WHEN Parliament replaced murder’s partial defence of provocation with the new partial defence of loss of control, it did so in part to give judges more control over proceedings. Judges should only leave loss of control to juries where there is “sufficient evidence … adduced to raise an issue” as to the defence, where “sufficient” means that a jury “could reasonably conclude that the defence might apply” (Coroners and Justice Act 2009 (hereafter “CJA”), s. 54(5)–(6)). In Gurpinar [2015] EWCA Crim 178, at [14], Lord Thomas C.J. explained that judges “must undertake a much more rigorous evaluation of the evidence before the defence could be left to the jury than was required under the former law of provocation”. But there is a fine line between “rigorously evaluating” the evidence and usurping the jury’s fact-finding role. Judges “should not reject disputed evidence which the jury might choose to believe” (Clinton [2012] EWCA Crim 2, at [46]).

Whether there is sufficient evidence of the defence arising depends on what counts as sufficient evidence. Unfortunately, Parliament formulated loss of control’s substantive elements in ambiguous terms. Some disambiguations make the defence surprisingly generous. This limits the ability of judges to play their intended gatekeeping role, for a jury may (reasonably) interpret the statutory language in a generous way.

The problem is illustrated by R. v Turner [2023] EWCA Crim 1626. Turner killed his wife by stabbing her 68 times. He claimed loss of control. The trial judge refused to leave the defence to the jury on the basis that no reasonable jury could find that Turner passed any of defence’s three stages. He was convicted of murder. By contrast, the Court of Appeal considered all three stages potentially surmountable. They quashed his conviction and ordered a retrial.

The first stage is that the defendant’s killing must result from a loss of self-control (CJA, s. 54(1)(a)). What was the evidence that Turner lost control? Well, he stabbed his wife 68 times. Does that not suffice?

The Court of Appeal has been wary of that inference. Ferociously inflicting many blows is not necessarily evidence of lost control (Goodwin [2018] EWCA Crim 2287, at [46]; Dawson [2021] EWCA Crim 40, at [23]). But in Turner the trial judge apparently went further and claimed that the “number of blows… suggested that there was, in truth, no loss of control” (as paraphrased by the Court of Appeal, at [44]).

This goes too far. Ferocious but self-controlled people exist, like gangland enforcers or sadistic abusers. Hence ferocity is not necessarily evidence of lost control. But for most people, even most killers, ferocity is, presumably, more likely if they lost control than if they did not. This seemed true of Turner, a “gentle giant” (at [30]) who reached his fifties with no prior convictions (at [53]). A fact is evidentially relevant if it alters the likelihood ratio between two hypotheses. So, Turner’s ferocity did provide some evidence of lost control.

Well, perhaps. In truth, the entire “loss of control” concept lacks “sharpness or a clear foundation in psychology” (Law Commission, Partial Defences to Murder (L.C. 290, 2004), at [3.30]). It is not a question requiring expert psychiatric assessment. (Distinct from diminished responsibility’s question of whether the defendant’s general ability to exercise self-control was impaired). Indeed, while the trial judge in Turner thought that the number of blows evidenced no loss of control, an expert witness in a different case this year opined that “a single stab wound indicated that the perpetrator was in control” (Ogonowska [2023] EWCA Crim 1021, at [33]). It seems that both too few and too many blows can be read as evidence against losing control!

Given that the underlying phenomenon of “losing control” is conceptually unclear, the question of what evidences this phenomenon is also necessarily unclear. But so long as a jury might reasonably accept whatever evidence is available, a judge should not withdraw the issue from them on this basis. And a jury might reasonably have inferred from Turner’s ferocity that he lost control. The trial judge was wrong to withdraw the defence on this basis.

The second stage of the defence is that there must be a qualifying trigger (CJA, s. 54(1)(b)). There were two candidate triggers for Turner’s killing. First, his wife’s infidelity. But this alone does not qualify: CJA, s. 55(6)(c)). Second, she (allegedly) threatened that he would never see their children again. (Technically her biological grandchildren, but they jointly cared for them under a Special Guardianship Order (SGO)). This trigger qualifies only if it “constituted circumstances of an extremely grave character, and caused D to have a justifiable sense of being seriously wronged”: CJA, s. 55(4).

Losing one’s children is a serious matter. Turner’s eldest child also had serious care needs. One implication of his wife’s alleged threat was that the SGO would cease, and the child could have reverted to institutional care (at [47]). These seem like extremely grave circumstances.

But does the threat of losing one’s children give one a justifiable sense of being seriously wronged? The trial judge noted that Turner’s wife had no power to fulfil her threat, for custody was a matter for the courts (at [46]); and parents anyway have no legal right to access their children come what may. Hence, he failed to pass the second stage. By contrast, the Court of Appeal pointed out that “a man, in the heat of the moment, might not be able to engage in the cool rationalisation of the judge’s reasoning” (at [47]). Hence, a jury were entitled to find this stage fulfilled.

The Court of Appeal did not spell out their (or the trial judge’s) reasoning as to whether and why these facts support their conclusion. But one way to explain the divergence is that the trial judge interpreted “seriously wronged” as “legally wronged”, and Turner’s wife did not explicitly threaten unlawful action. By contrast, the Court of Appeal perhaps interpreted “wronged” more broadly, to encompass extralegal moral wronging, and Turner’s wife’s (alleged) threat (at least arguably) rose to this bar.

As with the first stage, the second stage is sufficiently vague, and has one plausible interpretation which is sufficiently generous to defendants, that a judge should not remove the issue from the jury unless sure they would reject the defence on this basis.

The third stage is that a person of normal tolerance and self-restraint, in the defendant’s circumstances, might have reacted in the same or in a similar way to the defendant (CJA, s. 54(1)(c)). The trial judge dealt with this briefly: a tolerant person “might respond by a short-lived loss of temper or by saying “See you in court”. They would not stab their partner 68 times” (at [50]). True. But the question is not whether a person of normal tolerance (etc.) would have done precisely what the defendant did. As the Court of Appeal pointed out, the statutory language asks whether the tolerant (etc.) person “might have reacted … in a similar way to the appellant by stabbing his wife” (emphases in original, at [50]). Indeed, even that paraphrase is too restrictive: “similar” reactions are presumably not limited to stabbings. If the tolerant (etc.) person might have shot the victim, stabbing them is a fortiori.

What constitutes a “similar” reaction? And what degree of probability is connoted by saying that the tolerant person “might” have reacted thus? Read most generously to defendants, a jury might think that a “normal” degree of tolerance is compatible with some hot-headedness, that a “similar” reaction is one involving any sort of violence, and that “might” connotes a very small possibility of reacting in that way. On this view, the third stage would be passed if some minor violence was a remote possibility for a moderately tolerant person in the defendant’s shoes. That seems too generous. But a jury could reasonably interpret the statutory language in this way. And for this reason, once again, it is inappropriate for a judge to withdraw the defence from a jury on the basis of the judge’s own, more restrictive, interpretation.

In Turner the Court of Appeal failed to grapple with this interpretive difficulty. Instead, they rejected the judge’s ruling on a narrower basis. However we interpret “might have acted similarly”, we must always consider “the defendant’s [relevant] circumstances”. In withdrawing the issue from the jury, the trial judge failed to account for the circumstances of Turner’s wife’s infidelity, the older child’s special learning needs, and the importance to him of the SGO in ensuring the child’s welfare (at [50]). These were relevant circumstances which made his reaction more understandable. And for this narrow reason it was inappropriate to withdraw the issue from the jury.

Judges “cannot – tempting though it may sometimes seem – simply leave loss of control to the jury in order to seek to avoid generating a potential ground of appeal” (Goodwin [2018] EWCA Crim 2287, at [38]). Turner illustrates that this is easier said than done. Given the ambiguous and (potentially) expansive statutory language, withdrawing the issue will often encroach on the jury’s role – even if the defendant stabbed his wife 68 times because of a hollow threat.