Morally permissible killing and necessity in the law
The suggestion that one might be permitted to kill another person in exceptional circumstances where greater evil is avoided by killing than would otherwise occur is, first and foremost, a proposition about our extra-legal permissions, and one which the judges in Re A clearly found compelling. As is reflected in the conjoined twins case, the paradigmatic necessity scenario in moral theory is where the victim’s death is unavoidable in all eventualities but where his being killed can secure the salvation of others. Phillipa Foot famously argued that in some such scenarios, choosing to kill the victim can be morally permissible.26 Foot imagined the situation in   
which, due to some complication, nothing can be done to save the life of a fetus, but in which killing the fetus will save the pregnant woman who would otherwise also perish. She compares that scenario to the imaginary situation in which a person of generous proportions is blocking the mouth of a cave that is gradually filling up with water, and in which others are trapped. If nothing is done, the man is bound to drown with the others. But if the others are somehow able to blow him up, and only if they do this, they can move him out of the way and escape. Foot endorses the view that such an action would be permissible ifhe is certain to die soon anyway. Killing the man is the only way to save as much life as possible in the circumstances, and it does not change the outcome for him.
As with self-defence killing, the first condition we find in the moral version of the necessity justification is a proportionality constraint. In fact, we might say that proportionality is the essential component of necessity. Necessity killing is only justified because it averts more harm than it causes; it is the ‘lesser of two evils’. More specifically, and as is captured in the cave thought-experiment, killing a person out of necessity is only justified where it is the only means of preventing a greater number of deaths. All candidate scenarios for permissible necessity killing in moral philosophy share this feature. All of them present us with killings which, if carried out, will result in the greatest number of lives being saved. May I redirect a runaway trolley so that it kills one person instead of five? May Jim shoot one Indian to save nineteen others? The stipulation that killing the one will save the most amount of life is the most rudimentary condition of the necessity justification. That condition was iterated in Re A when Brooke LJ stated the three core requirements for the application of the necessity defence, which he deemed to be met in the case of the conjoined twins:
[T]hat the act was needed to avoid inevitable and irreparable evil; that no more should be done than was reasonably necessary for the purpose to be achieved and that the evil inflicted was not disproportionate to the evil avoided.
Killing Mary was the only way to avoid the ‘irreparable evil’ that was the death of both twins. It was not disproportionate, since the overall effect would be to save the most amount of life—one twin instead of no twins.
So the maximization of life is essential to the necessity defence. But this does not mean that necessity is only a numbers game. That is to say, the necessity justification is not made out just whenever it is true that killing one person can save more than one. Foot illustrated this point with another example. If a healthy homeless man walks into a hospital, it is not morally permissible for the doctors at that hospital to kill him and harvest his organs for transplants to save the lives of five others. How is this example materially different from the man blocking the mouth of the cave? Well, for one, the man in the cave is doomed to die soon in any event, whether he is blown up or not. In Re A, it was critical for the judicial reasoning that Mary was, sadly, ‘designated for an early death’. Because they were unable to save Mary in any scenario, the Court judged the doctors to be in need of an ‘escape route’ to at least save Jodie. But Mary’s poor prospects and the ability to save Jodie were not all that was required for the defence. When restating the ‘unique circumstances’ for which the decision could be an authority, Ward LJ enumerated the conditions that:
- 1. It must be impossible to preserve the life of X without bringing about the death of Y,
- 2. Y by his very continued existence, will inevitably bring about the death of X within a short period of time, and
- 3. X is capable of living an independent life but Y is incapable under any circumstances (including all forms of medical intervention) of viable independent existence.28
All three judges regarded it as essential for the permissibility of the operation that Mary could not be saved at all, as well as that killing Mary was the only way to save Jodie. However, Ward LJ’s conditions seemed to extend even beyond this. For him, the only killings for which the decision might be an authority would be those where the victim was by her very continued existence threatening to bring about the death of the other. In other words, the victim must present or constitute the threat to the lives of others, as well as being beyond help herself. In Re A this condition appeared to be met. Mary’s imposition on Jodie was the source of the threat to Jodie’s life. As Ward LJ opined, if Jodie could speak, she could rightly protest ‘Stop it Mary, you’re killing me’. Mary was, he said, ‘sucking the life-blood out of JodieM9
This feature of Re A resembles Foot’s cave scenario, where the large man is the source of the threat to the other lives, albeit an innocent one. His presence is the deadly threat to the greater number; it is not just the case that his death is a means of saving them, as it would be if something else was causing the blockage, and he could be fired in a cannon through the wall of the cave to unblock it. In contrast, the homeless man in the hospital does not present the threat to the other lives; their illnesses do. His death might be a means of saving them, but their imperil- ment does not owe to his presence. As we noted, the homeless man is moreover not himself destined to die very soon, like the man in the cave, and like Mary. However, it seems clear that the doctors would not be permitted to butcher the homeless man to save five others even if he were certain to die of a terminal illness  
Foot acknowledged that one may allow a person to die in order to save many more under less strict conditions. Doctors can surely withhold a scarce life-saving drug from one patient when the same dosage will save a greater number of them. Foot argued only that the rules of morally permissible killing place further conditions on the situation of the victim than merely that his death will preserve more life. As she explained further, the asymmetry between the conditions in which it is permissible to withhold aid and to kill is often a function of the relative strictness of positive duties to give aid and negative duties not to injure, a distinction introduced in the previous chapter. On Foot’s estimation, the driver of the runaway trolley may steer it into the one instead of the five for, being faced with a conflict of negative duties (not to kill either set of victims), the only rational action is to inflict the lesser injury. But conflicts between negative and positive duties play differently, she claimed. One may not murder someone even to bring food to one’s starving children—the weightiest of positive duties. The doctors may not murder the healthy homeless man to fulfil their duty to save the five other patients. Where a positive duty is pitted against a negative one not to kill, the negative duty wins out.
In keeping with Foot’s analysis of morally permissible killing, there was evidently far more to the Court of Appeal’s decision in Re A than a simple calculation of how to preserve the most amount of life. These victim-centred considerations were brought into the foreground by the way in which the Court distinguished the historical case R v Dudley & Stephens. There, a necessity defence to murder was withheld from a pair of sailors who, set adrift at sea for twenty days without water or food, killed and cannibalized another member of their party, believing it necessary to preserve their own lives.31 The judge in the case, Lord Coleridge, famously asserted that ‘the temptation to act which existed here was not what the law has ever called necessity’. He reasoned that granting the defence in such circumstances would give rise to a principle which ‘might be made the legal cloak for unbridled passion and atrocious crime’.32 A pressing question in Re A was how Dudley & Stephens differed from the conjoined twins scenario. Brooke LJ regarded the earlier decision as authority for the law’s ‘disapproval of the idea that in order to save himself a man is entitled to deprive another of the place of safety he has already secured for himself.’33 Put otherwise, it was the feature of choosing a victim—choosing whom to cannibalize for the sustenance of the others—a victim who, moreover, was not necessarily destined to die soon, which placed an act of intentional killing firmly out of reach of the necessity defence. 
In this respect, Mary’s terminal condition distinguished Re A from Dudley & Stephens. Mary was, unfortunately, doomed to die soon in any eventuality.35 This took the choice of whom to save out of the doctors’ hands. Nature had individuated Mary as the only twin whose death could result in the most life, and Jodie was the only twin that could be saved. In stark contrast, the sailors in Dudley & Stephens chose from a number of candidates (themselves included) whom to sacrifice to save the rest. The law’s conclusion that they were not entitled to do so made it abundantly clear that the ‘lesser of two evils’ defence will not countenance the destruction of a person who, apart from the choice to kill her, may well have survived. Given his distinguishing of Dudley & Stephens on this basis, it can be deduced that an action to avoid ‘inevitable and irreparable evil’ according to Brooke LJ’s test would not have included the killing of one of the twins if there were any choice over whom to kill. This would have been so where killing Mary could save Jodie, but where killing Jodie might equally save Mary. In this situation, making a choice one way or the other meets the proportionality constraint of the necessity doctrine, since it is the only way to save the most amount of life (one twin rather than no twins). But it will not meet the victim-centred constraints on necessity killing.
From all of this, we can draw the following conclusions. A necessity defence to homicide would not have been afforded to the doctors in Re A if either: 1. Mary had not been beyond saving; 2. her attachment to Jodie had not been the source of Jodie’s endangerment; and 3. if it were possible to save either twin through a procedure which killed the other, thus leaving nothing to choose between them. If any of these things were true, the doctrine of necessity as articulated in Re A could not have justified the operation.
Still working on the presumption that the fetus is a person, what do these conditions suggest about the scope for the application of the necessity defence if aborting a fetus is indeed to kill it? First, it is clear that only in the most extreme circumstances will an abortion situation come close to paralleling the genuinely taxing thought-experiments concerning the justification of killing in moral philosophy. The man in the cave scenario is not the parallel of the straightforward therapeutic abortion to save the life of the pregnant woman. For this, the fetus would have to be destined to die soon no matter what course of action was taken. In fact, only one abortion situation is even a candidate for justified killing through necessity: where the lives of both pregnant woman and fetus are endangered, and where killing the woman to save the fetus is futile, but aborting the fetus can save the woman. Here, as in Re A, there is no forbidden choice between lives, the ‘victim’ cannot be saved in any eventuality, the victim’s presence constitutes the threat, and killing the victim preserves the most amount of life. The same permission to kill the fetus would not extend to what Foot regarded as the ‘worse dilemma’ where the fetus is not destined to die in every alternative, but where the woman can still only be saved by killing it. That is not like the man in the cave or like Mary and Jodie. As Foot explained, ‘on a strict parallel with cases not involving the unborn’ it appears that intentionally killing the fetus-person would not be permissible here. She explained further:
Suppose, for instance, that in later life the presence of a child was certain to bring death to the mother. We would surely not think ourselves justified in ridding her of it by a process that involved its death.36
If fetus and woman are presumed to be of equal moral worth, it seems that only the scenario in which the fetus cannot be saved but aborting the fetus will save the pregnant woman meets both the proportionality and the victim-centred conditions on morally permissible killing. This conclusion might sound fairly radical. It means that therapeutic abortion to save the life of a pregnant woman is not even a hard case on the assumption that abortion is homicide; it is a case where abortion is clearly impermissible unless the fetus is incapable of sustained life in any event. Thus Foot suggested that those people who accept the justifiability of therapeutic abortion ‘probably equate the life of the mother against the unborn fetus as the many against the one’.37 They may be correct to do so, but this would mean surrendering the personhood proviso and the putative strength of Thomson’s position.
I have already adumbrated the ‘victim-centred’ constraint on necessity killing that one cannot choose one’s victim. Thus in Dudley & Stephens the sailors were not permitted to choose someone to cannibalize, even to save the greatest amount of life, whereas the doctors in Re A were permitted to kill Mary because nature had already marked her out for death. I argued that as regards therapeutic abortion, given the personhood proviso, the upshot is that abortion is not permissible where killing the fetus could save the woman, but where killing the woman or allowing her death could equally save the fetus. However, it is worth noting that Foot took the contrary view here. That is, Foot believed it permissible to choose between the life of the fetus and the pregnant woman where either one can be saved by killing the other, but where otherwise both will be lost. She argued that in choosing to kill either, one can be acting reasonably, so long as uncertainty does not obscure the absolute necessity of acting in such a way.
Consequently, Foot concluded that in principle, it may have been reasonable for the sailors in Dudley & Stephens to kill and eat the cabin boy, since it is better to save some than none, even if this requires choosing a victim (although she acknowledged that we might respect someone who would prefer to perish than commit such an ‘appalling action’). She equally thought that if the certainty were absolute, as it can be in the abortion case, it is better to choose to kill one than save none. Given the choice, she predicted that most people would kill the fetus and thereby save the pregnant woman. However, this can only be because they struggle to accept the personhood proviso. Taking the personhood proviso seriously, the most that comes out of Foot’s argument at this point is that it is equally permissible to kill the fetus to save the woman or to kill the woman so as to save the fetus. This does not make for a robust defence of therapeutic abortion.
The conjoined twins case also helps to clarify the implications of the pure bodily imposition of a fetus upon a woman, assuming that both are persons. Had Jodie’s life not been in danger, it would be unthinkable for the Court to have ordered the separation, knowing that Mary would die, on the ground that Jodie simply had the right to have her detached so as to end the bodily imposition. Mary’s physical dependency on Jodie could not have been justification enough were both capable of living a normal life span in their conjoined state. This is so despite the fact that the nature of their physical union was far more thoroughgoing, and, if they had survived a normal life span, far more burdensome for Jodie, the supportive twin, than nine months of normal pregnancy. Extrapolating from this to the abortion scenario, it appears that the fetus’s physical dependency on the pregnant woman could not suffice for a necessity defence to abortion.
This may raise new questions about Thomson’s violinist. Although his imposition does not threaten to kill you, here the common intuition is that you are permitted to have him detached merely to end the bodily imposition. Why (assuming still that unplugging the violinist is killing him and not just refusing to give life-sustaining aid) is the amended Mary and Jodie case, where the burdens for Jodie are far greater and indefinite, any different? Defending this difference would, I presume, require one to argue that unlike the unconscious violinist, the conjoined twins predicament does not clearly entail the unjust imposition of one person upon another. Although Mary’s parasitic living is a burden on Jodie, it may be difficult to describe her as being under a duty not to be where she was. One question this raises is whether the imposition of the fetus upon the pregnant woman better resembles the conjoined twins scenario or the violinist scenario in terms of the inherent unjustness of the weaker party’s imposition on the stronger one. However, our certainty that Jodie does not have the bare right to have Mary detached, at the cost of Mary’s life, puts immense pressure on Thomson’s claim that a pregnant woman has such a right in respect of a fetus-person.
All in all, the proportionality and victim-centred constraints on necessity killing bring us to the inescapable conclusion that, on the presumption that a fetus is a person and that abortion is killing, a simple therapeutic abortion to save the life of the pregnant woman fails to meet the conditions for the defence, let alone abortion for any less urgent reason, such as temporary physical discomfort, derailment of life plans, or emotional turmoil. The only abortion scenario that meets the Re A conditions is where killing the fetus is the only way of avoiding the deaths of both parties, and where there is no choice about whom to save. It is an implication of this that were the fetus to be granted personhood under English law, both the Bourne decision and the Abortion Act 1967, which widely legalized abortion, would be flatly inconsistent with the common law doctrine of necessity and murder.
-  R v Bourne  3 All ER 615, 617. 45 Re A (n 21) 1043.
-  26 Phillipa Foot, ‘The Problem of Abortion and the Doctrine of Double Effect’ (1967) 6 Oxford
-  Review 5.
-  Re A (n 21) 1052.
-  ibid 1018.
-  ibid 1016—17. Ward LJ contemplated the possibility that Jodie might have a plea of ‘quasi selfdefence’ against Mary, owing to the parasitic nature of Mary’s living and the consequences for Jodie,although he thought it inappropriate to label her an ‘unjust aggressor’. If this combination of factorswere sufficient to justify killing Mary, then the justification might extend analogously to the simple therapeutic abortion case. However, even Ward LJ did not regard Mary’s threat to Jodie as beingenough, without more, to justify killing Mary. Like the other judges, it remained integral to his decision that Mary could not be saved (as he said, ‘fairness and justice between the children’ must takeaccount of the fact that ‘Mary is beyond help’ (at 1010)). If Ward LJ would not have permitted theoperation but for Mary’s sealed fate, his quasi self-defence argument will not extend to the therapeuticabortion except where the fetus also is beyond help, at which point it might be argued the quasi-selfdefence ground becomes indistinguishable from necessity.
-  Of course, if the numbers are only large enough, some might hesitate to cling to the usualprohibitions. Would killing the healthy homeless man be morally impermissible if his biologicalmaterial could be used to cure a deadly disease that would save millions? Can a judge execute aninnocent man if it were the only way to prevent nuclear war? Such scenarios would cause most people to pause. However, we need not decide about this further question of whether the constraintson the victim’s situation (specifically, her own endangerment) can be relaxed if the numbers savedare great enough. The number of lives saved by an abortion never exceed one, and even this will notamount to the avoidance of greater harm unless more than one (woman and fetus) would otherwisebe lost.
-  R v Dudley & Stephens [1881—5] All ER 61. 32 ibid 67. 33 Re a (n 21) 1034.
-  Lord Coleridge in Dudley & Stephens also pointed to the criminal law’s denial of duress as adefence for murder, set down in R v Howe  2 WLR 568, as exemplifying the general legal principle that one may not take the life of another to save oneself, or even others, where the killing amountsto a choice between lives (Dudley & Stephens (n 31) 66).
-  Re A (n 21) 1051.
-  One of the Re A judges, Walker LJ, came close to acknowledging as much when he opined thatexploring the defensibility of abortion was of limited use as an analogy for how to respond to conflicting medical duties between actual persons, since fetuses are not regarded as such until delivery (Re A(n 21) 1066).