Roddy Dunlop KC

Introduction

For centuries, the requirement of corroboration has been regarded as one of the cardinal principles of the criminal law of Scotland. As Baron Hume put it, “no one shall in any case be convicted on the testimony of a single witness”. The purpose of the requirement was to guard against the risk of wrongful conviction by ensuring that the case against the accused was supported by evidence from more than one independent source. It was, as Lord Justice Clerk Aitchison said in Morton v HM Advocate 1938 JC 50 at 55, “an invaluable safeguard in the practice of our criminal courts against unjust conviction”.

That safeguard is now, in any meaningful sense, dead: at least in the prosecution of sexual offences. Through a series of three Lord Advocate’s References decided between 2023 and 2026, the High Court has progressively dismantled the substance of the corroboration requirement while maintaining its form. The culmination of this process is the decision in HM Advocate v PK [2026] HCJAC 42, in which the court held that forensic-scientific evidence confirming that the accused penetrated the complainer’s vagina can, by itself, corroborate the complainer’s account that she was raped. The consequence is that, in any case in which DNA evidence is recovered – which is to say, in practically any case involving sexual contact – the requirement of corroboration is met as a matter of law, regardless of whether there is any independent evidence whatsoever bearing on the presence or absence of consent.

This article argues that the law has taken a wrong turn; that the reasoning by which this result was reached is flawed; and that corroboration, as now understood, has been reduced to an empty formality which provides no real protection against wrongful conviction.

Before proceeding, however, a necessary disclosure: what has happened here has stemmed from a submission I made, as counsel for the respondent, in Lord Advocate’s Reference No 1 of 2023 [2023] HCJAC 40, 2024 JC 140. I must take ownership of that, and the reader of this article should understand this background in arriving at her own views as to what follows. I should also say that, whilst this article carries the concurrence of both the Vice Dean, Tony Lenehan KC, and the chairman of Faculty Services, Tony Graham KC – both of whom have orders of magnitude more experience in criminal cases than do I – it represents my own personal views, and should not be taken as the views of Faculty as a whole. This is an area in which reasonable minds may reasonably disagree.

The position before 2023

The law of corroboration in sexual offences was, for a quarter of a century, governed by the full bench decision in Smith v Lees 1997 JC 73. The court there held that the essential elements – the facta probanda – of any crime required to be proved by corroborated evidence. In a charge of rape, these essential elements included: identification of the accused; the fact of penetration; and the absence of consent. Whilst a single piece of circumstantial evidence might corroborate more than one of these elements, it did not follow that evidence which corroborated one element automatically corroborated them all.

Evidence of distress, for example, could corroborate that a sexual encounter had been non-consensual – because distress was inconsistent with a consensual act – but it could not corroborate the fact of penetration, because distress was equally consistent with a non-penetrative assault, or indeed with any number of distressing experiences that did not involve sexual contact at all. This approach reflected what Lord Gill described as “the true principle”: that “distress alone cannot be said to point to the crucial fact that penetration occurred” (121B-D).

The result was confirmed and elaborated on by another full bench in Fox v HM Advocate 1998 JC 94, which, whilst disapproving the more restrictive test which had been proposed in Mackie v HM Advocate 1994 JC 132, maintained the requirement to corroborate the crucial facts. As Lord Justice General Rodger stated, “so long as the circumstantial evidence is independent and confirms or supports the direct evidence on the crucial facts, it provides corroboration and the requirements of legal proof are met” (p.100G). Every member of the bench in Fox proceeded on the basis that the facta probanda required individual corroboration.

This was not a novel or controversial position. It had been stated by Lord Justice Clerk Macdonald in Lockwood v Walker 1910 SC (J) 3, adopted by Walker and Walker as the foundation of their treatment of the law of evidence, and applied by at least sixteen judges in cases spanning decades. It was the law that practitioners, judges, and jurors understood and applied daily.

Pausing there, one might have thought that the position was clear beyond any doubt. LJC Macdonald is renowned as one of the leading lights in the criminal law of Scotland. The author of A Practical Treatise on the Criminal Law of Scotland in 1867, he served as Solicitor General for Scotland (1876–80), then Lord Advocate in 1885–86 and again from 1886–88. Those roles placed him at the centre of the prosecution system, criminal legislation, and public criminal policy. His period as Lord Advocate was associated with the Criminal Procedure (Scotland) Act 1887, an important reform measure which simplified aspects of Scottish criminal procedure. He was appointed Lord Justice Clerk in 1888, and presided over the Second Division for a remarkable 27 years, standing down in 1915. It might be thought unlikely that he would get the law of corroboration wrong.

The same might be said of the judges who decided Smith v Lees, 87 years later. Lords Rodger and Ross require no introduction. Lord Gill would go on to become Lord Justice General. The expertise of Lords Sutherland and McCluskey in matters criminal is unquestionable.

To all of that, one might add the identity of the Solicitor General who argued the case for the Crown in Smith v Lees, and who deliberately (and doubtless advisedly) eschewed the more “ambitious” propositions advanced by his successor in PK: the then Paul Cullen QC, now – of course – the Lord Justice General.

As I submitted in Lord Advocate’s Reference No1 of 2023, even Homer nods: but it is frankly remarkable to find that so many Homers have been, as it were and as we are about to find out, asleep at the wheel.

The first reference: Lord Advocate’s Reference No 1 of 2023

The first step in the dismantling of the well-established framework was the full bench decision in Lord Advocate’s Reference No 1 of 2023 [2023] HCJAC 40, 2024 JC 140. There, the court overruled Smith v Lees and held that there was “no requirement to prove the separate elements in a crime by corroborated evidence”. What required to be proved by corroborated evidence was “the case against the accused”: namely, that the crime was committed and that the accused committed it. Distress could now corroborate the complainer’s account that she had been raped, including the element of penetration.

The court’s reasoning rested primarily on a reinterpretation of the Institutional Writers, a rehabilitation of the full bench decision in Gillespie v Macmillan 1957 JC 31 (which had been heavily criticised in the academic literature and was of uncertain authority), and the proposition that the law of evidence should be practical, clear, and simple.

There is much that might be said about whether the court’s reading of the Institutional Writers was correct. However, the real difficulty lay not in the historical analysis but in the logical consequences of the new rule. During the hearing, on behalf of the respondent I submitted that the Crown’s position would mean that, in any case where DNA evidence confirmed sexual intercourse, corroboration would be satisfied as a matter of law: because DNA evidence was “supportive” of the complainer’s allegation of rape in the sense of not being inconsistent with it. I advanced that submission intending for it to be, and assuming that it would be understood as being, reductio ad absurdum. The way I put it was as follows: “if the Lord Advocate’s submission is upheld then there is a sufficiency, for an allegation of rape, in any sexual interaction which results in a DNA finding – something which has been apparent in pretty much every human sexual interaction since the dawn of time. That runs contrary to every authority known to the law of Scotland.”

The court did not address this point. It answered the questions before it and moved on.

The third reference: HM Advocate v PK

What was advanced as reductio ad absurdum in 2023 is now the law. In HM Advocate v PK [2026] HCJAC 42, the court held, in answer to the Lord Advocate’s first question, that “forensic-scientific evidence of penetration of the complainer’s vagina by the accused’s penis can corroborate the complainer’s account of rape.” In answer to the fourth question, it held that “any piece of direct or circumstantial evidence, which before Lord Advocate’s Reference No 1 of 2023 could only corroborate one element of the crime, can now corroborate the fact that the crime has been committed.”

The court’s reasoning was as follows. Since the law no longer requires separate elements of a crime to be individually corroborated, evidence of penetration should be understood not as corroborating the element of penetration specifically, but as corroborating “the whole case against the accused”. If distress – which on its own says nothing about whether penetration occurred – can corroborate the whole crime of rape, then evidence of penetration – which on its own says nothing about consent – must equally be able to corroborate the whole crime.

The court dismissed the respondent’s argument that this reduced corroboration to an empty formality: “There require to be two sources of evidence. There must be independent support for the complainer’s account”.

Why the reasoning is flawed

The false symmetry between distress and DNA

The court’s central reasoning depends on a symmetry between distress and DNA evidence. It reasons: distress tells you nothing about penetration but can corroborate the whole crime; therefore DNA evidence of penetration, which tells you nothing about consent, can equally corroborate the whole crime.

But the symmetry is false. Distress has at least some connection to the question of whether an encounter was non-consensual. As the court in Smith v Lees recognised, a jury is “entitled to infer that the complainer was distressed because she was forced to submit to the intercourse and did not agree to it” (p.90C). Distress may not be uniquely referable to the commission of a crime, but it does carry some evidential weight bearing on the central question in a rape trial: whether what happened was against the complainer’s will.

DNA evidence of penetration, by contrast, has no evidential connection whatsoever to the absence of consent. It does not make the absence of consent more probable. It does not make the complainer’s account that she was raped more likely to be true. It tells the jury one, and only one, thing: that sexual intercourse occurred. In a case where the defence is consent – as it very often is – that is the one fact that is not in dispute. The DNA evidence is, on the central question in the case, entirely inert. To describe it as “corroborating” the complainer’s account that she was raped is to use the word “corroborate” in a sense so attenuated as to be meaningless.

Corroboration reduced to a formality

The practical consequence of the decision must be confronted directly. In virtually every modern prosecution for rape, DNA or other forensic evidence will be available confirming that sexual intercourse took place. In cases where the defence is one of denial, that evidence may have genuine evidential significance. But in “consent” cases, which constitute a substantial proportion of rape prosecutions, the fact of intercourse is admitted. The DNA evidence adds nothing to the case that the jury does not already know. Yet, under the law as it now stands, it provides a sufficiency of corroboration for the entire crime.

What this means in practice is that, in any case where DNA is recovered, the question of whether the complainer was raped begins and ends with her word. The only “corroboration” is evidence that sexual contact occurred, a fact which is either admitted or is perfectly consistent with innocence. The requirement of corroboration has become a purely formal requirement – dare I say it, a box-ticking exercise – which provides no real check on the reliability of the complainer’s account.

This brings me back to my own role in this. In Lord Advocate’s Reference No.1 of 2023, I warned in submissions that this was “a proposition which the Lord Advocate herself dares not advance: yet it is the logical consequence of the position which she takes”. The Crown, buoyed by success in Reference No.1 of 2023, and doubtless reinforced by the overruling of Morton in References No.s 2 and 3 of 2023, picked up a ball with which I had intended to skittle the argument, and ran with it. What I advanced as reductio ad absurdum has now been embraced, and become the law.

The abandonment of the rationale for corroboration

The whole point of requiring corroboration has always been to provide an independent check against the risk that a single witness may be mistaken, unreliable, or untruthful. As Hume explained, the law preferred to let the guilty escape “rather than run the risk of such an error, a risk which does not hold when there is a concurrence of testimonies”. Lord Sutherland put it well in Smith v Lees (at p.111G): “the only thing that can be said with any degree of certainty about corroboration is that it is such evidence as is necessary to enable the Crown to surmount the hurdle that proof of the commission of a crime and the identity of the perpetrator cannot rest on the evidence of a single witness”.

Evidence of DNA confirming penetration does not provide any independent check on the reliability of the complainer’s account that the encounter was non-consensual. It provides no check at all. It is not evidence that “confirms or supports” the allegation of rape in any meaningful sense. It is evidence that sexual contact occurred: no more, no less. To treat it as corroboration of the entire crime is to maintain the appearance of a safeguard while removing its substance.

The court’s failure to engage with the consequences

Remarkably, while the court in PK proposed that “cases in which evidence supporting penetration or sexual assault is the only source of corroboration of the commission of the crime will be rare,” it did not seriously engage with the question of what happens when it is. The court did not explain how, in a consent case where DNA evidence provides the only corroboration, the requirement of corroboration serves any function whatsoever. The court simply asserted that “this analysis does not reduce corroboration to an empty formality” without demonstrating why. That there are “two sources of evidence” is a statement about quantity, not quality. The entire point of the respondent’s argument was that the second source adds nothing to the first on the question that matters. The court’s response does not answer that argument; it merely restates its conclusion.

Indeed, even DNA will no longer be necessary in many cases. As already observed, in many cases the question of penetration will not be in issue: the accused may well, either at police interview or in the course of the preliminaries to trial, have accepted that intercourse took place, with the only question being whether or not it was consensual. Such an acceptance is consistent (on the logic of PK) with the complainer’s version of events. Even where no DNA was recovered, the acceptance of intercourse, in this brave new world, corroborates the crime of rape, as the court in PK recognised (at [63]). And it does not stop there. The existence on this planet of an accused’s daughter aged 19 years and 3 months would, presumably and on the reasoning found in PK, corroborate an allegation by his wife of rape 20 years ago. O tempora, o mores.

It may only be me, but I confess to some difficulty in distinguishing this situation from the very one conceded by the Crown as not affording corroboration: in PK (see [26]), it “was accepted that, to be corroborative, it was insufficient for evidence to confirm and support any minor aspect of the principal source of evidence, for example that the complainer had been wearing red shoes at the time of the offence”.

I do not wish to be thought to be belittling the importance of the subject matter by reference to footwear. Plainly, where there is a dispute as to whether intercourse (or other sexual interaction) between complainer and accused happened at all, that question is not a “minor aspect of the principal source of evidence”. But where intercourse is admitted, such that there is no contest on the point whatsoever, corroboration thereof adds nothing to the picture before the jury. To hold that evidence on a point which is admitted corroborates that which is in dispute is to render the notion of corroboration entirely meaningless.

Moreover, there are real questions as to whether it would be correct to think that these musings are unique to sexual offences. The cases under consideration address the law of evidence, not the law of rape. There is no obvious reason why the ratio should not apply elsewhere.

Take a case of robbery. The facta probanda in that crime are unquestionably twofold: (1) the taking of property, and (2) the use of violence or threats to accomplish it. Both are essential: without the violence, the crime is simple theft; and without the taking, the crime is assault. In Mongan v HM Advocate 1989 SCCR 25, discussed at length in Smith v Lees, the court quashed a conviction for robbery and substituted theft precisely because evidence of the complainer’s distressed state could not corroborate the violence element where theft was already proved. The Solicitor General (whose identity has already been referred to) in Smith v Lees asked for Mongan to be overruled, but the court declined.

On the logic of PK, that result would now be reversed. A says that B snatched her handbag at knifepoint. B, having been found in possession of the handbag, admits taking it, but says she picked it up after A dropped it. On the reasoning in PK, corroboration that B took the handbag, that being “consistent with” A’s account of robbery, corroborates the whole crime, including the use of the knife. Proof that the handbag was taken tells the jury nothing whatsoever about whether violence was used. Yet it “fits with” the complainer’s account in precisely the sense the court in PK describes at paragraph 61.

Indeed, the consequences might be even more alarming. In the example figured above, at least there is demonstrated, on corroborated evidence, a crime of some sort (theft by finding). But what if B’s story is that she found it on the ground after A dropped it, and took it because she owns an identical handbag and thought that it was hers. On the reasoning of PK, the admission of appropriation – consistent as it is with A’s complaint – corroborates the complaint entire. Indeed, arguably the mere fact of B being found in possession of the handbag (again, consistent with A’s story) would suffice: even if that possession was temporary, such as where, for example, B hands the bag in to a police station. Can that really be the law of corroboration in 2026? It would appear so.

The road not taken: Parliament

In Smith v Lees, Lord Sutherland acknowledged (at p.118B) that his view might “lead to insuperable difficulties for the Crown in proving a number of similar cases” but maintained that “the valuable safeguard of proper corroboration being necessary to establish any common law crime in Scotland should not be lost”.

When the Scottish Parliament considered (at the instigation of Lord Carloway, who years later would become the architect of all of this in Lord Advocate’s Reference No.1 of 2023) abolishing corroboration, it recognised the need for concomitant safeguards. That provision was designed to ensure that, if the safeguard of corroboration were removed, other protections would take its place. The proposals were ultimately abandoned.

What the court has now achieved by judicial decision is, in substance, the abolition of the corroboration requirement in the vast majority of sexual offence cases, but without any of the safeguards that Parliament considered necessary. A person may be convicted of rape on the basis of a complainer’s account, with no evidence of distress and the allegation supported by nothing more than DNA evidence showing, or an accused’s acceptance, that intercourse took place. To those who contend that corroboration is outmoded and no longer necessary, that may make perfect sense: but if we are going to jettison the protections of corroboration, we should be honest that this is what we are doing, and confront the consequences.

That being so, this is a matter which, if it were to be introduced at all, should surely have been effected by way of legislative intervention, not judicial innovation. Parliament can take a holistic view, consult widely, and introduce appropriate safeguards. The court cannot. As Lord Leggatt observed in Philipp v Barclays Bank UK plc, courts “proceed by reasoning from established principles and are under a duty to promote consistency and predictability in the law” and do not have “the institutional capacities of legislators”. And legislators have to answer for the consequences of their decisions, in a way that judges do not. If politicians are not prepared to take the ballot box risk inherent in a change of the law of this magnitude, it seems odd that it might be imported by way of a judicial side wind.

Conclusion

The trajectory from Smith v Lees to HM Advocate v PK represents a fundamental transformation of the law of corroboration in Scotland. What was for centuries an essential safeguard against wrongful conviction has been reduced, in the context of sexual offences, to a formal requirement with no substantive content. The reasoning by which this has been achieved (that any piece of evidence which is consistent with the complainer’s account can corroborate the entire crime) is logically indistinguishable from the proposition that no corroboration is required at all. If evidence that sexual intercourse took place can corroborate a complainer’s account that she was raped, then the word “corroborate” has lost all meaning.

As Lord McCluskey warned in Smith v Lees (at p.108D), “there is, in my opinion, no legitimate stopping place between the requirement for full legal proof of each of the facta probanda and the dispensing altogether with the need for corroboration”. The court has now arrived at precisely that place. It has dispensed with the need for corroboration in substance while retaining it in name. If this is to be the position, it should be the product of a deliberate legislative choice, accompanied by the safeguards that such a choice demands, rather than the consequence of a series of judicial decisions that have progressively emptied a fundamental principle of its content.

Roddy Dunlop KC
Dean of the Faculty of Advocates
25 September 2026