Abstract
Keywords
Facts
The appellant Rowe and three others were tried for joint enterprise murder. It was the prosecution's case that they had participated in a planned shooting to avenge the killing of one of their associates by a rival gang. The shooting resulted in the death of an innocent member of the public. He had been mistaken for a member of the rival gang and had been followed and then confronted by the four accused, one of whom shot him in the face.
The judge directed the jury that they had the alternative of returning a manslaughter verdict. Rowe was acquitted of murder but convicted of manslaughter. His co-accused were convicted of murder. Rowe and two of the others appealed their convictions.
Rowe's appeal was primarily based on there having been no proper basis for a verdict of manslaughter. Where the joint plan had been described as a shooting rather than an attack, it was impossible to conceive of circumstances in which an accessory could be not guilty of murder but guilty of manslaughter. This was because it would not be logically possible for someone to intend to shoot someone (at close range, and not merely to shoot at them) and, at the same time, intend to cause anything less than serious bodily harm.
Rowe's fellow appellants sought inter alia to rely on the argument (which Rowe adopted) that the judge was wrong not to direct the jury that encouragement or assistance must have had some effect on the relevant events to establish criminal liability. An accessory could not be held criminally responsible for a homicide to which he had not contributed. There needed to be at least some causal link between the accessory's encouragement and the principal's offence, albeit not a sine qua non. They submitted that there was a tension in Jogee [2016] UKSC 8 between the first part of [12] and the last two sentences of that paragraph. Whilst the first part suggested that once assistance or encouragement had been rendered, this did not need to have had any effect on the principal or his commission of the offence, the court's explanation of overwhelming supervening act (OSA) was concerned with a break in the chain of causation. They argued that it was therefore essential for the prosecution to show that the encouragement or assistance had had some effect upon the course of events relating to the principal offence.
Held, allowing Rowe's appeal against conviction (and dismissing the other appellants’ appeals), that the way in which the questions [for the jury] had been ultimately framed provided neither a sensible nor a secure route to a manslaughter verdict on the facts; Rowe's conviction for manslaughter could not be regarded as safe and needed to be quashed. Once the planned act which the secondary party encouraged or assisted had been characterised as a shooting, it was difficult to see that there could be circumstances in which a secondary party could say that the harm intended was some harm, but not really serious harm. Care should be taken to ensure that the natural desire to pose a series of simple questions did not override the imperative that the questions should, where necessary, be tailored to the individual circumstances of each defendant.
There was no tension within [12] of Jogee. The first part of [12], read together with [8] and [11], dealt with the conduct element of an accessory's liability for assistance or encouragement. But the last part of [12] dealt with an exclusion from that aspect of liability, namely an overwhelming supervening act. There was no logical reason why the legal test for such an exclusion needed to be treated as being relevant to the tests for establishing the conduct element of criminal liability as an accessory.
An overwhelming supervening act was no more than an exclusion. It was important not to abbreviate the relevant test. The focus initially was on whether nobody (not simply the defendant) standing in the defendant's shoes could have contemplated that the act in question might happen. That part of the test did not ask whether there was a causal link between the defendant's conduct and the principal's act. The second part of the test was whether the accessory's assistance or encouragement had ‘faded to the point of mere background’ or had become ‘spent of all possible force’ by an overwhelming supervening act (see Jogee at [12]). That language described conduct on the part of an accessory which had become irrelevant … or had become incapable of encouraging or assisting the commission of the offence by the principal. In such circumstances an accessory's encouragement of or assistance to the principal had ceased. The court concluded that this did not ‘make causation a test for satisfying the conduct element in the first place’ (at [136]).
Commentary
Rowe illustrates the difficulties in characterising a joint criminal enterprise and the implications of the characterisation exercise for the jury's route to verdict. The decision suggests that it can favour the defence to have a criminal incident described in narrow terms, especially where, as in Rowe's case, doing so effectively results in an all-or-nothing route to liability that might be difficult to prove against the defendant on the facts. In Rowe, characterising the principal offence as a revenge shooting rather than a revenge attack called for proof that all participants had assisted or encouraged the principal (to act) with an intent that the victim be caused GBH at least. The Court was surely right in concluding that characterising the incident in this way was incompatible with an intent that the victim be caused less than grievous bodily harm: it is indeed hard to imagine circumstances where an accessory assists or encourages the principal to shoot the victim at close range but without an expectation that doing so would result in serious injury.
By contrast, it will often be advantageous for the prosecution to insist on a broader description of the criminal incident. Rowe brings this into sharp focus: had the route to verdict required the jury to consider whether Rowe had participated in an unlawful attack, the manslaughter verdict would have been unassailable. This is because, as a matter of law, an unspecified attack is clearly compatible with an intent that less than serious injury be caused to the victim. Attacks can differ in scope and seriousness; the level of harm and endangerment participants can expect is not pre-empted by the label attached to the incident.
The issue of characterisation and its implications were not addressed in Jogee. In fact, the re-statement of the principles of accessory liability and the examples used by the Supreme Court to illustrate how they work simply rely on broad descriptions throughout: the relevant passages refer to participants to an unlawful attack (see e.g. Jogee paras [27], [33], [90], [96]) or a joint criminal venture (Jogee paras [78], [92]), [93]).
Perhaps this is how it ought to be, given the often fluid and fast-evolving nature of violent events and the jury's task to act as fact-finder. But the broader the characterisation of an incident, the easier it will be to prove participation (in homicide). An earlier example is Bristow & Ors [2013] EWCA Crim 1540 (decided pre-Jogee): the victim owned a vehicle repair business in a remote location. He lived on the premises. One night, his business was targeted by a group of burglars who arrived in two vehicles and left in three, the third vehicle having been stolen on the premises. One of the vehicles ran over and killed the victim as the burglars attempted to escape. In such circumstances (and in the absence of proof of murderous intent), the obvious approach would have been to treat the assault committed when the vehicle was driven at the victim as the base crime for a charge of unlawful dangerous act manslaughter (UDAM). However, the crux in Bristow was that the prosecution could not prove which of the burglars had been behind the wheel or, indeed, inside the vehicle that killed the victim. In their route to verdict, the jury was therefore invited to consider whether the burglary had been a dangerous act in the circumstances of the case. Bringing the fatal collision within a broadly defined burglarious episode allowed the prosecution to charge all the burglars with constructive manslaughter.
Bristow may be an extreme example (and one that raises questions not just as to the proper characterisation of criminal incidents but also as to the correct choice of base offence to ground an UDAM charge), but it shows that there is a slippery slope to liability that practitioners will need to beware of. Juries, too, will need careful instruction to prevent them going down that slope or, as in Rowe's case, from arriving at illogical conclusions.
Rowe is also noteworthy for yet another unsuccessful attempt to conflate the overwhelming supervening act principle (OSA) with causation. The court's conclusion that there is no tension within Jogee is defensible, to the extent that the apparent contradiction between an overwhelming supervening act principle expressed in language reminiscent of a break in the causal chain when no proof of a chain of causation is required for accessory liability to arise in the first place can be resolved. The court calls OSA an ‘exclusion of liability’, suggesting that it ‘describes conduct on the part of an accessory which has become irrelevant (…) or has become incapable of encouraging or assisting the commission of the offence by the principal [emphasis added]. In such circumstances an accessory's encouragement of or assistance to the principal has ceased’ (at [136)]. It is true that this sounds very much like a reference to the (lack of) effect or impact of the assistance or encouragement on the principal's commission of the crime (the very thing the prosecution does not need to prove according to para [12] of Jogee) – the accessory's conduct has either become immaterial for the principal's commission of the offence or lost its potential or capacity to assist or encourage him. However, the fact that assistance and encouragement do not need to be shown to have been causal (in the ‘but for’ sense – some unspecified connecting link is clearly presupposed by the very concept of accessory liability, albeit that the case law does not insist on proving this connection beyond proof that there was assistance or encouragement) for the principal's commission of the 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 (evidence of) a causal contribution is not required for accessory liability to arise, there can be no remoteness limits placed on the accessory's liability. Indeed, it would only deepen any sense of injustice arising from the absence of a proven causal connection if the law were not to allow accessories whose contribution can demonstrably be shown to have been superseded (in the manner prescribed by OSA) to escape liability: the more so when their liability derives from a constructive homicide offence.
Counsel's inverse reasoning that, if liability can cease due to remoteness concerns, then sufficient proximity (by way of a causal contribution) needs to be proved to begin with, was predictably rejected by the Court of Appeal. The argument failed because, whatever the doctrinally desirable position, Jogee is clear that the prosecution does not need to show any positive effect or impact of the assistance or encouragement on the principal's conduct or the outcome. Conversely, proof that any (putative) impact has ceased (due to an extraordinary and wholly unforeseeable act or event) can expunge liability. This is to prevent outcomes that are demonstrably unjust, seeing that the law of accessory liability holds accessories to account for the full offence, not just their own contribution. Remoteness (of which the idea of supervening acts forms part) limits responsibility for outcomes factually caused by an accused because factual causation at times overshoots the mark. It prevents the attribution of factually caused consequences where they were not foreseeable. Where factual causation is not required, as in complicity cases, there is greater scope for excessive attribution and thus a greater need for a limiting principle. Given that complicity attributes the principal's offence, in its totality, to the accessory, OSA recognises that circumstances around the principal's offending can be such that it will no longer be appropriate to attribute the resulting offence to the accessory, even though there may have been an act of assistance or encouragement to begin with. Specifically, if the principal offender and accessory did not share an intent to kill, and the lethal danger created by the principal's conduct was far outside of what could ordinarily be anticipated by anyone standing in the accessory's shoes, it may simply be unjust to treat the latter still as an accessory to that conduct and its consequences.
OSA thus serves to restrict accessory liability for constructive homicides, in much the same way as the fundamental difference rule did under the pre-Jogee case law (for a fuller version of this argument see B Krebs, ‘Overwhelming supervening acts, fundamental differences, and back again?’ J Crim L 2022, 86(6) 420–440). While the policy behind OSA as an exclusion of liability remains unarticulated, it is clearly grounded in ideas of fair attribution. The assessment of fair attributability rests with the jury. Perhaps the jury's job could be aided by characterising OSA as an exclusion of attribution. This label might bring home to the jurors that what OSA requires is an act (or event) that was so outside of the ordinary and foreseeable that it would simply be unjust to ascribe the resulting offence to the accessory.