Abstract
Jogee established that, if the victim was attacked in a manner different to that which the accessory envisioned, the accessory will not be liable for a resulting death, provided the fatal act amounted to ‘some overwhelming supervening act [OSA]… which nobody in the defendant’s shoes could have contemplated might happen and is of such a character as to relegate his acts to history’. Jogee further suggests that OSA has replaced the fundamental difference rule (FDR) which used to keep the foresight test in parasitic accessory liability (PAL) cases in check, by absolving accessories of liability for murder or manslaughter, where the principal had caused the victim’s death with a different and more lethal weapon or in a different and more dangerous manner than had been foreseen by the accessory. In Jogee, the UKSC appreciated this role of the FDR, and therefore, in a misguided attempt to appease the defenders of PAL, abolished it along with PAL. While they left OSA in place in its stead, the judgment is unclear about its scope, role, and function. Since then, OSA has been considered on several occasions by the Court of Appeal. However, there is still no clarity as to when OSA ought to be left to the jury. This paper will explore OSA, its role, scope of application, and relationship with the FDR. It will be argued that OSA is a concept of remoteness which, like the FDR before it, functions as a corrective to constructive liability for accessories to murder and manslaughter. It aims to attribute to the accessory only those consequences of the principal’s acts that are sufficiently linked to the risk that the accessory set or supported by his acts of assistance or encouragement. The paper concludes that, in fleshing out OSA, the courts ought to pay greater attention to this role and purpose, and the principles of fair attribution that underpin it.
Keywords
Introduction
Criminal incidents do not necessarily unfold as those involved expect. In English, 1 the appellant participated in a violent attack on a police officer. He thought that his co-offender might seriously injure the victim by beating him with a wooden post. In the event, the co-offender fatally stabbed the victim with a knife. In Gamble, 2 the defendants were charged with murder. They submitted that they had expected the victim to be subjected to a punishment beating or ‘kneecapping’ (which involves firing bullets in the kneecap). In fact, the principal offenders deliberately killed the victim by slitting his throat. In Tas, 3 the appellant expected to participate in a ‘punch-up’. He sought to have his manslaughter conviction overturned on the basis that he had not known of the knife one of his group had used to kill one of the opponents. In Grant, 4 the appellant was a passenger in a car that was driven around in search of two men. He anticipated that they might be caused serious harm on foot during a face-to-face confrontation. Instead, the driver ran both victims down, killing one of them in a deliberate hit and run attack.
Prior to Jogee, such scenarios attracted the application of the so-called fundamental difference rule (FDR). Under the FDR, the actions of the principal offender (P) remained attributable to an accessory (S) unless they were ‘of a type entirely different from actions which [S] foresaw as part of the attack’. 5 In the context of murder, 6 the FDR was applied to cases where the common purpose of the principal offender and his accessory was not to kill but to cause serious harm, 7 and had the effect that the accessory was not liable for murder if the direct cause of the victim’s death was a deliberate act by the principal which was of a kind, firstly, unforeseen by the accessory and, secondly, likely to be altogether more life-threatening than acts of the kind intended or foreseen by him. 8
In Jogee, the Supreme Court considered that the main role of the FDR was to put boundaries on the foresight test in parasitic accessory liability (PAL), 9 absolving accessories of liability for the victim’s death where the principal had caused it with a different and more lethal weapon or in a different and more dangerous manner than had been foreseen by the accessory. With the abolition of PAL, their Lordships considered that the need for a limiting principle such as the FDR was much reduced: 10 the accessory would now only be liable if the prosecution could prove that he had intended (to assist or encourage) the assault that resulted in death; mere foresight would no longer suffice. However, the Supreme Court still saw some scope for a narrower limiting principle – while abolishing the FDR, they left in place the so-called overwhelming supervening act principle (OSA) to fulfil that function. After Jogee, the fatal use of a hidden knife or murderous car-attack when a serious beating or fistfight had been expected will absolve an accessory of liability for death only if this was ‘caused by some overwhelming supervening act by the perpetrator which nobody in the defendant’s shoes could have contemplated might happen and is of such a character as to relegate his acts to history.’ 11
This paper will examine the concept of OSA, its requirements, scope of application, and relationship with the FDR, as discussed in Jogee and the post-Jogee case law. I will argue that the Supreme Court, in replacing the FDR with OSA, under-appreciated the role the FDR used to play as a safety valve for the grievous bodily harm (GBH) rule in murder, as well as in circumscribing the dangerousness requirement in constructive manslaughter. I will suggest that OSA ought to be understood as a concept of remoteness which, like the FDR, serves as a corrective to constructive liability for accessories to homicide. 12 As such, OSA aims to attribute only those actions by the perpetrator that are sufficiently linked to the risk that the accessory set by his intentional act of assistance or encouragement. In considering whether the jury needs to be given an OSA direction, the courts ought to pay greater attention to this role and purpose, and the principles of fair attribution that underpin it. I will further argue that OSA, like the FDR before it, should not be applied where principal and accessory shared an intent that the victim be killed. I will conclude that OSA has a greater (liability-limiting) role to play than is currently accepted by the Court of Appeal where the accessory intended to assist or encourage the principal to commit (no more than) GBH, or where the accessory intended to assist or encourage an assault (or any other unlawful act which carried a risk of harm to another person) that would have been much less dangerous as planned than as put into action by the principal offender.
The Origins of OSA
To set the scene, let us briefly consider the origins of OSA. These might be of older heritage, 13 but are commonly traced back to the Court of Appeal decision in Anderson and Morris. 14
The appellants, A and M, were appealing their convictions for murder and manslaughter respectively. They had been searching for the victim, W, in response to allegations that W had tried to strangle A’s wife. When they came across W, A attacked and stabbed him. M’s defence was that even though he may have taken part in a joint attack with A to beat up W, he had not known that A carried a knife; 15 nor had he intended GBH.
In his summing up the trial judge told the jury that they could convict M of manslaughter even though he did not know that A had armed himself with a knife, if they thought ‘there was a common design to attack [W] but it [was] not proved, in the case of [M], that he had [had] any intention to kill or cause grievous bodily harm, but that [A], without the knowledge of [M], had a knife, took it from the flat and at some time formed the intention to kill or cause grievous bodily harm to [W] and did kill him – an act outside the common design to which [M] is proved to have been a party – then [they] would or could on the evidence find it proved that [A] committed murder and [M] would be liable to be convicted of manslaughter provided [they were] satisfied that he took part in the attack or fight with [W].’ 16
The Court of Appeal considered this to be a misdirection and quashed M’s conviction for manslaughter. Lord Parker CJ concluded that ‘to say that adventurers are guilty of manslaughter when one of them has departed completely from the concerted action of the common design and has suddenly formed an intent to kill and has used a weapon and acted in a way which no party to that common design could suspect is something which would revolt the conscience of people today.’ 17 He suggested that ‘[c]onsidered as a matter of causation there may well be an overwhelming supervening event which is of such a character that it will relegate into history matters [meaning the accessory’s actions] 18 which would otherwise be looked upon as causative factors.’ 19
Both passages were cited in Jogee. 20 This and subsequent citations in the Court of Appeal have been blamed for potentially misleading practitioners about the role of causation in the post-Jogee law of complicity and having given rise to appeals that appear to be premised on the (mis)understanding that OSA engages causation principles 21 when (save for procuring) no causal link is required for accessorial liability to arise in the first place. 22
OSA, Causation and Remoteness
Part of the problem may be linguistic. When different people speak in terms of ‘causation’, they do not necessarily mean the same thing. Causation is a context-driven concept. In a narrow sense the term refers to a continuing physical chain reaction from conduct to result. 23 As a legal concept, however, it is often used in a wider sense, which includes evaluative, normative or moral considerations of accountability, to refer to only those criminal events which trigger legal responsibility. 24
Although the academic literature tends to distinguish between ‘causation in fact’ and ‘causation in law’, in the case law there is often no sharp distinction drawn between (factual) causation on the one hand and (legal) attribution on the other, 25 and references to causation might encompass either or both of these ideas. This can lead to misunderstandings.
The description of OSA in Anderson and Morris as a ‘matter of causation’ has the potential to mislead insofar as it might be (mis)understood to suggest that the basis for OSA is a lack of enduring causal impact of the accessory’s contribution: because of what the perpetrator did, the accessory’s act of assistance or encouragement no longer carries through to the principal offence. 26 Or, put another way, the complicitous conduct has lost its causal potency because of the principal’s subsequent intervention. On such a reading of OSA, there would be an obvious tension with the restatement of the general principles of complicity in Jogee which are clear that accessorial liability does not require a causal connection to begin with (although some connecting link falling short of causation is clearly presupposed). 27
Historically, OSA might indeed have been linked to an understanding of complicity as underpinned by a causal relationship. 28 As KJM Smith has argued, ‘although not completely unambiguous, the burden of these comments [meaning the above cited passages from Anderson and Morris] is reasonably clear: actions performed by the principal in furtherance of the common purpose incriminate the accessory because his association with the principal is a cause of or a reason for the principal’s conduct or its criminal consequences. If this causal relationship were not essential it would be a logical irrelevance to exculpate the accessory on the grounds of lack of causation where the principal acted beyond the scope of the common purpose’. 29 However, it seems clear that, by the time of Jogee, the law had moved on in that Jogee and the post-Jogee case law do not see causation as a necessary requirement of accessorial liability.
Jogee leaves no room for doubt that ‘once encouragement or assistance is proved to have been given, the prosecution does not have to go so far as to prove that it had a positive effect on [the principal offender]’s conduct or on the outcome’. 30 Grant recently confirmed that this pre-empts any attempt for OSA to be approached ‘through the lens of causation’ 31 (which in the context of that case would have involved arguing that the appellant’s encouragement of the principal to engage in a face-to-face confrontation on foot had not had a positive effect on what, in fact, occurred, namely a hit and run attack by car).
Nonetheless, the fact that an accessory’s act of assistance or encouragement does not need to be causal for the principal’s commission of crime does not preclude asking the question whether the accessory’s help was so far removed from the ultimately committed offence that it would be unjust to impose liability on him: it is a non sequitur to argue that because causation is not required, there can be no limits placed on the accessory’s liability.
This raises the question as to what kind of limitation OSA constitutes: one that goes to the accessory’s assistance or encouragement, in the sense that it negates the actus reus of accessorial liability? Or is it rather a remoteness principle, 32 which, whilst perhaps conceding that there was a relevant act of assistance or encouragement to begin with, prevents attribution of the principal’s criminal act and its result to the accessory on the basis that it would be unfair to regard the accessory’s contribution still as material? Or is it a special defence that applies in complicity cases, similar to, but distinct from, the defence of withdrawal?
There is one statement in Jogee which could be seen to support the view that OSA forms part and parcel of the test for assistance and encouragement. 33 There are also dicta in the Court of Appeal decision in Tas that appear to associate OSA with the defence of withdrawal, 34 lending some credibility to the suggestion that OSA might be a special defence to complicity. However, considered in context it is my argument that OSA is best seen as a remoteness principle that seeks to limit accessorial liability for constructive crimes (of homicide) 35 on grounds of (a lack of) fair attribution. As such, I would argue that OSA is concerned not so much with delimiting what counts as assistance or encouragement in individual cases, but with restricting the range of conduct on the part of the principal offender and ensuing outcomes that can fairly be attributed to an accessory, in light of the risks the latter set by his own contribution. This view is supported by passages in Jogee that indicate that OSA was meant to take the place of the FDR. 36
OSA in Jogee
In Jogee, OSA is first mentioned in the context of a historical overview of the principles of accessorial liability. Thus, at para [12], the Supreme Court suggests that ‘there may be cases where anything said or done by D2 [the accessory] has faded to the point of mere background, or has been spent of all possible force by some overwhelming intervening occurrence by the time the offence was committed. Ultimately it is a question of fact and degree whether D2’s conduct was so distanced in time, place or circumstances from the conduct of D1 [the principal offender] that it would not be realistic to regard D1’s offence as encouraged or assisted by it.’ The language used here is clearly evocative of remoteness (‘faded to the point of mere background’; ‘distanced in time, place or circumstances’ etc).
Subsequently, the Supreme Court makes OSA a prominent, if somewhat under-explained, feature of its restatement of the principles of accessorial liability: ‘If a person is a party to a violent attack on another, without an intent to assist in the causing of death or really serious harm, but the violence escalates and results in death, he will be not guilty of murder but guilty of manslaughter. So also if he participates by encouragement or assistance in any other unlawful act which all sober and reasonable people would realise carried the risk of some harm (not necessarily serious) to another, and death in fact results (…). As the Court of Appeal held in R v Reid (Barry) 62 Cr App R 109, if a person goes out with armed companions to cause harm to another, any reasonable person would recognise that there is not only a risk of harm, but a risk of the violence escalating to the point at which serious harm or death may result. (…).’
37
‘The qualification to this (…) is that it is possible for death to be caused by some overwhelming supervening act by the perpetrator which nobody in the defendant’s shoes could have contemplated might happen and is of such a character as to relegate his acts to history; in that case the defendant will bear no criminal responsibility for the death’.
38
‘This type of case apart, there will normally be no occasion to consider the concept of “fundamental departure” as derived from R v English. What matters is whether D2 encouraged or assisted the crime, whether it be murder or some other offence. He need not encourage or assist a particular way of committing it, although he may sometimes do so. In particular, his intention to assist in a crime of violence is not determined only by whether he knows what kind of weapon D1 has in his possession’.
39
It is clear from the context, and the last quoted passage in particular, that the Supreme Court expected OSA to replace the FDR. Indeed, the quoted passage goes on to consider one issue that was frequently a problem under the FDR, namely the impact of a difference in the weapon used by the principal or in the manner the principal committed the crime, as well as the impact of a lack of knowledge on the part of the accessory about the weapon the principal used on the accessory’s liability: ‘The tendency which has developed in the application of the rule in Chan Wing-Siu to focus on what D2 knew of what weapon D1 was carrying can and should give way to an examination of whether D2 intended to assist in the crime charged. If that crime is murder, then the question is whether he intended to assist the intentional infliction of grievous bodily harm at least, which question will often, as set out above, be answered by asking simply whether he himself intended grievous bodily harm at least. Very often he may intend to assist in violence using whatever weapon may come to hand. In other cases he may think that D1 has an iron bar whereas he turns out to have a knife, but the difference may not at all affect his intention to assist, if necessary, in the causing of grievous bodily harm at least.’
40
This paragraph places the issue of fundamental difference (which has often focussed on a difference or lack of knowledge of the fatal weapon used) squarely within the enquiry as to the accessory’s mens rea. It has been suggested that, consequently, under such an approach, it does not matter by which means or in what manner the principal attacks the victim – provided principal offender and secondary party were agreed that the victim should be caused GBH, then the secondary party cannot escape liability by arguing that death was brought about with unknown means or in a manner fundamentally different to anything he had intended or foreseen. As Umar Azmeh has argued: ‘It is hard to find a sound justification, legal or theoretical, for the following plea: notwithstanding an original plan between individuals that intended at the very least to cause grievous bodily harm, in which the secondary party was an enthusiastic participant, as the method of attack that ultimately resulted in death was different to that original agreement, the secondary party ought to have the opportunity of availing themselves of an OSA direction leading to an acquittal.’ 41
However, as I will explain below, the conclusion that post-Jogee there remains no scope for claims of fundamental difference where principal and accessory were both agreed that the victim be caused GBH is unpersuasive. 42 It is flawed in so far as it is premised on an understanding that the FDR existed because of the foresight test in PAL, and that it therefore ceased to be relevant once Jogee abolished PAL and changed the mens rea for all types of complicity from foresight to intention.
It is my argument that the FDR existed to delimit the reach of constructive liability for accessories (to murder). 43 This need remains post-Jogee, for GBH is a very broad category of harm, ranging from serious or very serious but non-lethal to life-changing and life-threatening injuries. 44 The level of risk that an accessory sets by his acts of assistance or encouragement cannot be entirely disregarded where the harm caused by the principal’s actions manifests a commitment to a much greater level of risk than may be manifested by the accessory’s participation. OSA thus might have a role to play even where the evidence suggests that the accessory intended to assist or encourage the principal offender to cause GBH, and concomitantly, it can be justified to give an OSA direction in such circumstances.
Similarly, in the context of manslaughter, OSA should have a role where the principal inflicted GBH when the accessory intended to support at the most the infliction of some harm. Just as, in murder, the infliction of GBH can be more or less likely to lead to the victim’s death, in constructive manslaughter, different types of assault carry different levels of dangerousness. This may make for an even clearer case of why there is a need to delimit the accessory’s liability for crimes of constructive liability: while under Jogee, escalating violence can give rise to liability for unlawful act manslaughter for an accessory who intended no more than a common assault or some harm, surely not every act of escalation should be attributable to him, even when an escalation of violence was (objectively speaking) generally predictable. As Nathan Rasiah has put it: ‘arguably, if D2 encourages or assists an assault, but D1 embarks on a murderous armed attack, there is a question as to whether D2 has encouraged or assisted the attack that has resulted in death that merits evaluation by a jury.’ 45
Fundamental Difference pre-Jogee
When claims of fundamental difference first reached the House of Lords, in English, the discussion focussed primarily on the impact of the accessory’s lack of knowledge of the fatal weapon and its potential to inflict deadly injuries. 46 Thus, Lord Hutton suggested that ‘if the jury considered that the use of the knife by [P] was the use of a weapon and an action on [P]’s part which [S] did not foresee as a possibility, then [S] should not be convicted of murder. As the unforeseen use of the knife would take the killing outside the scope of the joint venture the jury should also have been directed, as the Court of Appeal held in Reg. v. Anderson, that [S] should not be found guilty of manslaughter.’ 47 Considering the differing circumstances in which the issue may arise, he thought it ‘undesirable to seek to formulate a more precise answer to the question in case such an answer might appear to prescribe too rigid a formula for use by trial judges.’ However, he offered the following observation: ‘if the weapon used by the primary party is different to, but as dangerous as, the weapon which the secondary party contemplated he might use, the secondary party should not escape liability for murder because of the difference in the weapon, for example, if he foresaw that the primary party might use a gun to kill and the latter used a knife to kill, or vice versa.’ He concluded that ‘in a case where, although the secondary party may have foreseen grievous bodily harm, he may not have foreseen the use of the weapon employed by the primary party or the manner in which the primary party acted, the trial judge should qualify the test of foresight […] in the manner stated by Lord Parker C.J. in […] Reg. v. Anderson; Reg. v. Morris.’ 48
Whilst Lord Hutton’s explanations left open the possibility that the FDR could be invoked where the principal had acted in a different manner, when the issue was revisited by the House of Lords in Rahman, Lord Brown restated the FDR in a way that, arguably, narrowed the principle down to require the use of an unknown and significantly different weapon:
49
‘If [S] realises (without agreeing to such conduct being used) that [P] may kill or intentionally inflict serious injury, but nevertheless continues to participate with [P] in the venture, that will amount to a sufficient mental element for [S] to be guilty of murder if [P], with the requisite intent, kills in the course of the venture unless (i) [P] suddenly produces and uses a weapon of which [S] knows nothing and which is more lethal than any weapon which [S] contemplates that [P] or any other participant may be carrying and (ii) for that reason [P]’s act is to be regarded as fundamentally different from anything foreseen by [S].’
50
Subsequent Court of Appeal decisions accorded the FDR a broader scope of application, however, in that they accepted that the principle could apply not just in the context of different and more dangerous (deadly) weapons, but also where the perpetrator had engaged in conduct that was more perilous than anything the accessory had expected.
Thus, in Mendez and Thompson, Toulson LJ (as he then was) endorsed the following account of the FDR put forward by Counsel: ‘In cases where the common purpose is not to kill but to cause serious harm, [S] is not liable for the murder of V if the direct cause of V’s death was a deliberate act by P which was of a kind (a) unforeseen by [S] and (b) likely to be altogether more life-threatening than acts of the kind intended or foreseen by [S].’ 51 He emphasised that ‘[w]hat matters is not simply the difference in weapon but the way in which it is likely to be used and the degree of injury which it is likely to cause.’ 52 The key criterion for operation of the FDR therefore seems to have been the dangerousness exhibited by the principal offender’s actions in comparison to the level of risk assumed by the accessory. The pre-Jogee law treated S as appreciating that V may be killed as a result of the attack up to a certain point, namely up to the point where the means or manner employed by P were materially different (as in more lethal) from any manner or weapon whose use S had foreseen.
It is important to appreciate that the FDR did not apply where perpetrator and accessory shared an intent that the victim should be killed. As Lord Rodger said in Rahman:
53
‘It can, in my view, make no difference if A and B agree to kill their victim by beating him to death with baseball bats, but in the course of the attack A pulls out a gun and shoots him. B must still be guilty of murder: since he intended to bring about the death of the victim, B cannot escape guilt on the ground that he did not foresee that A would kill him by using a gun instead of a baseball bat. The unforeseen nature of the weapon is immaterial. If, instead of a baseball bat, in the course of the attack A unforeseeably used an explosive and killed people in addition to the intended victim, then B would still be guilty of the murder of their intended victim — but not, I would think, of the murder of anyone else who was killed by the explosive.’
The reason for this is that it is not inapt or unfair on an accessory to be held to account for an outcome he intended to assist or encourage, even though he did not, and could not, foresee the precise way in which it came about. As Antje Du Bois-Pedain has argued: 54 ‘foreseeable consequences […] are attributable because they give the actor a reason not to engage in conduct that will foreseeably bring about [or, one might add, assist or encourage] a prohibited result. When, by contrast, the possibility of a particular conduct leading to a particular consequence is not one that the actor can be expected to perceive, then that gives him no reason not to perform the conduct, which is why it would be unfair to hold him responsible for it. But where D intends to bring about the very consequence that did occur (albeit that it was highly improbable, or came about only via a freaky causal sequence), he already has every reason he needs not to engage in the conduct: he shouldn’t try to bring about this prohibited consequence.’ 55
However, against the background of PAL the FDR assumed a wider role not exclusively focussed on constructive liability. 56 The Court of Appeal in Rahman had thought that ‘[n]othing in Powell and English suggests that a defendant who realised that one of the attackers might kill with intent to kill can enjoy the benefit of the [fundamental difference] rule. The liability of such a person is not “parasitic” […]. He realises that the crime of murder may be committed.’ 57 However, when the House of Lords revisited the issue on further appeal, it accepted that an accessory could still rely on the FDR even where he had foreseen that the principal offender might kill the victim with intent. In the words of Lord Brown: ‘[T]he act was the act of killing and the only question arising pursuant to the English qualification is whether the possibility of killing in that way (rather than in some fundamentally different way) was foreseen by the accessory—whether the act which caused the death was, as Sir Robin Cooke had put it in Chan Wing Siu, “of a type” foreseen by the secondary party.’ 58 This approach was subsequently confirmed by the Court of Appeal in Yemoh which found that, provided the defendant’s state of mind was less culpable than intending the victim to be killed, 59 the FDR was engaged if the defendant: a) realised that one of the attackers might kill with intent to kill or cause really serious bodily harm; or b) intended that really serious bodily harm would be caused; or c) realised that one of the attackers might cause really serious bodily harm with intent to cause such harm.’ 60
The above cases and dicta muddy the waters somewhat. If it is right that the FDR was concerned with the risk of death assumed by the accessory, it is difficult to see why it was relied upon to absolve an accessory of liability who foresaw the possibility of an intentional killing but did not foresee the precise way it was brought about. The explanation is likely to be the over-inclusivity of PAL: to hold an accessory liable for actions by the principal foreseen but not intended or authorised is in itself difficult to justify (and this led, of course, to the abolition of PAL in Jogee). It is not surprising, in these circumstances, that the courts would be looking for corrective mechanisms such as the FDR, thereby diluting its original rationale. The FDR thus assumed, at least on the surface, a new and different role: as a limit to PAL. This, in turn, explains why, in Jogee, the Supreme Court felt compelled to abolish it along with PAL, or at least to return it to its origins in the form of the overwhelming supervening act principle. It will be remembered that Anderson and Morris concerned unlawful act manslaughter and was decided before the common law had taken the ‘wrong turn’ in Chan Wing-Siu. 61 Taking the FDR back to its origins in that case thus essentially means that its original focus on constructive liability is restored. An accessory will now only be liable if he intended to assist or encourage the assault that was committed, and not merely because he foresaw it. The FDR is no longer needed to counteract the excessively wide scope of the foresight test, and this led the Supreme Court in Jogee to all but abolish it. But the abolition of PAL does not require the original role of the FDR, as a corrective in cases of constructive liability, to be diminished in any way. Some of the passages on the FDR in Jogee thus appear to overshoot the mark.
It might be objected that, when the mens rea for complicity is intention and no longer foresight, the maxim ‘intended consequences are never too remote’ should now lead to accessorial liability for fatal outcomes whenever principal and accessory were agreed that the victim should be caused GBH (since an intent to cause GBH suffices for murder). This made the Supreme Court assert that there normally is ‘no occasion to consider the concept of “fundamental departure”’ 62 and that what matters is the accessory’s intent to assist or encourage the intentional infliction of GBH at least. 63 Yet the objection overlooks the constructive nature of liability for murder: the accessory may intend the infliction of GBH, but not death. The GBH rule makes the perpetrator liable for (possibly) unintended consequences of his criminal conduct, but it is up to him to decide what risk of such consequences to set (by use of weapon, extent of violence employed etc). Where the principal departs from the agreed plan, using more serious violence, the accessory has no influence over that risk, and this is where the FDR, or now OSA, should come in.
If the UKSC meant to abolish the FDR in both its aspects (as a limit to PAL and a corrective to constructive liability), why did it leave scope for a principle of OSA? 64 It is suggested that it made perfect sense to abolish the FDR as a limit to PAL (given that PAL was being abolished), but not as a corrective to constructive liability of accessories. In the following, this argument will be explored in a little more detail.
Fundamental Difference in Jogee
In Jogee, the Supreme Court seems to have taken the view that the FDR was required against the background of PAL: 65 if all that was required to fix an accessory with liability for an offence committed by the principal was foresight that the principal might commit that crime, the law needed a robust corrective to limit that very wide principle. Foresight was meant to be sufficient for liability up to a point, that point being where the offence was committed by means that were precisely not foreseeable.
While it is understandable that the FDR was thus relied on as a corrective to PAL, using it in this way was never entirely logical: if the test required for PAL was foresight, and death was foreseen, surely the way that death was brought about should not have mattered, far less absolved the accessory of liability? This function of the FDR was thus more based on policy than logic: PAL was too wide and needed to be kept in (at least some) bounds. The unfortunate effect of relying on the FDR in this way, was, I would suggest, to obscure its original point, namely that the mens rea for murder is constructive: if a perpetrator intends to inflict serious harm, we treat him, by way of legal fiction, as if he intended to kill.
Yet while the question whether or not the victim has been killed is binary (he is either dead or he is not), there is a sliding scale of seriousness of injuries that an accessory might have foreseen or intended, and in that context the manner in which those injuries are to be brought about assumes primary importance. So, prior to Jogee, an accessory who could foresee that the principal might inflict serious harm on a victim, but nevertheless went along with the criminal enterprise, might escape liability if, rather than using the (foreseeable) knife or baseball bat, the principal used a machine gun in inflicting the (unsurprisingly) fatal injuries on the victim. The abolition in Jogee of PAL does nothing, however, to address this problem. All it does is replace the language of foresight with the language of intention. An accessory might intend, rather than merely foresee, the infliction of serious harm by way of a knife or a baseball bat, and might, rightly, be convicted of being an accessory to murder if that harm results in death. It is not at all obvious, however, that he should also be so liable if the principal instead uses a machine gun or rocket launcher. This is because the risk of death is lower where GBH is inflicted by knives than where it is inflicted by a machine gun or rocket launcher. The principal sets that risk by his choice of weapon and is rightly held liable even where his intention was merely to inflict serious harm; yet the accessory, who anticipated the use of a less lethal weapon, sets a lower risk of death, notwithstanding his proven intention that the victim be caused serious harm. The FDR used to reflect this, though its use to limit the otherwise unchecked widening of accessorial liability by PAL sometimes obscured this. Stripped of that function, for which it was never logically suited, the principle, whether we call it FDR or OSA, continues to be focussed on that original rationale.
Fundamental Difference post-Jogee
The point I am making is that the insistence on an accessory’s intention in Jogee and the abolition of PAL did nothing to solve the puzzle of how to deal with accessories to constructive crimes such as murder or unlawful act manslaughter. The Supreme Court in Jogee does not appear to have appreciated this, and neither does the Court of Appeal in subsequent cases: in Grant, 66 the Court of Appeal is adamant that the FDR was more generous towards accessories than the OSA test that was meant to replace the FDR in Jogee. The language used by the Supreme Court in Jogee (repeated, mantra-like, by the Court of Appeal in Grant), however, is not distinguished by its clarity or precision, speaking in terms of so far removed, ‘in time, place or circumstances’, from the ultimate offence so as to relegate the accessory’s involvement to history. On one view, OSA is a very demanding remoteness principle: mere lack of foreseeability will not do, what is required is a radical fissure between the accessory’s involvement and the ensuing offence. This could make some sense, particularly against the background that (procuring aside) causation need not be established for accessorial liability: with only some link or connection, short of but-for causation, required, it makes sense to mirror the same relaxation of general principles when it comes to remoteness. This approach ignores, however, that OSA was meant to replace the FDR. 67
Unfortunately, in Grant, the Court of Appeal read various obiter passages in Jogee as if they formed part of a statute, thus missing the chance to put OSA back on a secure footing. The court considered Counsel’s example of a team of assassins setting out to kill a victim. They envisaged that the victim would be killed by a gun, but in the event the killing is brought about by one of them using explosives. In such a case, Counsel had submitted to the trial judge, an OSA direction ought to be given. The Court of Appeal disagreed, but did not explain precisely why, other than to repeat the dicta from Jogee that seem to lay down a very narrow definition of OSA. It failed to point to the crucial factor that would have justified not letting OSA go to the jury in this hypothetical, namely that principal and accessory shared an intention to kill. Where their intention falls short of this, and where the accessory merely intends that the victim be caused serious harm, the weapon chosen by the principal matters, or at least it should.
One reason why the Court of Appeal rejects this argument (that it should matter whether there is an intention to kill or merely to inflict GBH) is the perceived danger of accessories ‘talking their way out’ of liability in cases of escalating violence. Where a victim is killed by a stab wound, other participants should not be allowed to argue that all they contemplated was the use of fists. I would argue that this danger is overstated. As the Court of Appeal recently stressed in another case, Lanning and Camille, 68 it is important to consider the extent to which outcomes can be attributed to accessories against the relevant factual matrix.
The incident in Lanning and Camille started with an exchange of words between L and V1 and V2 who were standing on the opposite platform from L. L crossed over to their platform. On his way there he was joined by C. C was unarmed, but L proceeded to brandish a sheathed knife towards V1. The argument evolved into two fights: one involving L and V1, the other C and V2. Whilst C and V2 exchanged punches and kicks, V1 was fatally wounded by L’s knife.
L admitted that he had unlawfully killed V1 but denied he had intended to kill or cause serious harm. On his account, he had been wielding the knife in order to scare V1 and the fatal stabbing had occurred when he was holding V1 in a bear hug and V1 had turned onto the blade during their struggle. C was accused of having encouraged or assisted L to attack V1, intending that V1 would be killed or caused really serious harm. Alternatively, the case against him was that if he had encouraged or assisted L to attack V1 but had not intended that V1 would be killed or caused really serious bodily harm, he was guilty of manslaughter. C was convicted of manslaughter. He appealed on the basis that the judge had erred in not directing the jury that they could acquit him of manslaughter if the actions of L in killing V1 amounted to an overwhelming supervening act.
It was submitted on C’s behalf that Jogee did not preclude the possibility of the judge leaving OSA to the jury in all cases in which a knife was used during an escalation of violence during a fight. In rejecting C’s submissions, the court followed the distinction in Tas between acts that constitute a ‘mere escalation which remained part of the joint enterprise’ 69 and, therefore, attributable to the accessory, and acts that amount to an OSA (because these were, one might add, in effect an excess rather than an escalation). The court agreed with the trial judge’s conclusion that if ‘one or more of the perpetrators are unaware that one of their number has produced and then used a knife, that would not constitute an OSA’ because ‘once you have agreed to take part in an unlawful assault, it is an escalation of the violence to which you have signed up to and that is one of the perils of signing up’. 70
Surely, though, an accessory does not sign up to each and every possible escalation. Even where he is reconciled to the risk of serious violence (which the appellant in Camille was not, but the appellant in Grant was), the extent and nature of that risk cannot be wholly irrelevant. Would Camille have been decided the same way if L had produced a machete rather than a knife? Or a machine gun? Or a rocket launcher? The level of risk that infliction of what amounts to actual or grievous bodily harm will actually result in death differs depending on the nature of the weapon involved and how it is being used. The FDR used to take this into account, and OSA would do well to do the same.
Yet how can this be done without ending up with the kind of intricate weapons-focussed litigation that bedevilled the FDR and the pre-Jogee case law? Here, Camille might point the way: a significant feature of the court’s approach is the distinction it draws between the uncontemplated but commonplace and the uncontemplated and extraordinary. The court’s decision (that the trial judge was entitled to conclude that there was an insufficient factual basis for OSA to be left to the jury) and reasoning place a lot of reliance on the fact that, notwithstanding C’s personal lack of knowledge that L carried a knife, the weapon was one that is frequently present and used in incidents of a similar nature. As the court explained, ‘in today’s social climate’ where ‘knives are produced in situations of this kind with a high degree of frequency leading to serious injury or death’, 71 it did not follow that simply because C was unaware that L carried a knife ‘nobody in the defendant’s shoes could have contemplated’ 72 a stabbing might happen. This clarifies that OSA looks to the wholly unexpected, not from the defendant’s point of view (as the FDR did), but ultimately from the perspective of general life experience (which is then attributed to someone in the defendant’s position). And even then, the wholly unexpected needs to be so momentous as to consign the accessory’s deeds to history – a taxing two-limbed test.
Lanning and Camille accepts that setting out en groupe to inflict street violence does, unfortunately, nowadays almost guarantee the use of knives, and an accessory is going to find it extremely hard to argue that, in setting out to inflict really serious harm by fists and feet, he discounted the risk of knives being used. It is implicit in the Court of Appeal’s judgment in Lanning and Camille, however, that had the principal used a rocket launcher or a lethal nerve poison, the accessories would have stood a better chance of getting off the hook.
This is a sensible stance to take, and it is unfortunate that the Court of Appeal in the subsequent case of Grant did not engage with it when it insisted that OSA did not apply where, instead of being subjected to serious violence by way of a person-to-person fight, the victim was run over with a car. On those facts, it is at least arguable that there was a material departure from the original plan such as might impact on an accessory’s liability for the fatal outcome. At the least, it would have been prudent to put the issue in front of the jury and to let them consider whether what had occurred had been (objectively) unexpected and extraordinary, thus meeting the Supreme Court’s test of whether the death was ‘caused by some overwhelming supervening act by the perpetrator which nobody in the defendant’s shoes could have contemplated might happen and is of such a character as to relegate his acts to history.’ 73
It is telling that the Court in Grant admits that it cannot readily think of an example where OSA would actually apply. This shows how incredibly high the bar has (apparently) become for OSA to be even left to the jury. This cannot be what the Supreme Court had in mind. In Jogee, there are clear indications that the OSA principle was concerned with fair attribution and that this was a proper question to be left to the jury: ‘Ultimately it is a question of fact and degree whether D2’s conduct was so distanced in time, place or circumstances from the conduct of D1 that it would not be realistic to regard D1’s offence as encouraged or assisted by it.’ 74
OSA and Fair Attribution Principles
The OSA test is underpinned by fair attribution principles. As Lord Toulson, one of the authors of the judgment in Jogee, had previously observed extra-judicially: ‘Questions of remoteness are inextricably interwoven with a sense of what is fair’. 75 He thought that it was ‘for the jury, applying their common sense and sense of fairness, to decide whether the prosecution have proved to their satisfaction on the particular facts that P’s act was done with D’s assistance and encouragement.’ 76 Lord Toulson did not have any reservations about the jury’s ability to determine this issue: ‘It is inevitable in practice, and I do not see it as objectionable, that a jury’s decision whether P’s act was done with D’s assistance or encouragement should involve their judgment whether it is a just conclusion on the facts. […] Aside from the mental component of secondary liability, in such cases the approach to causation is inevitably influenced by a moral component whether it is just to consider the defendant culpable for what occurred.’ 77
However, given the number of appeals that have raised this issue to date, it is debateable whether juries might not need more guidance on the issue and trial judges a clearer sense of when the principles of OSA are potentially engaged and to be left to the jury.
As I have pointed out above, the OSA principle, as the FDR before it, is best regarded as a corrective in cases where it would not be fair to attribute the consequences of the principal’s actions to the accessory. But where is the line to be drawn? In particular, should the rule focus on the accessory’s subjective perception of the relevant risks (as the FDR did), or should an objective test be applied (which is what OSA seems to be doing)? In answering this question, it is useful to reconsider the underlying basis of constructive liability in English law. We are mainly concerned with two types of homicide here: murder and unlawful dangerous act manslaughter.
The mens rea for murder is intention to cause the victim really serious harm. In contrast to the rules that obtain in many other jurisdictions, it is not necessary for the prosecution to prove that death resulted from the defendant’s negligence: 78 in that sense, liability is strict in English law. The defendant cannot say, for example, that he did not have the requisite medical knowledge to appreciate that stabbing a person in the thigh can puncture the femoral artery and result in his death.
Similarly, the law of unlawful act manslaughter focuses on the objective dangerousness of the unlawful act in question 79 – again, the defendant cannot say that he did not appreciate the danger (of personal harm). But where the act is not objectively dangerous, and nevertheless results in death, the defendant will be absolved of liability for the fatal outcome.
Where an accessory is involved, things become more complicated. Where the principal does not depart from the plan, both principal and accessory share in the risk that the outcome might be fatal. But from the accessory’s point of view, there is also the risk that the principal might depart from the plan, rendering the risk of a fatal outcome greater. Where that risk is foreseeable, holding the accessory liable for the fatal outcome can be justified. Not so, however, where it is not, where the principal commits, in other words, an ‘overwhelming supervening act’. Thus, in setting out for a fight with members of an opposing group, all involved should, this is, unfortunately, the culture currently prevailing in our inner cities, realise that there is a risk that knives will be used and that the use of knives might kill. OSA only comes into play where the risk of an escalation was not and could not have been appreciated by someone in the accessory’s shoes – again, the example of a group member getting a machine gun or rocket launcher out of the boot of his car is useful here. That risk was not and should not have been appreciated by the accessory and, notwithstanding he shared the intention to inflict really serious harm on the opposing group members, the accessory should not be held liable for the materialisation of that risk. In such a case, the outcome is simply not a manifestation of a risk that he set by his own contribution.
OSA in Context: Case Law Examples
We have seen that what the court (when considering OSA) examines is not a matter of causation (since no factual causation is required in the first place) but a matter of fair attribution. We have further seen that the concept of OSA is designed to delegate the assessment of fair attributability to the jury. However, the current test can be criticised for its vagueness: to date, while the Court of Appeal has repeatedly found that certain acts by the principal did not engage OSA principles, its jurisprudence has been unable to express with clarity what feature of unusual or unexpected acts ought to affect the jury’s sense of justice such that it absolves the accessory of liability for a fatal outcome. The only conclusion that can be drawn with certainty from the case law to date is that what is, and what is not, fairly attributable is subject to social shifts: as explained above, the post-Jogee case law suggests that the objective foreseeability or commonality of an act is a highly influential factor in determining that an act will not be treated as an OSA, and that what is considered to be foreseeable or sufficiently commonplace can change over time. Indeed, recent cases suggest that some acts and harms that were seen as non-attributable under the FDR would today be seen as attributable, as I will now explain with reference to three ‘fundamental difference’ cases (two from the pre- and one from the post-Jogee periods).
a) OSA and Gamble
The Northern Irish case of Gamble 80 was frequently discussed in the pre-Jogee case law and academic literature. As explained below in more detail, the case attracted an application of the FDR which absolved two accomplices that had been charged with murder from liability for the victim’s death. In the post-Jogee case of Grant, the Court of Appeal recently suggested that were the defendants in Gamble to be tried in England and Wales today, they would be guilty of murder and unable to rely on the concept of OSA.
The four accused in Gamble were all members of a terrorist organisation who had a grievance against another man as they suspected he had given information about their activities to the police. They decided to inflict punishment on him. Two of them, Douglas and McKee, expected that their victim would be subjected to a severe punishment beating or ‘kneecapping’ (involving the firing of a bullet into his kneecap). During the attack the victim was, however, murdered by the other two accused who cut his throat with a knife; they also shot him (though not in the knee), and two of the bullet wounds would have been fatal had his death not been caused by the cutting of his throat. Douglas and McKee had not foreseen that the victim would be killed with a knife or by firing of bullets into a vital part of his body. The prosecution argued that the joint enterprise of committing GBH, combined with the GBH-rule in murder, was sufficient to convict them of murder, notwithstanding that they had not foreseen the acts which actually caused death.
Carswell J (Gamble having been a non-jury trial) rejected this argument, however, and Douglas and McKee were acquitted of murder (but convicted of wounding with intent to cause the victim GBH). The judge put forward the following explanation: ‘When an assailant “kneecaps” his victim (…) there must always be the risk that it will go wrong and that an artery may be severed or the limb may be so damaged that gangrene sets in, both potentially fatal complications. It has to be said, however, that such cases must be very rare among victims of what is an abhorrent and disturbingly frequent crime. Persons who take a part in inflicting injuries of this nature no doubt do not generally expect that they will endanger life, and I should be willing to believe that in most cases they believe that they are engaged in a lesser offence than murder. The infliction of grievous bodily harm came within the contemplation of Douglas and McKee, and they might therefore be regarded as having placed themselves within the ambit of life-threatening conduct. It may further be said that they must be taken to have had within their contemplation the possibility that life might be put at risk. The issue is whether it follows as a consequence that they cannot be heard to say that the murder was a different crime from the attack which they contemplated, and so cannot escape liability for the murder on the ground that it was outside the common design. To accept this type of reasoning would be to fix an accessory with consequences of his acts which he did not foresee and did not desire or intend. (…) Although the rule remains well entrenched that an intention to inflict grievous bodily harm qualifies as the mens rea of murder, it is not in my opinion necessary to apply it in such a way as to fix an accessory with liability for a consequence which he did not intend and which stems from an act which he did not have within his contemplation. I do not think that the state of the law compels me to reach such a conclusion, and it would not in my judgment accord with the public sense of what is just and fitting.’ 81
Gamble has proved a contentious decision that has attracted much debate, 82 often focussing on whether it was the difference in the perpetrators’ intentions that absolved the defendants of liability for murder (the perpetrators having acted with intent to kill when the defendants had expected the victim to be attacked with intent to cause him GBH) or whether they were acquitted because they had not contemplated the specific act or weapon used, i.e. the fatal use of a knife as opposed to a (non-fatal) punishment beating or ‘kneecapping’. To the extent that discussions of Gamble are focussed on the change in the perpetrators’ mental state, I would argue that the focus is on the wrong parties’ mens rea: it cannot matter to the accessories’ liability whether the principals acted deliberately with intent to kill (rather than with intent to inflict GBH). 83 What matters is the accessories’ mens rea. Clearly, where the accessory intends to encourage a killing, any appeal to OSA would be misplaced, however differently death is brought about to what the accessory envisaged. OSA only becomes relevant where the (accessory’s) intention was to inflict GBH short of death, and here the means employed are, I would suggest, crucial.
In Grant, by contrast, the court concluded that there would be no OSA in Gamble because the defendants had participated with intention that the victim be inflicted GBH, and since that satisfied the mens rea of murder, the deliberate killing of the victim was not very different in kind from what they had contemplated: ‘on a charge of murder, if the accessory intentionally assisted or encouraged the perpetrator and intended that the perpetrator should cause grievous bodily harm with intent, he or she will have satisfied the elements of the offence of murder. The precise manner in which the victim happens to be killed and whether the perpetrator intended to kill as opposed to inflict really serious harm are by the way, so long as the encouragement or assistance of the accessory has not been “relegated to history”’. 84
However, this type of reasoning fails to appreciate what OSA is meant to achieve. Specifically, it fails to take into account that GBH is a very broad category and that the risks assumed by the accessory can vastly differ from the risks actually set by the perpetrator. As argued above, where an accessory assists or encourages GBH, the degree of GBH intended cannot be wholly ignored. Contrary to what the Court of Appeal suggests there are good reasons to think that Gamble is precisely the kind of case where OSA ought to be left to the jury: the seriousness of the injury and associated risk of death are substantially higher where the victim’s throat is cut than where he is (repeatedly) shot in the knee (even assuming that the latter could potentially be followed by life-threatening complications).
The focus in Jogee on intention (to do GBH) does not absolve of the need to enquire whether the GBH expected was roughly comparable to the GBH done. As Lord Bingham (discussing Gamble in Rahman) 85 has observed: ‘the violence in fact inflicted with the knife was of an entirely different character in an entirely different context from that which [the defendants] had foreseen and, in that sense, bargained for.’ ‘Bargained for’ is an apt way to describe an accessory’s intention, and whilst Jogee was right to make this the centre piece of accessorial liability, the constructive nature of murder and manslaughter can still lead to over-inclusiveness, so that OSA is needed as a safety valve.
It is true that the ‘kneecapping’ the defendants in Gamble expected to be involved with could possibly also have resulted in death. However, this possibility became all but certain to eventuate once it was replaced by throat-slitting, and this surely must make a difference. On the argument put forward here, Gamble should today be decided in exactly the same way it was. In any event, the issue should be left to the jury (Gamble having been the decision of a Diplock court, this did not arise in the actual case, of course).
b) OSA and English
In contrast, a fundamental difference scenario that would almost certainly be decided differently today is English, 86 one of the key cases where the FDR was applied to exonerate the accomplice for the victim’s death: P and S had jointly attacked a police officer, V. S knew that P might intentionally cause GBH to V with a wooden post. In the event, however, P killed V with a knife. At trial, S alleged that he did not know that P was carrying the knife. Nonetheless S was convicted of V’s murder. On appeal, his conviction was quashed. The jury had been instructed on PAL as set out in Hyde: 87 ‘[i]f B realises (without agreeing to such conduct being used) that A may kill or intentionally inflict serious injury, but nevertheless continues to participate with A in the venture, that will amount to a sufficient mental element for B to be guilty of murder if A, with the requisite intent, kills in the course of the venture.’ The jury had not, however, been directed to consider whether the killing with a knife was an act of a fundamentally different kind to any act foreseen by the appellant. Relying on the aforementioned passage in Anderson and Morris (that ‘to say that adventurers are guilty of manslaughter when one of them has departed completely from the concerted action of the common design and has suddenly formed an intent to kill and has used a weapon and acted in a way which no party to that common design could suspect is something which would revolt the conscience of people today’), 88 Lord Hutton concluded that in English ‘the unforeseen use of the knife would take the killing outside the scope of the joint venture’. 89
It is unlikely that a serious act of violence which progressed from a fight with blunt instruments into a knife crime would still be regarded as a supervening act under OSA. As the Court of Appeal observed in Lanning and Camille, 90 ‘in today’s social climate’ where ‘knives are produced in situations of this kind with a high degree of frequency leading to serious injury or death’, it did not follow that simply because Camille was unaware that Lanning carried a knife ‘nobody in the defendant’s shoes could have contemplated’ a stabbing might happen.
It follows that the accomplice in English would have a hard time persuading the court that the foreseeable risks excluded a stabbing, given how common knife crime seems to have become. And this, I would argue, is defensible, and in line with the ‘risk setting’ rationale explored earlier in this paper: an OSA will only prevent legal attribution if its occurrence is extraordinary rather than an (objectively) predictable intervention, even if it was not foreseen by the defendant himself.
c) OSA and Tas
The conclusion that English would today be decided differently is supported by Tas, 91 albeit in the context of unlawful dangerous act manslaughter. The case concerned another (post-Jogee) incident of group violence that had escalated into a fatal stabbing.
The appellant, T, had gone out with others to commit what looked likely to be an assault from the outset. He sought to escape his manslaughter conviction by arguing that he did not know of the knife the principal used to kill the victim. Lack of knowledge of the knife, he argued, should relieve him of liability for manslaughter even if he had intended that some harm, though not serious harm, be inflicted on the victim. 92 The use of a hidden knife, when a ‘punch up’ or argument had been anticipated, could constitute a supervening event. 93
The Court of Appeal did not accept this argument. Referring at length to the relevant passages in Jogee, it pointed out that the focus on intention reduces ‘the significance of knowledge of the weapon so that it impacts as evidence (albeit very important if not potentially irresistible) going to proof of intention, rather than being a pre-requisite of liability for murder.’ 94 If it was not necessary that the accessory knew of the weapon in order to bring home a charge of murder such a requirement (of knowledge of the weapon) should not be re-introduced through the supervening event concept for manslaughter. 95
The appellant remained liable for manslaughter because in the Court’s view the conduct which caused death was a foreseeable escalation, in that, even though the defendant might not have been aware of the presence of the knife with which the victim was killed, he had taken ‘the risk that the others involved (…) would go further than to inflict “some harm”’. 96 In the Court’s view, lack of knowledge of a weapon did not transform the incident from a simple (foreseeable) escalation to a supervening act. The Court supported this by drawing an analogy with situations where a group of people go out unarmed but where weapons of some sort (eg bottles) are then found on the ground. The Court did not think it mattered whether a weapon was there all along or found in the middle of an altercation. If an escalation was objectively foreseeable, it makes no difference whether it was achieved with whatever weapon came to hand in the heat of the moment or with weapons that were brought along and remained concealed up to the point of escalation.
The result in Tas can clearly be supported. In the circumstances of the case (which involved one group of students going in search of another group in response to one of the latter group having earlier in the day glared angrily at one of the former) it was probably foreseeable that the principal offender, or one of his group, would be carrying a knife. The facts bear all the hallmarks of a group going out in deliberate pursuit of a confrontation. It does not follow that simply because the defendant was unaware of the knife ‘no one in the defendant’s shoes could have contemplated’ a stabbing might happen, given the prevalence of knife crime.
If OSA is about the risks that an accessory is deemed to have assumed by his intentional support of the principal offender, then the risk set by the accessory includes risks that were objectively predictable. Arguably, an escalation of violence into knife crime is now a risk that people who participate in group assaults ought to have ‘on their radar’, wherefore it will not be inappropriate or unfair to hold them to account for the consequences of such actions.
Instructing Juries
To date, the Court of Appeal’s attempts to explain the circumstances in which an unusual or unexpected act by the principal offender ought to affect the jury’s sense of justice in such a way as to absolve the accessory of liability for a fatal outcome have not been distinguished by their clarity. Perhaps it might help to instruct jurors more explicitly in terms of the risk objectively set by the accessory’s assistance or encouragement. The test set out by Carswell J (as he then was) in Gamble might be useful here, at least in murder cases, though it may require some adjustment so as to reflect the objective nature of the risk set by the secondary party. His Lordship said: ‘Although the rule remains well entrenched that an intention to inflict grievous bodily harm qualifies as the mens rea of murder, it is not in my opinion necessary to apply it in such a way as to fix an accessory with liability for a consequence which he did not intend and which stems from an act which he did not have within his contemplation.’
97
The jury should thus ask itself:
Did the accessory intend to assist or encourage the principal offender to commit an assault with the intent to inflict GBH on the victim? Was the assault committed by the principal of the type of attack that the accessory did or that somebody in the accessory’s shoes should have contemplated might happen?
If the answer to the second question is ‘no’, then the jury should only find the accessory guilty of having assisted and encouraged the type of assault that should have been contemplated, in other words, the accessory should not fairly have the lethal consequences of the assault attributed to him, given that this resulted from an act the risk of which he did not set or support.
OSA’s Scope of Application
A final point to make about OSA concerns its scope of application. As the successor to the FDR (the key cases of which primarily concerned murder) and mindful of its origins in Anderson and Morris (a manslaughter appeal), one would expect OSA to be applicable in constructive murder as well as manslaughter. Latterly, this has become the subject of some controversy, however, with some commentators suggesting that OSA may be limited to constructive manslaughter, as it was in this context that OSA was put forward as a qualification to the principles of accessorial liability in Jogee. 98 The point can now be taken to have been settled by Grant which considered OSA in the context of a murder conviction. Were OSA not applicable to murder, the Court could simply have said so and dismissed of the appeal on that basis. In any event, as I have sought to demonstrate throughout this paper, OSA, as a limiting device for constructive liability in complicity, can only serve its function if it is applicable to both constructive murder and constructive manslaughter.
Conclusion
The liability-limiting concept of OSA is both the precursor to and successor of the FDR. In Jogee the Supreme Court considered that the FDR’s role was to keep the foresight test in PAL cases in check, by absolving accessories of liability for homicide, where the principal had caused the victim’s death with a different and more lethal weapon or in a different and more dangerous manner than had been foreseen by the accessory. Therefore, and in a misguided attempt to appease the defenders of PAL, the Supreme Court declared the FDR redundant along with PAL. I have argued that, in doing so, the Court under-appreciated the role the FDR used to play as a corrective to the GBH rule in murder and in circumscribing the dangerousness requirement in constructive manslaughter.
As the FDR before it, OSA is applicable to constructive murder as well as constructive manslaughter (and this is supported by recent cases in the Court of Appeal). In deciding whether OSA should be left to the jury, this paper has advocated that greater emphasis should be placed on the need to keep constructive liability within reasonable bounds. The courts simply must not lose sight of the fact that OSA has replaced the FDR and, as such, has taken over its function as a corrective to prevent overreach of constructive liability in complicity. The need to delimit constructive liability for secondary parties in murder and manslaughter cases survives the abolition of PAL. In fact, as I have sought to demonstrate, OSA was not even fashioned to counter-act PAL in the first place; the need to delimit constructive liability in homicide and complicity both pre- and post-dates the FDR.
I have highlighted that it is important to stress that what the court (when considering OSA) examines is not a matter of causation (since no factual causation is required in the first place) but a matter of attribution, lest OSA be conflated with mens rea analysis or causation. Notwithstanding its historic inception as (possibly) a ‘matter of [factual] causation’, OSA is not to be approached as a causation principle. It rather is a defensive claim going to remoteness. The post-Jogee case law can be criticised, however, for having set an impossibly high hurdle for when this remoteness principle is engaged. Part of the problem is that the language of ‘overwhelming supervening act’ obscures what is in fact an evaluative question of whether the perpetrator’s act and resulting outcome can be fairly attributed to the accessory. In Jogee, the test was designed to delegate the assessment of fair attributability to the jury. However, it can be criticised on grounds of vagueness: to date, the Court of Appeal has been unable to express with clarity what kinds of unusual or unexpected acts ought to affect the jury’s sense of justice such that it absolves the accessory of liability for a fatal outcome. What’s more, the Court of Appeal has been unable to say even when the test ought to be left to the jury. The threshold for the principle to be engaged is a very high one.
The challenge now facing the courts is two-fold: first, to come up with appropriate criteria to delimit the criminal liability of accessories for results that were so far from the ordinary or expectable as to leave doubt about the justice of imposing such liability on a secondary party, and, secondly, to define the extent of such limitation in a way that enables consistent application.
Finally, although there is no easy way to prove this empirically, judging by anecdotal evidence and the number of appeals that have been based to date on a trial judge’s reluctance to leave OSA to the jury, the concept seems to be too rarely brought to the attention of the jury. Mindful of its purpose as a safety valve for constructive liability (in homicide), and the trust that the authors of Jogee must have had in the jury’s ability to come to just and defensible conclusions when considering questions of fair attribution, trial judges should be more willing to leave the issue to the jury.