evidence may provide the necessary corroboration for the allegation of rape.
The issue is not whether the evidence is reliable. It may be scientifically impeccable. The issue is what it actually proves.
In both situations, the corroboration requirement can be satisfied even though the second source does not independently support the conclusion that a crime was committed.
Corroboration begins to look less like a requirement for independent support for guilt and more like a requirement for two sources of evidence which, when combined, fit the prosecution case.
That is a significant change.
What happened to Hume’s principle?
Hume’s underlying concern was straightforward: even an apparently credible and reliable witness can be wrong. The consequences of criminal conviction are sufficiently serious that the law should not permit a single person’s testimony, however convincing, to determine guilt without independent evidential support.
Hume’s rationale was that it was better for a guilty person to escape punishment than for an innocent person to be convicted.
The recent decisions do not formally abandon that principle. The presumption of innocence remains, and guilt must still be established beyond reasonable doubt.
Nevertheless, they have removed an evidential restriction which previously prevented conviction in certain circumstances where an essential element lacked independent corroboration.
That creates a legitimate concern about increasing the risk of wrongful convictions.
The standard of proof and corroboration serve different purposes. The former determines how sure a jury must be before convicting. The latter determines whether the prosecution has the necessary evidential foundation for conviction in the first place.
A jury may be entirely convinced by one witness, particularly where other evidence appears consistent with that witness’s account. Corroboration was intended to provide an additional safeguard against the possibility that the central account, however persuasive, might be wrong.
If evidence which does not independently support the disputed criminality can satisfy corroboration, what meaningful additional protection does the requirement provide?
Conclusion
I am not suggesting that the old law was perfect. In my view it could be overly restrictive and create significant difficulties for the prosecution, particularly in relation to sexual offences. I am not necessarily opposed to an approach under which the case as a whole must be corroborated, rather than each essential element separately.
For me, however, that should require evidence from two independent sources that both point towards guilt—not simply evidence that is consistent with the allegation.
My principal concern is that the recent judgments have significantly increased the risk of wrongful conviction by removing the substance of a vital safeguard. That concern cannot be considered in isolation from the majority required to convict.
In Scotland, a conviction can now be returned by ten of fifteen jurors, leaving five unconvinced of guilt. By contrast, In England and Wales where there is no requirement for corroboration, juries are initially required to reach unanimity, with a majority of at least ten of twelve permitted only if the judge allows it.
In my view, this requires Parliament’s attention. If corroboration no longer provides the protection it once did, then Parliament should consider whether the remaining safeguards—including the majority required for conviction—are sufficient. A protection of this importance should not be substantially diminished without a wider assessment of the risk of convicting the innocent.
If you have any queries, please get in touch with the Brodies criminal defence team.
Contributor
Senior Associate