Beyond Doubt: The Case Against ‘Not Proven’

Scotland,unusually,has three verdicts in criminal trials:guilty,not guilty,and not proven.The not proven verdict, regarded by many as an intermediate option between the other two, has been the subject of a long-running debate as to whether it should be abolished. In this article we argue that it should. Drawing on empirical evidence from two recent studies, we cast doubt on the arguments most often made in its favour – that it serves a valuable communicative function, protects against wrongful conviction, and/or increases juror satisfaction. There is no consensus on its meaning or appropriate application in any given case, and it risks both stigmatising an acquitted accused and diminishing complainers’ opportunities for closure. It is doubtful that it prevents wrongful conviction,but even if it does,there are more effective measures in this regard.


INTRODUCTION
Scotland, unusually, has three verdicts in criminal trials. In addition to the options of 'guilty' and 'not guilty', familiar throughout the common law world, jurors have a third option: 'not proven'. The not proven verdict has exactly the same effect as one of not guilty, and there is in law no distinction between the two verdicts of acquittal: jurors are simply told that they have two alternative options with the same consequences. The absence of such a distinction reflects the fact that the third verdict's existence is a matter of historical accident rather than conscious design.
The lack of any formal distinction does not mean, however, that jurors are bound to treat the two verdicts as identical or that their choice between the two will be regarded by others as without meaning. Not proven has long been regarded by many as an intermediate option between the other two verdicts, allowing the jury to signal their belief in the guilt of the accused alongside a conclusion that the standard of proof has not been met. Debate on the appropriateness of this third option has run since at least 1846, and the verdict has had both fierce proponents and opponents.
The two main arguments that have been consistently made in favour of the not proven verdict are, first, that it serves an important communicative function and, second, that it acts as a safeguard against wrongful conviction. The first argument has most commonly been made in sexual offence cases, the argument being that the verdict signals to the complainer that, while the evidence did not meet the criminal standard of proof required to convict, the jury did not disbelieve her. The second argument is a more general one: that the availability of the verdict means that, in a case where the evidence does not quite meet the standard of proof, jurors will use the not proven verdict where otherwise they might have been tempted (wrongly) to convict. Both of these arguments, in different ways, also link to a further claim sometimes made in favour of the not proven verdict, which is that the availability of this option is valued by jurors. One of the difficulties with these arguments is that they are premised on empirical claims about how jurors understand and apply the verdict, and how others interpret it, for which data has not previously been available.
Our contribution in this paper is two-fold. First, we supply some of the missing empirical data to enable these arguments to be properly evaluated. This comes from two sources -a large-scale and pioneering mock jury deliberation study, which explored in detail how juries operated both with and without this third verdict, and a programme of interviews with complainers whose cases had concluded with a not proven verdict. This data allows us to refute the main arguments for the not proven verdict in a way that has not been possible until now. Secondly, building on that data, we also make a substantive case against the not proven verdict, on the basis that it is unjust to have a stigmatic acquittal verdict and that the verdict risks a loss of public confidence because it allows (or at least is seen to allow) jurors to use it as a compromise rather than properly discharging their function. We argue ultimately that the not proven verdict should be abolished in Scotland.
The article proceeds as follows. First, we set the scene by outlining a brief history of the not proven verdict, before discussing our data sources. We go on to examine the main arguments that have been made in favour of the not proven verdict,in the light of our empirical data.We argue that they are all either unsupported by this data or can be refuted by powerful normative arguments. We then draw together the empirical and normative arguments against the not proven verdict, and make the case for the verdict's abolition. Finally, we briefly examine, but reject, the possibility of retaining the not proven verdict (alongside a verdict of proven) in a binary verdict system.

SETTING THE SCENE
Scotland's three-verdict system is, as noted earlier, a matter of historical accident rather than conscious design. 1 A 17 th -century procedural change meant that juries returned 'special verdicts' stating whether individual facts were proven or not proven rather than declaring on the guilt or innocence of the accused, which was a matter for the trial judge based on the terms of the special verdict. A 1728 case, the trial of Carnegie of Finhaven, re-established the right of the jury to return a verdict of not guilty, but the language of 'not proven' remained and became an alternative form of general verdict in Scots law. While the not proven verdict is unique to Scots law, it is not the only differentiated verdict system in use. 2 Mindful of the dangers of uncritical legal transplants, the case we put forward here is made primarily in the Scottish context, but the arguments we make do have wider significance for the broader debate about differentiated verdict systems. 3 In modern Scottish practice, juries are simply directed that there are two verdicts of acquittal open to them, that these have the same effect, and that it makes no difference which verdict of acquittal they choose. 4 The appeal court has consistently dissuaded trial judges from attempting to offer any explanation of the difference between not guilty and not proven when charging juries. 5 While historically it appears that juries had a strong preference for the verdict of not proven over not guilty,using it in as many as three-quarters of acquittals in the late nineteenth century,that practice has shifted over time and it amounts for around 30 per cent of jury acquittals today, with particularly high use apparent in contemporary sexual offence trials. While the not proven verdict has long been controversial, 7 with its possible abolition having been raised in parliament on a number of occasions and considered by official committees, its position has remained relatively secure at least until now. Increased impetus for change has come in recent years through a high-profile murder case resulting in a not proven verdict in the 1990s, 8 and then increased concern about the use of the verdict in sexual offence cases (against a background of low conviction rates) in particular. 9 Following a two year programme of mock jury research in Scotland 10 -the largest single mock jury deliberation study in the United Kingdom to date -the Scottish Government has committed to engaging in 'serious discussions' about 'whether we should move from a three verdict system to a two verdict system', 11 while abolition or review of the not proven verdict featured in several manifestos in the 2021 Holyrood election. 12 And yet, debate amongst commentators in Scotland about the merits and demerits of the not proven verdict is not new. In 1846, Lord Cockburn published a critique of the verdict, arguing that it is incompatible with the presumption of innocence and that it casts unjust stigma on an accused against whom guilt has not been proven, 13 arguments that have been repeated by abolitionists until the present day. 14 So too, the positive case for the verdict -especially in terms of its role in protecting against wrongful conviction -has a long history. 15 The debate has tended to repeat itself in a loop ever since, rather than evolving over time. 16 A key barrier to progress has been that many of the propositions advanced by retentionists and abolitionists alike are speculative. For example, without knowing whether the availability of the not proven verdict results in an increased acquittal rate, and thus might provide the protection against wrongful conviction that it is argued to do, it is impossible to evaluate that proposition. In order to advance the debate, robust data is needed about how the not proven verdict operates in practice -data that has, until now, been lacking. There had previously been a small number of studies exploring the manner in which (mock) jurors 7 The first sustained criticism appears to be in an influential article by the judge use the not proven verdict, 17 but the small scale and methodological limitations of these studies means they have not provided a sufficiently reliable grounding to justify legislative change. 18 A key contribution of this article is thus to supply new, and more robust, data that can inform the debate in a way that has not hitherto been possible. There are two main sources of original data that we draw upon, each of which is described below.
It is, of course, worth noting that while one of the two data sources that we draw on is a study of the use of not proven by mock juries, the verdict is not exclusively used by juries in Scotland. Juries decide on the guilt of the accused in all solemn cases, but in summary cases this is determined either by sheriffs 19 (if they are heard in the sheriff court) or lay justices (if they are heard in the justice of the peace court). The verdict is, however, used by jurors with greater frequency than by judges sitting alone. Figures released by the Scottish Government demonstrate that over 2015-2020, the not proven verdict accounted for 17.5 per cent of acquittals overall, but 30.5 per cent of all acquittals by juries and 14.7 per cent of all acquittals in summary cases. 20 The empirical evidence that we present here from the jury study is obviously specific to its use in the solemn context, but as we explore below many of the arguments that it gives rise to in favour of abolition are also equally pertinent, regardless of whether the verdict is to be determined by lay people or judges.

The Scottish Jury Research
The first data source we draw on is a major programme of research into mock jury decision-making, undertaken in Scotland between 2017 to 2019. 21 The focus of the study was to investigate (a) the difference the three unique features of the Scottish criminal jury (which, aside from three verdicts, also has 15 members and makes decisions by a simple majority) make to juror verdict preferences and (b) the way in which jurors understand the not proven verdict. 22 In this study, 23 64 mock juries (comprising 863 participants) watched a filmed rape or assault trial of around an hour long and then deliberated for up to 90 minutes in an attempt to reach a verdict. We took particular care to emphasise the solemnity of proceedings and jurors took the standard juror affirmation before viewing the film. The filmed trials involved professional actors in the roles of witnesses and advocates, with a senior judge giving legal directions, taken from the Judicial Institute's Jury Manual, 24 replicating the directions juries would hear in a real trial. 25 Experienced legal practitioners advised on the design and realism of the trial scripts and closing speeches, 26 and were on hand during filming to assist with directing actors in appropriate delivery and courtroom performance. The jurors were recruited from the general public to be representative of the local population in terms of age, gender, education level and work status.
Potential participants were approached by professional recruiters from Ipsos MORI Scotland through door-to-door or street recruiting, rather than relying on any pre-existing lists of 'frequent' market research volunteers, and were not solicited to participate in response to any advertisement for participants on social media. Juries were run on Saturdays over a two-month period in an effort to widen the pool of potential participants. When approached by recruiters, potential participants were simply told that the study was 'to learn about the Scottish jury system' and that it would involve a four-hour session during which they would watch a trial video and then be asked to deliberate as a group towards a verdict. 27 After watching the trial, half of the juries deliberated with a choice of three verdicts (guilty, not guilty and not proven) and half with two (guilty and not guilty), 28 permitting an assessment of the difference that the availability of three verdicts made. The deliberations, which were undertaken without a researcher in the room, were audio and video-recorded so that the team could thematically code transcripts of discussions and analyse wider discursive and group dynamics. Alongside this, jurors completed questionnaires pre-and post-deliberation which, together with their comments in deliberations, provided information about how they understood the not proven verdict and the circumstances in which it would be fitting.
It is important before proceeding to reflect a little further on our use of the mock jury method.This was necessary to overcome the limitation that the Contempt of Court Act 1981 places on discussing the content of jury deliberations 23  with jurors in 'real' trials, 29 but even if this limitation were relaxed, we simply would not have been able to undertake this research with jurors in real trials. The mock jury method allows researchers to hold constant factors within the trial scenario to isolate variables across different deliberating groups. This was particularly important here, since it allowed us to compare juries which had the not proven verdict to juries which did not, and to be more confident that any differences could be attributed to this variation, rather than other contextual factors.
All research methods, of course, have their limitations, and it is important to bear these in mind. 30 Mock jury studies have been criticised in particular for their lack of realism, which limits the extent to which their findings can be generalised. 31 The nature of our project and funding that we received from the Scottish Government meant that we were able to take steps to replicate the trial experience as far as possible, and the steps we took in this respect are set out above. One thing that we could not overcome in the experimental context was the fact that participants knew that they were role playing, which might mean that they did not approach their task as seriously as real jurors would have done. It has recently been suggested that there are 'fundamental differences between real jurors and volunteers', 32 that severely limit the reliance that can be placed on mock jury studies. However, we would dispute the claim, as we have done in detail elsewhere. 33 Mock jurors and 'real' jurors are not fundamentally different populations. Our jurors were all eligible for jury service and the compulsory nature of jury service means that they could easily end up as real jurors in real cases. Moreover, while it is true that we cannot be sure that our mock jurors would have behaved in exactly the same way in a real trial, the seriousness with which they took their deliberations -with participants spending considerable time discussing the demeanour of the accused and the complainer in a way that indicated they had suspended disbelief and forgotten that the roles were being played by actors, and several remarking to another how stressful they had found the deliberative process -gives us some confidence. 34 Finally, we do need to recognise that there are limitations to the research in terms of sample size. Although the study was substantial, the total number of juries (64) was still relatively small, particularly when split across trial type (rape / assault) and study variables. This means that, at the jury level, anything other than very large differences in verdict patterns between juries was unlikely to be significant. 35 Indeed, this proved to be the case in the context of the not proven verdict, as there were no significant differences in jury verdicts according to whether the jury had two or three verdicts available to them. 36 However, the fact that we had data on individual juror preferences for a large sample of jurors (863) meant that we were able to identify significant differences at juror level. Translation from juror-to jury-level findings is not straightforward, but our findings in this respect do provide an indication of the direction of change in the pattern of jury verdicts that is likely to result if the not proven verdict was to be abolished, if not the exact magnitude of that change.

Experiences of receiving a not proven verdict
The way in which jurors utilise the not proven verdict is not the only factor that merits further scrutiny and a more robust evidence base. In considering what a criminal justice process fit for the 21 st century looks like, it is also important to reflect on the 'human' impact of receiving a not proven (as distinct from not guilty) outcome. Though this has ramifications for an accused and a complainer alike, it has been the potential impact on complainers in the specific context of sexual offence trials that has garnered most attention in recent debates, linked in particular to the campaigning of Miss M and Rape Crisis Scotland. 37 Given the recurrent rhetoric from both Westminster and Holyrood governments that they are committed to 'improving the rights, support and experience of victims' 38 and developing a justice system with 'the needs of victims and witnesses at its heart', 39 it is right that these perspectives are afforded an increased prominence, without of course detracting from the rights and protections afforded to the accused. Thus, we draw here on a series of ten semi-structured interviews and one focus group discussion (involving five participants) conducted by one of the authors with individuals whohaving brought complaints (mostly, but not exclusively, of rape) -received a not proven verdict. Though this is a small sample, limited by its focus on only one constituency of trial party, the themes that emerged across research interviews were pronounced and consistent. They highlighted the challenges encountered in giving meaning to and making sense of a not proven verdict, and the complex ways in which its return had impacted upon complainers' mental health, emotional well-being, capacity for closure, and broader confidence in the justice system. 40 The focus on rape here in particular is not intended to presume that the experiences of complainers in other types of trials would 35 Statistical significance refers to tests that produced P-values of <=0.05. This means that the probability of such a difference occurring in our sample when there is no actual difference in reality is less than five per cent. 36  necessarily be the same; but it reflects the fact -which we discuss further below -that it is in the context of sexual offences that the performance and legitimacy of the not proven verdict has recently been most contested.

THE COMMUNICATION ARGUMENT
The first argument that has been made in favour of the not proven verdict (primarily in academic literature, rather than policy discourse) is that it serves a communicative function. Having two acquittal verdicts, it is said, provides more nuanced information than a single verdict would. 41 The assumption usually made by the verdict's proponents is that the message conveyed by a not guilty verdict is that the accused is factually innocent, whereas the message conveyed by a not proven verdict is something 'less' than this; that the accused might be (or probably is) guilty, but that the prosecution has not proved this beyond reasonable doubt. 42 Sometimes this argument is made in terms of greater clarity being a good in itself -that it is intrinsically valuable to be transparent and honest about what the jury thinks. 43 More commonly, it is made in terms of the advantages that could flow from the communication of this information. There are various constituencies that might benefit, which we group here into two categories: the parties to the trial 44 and the wider community.

The parties to the trial
A not proven verdict might communicate information to various parties to the trial, but the main focus has been on the acquitted person and the complainer. 45 In relation to the former, the argument is that the verdict warns the acquitted person that they only narrowly escaped conviction and puts them on notice that if they repeat the behaviour in question they may not be so 'lucky' next time. 46  message is 'not guilty, but don't do it again'. 47 There was some evidence from the deliberations in the Scottish Jury Research that jurors actively wished to convey this message. One juror stated, for example, that their personal choice of the not proven verdict over a not guilty verdict was intended to send a message to the accused that they doubted his story: It's like you were saying though, you're at a situation where you're not really sure, so it should be a not proven, which lets him know that there is a bit doubt of what's going on … (M07E) 48 Another juror was similarly clear about the message they thought would be communicated to the accused by the verdict: First juror: There's not enough evidence to back it up, that's the difference between not guilty and not proven.
Second juror: But, they still get away with it.
First juror: Yes, they get away with it but in their mind they know. (M05H) It might also be argued that the availability of three verdicts allows jurors to use the not guilty verdict to signal to the acquitted person that they positively believed in their innocence. There was some evidence of this in the jury deliberations too, where jurors stated that they supported a not guilty verdict over one of not proven because they did not want to send the message that they doubted the accused's account, as the following exchange illustrates: First juror: And you're the only one that's saying not guilty?
Second juror: Well basically I'm not prepared to imply that someone's committed rape. (M08G) Another juror argued for a not guilty verdict by stating of the not proven verdict: He has done absolutely nothing wrong and it's going to be hanging over him for the rest of his life, not proven on a rape is, no way. (M07E) Whether any of these messages are received by accused persons acquitted via a not proven verdict is unclear. Those who represent accused persons have suggested that, in their professional experience, acquitted persons care little about whether they have been acquitted via not proven or not guilty. 49  acquitted person's reaction at the time of acquittal, which is when it tends to be communicated to or observed by their legal counsel, says nothing about their experience of living with the verdict in years to come. 50 Robust empirical evidence regarding the longer-term impacts upon an accused of receiving a not proven verdict remains elusive, but the prospect that a lingering stigma might attach to this verdict is something we will return to further below. The argument that is much more commonly made, however, is that the not proven verdict has an important communicative function for the complainer: unlike a not guilty verdict, a not proven verdict need not cast doubt on her honesty or reliability. 51 This is one of two key claims that has consistently been made to justify the retention of the not proven verdict and, as such, is worthy of more detailed consideration. 52 We consider it primarily in the context of rape and sexual offence cases, as this is the arena in which this claim is almost always made.
The claim is that a not proven verdict sends the message to the sexual offence complainer that she was believed, but that there was ultimately felt to be insufficient evidence to support a conviction. 53 This belief in the reassurance offered by the verdict is sometimes asserted in the context of the corroboration requirement. 54 As Lord McCluskey put it: What I believe … is that the not proven verdict is very valuable in certain everyday situations. A clear example is that of a rape case in which the jury unreservedly accept the evidence of the woman that she was raped by the accused, but are unable to find any corroborative evidence. They have to acquit the accused. If they say 'Not Guilty', the impression may well be given that the woman was not believed. A verdict of not proven more accurately shows that sufficient evidence as required by Scots law was not found. Similarly, where the victim of a crime is accepted as fully reliable but the only corroboration is provided by evidence that is manifestly bogus,then the jury may feel that the proper course is to deliver a verdict of acquittal that leaves the victim untainted but rejects the bogus corroboration. 55 There is evidence from the jury deliberations that jurors did sometimes use the not proven verdict to send precisely this message, as the following quotes from one of the rape trial juries illustrate: … obviously if we say not guilty in this case we would be implying that we think, for example, that [the complainer] is not a credible witness, that she is lying. We might feel more comfortable saying not proven because we don't want to make that implication. (M03F) I think he was guilty, but I would probably have to go for a not proven because I think the fact that he couldn't prove, beyond whatever reasonable doubt is, that she got the bruises from an attack means that really reluctantly I would probably have to go not proven. Although on the basis of her testimony and his, I definitely believe her. (M03F) Like Lord McCluskey, the mock jurors sometimes framed this message around corroboration. This usually occurred in the rape trial, where the complainer's testimony was accompanied by evidence from a forensic examiner who confirmed that the complainer's external injuries were consistent with rape, but acknowledged that alternative explanations could not be ruled out. While this sufficed in law to satisfy the corroboration requirement, jurors sometimes misunderstood this, thinking instead that the doctor's evidence had to be unequivocal in order to corroborate the complainer's account. 56 When jurors relied on this assessment to support their return of a not proven verdict, they often accompanied it with a desire to stress that they did not disbelieve the complainer's account: Yes, I'm going not proven, because I think that she is credible and reliable and I believe her testimony, but as the judge said that we need to have corroboration and I don't think it was beyond reasonable doubt corroboration. (M05H) To the extent that this is the message intended to be communicated by jurors through their return of a not proven verdict, it was far from clear in the interviews conducted with complainers that this message was received; or that it had offered the kind of consolation intended if it was. For many complainers, while the return of a not proven verdict did signal that the jury had entertained a level of belief in the veracity of their account -and indeed, they were often specifically advised by third parties, including criminal justice officials, to take this meaning from the decision -it offered little reassurance. Some noted that 'knowing that there are men and women on that jury that believed me just a little bit makes it so hard' (Participant 4). What this communicated, in their understanding, was that they were on the cusp of securing a conviction, and this provoked a stream of new -unanswerable -questions about what they could have done differently to 'tip the balance'. More than one complainer likened the criminal trial to a 'test'. Participant 1, for example, observed that receiving a not proven verdict at the end of it is 'like finding out you've failed a test with a 50 per cent pass mark at 49 per cent … I'd rather have failed it at 10 per cent because it's easier to say "wow, they really didn't believe anything I said"'. This sense of frustration and loss associated with a not proven verdict often became more heightened for complainers over time. Thus, whilst in the initial stages of the decision, some interviewees found consolation in the communication of a level of belief from the jury, that comfort diminished as they came to the conclusion that acquittal via not proven still had no ramifications for the accused. As Participant 9 put it: … momentarily, it did help in the sense that there is a bit more belief there in you … (but) it was only me saying not proven and him more saying acquitted … (and) at the end of the day, it actually meant nothing because it had no impact on his life. So too, as complainers' confusion and disappointment in relation to jurors' reasoning grew over time, it provoked for many a deeper sense of mistrust in the justice system and a feeling of lack of protection that they carried for years thereafter. Concern about the potential legacy of the not proven verdict in this respect was acknowledged in jury deliberations, but outweighed for some by questions of evidential sufficiency. As one juror in the rape trial put it: This would be a real kick in the teeth [to the complainer] going home and given a not proven … because I genuinely believe that he did do it, but I don't believe the evidence is there, that's the problem. I believe probably she is failed by the justice system taking it to court without any more evidence. (M05F) A consistent and overarching theme in interviews was that complainers felt let down by the justice process: this manifested itself in a variety of ways but particularly pertinent for current purposes is that many complainers reflected on a sense of failure and betrayal by police and prosecutors whom they felt had either omitted to prepare them for the possibility of a not proven verdict or appeared uncomfortable, incapable and/or disinterested in engaging in discussion afterwards to help them try to make sense of what happened in the jury room. Reflecting on the implications of that experience, participants often remarked that, in important ways, they felt that a not proven verdict had -if anythingbeen worse or more damaging to them than a not guilty verdict would have been.
Of course, there is a counter-factual at play here, and as some complainers themselves acknowledged, it is impossible to know for sure how they would have felt about an alternative outcome: still, the sense that not proven was at best a 'back-handed compliment' (Participant 1) from the jury was widely shared. One of the interviewees reflected, for example, that she felt that throughout the trial in her case, she had seen the law 'being manipulated so that people (in the jury) didn't need to make a decision' and this had made her feel 'very unsafe' in the world and 'with no faith in the justice system'. She stated that: … with the not proven verdict, people are saying well it is better to receive because you get an acknowledgment that the jury believed you, but I would argue that no, not really, because at least if I thought that everyone was wildly not believing I would at least be able to trust life. I know what happened: being told that it did but that no one is going to get convicted for it just leaves me thinking what sort of society do you live in? (Participant 2) Participants were unanimous in the view that, as they sought to make sense of the jury's decision and begin the process of moving on with their lives, the not proven verdict had presented them with additional challenges that would not have been posed -in the same way at least -by a not guilty acquittal. That is not to say that a not guilty verdict would not have been experienced by them as 'devastating' (Participants 1, 3 and 7) at the time. However, complainers opined that the mixed messages received by a not proven verdict, together with the lack of closure this afforded, had made their recovery harder, 'leaving you powerless and in limbo' (Participant 7) and making 'the system look hollow' (Participant 8). Thus, while the jury might 'think they are giving something to a survivor' with a not proven verdict, as one participant, who recounted the difficulties she still experienced in moving on several years after the trial, articulated it 'they're actually making it worse … really it's like a slap in the face' (Participant 8).
In this way, in a context in which any acquittal is apt to be difficult and disappointing for complainers, a not proven decision can be experienced as worse than not guilty even though, on the face of it, the latter might be thought less desirable because it communicates a more categorical exoneration. Importantly, those participants who took part in the interviews were not suggesting that removal of Scotland's third verdict would have ensured convictions in their cases. Certainly, there were some complainers who were confident that this would have been the case given the strength of the evidence they felt that the Crown had presented; but there were others who were more circumspect about the prospects for conviction given the rules on evidence and requirement for proof beyond reasonable doubt. Still, they emphasised that a not guilty verdict would have been preferable because of its more categorical nature and they hypothesised that juries confronted solely with this starker choice would have been inclined to deliberate more robustly. This question of whether the mere existence of a third verdict impacts upon the approach taken in the jury room, and jurors' levels of satisfaction and confidence in their verdict outcome, is one we consider below.

The community
It has been argued that the not proven verdict sends a message to the community about the acquitted person that facilitates better informed decisions. 57 In a two-verdict system, 'not guilty' covers everything from the person who definitely did not commit the crime with which they were charged, to the person who probably did (but where the case against them was not proved beyond reasonable doubt). The three-verdict system facilitates a more nuanced message -separating the probably guilty from the probably innocent. There are different constituencies within the community who might benefit from this 57 Laudan, n 3 above, 4-5; Wansley, n 3 above, 324; Allan, n 43 above, 226.
information. The argument is made most commonly in relation to employers, 58 who may not wish to take the risk of hiring an offender (the assumption here being that such a person might re-offend). 59 The not proven verdict enables them to avoid (or at least lessen) that risk. 60 However the argument might equally apply to friends, family, acquaintances and so on -someone entering a romantic relationship, for example, may be more hesitant in doing so if their prospective partner had been acquitted of domestic abuse via a not proven rather than not guilty verdict.
Whatever the constituency, there are two main difficulties with this communicative argument -that the verdict lacks a clear definition (thus diluting its communicative message) and that, to the extent that this objection could be met through better defining the verdict, this would result in unjust stigma for the acquitted person.

A Verdict that Defies Definition
The first difficulty, then, is that in terms of how the not proven verdict currently operates in Scotland, it cannot be assumed that the verdict is being used to communicate any particular message about the acquitted person. As we have already seen, while different constituencies might impose their own interpretations and intonations upon it, the verdict has no settled meaning other than it being one of two verdicts of acquittal. In strict legal terms, it does not signify any particular probability that the acquitted person committed the offence charged. This leaves space for different -and potentially inconsistent -understandings of the verdict, as well as general confusion over the verdict's 'meaning' and how it might be differentiated from a verdict of not guilty.
This was illustrated repeatedly in the jury deliberations, where jurors stated on a number of occasions that they did not understand the verdict or asked other jurors what it meant. The understandings of the not proven verdict that jurors did arrive at varied although, given that the verdict is simply one of two verdicts of acquittal, very few were legally incorrect. 61 The most prominent was that jurors regarded not proven as the verdict to be used if they believed or suspected the accused was guilty, but felt that this had not been proved beyond reasonable doubt. This was apparent from the questionnaires that jurors completed, where a majority (70 per cent) of respondents thought that if a jury thinks the accused 58 S.L. Bray, 'Not Proven: Introducing a Third Verdict' (2005) 72 University of Chicago LR 1299, 1309; Phalen, n 45 above, 416. 59 Of course, this is an assumption that can be contested. Calculating rates of re-offending is a complex exercise, and the answer will vary according to method of calculation and the nature of the baseline offence. Nonetheless, statistics from England and Wales suggest that only a quarter of convicted persons re-offend within a year of conviction (in the case of a community penalty)/release from prison: Ministry of Justice, Proven Reoffending Statistics Quarterly Bulletin, England and Wales, January 2018 to March 2018 (2020). 60 Assuming, that is, that they would be aware of the acquittal at all. In practice, the two main ways in which this information may become known are via media reporting of the trial or via the background checks that apply to those working with vulnerable groups: for discussion of the latter see Scottish Jury Research, n 10 above, 50. 61 There were some exceptions, which we discuss below. is guilty, but do not think the evidence proves it beyond reasonable doubt, they should return a verdict of not proven. Just seven per cent said that it does not matter which of not proven or not guilty is returned, and 12 per cent that the jury should return a verdict of not guilty.
This theme also arose in the jury deliberations. As some of the jurors put it: First juror: Not guilty is completely different to not proven.
Second juror: Not proven is guilty, but you can't prove it. One juror stated that 'it means we think you did it, but we can't nab you for it' (M03H); another described the not proven verdict as 'a bit of a get out of jail card' (M05B). Others referred specifically to the stigma that they felt would follow a not proven verdict: 62 First juror: If this guy hasn't raped her I doubt he wants to jolly about with a not proven verdict hanging over him because that is still … Second juror: Yes, the effect is the same that he will walk free, but it's not the same.
The only thing with not proven is there is always that bit of doubt in people's minds, it doesn't matter who it is. (M06A) In a similar vein, some jurors saw the not proven verdict as the verdict to be used when the accused has not 'proved their innocence'. There is, of course, no requirement to 'prove' innocence in legal proceedings, and all of the juries were directed accordingly by the trial judge. Nonetheless, it was apparent that some jurors felt not proven should be used when innocence has not been demonstrated and distinguished between the not proven and not guilty verdicts in precisely these terms (that is, that not guilty indicates that the accused has proven their innocence, whereas not proven indicates that they have not). As some of the jurors put it: The difference between not proven and not guilty is not guilty is you believe there is evidence to prove that he is not guilty. (M08C) [Not proven means] there is not enough proof that he has done it, but there is also not proof, well no proof that he didn't do it, if that makes sense? (M05F) 62 We discuss the issue of stigma further below.
You will see a lot of people who will say rightly or wrongly, oh that guy's got a not proven, but there is a huge big bit of doubt about whether he was guilty or not. Whereas if you get a not guilty then there is absolutely no evidence and you're considered innocent. (M04H) First juror:Can we just talk about the difference between not proven and not guilty? So, is not guilty when you think … Second juror: You're 100 per cent sure that that person is not guilty. Another stated that: I went for not proven,because … both [the complainer and the accused] were too convincing, the evidence I felt was far too convincing, she really convinced me that he was guilty, he really convinced me that he was not guilty, and really the only way I can sum that up I have to say that it was not proven. (M03F) An example from the deliberations of the verdict being used as a collective compromise was the juror who responded to an impasse in the deliberations by suggesting 'will we just say it's not proven?' (M01F). In another case, a juror left the jury room and sought out one of the researchers to ask for clarification as follows: Juror: Can I ask a question?
Juror: Right, not proven and not guilty, does not proven mean that it is half and half?
Researcher: I can't answer anything like that.
Juror: You can't, all right.
Researcher: You just have to go on what you remember the judge saying about that. 63 Juror: Sorry, when I say not proven half and half, I mean six people saying guilty and six not guilty, does that mean not proven because there is not a majority? (M07B) The point was also demonstrated in the questionnaires that jurors completed where, when asked which verdict should be used 'when the jurors need to compromise to decide on a verdict', significantly more jurors selected not proven than not guilty. 64 The understanding of not proven as a group compromise verdict is not, however, the same thing as using the verdict to signify that the accused is probably (or may be) guilty. In the former scenario, jurors are using the verdict to avoid further -and possibly difficult -debate that ultimately might have changed the minds of some jurors. In the latter, jurors have fully debated the issues and wish to express a collective belief about the possible guilt of the accused. The verdict cannot, therefore, be assumed to send the message that proponents of the communicative argument assume it does, which weakens that argument considerably.
The absence of a clear and settled meaning for the not proven verdict not only undermines a case for its preservation grounded on its communicative function, but also poses more fundamental questions about the acceptability as a matter of justice of retaining a verdict that cannot be legally defined or explained. 65 Historically, trial judges did sometimes attempt to explain the difference between not proven and not guilty to juries, 66 but more recently the appeal court has warned trial judges that it is 'highly dangerous' 67 to try to do so. Trial judges now tend to stick to the formulation in the Jury Manual, which states only that: Not guilty and not proven are verdicts of acquittal and have the same effect. An accused acquitted of a charge cannot be prosecuted again on that charge, save in exceptional circumstances, and it makes no difference whether the acquittal verdict is not guilty or not proven. 68 It might be argued that, as a matter of justice, it does not matter if the difference between the two verdicts cannot be explained, as long as jurors are not 63 Jurors were directed on the verdict as they would have been in a real Scottish trial: 'Finally, I need to tell you that there are three verdicts you can return on this charge: not guilty, not proven, or guilty. Not guilty and not proven have the same effect, acquittal, which means that the accused cannot be tried again for the same offence.' The latest guidance for Scottish judges now recommends saying that 'an accused acquitted of a charge cannot be prosecuted again on that charge, save in exceptional circumstances', which reflects the Double Jeopardy labouring under misapprehensions and are always acquitting when there is reasonable doubt. It is not clear, however, that we can safely assume this. While legal misunderstandings of the verdict were uncommon in the Scottish Jury Research, given the paucity of a legal definition to start with, they did nonetheless arise. 69 Most commonly this related to the possibility of retrial following a not proven verdict, with some jurors mistakenly thinking that the distinction between not proven and not guilty was that if fresh evidence arose a retrial could occur in the former but not the latter. 70 Still, of course, it might be responded by defenders of the not proven verdict that this is not so problematicmisunderstandings (certainly of this sort) can be corrected; and there was certainly evidence from the Scottish Jury Research that legal direction did improve juror understanding of this issue. The proportion of jurors in the two-verdict condition (who had not been directed on the not proven verdict) who held this mistaken view was significantly higher than in the three-verdict condition (where jurors were directed that the position regarding retrial was the same for the not proven and not guilty verdict). 71 Though that instruction was given orally, research suggests that written directions might improve juror comprehension further. 72 In any event, we do not think it is acceptable to simply take the view that it does not matter that the not proven verdict has no legal meaning, as long as jurors do not misunderstand it and always return an acquittal verdict when the standard of proof has not been met. Apart from anything else, if the not proven verdict cannot be legally distinguished from the not guilty verdict, this makes it very difficult for legal professionals to explain it to those who are affected by it. Several complainers spoke at length in their interviews about feeling 'let down' (Participants 5, 7, 8 and 9) by the inability (often perceived to be borne out of indifference or discomfort more than lack of explanatory resource) of police or prosecutors to provide a meaningful explanation of what not proven means or why the jury might have chosen it. Participants often reported having no prior appreciation of the meaning of not proven, and feeling insufficiently prepared for the possibility of it being returned in their case. As one put it, 'I didn't even know that it existed, to be honest … maybe that's ignorance on our part but we didn't even know that it was a possibility' (Participant 7).
Many complainers turned to criminal justice personnel after their trial in pursuit of greater clarity: at least half made inquiries about accessing court transcripts, while others secured meetings with the Crown Office. All who 69 See Scottish Jury Research, n 10 above, 45-48 (concerning re-prosecution) and 48-50 (concerning criminal records). 70 In fact, while the Double Jeopardy (Scotland) Act 2011 does, in limited circumstances, allow the prosecution to apply for permission to re-prosecute following an acquittal, such an application for re-prosecution can be made regardless of whether the verdict is not guilty or not proven. 71 Scottish Jury Research, n 10 above, 46-47 (data from the post-deliberation questionnaires had done so reported in their interviews that they found their treatment unsatisfactory. Many continued to be confused about the meaning and implications of the verdict, and the vast majority asked the researcher for clarification on that confusion during the interviews. Participants spoke about feeling condescended to by officials who gave what they felt to be 'very political answers' to questions, and who seemed such uncomfortable interlocutors in the process that, as one complainer put it, 'if there had been an eject button in that room, she [the procurator fiscal] would have pressed it to get out' (Participant 1).
Other complainers recounted frustrating conversations where -in response to their questions -prosecutors kept repeating that not proven was a complete acquittal without explaining how to distinguish it from not guilty. Indeed, one reported that she was told by a very senior official that: 'no one really knows what not proven means and it's quite misunderstood by different organisations, so it is not very helpful for me to give you a different interpretation of what not proven means' (Participant 8). Though that may be true, this opacity of response was difficult to cope with, not least since it underscored many complainers' impression of being ignored or instrumentalised in the justice process: 'nobody explains what it means and it's as if you're invisible, or you don't matter' (Participant 7). It also -one might infer -places criminal justice officials in an unnecessarily difficult position where the absence of clarity that surrounds jury deliberation in any case is compounded by uncertainty as to what would be an appropriate interpretation and application of the not proven verdict, making it almost impossible satisfactorily to answer complainers' questions.
Of course, it might be said at this point that the difficulties of explaining a verdict that has no consistent meaning would disappear if the not proven verdict was given a definition and this was clearly communicated to -and understood by -all concerned. This requires us, however, to agree what that definition would be. The most obvious candidate is the definition assumed by those who argue for the verdict's communicative function -that not guilty signifies innocence, whereas not proven signifies only that guilt has not been proved beyond reasonable doubt. 73 Indeed, it was sometimes assumed historically that this was the distinction between the two verdicts. 74 We discussed above the concern that, even if this meaning were more clearly communicated through the use of a not proven verdict, for complainers at least, its value as an expression of belief in the veracity of their claim may be limited, and outstripped by a 'hollowing out' of the justice process. This is not the only concern associated with this 'solution', however, as we explain further below.

The Presumption of Innocence and Unjust Stigma
Let us assume, then, that we could overcome the difficulty of having two acquittal verdicts that cannot be distinguished from each other by defining the not proven verdict as the verdict to be used when guilt has not been proved beyond reasonable doubt, reserving the not guilty verdict for cases where the jury are convinced of the accused's innocence. The difficulty with this is that it attaches to the not proven verdict an unjust stigma 75 that, as a result, runs contrary to at least the principles underlying the presumption of innocence. 76 Even without such a formal definition, the verdict as it is presently used has regularly been criticised on the basis of the unjust stigma it places on the acquitted person -it was,for example,described by three dissenting members of the Thomson Committee as a 'second class acquittal'. 77 We saw earlier that this was also a theme that arose in the mock jury deliberations, with participants seeing it as communicating something less than innocence when contrasted with not guilty.
At this juncture it could be countered that stigma can arise upon acquittal in a two-verdict system too, 78 a point sometimes made in relation to the OJ Simpson trial in the US 79 or those initially acquitted of the murder of Stephen Lawrence in the UK. 80 But these are extreme cases and stigma following acquittal is far more likely to arise in a differentiated acquittal system, such as Scotland's. If 'innocence' vs 'not proven beyond reasonable doubt' were formally adopted as the distinction between not guilty and not proven, this stigma would become more pronounced, and -crucially -it would be a stigma formally sanctioned by law.
The extent to which accused persons acquitted by a not proven verdict actually experience stigma is a point we considered earlier. Robust evidence on this matter is lacking. Interviews with complainers indicated that they did not consider the not proven verdict as stigmatising to accused persons, and those who retained a degree of contact -direct or indirect -with their alleged Stephen Lawrence was a black teenager who was stabbed to death in 1993 in an unprovoked attack while he was waiting for a bus in London. Charges initially brought against two suspects were dropped by the Crown Prosecution Service. In 1996, a private prosecution of three men for his murder resulted in their acquittal. Changes to double jeopardy law in England and Wales allowed a second prosecution of two of the men for murder in 2011, and they were convicted. assailant reported little indication of any repercussion or remorse to differentiate the outcome from a not guilty acquittal. But, of course, the vantage point from which this assessment is made is framed through the lens of complainers' own experience and disappointment at the trial outcome, meaning that this evidence does not definitively answer the empirical question. Whatever the answer to that question, there is a normative argument to be made, which is that the criminal justice system should not sanction an acquittal verdict that is stigmatic, even if acquitted persons do not always -or perhaps even do not often -experience it in that way. This argument has sometimes been made in the context of the presumption of innocence and we consider it further below.

Does the Not Proven Verdict Implicate the Presumption of Innocence?
The presumption of innocence can be understood in two senses: first, the rule that the prosecution bears the burden of proof to establish guilt beyond reasonable doubt and,second,that the treatment of the accused should as far as possible be consistent with their being innocent. 81 Proponents of the not proven verdict will sometimes argue that it raises no conflict of any sort with the presumption of innocence, that being merely a statement about the burden of proof. 82 While such a narrow reading of the presumption is not uncommon, 83 it ignores two things. First, the jurisprudence of the European Court of Human Rights makes clear that the presumption, as guaranteed by Article 6(2), has consequences beyond the incidence and standard of the burden of proof. Secondly, the presumption can be understood in a broader, moral, sense. 84 Bare compliance with domestic rules of procedure or the strictures of the Convention do not preclude an argument that, as a matter of policy or morality, the not proven verdict is inappropriate because it treats the acquitted person with suspicion despite the state having failed to prove their guilt beyond reasonable doubt.
The not proven verdict is vulnerable to an argument that it contravenes Article 6(2), albeit it may be spared from such a finding by its own ambiguity. ECHR jurisprudence holds that the presumption will be violated where, following an acquittal, suspicions of guilt are voiced by a court. So, for example, violations have been found where individuals acquitted of criminal offences have had claims for compensation or costs rejected on the basis that suspicions remained as to whether they committed the offence. 85 There is a violation in such cases even where the subsequent decision involves no new evaluation of the case but is simply based on the file from the original proceedings. 86 This body of case law is,of course,distinguishable from the not proven verdict itself.First,the not proven verdict is not a separate finding by a court.It therefore does not,as the European Court of Human Rights has framed the problem,'cast doubt on the correctness of the acquittal': 87 it is the acquittal. 88 In that sense, it is no different from a reasoned verdict which notes the strength of the case against the accused while at the same time concluding that guilt has not been proven beyond reasonable doubt. Moreover, the absence of any formal definition of the not proven verdict means that, strictly speaking, it can be read as casting no doubt whatsoever on the accused's innocence: as a matter of law it has no meaning beyond, as with not guilty, a statement that guilt has not been proven beyond reasonable doubt. While, as we have demonstrated above, this is not how the verdict is in practice understood, it remains the position as articulated by the courts.
For these reasons, it is unlikely that the not proven verdict would be found to contravene Article 6(2). But at the same time, this illustrates the ambiguity and even futility of the verdict: it cannot be relied upon in any sense as casting doubt on the innocence of the accused and any official statement that it did would itself risk violating the presumption. 89 It complies with the Convention only because it operates by a nudge and a wink: as one mock juror put it, 'you walk away innocent, but everybody knows'. 90 It is neither honest nor principled for a legal system to operate in this fashion even if it clears the bar of minimal compliance with the letter of the Convention. 91

Where does this leave the communication argument?
As the evidence above demonstrates, jurors may believe that they are sending a particular message through their choice of verdict, but that message is not clearly received. Certainly in respect of sexual offence cases, where the not proven verdict is disproportionately utilised, 92 there appears to be evidence of a mismatch between what jurors may believe they are communicating and the message which is actually heard, especially by complainers. More broadly, the lack of any clear and settled meaning for the verdict undermines any potential communicative function to the community. To the extent that the communicative argument holds sway, it is because the verdict stigmatises the accused, carrying a meaning which no-one is willing authoritatively to articulate and which, if they were prepared to articulate it, would be regarded as clearly improper in its failure to comply with the principles underlying the presumption of innocence.

THE PROTECTION AGAINST WRONGFUL CONVICTION ARGUMENT
A second argument in favour of the not proven verdict is that it might provide some protection against wrongful conviction. 93 It has become increasingly apparent since the advent of DNA testing that wrongful convictions do occur. 94 When they do they have profound consequences not only for the wrongfully convicted person, 95 but also for others involved in the case 96 and for the community more broadly. 97 The argument is that when jurors are faced with a borderline case that does not quite reach the threshold of proof beyond reasonable doubt, they are likely to opt for a not proven verdict in a three-verdict system, whereas faced with an identical case in a two-verdict system, they may be tempted to convict. 98 The point has sometimes been dismissed on the basis that juries should only ever convict if they are sure of guilt beyond reasonable doubt. 99 If not proven were to be abolished, on this argument, all not proven verdicts would simply become not guilty verdicts. Using the data generated from the Scottish Jury Research, however, we can see that this is very unlikely to be the case. In the 64 mock juries, 32 had the choice of two verdicts (guilty and not guilty) whereas 32 had the choice of three verdicts (guilty, not guilty and not proven). Jurors were asked for their verdict preference following the trial -both before and after deliberation. The research found that individual jurors were significantly less likely to favour a guilty verdict when the not proven verdict was available, both before and after deliberation. 100 In other words, the availability of the not proven verdict did indeed increase the proportion of jurors voting to acquit. 101 The difficulty with this, however, is that we cannot simply assume from an increased acquittal rate that the not proven verdict prevents wrongful conviction. We have no way of knowing whether the people who are acquitted on a not proven verdict (who would have been convicted in a two-verdict system) are factually innocent. Indeed, it is equally plausible to argue that the two-verdict system facilitates the acquittal of the factually guilty. This argument has featured most prominently in relation to sexual offences. Indeed, a collaboration between Miss M and Rape Crisis Scotland has specifically called for the abolition of not proven as a verdict across all offence types, but particularly based on an insistence that it operates as a profound barrier to conviction in cases of rape and sexual assault. 102 This is partly because of well-established challenges in securing robust corroborative evidence in sexual offence cases: while a succession of moments of judicial creativity have allowed more flexibility in respect of what constitutes corroboration, 103 the possibility remains that jurors will adopt a more restrictive interpretation. Within this more restrictive framing, though jurors may believe a complainer's account, they may nonetheless conclude the evidential threshold required for conviction has not been met. This was certainly a theme that concerned several jurors who deliberated upon the rape trial scenario in the Scottish Jury Research, where -as noted above -medical testimony that did, as a matter of law, satisfy corroboration requirements was found by many to be insufficient. 104 More widely, the concern has been raised that -in a context in which research has repeatedly documented the risk of unfounded myth and stereotype being relied upon to inform deliberation in rape cases 105 and the nature of the offence requires jurors to make determinations about consent and belief in consent framed by a complicated structure of socio-sexual norms 106 -'the existence of the not proven verdict gives juries in rape trials an easy out and contributes to guilty people walking free'. 107 Certainly, as we discuss below, there was evidence in the Scottish Jury Research that the opportunity to opt for a not proven verdict was seen, by some jurors, to offer a kind of comforting compromise position, one that absolved them from the challenges of making a categorical determination. There was also evidence from the interviews with complainers that this feature of how jurors might undertake their deliberations was far from lost on them. Indeed, as we also explore further below, while many complainers appreciated the reasons why -at a human level -there might be an attraction and appeal to 'settling' for a not proven verdict, they saw it as an abandonment of civic duty.
The argument about protection against wrongful conviction is, thus, more complex than it might initially seem. Further, even if commentators are right to be worried about the prospect of wrongful conviction more than the acquittal of the factually guilty, and even if we accept that the not proven verdict might act as a safeguard, it does not follow necessarily that it should be retained. There are a myriad of measures that can be taken to address the causes of wrongful conviction aside from operating a three-verdict system -measures that do not have the inherent difficulties associated with not proven that we have identified above. 108 As Duff puts it,'no other country in the world feels that giving the jury a choice of three verdicts is a solution to this problem'. 109 Indeed, it might be argued that the existence of the not proven verdict has actually been detrimental to the prevention of wrongful conviction in Scotland, as it has been used to argue that Scots law need not introduce any of the alternative safeguards found elsewhere. 110 This potentially leaves Scotland with weaker protection against wrongful conviction overall compared to other jurisdictions.

THE JUROR SATISFACTION ARGUMENT
The final substantive argument that has been made in favour of the not proven verdict is that it increases juror satisfaction. 111 There is some evidence to support this from the Scottish Jury Research. Jurors who were asked to choose between two verdicts were more likely to say they had been dissatisfied with the experience of being a juror than jurors who had three verdicts available. 112 There are two possible reasons why this might be the case. The first is that the not proven verdict allows jurors to express their feelings more clearly than is possible in a two-verdict system. 113 As we discussed above, it allows them to communicate (or at least to believe that they have communicated, since the question of whether it is received or welcomed is another matter) a particular message, whether this be a doubt in the accused's innocence or a message to the complainer that they did believe her account.
The second reason is that jurors prefer to have a 'compromise' option, 114 rather than having to make a choice between the stark alternatives of not guilty and guilty. This might absolve the individual juror from making a harder choice or allow the jury as a collective to reach a verdict more easily and with less conflict than would have been the case under a two-verdict system. We noted above that there was certainly evidence in the Scottish Jury Research to support the contention that not proven is seen as a compromise verdict; an easy middle ground. It is worth adding here that jurors who held this view sometimes expressed relief that the not proven verdict was available as a way of ending deliberations, and this is illustrated by the following exchange between two of the jurors in the rape trial: However, while jurors might be happy with the availability of the not proven verdict as a compromise verdict, that does not mean it is a desirable state of affairs. Its existence may, as Lord Cockburn opined back in 1846, tempt jurors 'not to look steadily at the evidence, and to give it its correct result; but to speculate about the possibility of soothing their consciences, or their feelings, by neither convicting nor acquitting, but steering between the two.' 115 Indeed, some of the mock jurors recognised this, describing the not proven verdict as a 'cop-out' (M01F, M06G, M05F), 'an easy get out' (M08B) or 'a fudged decision' (M07D), that excused the jury from deliberating or examining the evidence more fully or from making a more difficult choice. One juror described it as 'a very unfair scapegoat for the victims' (M05F). Another put it as follows: It's like if you do these things at work where you have to kind of like you agree, you disagree, blah, blah, blah, and most people go for the middle ground because it's the easiest option and I think the danger of a situation like this, is that because you're not 100 per cent sure, which none of us are, you're just thinking it's safer just to say not proven. (M08G) For many of the complainers that took part in interviews, all of whom supported the removal of the not proven verdict, it was this which was at the forefront of their minds. During discussions, they frequently acknowledged that it 114 Bray, n 58 above, 1315. 115 Lord Cockburn, n 7 above, 206. is often 'hard' and 'horrible' (Participant 1) to have to act as a juror and decide another person's fate. However, they also saw the not proven verdict as affording an opportunity for those uncomfortable with this challenging role to 'ease their conscience' and 'sit on the fence' (Participant 7). Complainers spoke of the trial process -both in general and in their case -as being one in which defence counsel were allowed to 'manipulate' jurors' unease. As one interviewee put it, 'you put people in a really high stress situation, force them to make a decision and then tell them they don't have to' (Participant 2). Others agreed, suggesting that this 'defeats the point of having a jury' (Participant 1), 'leaves a cloud' (Participant 7) and 'is not what justice is about, whether you are the attacker or survivor' (Participant 6).
Irrespective of whether they thought the abolition of not proven would have changed the outcome in their own trial, the complainers in this sample believed unanimously that its removal would be likely to increase the stakes for jurors and would encourage more robust and careful deliberations. As one put it: Jurors may look at the evidence with more of an open mind and look at things more clearly instead of having this cushion, this safety net … if it was guilty or not guilty, they might look at it more authentically (Participant 7).
Another observed that 'if they actually had to make this cold decision between guilty and not guilty, it might mean they would think about it a bit more' (Participant 8). In an important sense, then, whether or not the not proven verdict is in fact often used by jurors as a 'get out' (Participant 7), the perception that it might be used in this way can undermine trial parties' and public confidence in the evaluative and deliberative processes that precede its return in specific trials, and by extension confidence in the rigour of the Scottish criminal justice process more broadly.
That said, it would clearly not be ideal if jurors were so dis-satisfied they were unwilling to serve, or did so only grudgingly. There is no evidence, however, to suggest that the availability of the not proven verdict makes very much difference to rates of juror satisfaction. 116 Issues such as court facilities, information provision and waiting time are likely to be far more important in improving the juror experience, 117 as might the provision of support and counselling at the conclusion of emotionally disturbing trials. 118 To the extent that the not proven verdict serves any useful function in the criminal justice system, we have argued in this article that its merits are significantly outweighed by its demerits. Three longstanding arguments in its favour -that it performs a communicative function, that it safeguards against wrongful conviction, and that it enhances juror satisfaction -can now be evaluated in the light of rigorous empirical evidence. None of these arguments, either individually or taken together, justifies the continued existence of the verdict. They are either unsupported by the empirical evidence that we have presented, or are outweighed by normative arguments against the verdict.
With regard to the possible communicative function of the verdict, while jurors appear to believe that they are sending a particular message through their choice of verdict, it is not clear that it is received in this way. In sexual offence cases in particular, there is a mismatch between what jurors believe they are communicating to complainers and the message which is actually heard. In terms of communication to the wider community, the lack of any clear and settled meaning for the verdict, and differing juror understandings as to what it signifies and when it should be used, undermines any potential communicative function and makes it difficult for criminal justice professionals to explain to complainers what they should make of the verdict.
In terms of the argument that the verdict operates as a safeguard against wrongful conviction, there is indeed evidence that the not proven verdict may reduce the propensity of jurors to convict. However, this does not demonstrate that it operates as a safeguard against wrongful conviction in the way often claimed: indeed, it may equally result in the acquittal of the factually guilty. The use of the verdict is particularly prevalent, but also particularly problematic, in sexual offences, where it may enable juries to give weight to myths and stereotypes in avoiding verdicts of conviction. And while there is no clear evidence that the verdict does in fact safeguard against wrongful conviction, its existence has been used to justify Scots law not introducing other measures which would, meaning that it may in fact be actively harmful in this regard.
In terms of juror satisfaction, there is some evidence to support the claim that a not proven verdict increases juror satisfaction, or at least that its removal might increase juror dissatisfaction. However, this is a weak argument when set against two key arguments that can be made against the verdict.
The first of those is in terms of the stigma that attaches to the verdict. While it is, perhaps, compatible with the presumption of innocence in strict ECHR terms, this bare compliance with Convention rights is no justification for the not proven verdict. It stigmatises the accused, operating by a nudge and a wink, carrying a meaning which no-one is willing to articulate and which, if they were prepared to articulate, would be seen as unjust and improper. In empirical terms, we do not know to what extent stigma is in fact experienced by those acquitted by a not proven verdict, but regardless of this, there is a normative argument that an acquittal verdict should not be stigmatising, and that in itself is a sufficiently powerful argument against its retention.
The second argument that can be made against the not proven verdict that it risks a loss of public confidence in the criminal justice system, as it allows jurors to use it as a compromise verdict to bring deliberations to an end, rather than engaging in more rigorous discussions. There is empirical evidence from the Scottish Jury Research that the verdict operates in precisely this way, with participants using it to bring deliberations to a premature end. There was also evidence that this use was 'read into' the verdict outcome by complainers, undermining their belief that jurors discharged the weighty responsibility placed upon them with appropriate diligence.
There is, however, a further point that needs to be considered. Even if it is accepted that Scots law should depart from the existing three verdict system -and we have argued that it should -the dying gasp of the not proven verdict takes its form in the argument that it is in fact the not guilty verdict which should disappear, and that not proven should be retained (perhaps alongside a verdict of proven rather than guilty). This argument has not been fully articulated in academic literature but when the Scottish Government held a series of engagement events following publication of the Scottish Jury Research, it was a view commonly expressed by legal practitioners. 119 It also emerged as the preferred model for a redesigned verdict system (when contrasted against the current three verdicts or a move to guilty/not guilty) in an online survey of 78 practitioners. 120 In support of this proven/not proven alternative, the reasoning offered 'virtually every time' during the engagement events was 'that a jury's role is not to determine a person's guilt or innocence, but rather to assess whether the Crown has proven the charge'. 121 This represents, however, a very denuded conception of the jury's role, which is not limited simply to questions of 'what happened'. Rather it frequently involves an evaluative role -such as,for example,determining whether the use of a particular level of force in self-defence was reasonable -a point recognised by early decisions of the Scottish courts which suggested that a judge might legitimately withdraw the option of 'not proven' from the jury in cases which turned solely on such a question. 122 Notwithstanding, it would be odd, given the demonstrable problems of the not proven verdict, if a decision were taken to retain it rather than not guiltyagainst which no case has been made. Although a recent survey of practitioners has suggested a preference for a proven/not proven model, it also revealed that one-third of respondents favoured abolition of the not proven verdict, precisely because of concerns about juror misunderstanding or avoidance of more difficult, binary decisions. Indeed, three-quarters of respondents said they believed jurors would perceive the verdict to mean 'innocent in law, but not in community', 123 underscoring rather than diminishing the concerns raised above. We are not starting from a clean sheet here. A single acquittal verdict of not proven is likely to carry a residual element of stigma that is incongruent with the principles underlying the presumption of innocence. It is also unlikely to be well understood by anyone outside the jurisdiction, who may attach a stigmatic meaning to it through lack of understanding, especially as the verdict would be out of step with the use of not guilty by the vast majority of other legal systems.
Though it was not tested in the practitioner survey or stakeholder engagement discussions, it may be that some proponents of the proven/not proven argument do not in fact intend that the terms guilty and not guilty would disappear entirely: one possibility is that a jury would declare individual charges proven or not proven, and that the judge would consequently pronounce the accused guilty or not guilty. But this is still vulnerable to the objection made above -that this is not reflective of the jury's role -and adds an unnecessary layer of complexity to the criminal trial. As such, it is also a proposal that we would resist.

CONCLUSION
Debate about the not proven verdict has been ongoing for over a century and a half. Despite several attempts to abolish it, the verdict remains a feature of Scots law. The underlying debate, however, still echoes Lord Cockburn's analysis in 1864. Critics of the not proven verdict in Scotland, such as Cockburn, tend to argue that it is incompatible with the presumption of innocence and casts an objectionable stigma on an accused against whom guilt has not been proven, whilst encouraging jurors to avoid the proper discharge of their function. 124 By contrast, proponents of the not proven verdict claim that it serves an important communicative function (in current debates often tied primarily to complainers in sexual offence cases) and might prevent wrongful conviction. 125 What has been lacking to date is robust empirical data to support -or dispute -these claims. This article makes an important contribution, both by building the evidence base to demonstrate that the not proven verdict lacks many of the merits its proponents have claimed for it, and by articulating the normative and empirical arguments that can be made more proactively against the verdict.
Empirical data from the Scottish Jury Research demonstrates that jurors are inconsistent in their interpretation and application of the verdict, but appear to often believe that, in selecting it over a not guilty acquittal, they are sending a particular message, to the trial parties and community at large. Research with rape complainers,however,shows that this message is not being received,despite 123 Curley et al, n 120 above, 4-5. 124 Lord Cockburn, n 7 above, 206. not proven's disproportionate use in sexual offence cases. Moreover, the lack of any clear and settled meaning for the verdict undermines any claim to a broader, community communicative function. While our empirical evidence suggests that if the not proven verdict were removed more convictions might result, this in itself tells us nothing about the 'proper' rate of conviction and, in the context of sexual offences in particular, the concern that the verdict may actually be enabling the acquittal of the factually guilty has been acknowledged as meriting review. 126 Our empirical evidence does indicate that the availability of the not proven verdict might improve juror satisfaction, but this is not a weighty enough argument for its retention, when set against the arguments for its abolition. It is inappropriate and unjust for the legal system to sanction an acquittal verdict that stigmatises the accused because it carries -for some -the implication that the accused person is not truly innocent. It is not acceptable to have a verdict that cannot be explained to those who experience it, whether as accused or complainer, and evidence that juries may use it as a compromise to bring deliberations to a premature end does not inspire confidence in the criminal justice system.
It may be attractive to Scots lawyers to believe that the Scottish system has, whether by chance or native genius, happened upon a uniquely desirable system that the rest of the world has been too blind to recognise. The arguments that have been made in its favour are, however, either not borne out by the empirical evidence,or can be countered by more compelling normative arguments against its use. It is, after over a century of debate, time to consign the not proven verdict to history. This is not to say that there are not further decisions to be taken regarding the other unique features of the Scottish jury system -in particular the use of majority verdicts,rather than a system of unanimity or near unanimity. Indeed, the inter-connectedness of the not proven verdict with other aspects of the Scottish criminal justice system has, in the past, operated as a barrier to reform, on the basis that these problems are too complex to resolve. 127 This complexity is, perhaps, over-stated, given that other jurisdictions have managed to resolve them without any apparent difficulty. But regardless of this, the issue of other changes that might need to be made if the not proven verdict were to be abolished is an ancillary one. It should no longer be used as an excuse to retain a verdict that is unjustly stigmatic and allows juries to avoid the proper discharge of their function. 126 See for example BBC News, 'Scottish election 2021: Sturgeon considers scrapping not proven verdict' 4 April 2021 at https://www.bbc.co.uk/news/uk-scotland-scotland-politics-56630482 (last visited 12 July 2021); R. Mackie, 'The case for abolishing the 'not proven' verdict' The Scotsman 5 April 2021 at https://www.scotsman.com/news/crime/ the-case-for-abolishing-the-not-proven-verdict-3190031 (last visited 12 July 2021). 127 See Chalmers et al, n 1 above, 167.