Sandy-Steel-Rationalising-Liability-for-Omissions-in-Negligence (1) - 1

Forthcoming in 2019 Law Quarterly Review. Please cite the published version.
Uploaded by : Ahmed I H
Lecturer - Newcastle Law Academy
Mob : +8801911080106
RATIONALISING OMISSIONS LIABILITY IN NEGLIGENCE
Sandy Steel*
Associate Professor, University of Oxford
It is true that the categories of negligence are never closed…and it would be open to the court
to create a new exception to the general rule about omissions...[The court looks] to see
whether there is an argument by analogy for extending liability to a new situation, or whether
an earlier limitation is no longer logically or socially justifiable. In doing so it pays regard to
the need for overall coherence. 1
I. Introduction
The general rule in tort law, and in the tort of negligence in particular, is that there is
no liability in respect of a pure omission – normally defined as a mere failure to act.
Thus, a person is generally under no tortious duty to provide positive assistance to a
stranger in peril. 2 Various exceptions to this rule are recognised, for instance, where a
person has assumed a responsibility to act, is in a special relationship with the person
injured, or has failed to alleviate a source of risk they have created.
The possible justifications for this general rule have been much examined. 3 This
article focuses on two questions that have received considerably less attention. The
first is: how, precisely, ought the general rule in the law of negligence to be
understood? The traditional formulation of the general rule is in terms of a distinction
between ‘acts’ and ‘omissions’. It will be shown that this distinction is at best a rule
of thumb. It argues that there are ‘acts’ which deserve the same legal treatment as
‘omissions’. It suggests a revised general rule which focuses upon whether the
defendant’s conduct makes the claimant worse off relative to a particular baseline.
Second, the article examines the exceptions recognised to the rule as it is currently
understood. Lord Toulson adverts in the quotation above to the need for overall
coherence in the recognition of these exceptions. There does not yet appear to be a
systematic analysis in the literature of whether the existing exceptions are coherent. 4
This article contributes to this analysis by examining one major issue of ‘coherence’
*With thanks for helpful comments or discussion to Jane Stapleton, Jeffrey Hackney,
the participants at the Edinburgh-Oxford Private Law Exchange, in particular John
Blackie, the participants at the Private Law Workshop at Hong Kong University, and
the anonymous reviewer. The usual disclaimer applies.
1
Michael v Chief Constable of South Wales [2015] UKSC 2, [2015] A.C. 1732, at
[102] per Lord Toulson.
2
Subsequent references to ‘tortious duty’ or ‘duty’ should be read as references to
duties of care in negligence, unless stated.
3
See, e.g., J. Kortmann, Altruism in Private Law: Liability for Non-feasance and
Negotiorum Gestio (Oxford: Oxford University Press, 2005).
4
For some helpful analysis of this issue, see J. Kortmann, Altruism in Private Law:
Liability for Non-feasance and Negotiorum Gestio (Oxford: Oxford University Press,
2005), ch 5; R. Stevens, Torts and Rights (Oxford: Oxford University Press, 2007), at
pp.9-16; E. Quill, “Affirmative Duties of Care in the Common Law” (2011) 2
J.E.T.L. 151.
1
in this area, namely, whether each exception is consistent with the rationale for the
general rule. The question it addresses is whether the law can consistently be
committed to the general rule while also admitting exceptions in the situations it
currently does. I show that the existing exceptions can be considered consistent with
the general rule. The article pursues a third, subsidiary, aim: by examining the
rationales of each exception, it resolves some doctrinal controversies as to their scope.
The article proceeds as follows. The first section focuses upon the question of how
the general rule on ‘omissions’ should be understood. The second section moves on to
consideration of the scope of the exceptions and their coherence with the general rule.
The analysis of each exception is not comprehensive: it focuses on the central issue of
whether the exception can be considered consistent with the general rule. A brief
conclusion ensues.
II. The General Rule
The general rule is conventionally formulated in this way: 5
General rule. A is not liable to B in the tort of negligence for a pure omission.
An omission is a failure to act. In Robinson v Chief Constable of West Yorkshire

Police, Lord Reed described these as the “commonly given examples” of pure
omissions: “someone who watches and does nothing as a blind man approaches the
edge of a cliff, or a child drowns in a shallow pool”. 6 The classic example of an
impure omission, which is treated analogously to an act, is a failure to brake or to
indicate while driving. 7
The General rule is purely negative: it says that, generally, A is not liable in
negligence to B for a pure omission. It does not logically entail, then, that A is
generally liable to B for an act. However, this is typically held to be true in relation to
reasonably foreseeable physical damage. 8 Thus, we can formulate the following
closely connected rule:
Acts. A is generally under a duty of care to B in the tort of negligence in

respect of reasonably foreseeable physical damage caused by A’s actions.
In what follows, I will first consider a problem with the General rule which arises due
to its interaction with Acts. This is the problem that some acts deserve the same duty
of care treatment as pure omissions. Furthermore, they deserve this treatment for the
same reason that there is generally no duty of care in respect of a pure omission. This
5
See, e.g., Smith v Littlewoods Organisation [1987] A.C. 241, at 270-271, [1987] 1
All E.R. 710; Robinson v Chief Constable of West Yorkshire Police [2018] UKSC 4,
at [69], [71]-[72]. M. Lunney, D. Nolan and K. Oliphant, Tort Law: Text, Cases and
Materials (Oxford: Oxford University Press, 2017), at pp.469-471. The rule also
applies in other torts: see, e.g., Iqbal v Prison Officers Association [2009] EWCA Civ
1312, [2010] Q.B. 732, at [15]-[23] (false imprisonment).
6
[2018] UKSC 4, at [72].
7
Mitchell v Glasgow City Council [2009] 1 A.C. 874, [2009] 3 All E.R. 205
at [76] per Baroness Hale; J. Steele, Tort Law: Text Cases and Materials, 4th edn.
(Oxford: Oxford University Press), at p.170.
8
N J McBride and R Bagshaw, Tort Law, 6th edn, at X.
2
problem entails that the General rule needs revision in some way to accommodate
cases where an act causes foreseeable physical injury but should generate no duty of
care: some such cases ought to fall within the protection of the General rule. In light
of this, the article defends an alternative account of the conduct that ought to fall
within the General rule. It argues that A should generally not be liable to B in the tort
of negligence if A’s conduct is such that it does not cause B to suffer a damage which
B would not have suffered had A’s body or resources not been present at the time of
the careless conduct. It then considers an alternative formulation based upon creating
a risk and not alleviating a risk. It shows that this formulation either adds nothing to
the account proposed or should be rejected.
Beyond acts and omissions
The function of the general rule is to identify a category of conduct in respect of

which a special reason is needed for the imposition of a duty in virtue of the nature of
the conduct. But the distinction between acts and omissions is at best itself only a rule
of thumb for identifying situations where the defendant’s conduct is such that there
must be special reasons for the imposition of a legal duty.
This is because there are situations where careless acts which foreseeably lead to
physical harm are not ones in respect of which a duty of care is owed and are akin to
cases of pure omission. Suppose that I set up a direct debit which provides funds to an
exceptionally talented brain surgeon, who uses those funds to perform operations that
are unavailable through the National Health Service. Having carelessly misread a
newspaper article as saying that the surgeon had negligently performed certain
operations, I instruct my bank to cancel the payment. Suppose that had that payment
been made, the surgeon would have been able to save one life. Here I have done a
careless act – instructing my bank to cancel a payment – which foreseeably led to
another’s death. Call this case Cancellation. 9 It seems highly unlikely that I am under
a duty of care with respect to the death in Cancellation. 10
This intuitive verdict can be strengthened by appeal to the basis of the general rule.
Suppose, plausibly, that the underlying basis of the general rule is primarily a concern
of freedom. 11 Some types of duty are more invasive of a person’s freedom than others
because they require that person to give up a wider range of (valuable) options or to
focus upon a specific end. This explains the intuitive classification of Cancellation as
a no duty situation: requiring the defendant to continue to make the charity payment
requires a very specific action on the part of the defendant and requires the defendant
to make its resources available for the use of another person.
a. Making worse and failing to make better
9
This case is modelled on one given in S. Kagan, The Limits of Morality (Oxford:
Oxford University Press, 1989), at p.108. See further J. McMahan, “Killing, Letting
Die, and Withdrawal of Aid” (1993) 103 Ethics 250.
10
At least on the assumption that my giving did not lead to the recipient refusing an
alternative source of aid. See below, X.
11
See further below, X.
3
An alternative formulation of the general rule, which has some judicial and academic
support, is as follows: 12
No benefit: A is not generally liable to B in negligence for failing to confer a

benefit upon B, but only for making B worse off.
This formulation is appealing in that it tracks an intuitively important moral

difference between affecting the world for the worse and failing to improve the world;
it thus seems to make the general rule morally intelligible. 13
The central difficulty with this formulation is how to specify the baseline
relative to which one judges that a person is made worse off or merely not
benefitted. 14 Consider Cancellation. It might be said that my instruction makes the
victim worse off than they would have been had I not given the instruction: had I not
given the instruction, they would have lived. Alternatively, it could be said that my
act merely causes the victim not to be benefitted because, had I not given to charity at
all, they would still have died. As yet, the formulation provides no guidance on which
baseline to use. 15
Here, then, is one way of defining the baseline:
No Benefit 1: A is not generally liable to B in tort for failing to make B better

off than had A not existed.
This reaches the intuitively correct answer in Cancellation: the cancellation of the
payment does not make the victim worse off than had I never existed. But No Benefit
1 survives only the least scrutiny. Suppose that I saved your life in circumstances
where no one else would have. Then I carelessly run you over in my car. My
carelessly running you over does not, or need not, make you worse off than if I had
never existed. But it is clearly conduct which should give rise to liability without
special reason needing to be shown (in virtue of the kind of conduct that it is).
A different suggestion is as follows: 16
12
See Robinson v Chief Constable of West Yorkshire Police [2018] UKSC 4, at [69]
per Lord Reed; D. Nolan, “The Liability of Public Authorities for Failing to Confer
Benefits” (2011) 127 L.Q.R. 260, at 260, claiming that it is normally more
straightforward to ask if “the defendant has made things worse for the claimant or
simply failed to confer a benefit on him or her”. See, similarly, R. Stevens, Torts and
Rights (Oxford: Oxford University Press) at p.9.
13
See, articulating the idea that acts are more morally problematic because they
involve interventions on the status quo, T. Honore, “Are Omissions Less Culpable?”
in P. Cane and J. Stapleton, Essays for Patrick Atiyah (Oxford: Oxford University
Press, 1991) 31, 41-42.
14
This crucial issue is often given little attention in the tort literature. See, e.g., R.
Stevens, Torts and Rights (Oxford: Oxford University Press) at p.9.
15
Moore makes this criticism: M. Moore, Act and Crime: The Philosophy of Action
and its Implications for Criminal Law, at p.28
16
This suggestion and the next are discussed in K. Draper, “Rights and the Doctrine
of Doing and Allowing” (2005) 33 Philosophy and Public Affairs 253, 260-262.
4

off. A’s conduct makes B worse off if and only if there is a time such that, had
A not existed at that time, B would have been better off.
This deals with the careless car collision example. There is a time (the collision with
B) when, had A not existed at that time, B would have been better off. But this
formulation reaches the incorrect result in Cancellation: there is a time, namely, when
I order the cancellation, at which, had I not existed then, B would have been better
off: thus, Cancellation is incorrectly treated as a case of making worse.
A third suggestion:

off. A’s conduct makes B worse off if and only if it makes B worse off than
the position B is entitled to be in with respect to A.
This suggestion is uninformative. This proposal says: a person is entitled to be made

no worse off than they are entitled to be with respect to another person. A failure to
benefit is a failure to improve a person’s situation beyond the situation they are
entitled to be in. This then raises the question of what entitlements people have
against each other, which is precisely the issue at stake. The General rule is supposed
to tell us what position people are generally entitled to be in relation to each other’s
careless conduct. No progress has been made on this question with No Benefit 3.
Consider, then:
No Benefit 4: A is generally only liable in negligence to B if A’s conduct

makes B worse off than B would be independently of the presence of A’s
body or resources at the time of the careless conduct.
This formulation is appealing. It deals with Cancellation. In this case, the defendant’s
act does not make the victim worse off with respect to what they had or would have
had independently of the presence of the defendant’s body or resources. Had the
payment not been made, the victim would have died in any event. This formulation
also deals with the careless collision example: A’s carelessly colliding with B makes
B worse off than B would have been had A’s body or resources not been present at
that time, even if, at an earlier time A’s body or resources benefitted B.
Unfortunately, however, No Benefit 4 faces two problems. These are the
problems of (i) overinclusiveness and (ii) underinclusiveness. It will be argued that
No Benefit 4 can meet these problems with a minor revision.
The problem of overinclusiveness
Consider OLL Ltd v Secretary of State for Transport. 17 In this case May J held that no
duty of care arose on the part of the defendant coastguard when it misdirected other
rescue services in the coordination of a rescue attempt. Thus, although the defendant
did something which made the victim worse off than they would have been had the
defendant not been present, they were under no duty of care. Rather, a duty only arose
not to “inflict direct physical injury” and misdirecting “other rescuers” could not be so
17
[1997] 3 All E.R. 897.
5
classified. 18 The objection, then, is that No Benefit 4, in combination with a revised

version of the Acts principle – which focuses upon whether the presence of A’s body
or resources made B worse off – would entail that a duty should have been owed in
OLL.
It may, however, be argued that OLL is consistent with the general principle
that one is under a duty of care not to render persons foreseeably physically worse off
than if one had not intervened. The argument is that the defendant coastguard did not
render the victims worse off than if they had not intervened because all public rescue
services should be treated as one single legal entity. 19 The defendant’s act did not
render the claimant worse off than if no public rescuer had intervened. This is a
somewhat strained interpretation of OLL, however, as May J does not distinguish
between public and private rescuers and one private rescuer was alleged to have been
misdirected in the case. 20 Moreover, there seems to be no normative justification for
grouping together all public rescue services in so far as they are distinct legal persons.
The better view is that OLL is incorrect. A further reason may be given. It is
arbitrary to distinguish between cases where a person prevents an object from
preventing a risk materialising and cases where a person prevents a person from
preventing a risk materialising. It is clear that there can be a duty of care in the former
type of case. If the fire brigade turn off a sprinkler system with the result that damage
occurs which would not have occurred if they had not intervened at all, then they can
be liable in respect of that damage. 21 It is not clear why one should distinguish
between turning off a sprinkler and misdirecting a person.
The problem of underinclusiveness (1): harming and benefitting
Let us now consider a set of examples which appear to show that No Benefit 4 is
underinclusive in what it treats as a making worse. Consider, first:
No rescue. A walks past B, an unconscious stranger, who is drowning. B dies.
No Benefit 4 would classify this as a case where there is generally no duty of care in
negligence. Clearly, this is correct. Now consider:
Aborted rescue. A sees B, an unconscious stranger, who is drowning. A begins

to wade in to the water to assist, but failing to see an obvious sharp object,
cuts A’s foot. A consequently aborts the rescue. B dies.
No Benefit 4 would again correctly classify this case as one where there is generally
no duty of care in negligence. Now consider:
18
[1997] 3 All E.R. 897, 908. Compare, also, Darnley v Croydon Health Services
NHS Trust [2017] EWCA Civ 151, [2018] Q.B. 783, where a receptionist gave
misleading information as to waiting times to the claimant visitor to an Accident and
Emergency ward, with the result that the claimant did not wait to see a triage nurse.
By a majority, the Court of Appeal held that no duty of care arose.
19
R. Bagshaw, “The Duties of Care of Emergency Service Providers” [1999]
L.M.C.L.Q 71, 83
20
[1997] 3 All E.R. 897, 908.
21
Capital & Counties v Hampshire County Council [1997] Q.B. 1004, [1997] 3
W.L.R. 331.
6
Damaging rescue. A sees B, a lightweight, drowning, unconscious, stranger.

A decides to rescue B. A is able to do this by leaning across a dock and
pulling B out of the water. A notices that a sharp pipe protrudes from the dock,
near B’s head. In order to obtain a dramatic video of the rescue for A’s social
media account, A throws B in the air, unintentionally striking B’s head against
the pipe, causing a severe brain injury. Had A not intervened, B would have
died shortly after the brain injury was inflicted.
Carelessly striking a person’s head against a sharp pipe is conduct which would
attract a duty of care without special reason needing to be shown. The difficulty here
is that No Benefit 4 does not reach this result. If A’s body or property had not been
present at the time of the careless conduct, we may suppose that B would be dead. A’s
causing B’s head to be damaged does not therefore make B worse off compared to
A’s body or propery being absent. This suggests the following modification of No
Benefit 4:
No Benefit 5. A is generally only liable in negligence to B if A’s conduct

causes B damage that B would not have suffered but for the presence of A’s
body or resources at the time of the careless conduct.
This formulation reaches the correct result in Damaging rescue because to suffer
‘damage’ is to be caused to be in a particular negative state. It does not require that
the negative state one is in be worse than the state one would have been in had the
careless conduct not occurred. 22 Thus A causes B to be in a particular negative state –
the state of having a brain injury – that B would not have been in but for the presence
of A’s body or property. It is irrelevant, so far as the issue of whether B suffers a
damage is concerned, that B would have been in an even worse state (death) if A had
not intervened at all.
The problem of underinclusiveness (2): overdetermination
Suppose that A and C negligently fire in B’s direction and each bullet strikes B’s head
at the same time (Fire). This is a situation were B’s damage is overdetermined: there
are multiple sufficient causal conditions for its occurrence. 23 According to No Benefit
5, this appears to be a situation where A and C would not generally be liable to B in
negligence. This is because, if A had not been present, B’s damage would still have
occurred, because of C’s firing. Similarly, if C had not been present, B’s damage
would still have occurred, because of A’s firing. Clearly, however, whatever objection
there is to holding A and C liable here, it should not stem from the general rule. It
therefore seems that No Benefit 5 is underinclusive: it fails to find that A and C would
be generally liable in overdetermination situations such as Fire.
This problem is surmountable. It is possible to argue that, in almost all cases,
A and C will have inflicted different ‘damage’ upon B. Even if the bullets
22
See D. Nolan, “Rights, Damage, and Loss” (2017) 37 O.J.L.S. 255, 270-274. It only
follows that A wrongs B here, not that A is liable to pay substantial compensation to
B.
23
This is a somewhat loose definition of overdetermination. See further S. Steel,
Proof of Causation in Tort Law, (Cambridge University Press, 2015), at X.
7
simultaneously strike at one particular part of B’s body, there will most likely be a
sub-part of B’s body which is negatively affected only by each bullet, considered
individually. It will, then, be possible to say under No Benefit 5 that A’s conduct
caused a damage to exist to which would not otherwise exist.24
The problem of underinclusiveness (3): removal of one’s own property
Suppose B is drowning and succeeds in climbing into A’s boat, which is the only
nearby object B can employ to save himself. If A’s boat had not been present, B
would have died. Suppose there is a trap door on A’s boat. According to No Benefit 5,
there would be no duty of care on the part of A not to press the trap door button
because doing so would not cause the person to suffer a damage they would not have
suffered but for the presence of the boat owner’s body or resources. Call this case
Boat. It is difficult to defend the conclusion that there is no duty of care in Boat. 25 It
seems clear that A would owe a duty of care not to dislodge B from A’s property in
such a way as to cause them reasonably foreseeably physical damage, even if this
does not render them worse off than they would be had that that person and their
property not been present.
The response to this apparent counterexample to No Benefit 5 is that the
conduct in this example should be recognised as an exceptional situation where a duty
of care is owed despite it being a case that A has not rendered B worse off than B
would have been had A’s body and resources not been present. The duty to take care
not to exclude a person from their current use of one’s property when they are in dire
need should indeed be considered normatively unusual. 26 This is because it in effect
requires a person to bear costs to allow their resources to be used to benefit another
person; part of the normative justification of the General rule is precisely this
concern. 27 Thus, if the law were explictly to adopt No Benefit 5, it ought to recognise
a new exception to the General rule in situations where A’s careless conduct causes B
to be damaged by removing B from the active protection of B’s property. 28
24
A vaguer version of No Benefit which appeals to whether the conduct ‘typically’
increases the probability of damage that would not otherwise occur without the
presence of A’s body or resources would also offer a solution. A’s conduct typically
increases the probability of damage of that description even if it does not in the
particular case.
25
If the boat owner intentionally pressed the trap door button and the person died, this
would be murder.
26
Such a duty appears to be recognised by some US jurisdictions: e.g. Ploof v Putnam
(1907) 83 Vt. 494. See also Depue v Flatau (1907) 100 Minn. 299.
27
See below, X. The normatively exceptional nature of this duty is also evidenced by
the fact that it would be permissible for A to eject B from A’s boat if it were
necessary to save A’s life. This would not be so if A’s conduct made B worse off than
B would B independently of the presence of A’s body and resources.
28
It is a further question, not addressed here, whether this novel exception would
itself be consistent with the general absence of a duty of easy rescue. For some
doubts, see S. Steel, “Saving Private Wrongs” (2016) 14 Jerusalem Review of Legal
Studies 1, 16.
8
In summary, No Benefit 5 provides an appealing formulation of the general

rule. This formulation leads to the correct results in simple case where a stranger fails
to rescue a drowning child and in the more complex problem case of Cancellation. 29
Creation of risk
Another alternative formulation of the general rule is the distinction between creating
a risk and failing to alleviate a risk that one did not create. 30 Consider:
Creation of risk: A is not generally liable to B in negligence in respect of a

risk which A did not create.
On the face of it, Creation of risk generates the correct result in the paradigm pure
omission case where A fails to rescue B, a drowning child. If A were held liable in
this case, then A would be held liable in respect of a risk – the risk of death by
drowning – which A did not create.
However, to assess this formulation rigorously, one needs a clearer idea of
what it is to create a risk. A risk is a probability of a negative outcome, for our
purposes, a probability of damage. To create a risk, on a simple view, is to be a cause
of a risk. An immediate difficulty is that one may be said, by walking past the
drowning child, to be a cause of a risk of damage to the child: had one rescued instead
of walking past, the risk of the child’s drowning would be lower. Similarly,
Cancellation, I act in such a way that increases the probability of a negative outcome,
compared to the situation where I did not do that act. Therefore, if risk creation is
understood in this simple way – to engage in conduct which is a cause of an increased
probability that a damage will occur - Creation of risk must be rejected.
Once the simple model of risk creation is set aside, a defender of Creation of
risk seems to have two possible options. A first option is to insist that, in determining
whether A’s conduct creates a risk, one compares A’s actual conduct, not with the
counterfactual where A gave assistance, but with some other counterfactual. For
instance, one could compare the risk of damage to the victim given A’s actual conduct
compared to the situation where A was absent. 31 The risk to the drowning child is the
same regardless of whether A is present and walking away or absent entirely.
However, once this counterfactual route is taken, the problems raised in the section on
29
A further advantage is that No Benefit 5 leads to the correct result in cases of
impure omissions, such as a failure to brake (see above X) without the difficulty of
applying the somewhat unclear concept of an ‘impure’ omission.
30
Ripstein invokes the idea of creation of danger to explain some apparent cases of
liability for omission: see A. Ripstein, Private Wrongs (Cambridge, Mass., Harvard
University Press, 2016), at pp.94-95. See also his invocation of this idea in A.
Ripstein, “Private Authority and the Role of Rights: A Reply” (2016) 14 Jerusalem
Review of Legal Studies 64, 76. See also US Restatement Third of Torts, §7(a): “An
actor ordinarily has a duty to exercise reasonable care when the actor’s conduct
creates a risk of physical harm”; §37.
31
Stapleton analyses the notion of ‘creation of risk’ in the US Restatement Third of
Torts as employing a similar counterfactual: J. Stapleton, “The Risk Architecture of
the Restatement (Third) of Torts (2009) 44 Wake Forest Law Review 1309, 1312.
9
making worse and failing to benefit re-emerge. This version of Creation risk faces the
same difficulties as No Benefit 5, since it employs the same counterfactual question –
what would have happened but for the presence of the defendant’s body and
resources?
A second option for the defender of Creation of risk is to build into the idea of
causing damage through creating a risk a special understanding of the sequence that
must exist in order to be causally responsible for damage through creating a risk of it.
Those who appeal to the idea of creation of risk have not presented a clearly worked
out version of this account. 32 Ripstein says that the relevant distinction here is
between ‘doing something to’ someone and ‘failing to do something for’ someone. 33
Perhaps this could be understood as a further way of elaborating the idea of creation
of risk: to create a risk of harm that materialises is to do something to a person. 34
Intuitively, it seems that I do not do something to the victim in Cancellation. So,
albeit that the idea of doing something ‘to’ someone is extremely vague (how is it
different from negatively affecting that person?), this account seems correct so far.
But suppose that I negligently destroy a bridge on which you are standing. The bridge
belongs to T. You fall and drown. This is obviously a case where I ought to be liable
without special reason being shown. 35 But do I do something ‘to’ you here, given that
I do not touch your body or your property? Ripstein is obviously committed to saying
‘yes’, but it is not clear how the idea of doing something to someone as opposed to
failing to do something for someone explains this result.
In summary, this section has shown that a bare appeal to the idea of creation
of risk is not illuminating. This is because more analysis is required before the idea
has any real content. This section considered two ways in which that content could be
provided. The first was the proposal that one creates a risk if one increases the
probability of damage compared to a situation where one is not present. This proposal
adds little or nothing to No Benefit 5. The second was the suggestion that creating a
risk involves a particular way of causally producing an outcome. This suggestion was
found to be unhelpfully vague.
III. EXCEPTIONS TO THE GENERAL RULE
This part moves to consideration of recognised exceptions to the general rule, as the
general rule is currently formulated. 36 The primary aim of the discussion is to assess
whether the exceptions are consistent with the general rule. However, the analysis
32
See, e.g., A. Ripstein, Private Wrongs, at pp.55-58.
33
A. Ripstein, Private Wrongs, at p.55.
34
A. Ripstein, Private Wrongs, at pp.55-58.
35
This is also a problem for the account offered in P. Benson, “Misfeasance as an
Organizing Idea in Private Law” (2010) 60 U.T.L.J. 731, 743-744. Benson argues that
the act-omissions distinction is a proxy for the normative distinction between
exclusionary rights and non-exclusionary rights. One does not generally have non-
exclusionary rights. But it is not clear in what sense you have an exclusionary right
that I not destroy another person’s bridge: you do not have a (non-contractual)
entitlement to exclude a person from another’s bridge.
36
Not all of the exceptions would be ‘exceptions’ under the proposed revised rule;
here I follow the law’s classification.
10
will also consider some of the central uncertainties surrounding the scope of each
exception and how they ought to be resolved.
Each of these aims will be served in some measure by having an account of the
rationale of the general rule. It will not be possible to provide a complete defence of
the rule here. Instead, I will set out the most powerful considerations in favour of the
rule. Those who disagree with these considerations may still find the question of how
far the exceptions can be rationalised, consistently with commitment to the general
rule, of some interest.
Positive duties are more difficult to justify because they are more invasive of a
person’s freedom. 37 There are different ways of arguing for this claim. First, it is
sometimes said that positive obligations close off more alternative options than
negative obligations: a person’s being obligated to do a particular thing removes their
freedom to do everything but that thing, while a person’s being obligated not to do a
particular thing leaves them free to do everything except that thing. 38 This version of
the freedom argument has some force, though it can be overstated. Requirements not
to impose harm can also remove a large number of alternative options. Being required
to take care not to harm others while driving requires levels of attention that rule out
various courses of conduct. Or suppose someone is arranged such that when they stop
breathing, another person suffers harm. Imposing an obligation on that person to
continue breathing may remove very few options. Nonetheless, this seems to be a
relatively general distinction between the two kinds of obligation.
Second, Kantian theorists tend to draw attention to the idea that positive
obligations require a person to serve another’s ends. Ripstein writes that a person is
“not entitled to determine the purposes for which the defendant’s means are used”. 39
But in what sense do positive obligations give such an entitlement when negative
obligations do not? Could it not be said that a negative obligation gives a person an
entitlement to demand that a person not use their body in injurious ways? In my view,
there is a real distinction here. It is this. A positive obligation requires a person to act
with a specific intention – typically, to assist another. A negative obligation does not.
Consider the obligation not apply force to another person without their consent. I can
go about my life, complying with this obligation, without having any specific
intention. There is thus a sense in which a positive obligation imposes upon a person
in a way which a negative obligation does not. This framing of the point focuses upon
the defendant’s freedom. But the point can be put in claimant-centred terms, too.
When a person acts upon another, they impose upon them in a way they do not when
they do not assist them. They become implicated in the other’s life in a way they do
not when they fail to assist them.
Third, it has been argued that, if we do not recognise something like the general
rule, we cannot truly own our bodies or external objects. 40 To own these things,
morally, is to have a right to exclude others from them in accordance with one’s
desires. If our duties to benefit were as stringent as our duties not to harm, we would
37
For a well-known judicial statement of this point, see Stovin v Wise [1996] A.C.
923, at 943, [1996] 3 W.L.R. 388, per Lord Hoffmann.
38
P.J. Fitzgerald, “Acting and Refraining” (1967) 27 Analysis 133, 139.
39
A. Ripstein, Private Wrongs, at p.56.
40
See F. Woollard, Doing and Allowing (Oxford: Oxford University Press, 2015). For
a rejection of most other explanations of the objection to liability embedded in the
general rule, see S. Tofaris and S. Steel, “Negligence Liability for Omissions and the
Police” (2016) 75 C.L.J. 128, 129-133.
11
have a duty to give away our resources to the extent that would make the world better
overall. Whenever ‘our’ resource could be better used elsewhere, then we would be
obligated to allow it to be used in that way. This would mean that I would have to
allow my body to be used whenever the costs to me are less than the benefits to
others. But this, it is argued, is to deny that I have a special connection with my body
which means that, within limits, my say-so is what counts, rather than the overall tally
of costs and benefits. If my body or property can permissibly be used whenever
benefits are greater than the costs, then my say-so has, as such, no normative force at
all. But this veto based on one’s say-so is the hallmark of owning in morality. If this
argument is right, then we need some moral protection from the needs of others, if we
are to own our bodies. 41
Fourth, and importantly, it is generally more difficult to justify making a person
worse off than failing to benefit them. Other things being equal, it is worse harmfully
to intervene upon the world, rather than merely to fail to improve it. 42 This argument
is not based upon a concern for individual freedom.
These considerations support the idea that it is more difficult to justify positive
than negative obligations. In my view, they do not support the law’s position that
there is no private duty of easy rescue. Sometimes considerations of extreme need
trump considerations of freedom. It is consistent with one having ownership of one’s
body that someone’s one must accept minor incursions in order to prevent great
suffering (just as the existence of some planning regulations is consistent with
meaningful ownership of land). 43 It might seem, then, that the law’s exceptions to the
general rule will, on my view, obviously be consistent with its rationale. Given that I
believe that a duty of easy rescue would be consistent with the general rule, is it not
clear that the existing exceptions, which are usually thought to be less onerous, in
some sense, than a duty of easy rescue, will be consistent with it? Even if this is true,
however, there is still a question of whether the law’s own understanding of the
general rule is consistent with the exceptions made to it. We can ask: given that the
law takes a duty of easy rescue between strangers not to meet the justificatory burden
for positive duties, are the situations where it recognises positive duties consistent
with this stance? We can now turn to these situations.
1. Assumption of Responsibility
It is well established that an ‘assumption of responsibility’ can generate a duty to take

care in relation to a pure omission.44 The conditions under which an assumption of
41
The liberty argument does not entail that there are no positive rights between
strangers, as some authors claim. This depends on making out the further argument
that having a special authority over one’s own body is never consistent with having a
stringent obligation to aid another. Compare A. Ripstein, Private Wrongs, ch. 3.
42
See T. Honore, “Are Omissions Less Culpable?” in P. Cane and J. Stapleton,
Essays for Patrick Atiyah (Oxford: Oxford University Press, 1991) 31, 41-42.
43
Ripstein accepts the justifiability of a public law duty of rescue, however: A
Ripstein, “Three Duties to Rescue: Moral, Civil, Criminal” (2000) 19 Law and
Philosophy 751.
44
See, e.g., Robinson v Chief Constable of West Yorkshire, at [69] per Lord Reed, at
[115] per Lord Hughes.
12
responsibility will be held to arise are not entirely clear. 45 The authorities, however,
support the following principle:
Reliance. If A knows or ought to know that A’s conduct has led B reasonably
to rely upon A to do an act and that this reliance may result in detriment to B,
and B has in fact relied upon A’s conduct, A will be under a duty of care to
protect B from suffering detriment in reliance upon A’s conduct.
Welsh v Chief Constable of Yorkshire Police is illustrative. 46 A Crown Prosecution

Service (CPS) solicitor agreed to inform the Magistrates’ Court that the claimant’s
charges had earlier been considered in sentencing in the Crown Court. In reliance
upon this, the claimant did not attend his bail hearing for those offences in the
Magistrates’ Court and was consequently arrested. The CPS was held to have
assumed responsibility to inform the Magistrates’ Court. 47 This was on the basis that
the claimant had relied upon the assurance that the information would be passed on. 48
Similarly, if the police assure a person – an informant – that they will take care of the
informant’s physical safety in return for that person providing information, the police
will be held to have assumed responsibility to protect that person. 49 The assurance
induces the person to rely. The reliance is constituted by the provision of information
that endangers their safety.
It is not necessary that A cause B to rely by promising B that A will do an act,
although this is clearly one way in which B may be induced to rely. Mercer v South
Eastern & Chatham Railway Companies’ Management Committee illustrates this. 50
The defendant railway company had, independently of any legal obligation to do so,
continued a practice of locking railway gates when trains were passing. The claimant
was injured on one occasion when the defendant omitted to lock a gate. Lush J stated
that the defendant “ought to have contemplated that if a self-imposed duty is
ordinarily performed, those who knew of it will draw an inference if on a given
occasion it is not performed”. 51 Here, then, the company made no promise that it
would continue to lock the gates, but it was held liable on the ground that it induced a
reasonable belief that the gate would be locked if a train were coming.
Reliance is consistent with the underpinning rationale of the general rule. This
is for three reasons. First, in most cases, the defendant has a reasonable opportunity to
avoid coming under a duty by virtue of Reliance. The defendant needs only to take
steps to avoid inducing reliance upon their conduct. For instance, in Mercer, the
defendant could have given notice that railway gate would not always be locked when
trains come. Second, by inducing reliance in circumstances where this may lead to
45
See generally K. Barker, “Unreliable Assumptions in the Modern Law of
Negligence” (1993) 109 L.Q.R. 461; A. Robertson and J. Wang, “The Assumption of
Responsibility” in K. Barker, R. Grantham, W. Swain (eds) The Law of
Misstatements: 50 Years on from Hedley Byrne v Heller (Oxford: Hart Publishing,
2015).
46
[1993] 1 All ER 692.
47
[1993] 1 All ER 692, 703.
48
[1993] 1 All ER 692, 703.
49
Swinney v Chief Constable of Northumbria Police [1997] QB 464; An Informer v A
Chief Constable [2013] QB 579.
50
[1922] 2 K.B. 549.
51
[1922] 2 K.B. 549, 554.
13
detriment, the defendant creates a risk of harm to the claimant by their action which
the defendant fails to alleviate by protecting the claimant from the ensuing detriment.
Third, the defendant makes the claimant worse off than they would be independently
of the defendant in cases of Reliance.
There is support, however, in the authorities for the proposition that a duty of
care can arise in virtue of an assumption of responsibility even if Reliance is not
satisfied. Consider Morcom v Biddick’s Estate. 52 Biddick (B), an 80 year old man,
said to Morcom (M) that B that would keep the latch to the loft hatch in B’s house in
the closed position by holding it in place with a pole while M was carrying out work
on the hatch from inside the loft. B stopped doing this for a few minutes to answer his
phone. M fell through the hatch and suffered injuries. B was held liable on the ground
that he assumed responsibility to M. 53
There are two features to note. First, B’s assumption of responsibility was said to
be constituted by his “undertaking to ensure that the latch remained closed”. 54
Second, the Court of Appeal found that M did not rely upon B’s undertaking: 55
Once Mr Biddick took upon himself the task of ensuring that the latch
remained closed it seems to me that he assumed a duty to perform that task
carefully, even if Mr Morcom did not see Mr Biddick’s role as an element in
his own safety.
Furthermore, Lord Justice McCombe interpreted other authorities as suggesting that

“specific reliance upon the care exercised by the defendant is not a necessary element
of liability [in relation to assumption of responsibility]”. 56 Thus, Morcom supports the
proposition that an assumption of responsibility can exist independently of Reliance:
the duty in this case arose in virtue of a promise to do an act, a careless failure to
perform the promise, and independently of any reliance upon the promise.
A second argument supports the claim that an assumption of responsibility
may generate a duty of care in relation to a pure omission beyond the Reliance
principle. The argument is that a duty to provide positive assistance may arise in
virtue of an assumption of responsibility where there is no possibility of reliance,
because the beneficiary of the duty is unconscious. 57 In Barrett v Ministry of Defence,
an officer, A, was held to have assumed responsibility to B, a fellow officer, after B
became unconscious due to intoxication, when A took B to B’s bunk and placed him
in the recovery position. 58 The defendant was held liable for the subsequent careless
failure to attend to B, resulting in B’s death. There was no possibility, then, that B
could have altered B’s conduct as a result of A’s assurance or conduct. The same is
true in any case where a hospital doctor accepts an unconscious person as a patient,
52
[2014] EWCA Civ 182.
53
Ibid [42] per McCombe LJ.
54
Ibid [42].
55
Ibid [53] per McCombe LJ.
56
Ibid [50].
57
L.M.C.L.Q 71, 78.
58
[1995] 1 W.L.R. 1217, [1995] 3 All E.R. 87; the finding of an assumption of
responsibility was not challenged on appeal.
14
and yet the doctor’s duty to take care to improve the patient’s position arises in virtue
of an assumption of responsibility. 59
A possible response to these examples is that reliance can indeed be found in
each case. 60 The person who brings an unconscious patient to hospital relies upon the
hospital’s undertaking to treat accepted patients with reasonable care. Had this
undertaking not been given, the patient would have been taken to another hospital. In
Barrett, had B not placed the A in his bunk, others would have attended to him. 61 The
difficulty with this response is that, as Bagshaw observes, the issue of reliance is
never investigated in cases against hospital doctors. 62 Furthermore, if the law were
operating a presumption of reliance in these cases, one would expect it to be
articulated as such.
A third argument is that there are House of Lords authorities on the
assumption of responsibility concept beyond the context of liability for pure
omissions which are irreconcilable with a requirement of reliance. 63 For example, in
White v Jones, a solicitor was held to have assumed responsibility to the intended
claimant beneficiaries of a will, despite it being the case that they had not taken steps
they would otherwise not have taken as a result of an assurance given to them. 64
However, there is also support for the proposition that reliance is a necessary
condition for a duty of care to arise in virtue of an assumption of responsibility. In M
v Commissioner of Police for the Metropolis, the claimant’s case that police officers
had assumed responsibility to her in relation to a decision not to continue an
investigation was struck out on the ground that there was no reliance. 65 Lord Justice
Jacobs described reliance as a ‘vital’ part of a claim based on an assumption of
responsibility. 66 The Supreme Court decision in Michael v South Wales Police might
also be taken to support a reliance requirement. 67 In this case, an emergency call
operator carelessly failed to communicate a threat to the victim’s life to the relevant
police authority with the result that, the police failing to arrive in time, the victim was
murdered. The call operator had indicated to the victim that police would want to call
her back, but made no explicit communication that the police would arrive timeously.
The Supreme Court held that the police had not assumed responsibility to the victim
59
Capital & Counties v Hampshire County Council [1997] Q.B. 1004, 1034-1035.
60
Reliance by a third party with detrimental affect to the victim is considered
sufficient ‘reliance’ for the purposes of an assumption of responsibility: see Sherratt v
Chief Constable of Manchester Police [2018] EWHC 1746, at [80]. Such cases could
simply be considered ones where the defendant must take care in order to prevent the
claimant being rendered worse off as a result of their action.
61
For this reconciliation of Barrett with a reliance requirement, see A. Robertson and
J. Wang, “The Assumption of Responsibility” in K. Barker, R. Grantham, W. Swain
(eds) The Law of Misstatements: 50 Years on from Hedley Byrne v Heller (Oxford:
Hart Publishing, 2015), X.
62
L.M.C.L.Q 71, 83.
63
These are not pure omissions cases because in both the defendant likely made the
claimants worse off than they would be had they not acted at all.
64
White v Jones [1995] 2 A.C. 272, [1995] 1 All E.R. 691. See also Spring v
Guardian Assurance [1995] 2 A.C. 296, [1994] 3 All E.R. 129.
65
[2007] EWCA Civ 1361.
66
[2007] EWCA Civ 1361, at [27].
67
[2015] UKSC 2.
15
through the call operator. Lord Toulson’s discussion of assumption of responsibility

implies that an assumption of responsibility coupled with reliance is necessary for a
duty of care to arise: “The underlying principle rested on an assumption of
responsibility by the defendant towards the plaintiff, coupled with reliance by the
plaintiff on the exercise by the defendant of due skill and care”. 68
We cannot, however, definitively conclude from Michael that reliance is now a
requirement. The central focus of Lord Toulson’s analysis was that there was no
‘assurance’ sufficient to give rise to an assumption of responsibility on the facts. 69 He
differentiates Kent v Griffiths, 70 where an ambulance service was held to owe a
positive duty of care upon accepting an emergency call and the call operator’s
assurance made the victim’s position worse than it otherwise would have been,
because “the call handler gave misleading assurances that an ambulance would be
arriving shortly”. 71 This differentiating feature focuses upon the content of the
communications between the parties and not the presence of reliance in Kent. 72 A
requirement of reliance is not part of the ratio, then, of Michael.
The upshot of the analysis is that there are conflicting views in the authorities as to
the requirement of reliance. It is clear that the duties of care of medical professionals
who accept patients for treatment do not depend upon the existence of reliance. But,
beyond this, reliance is sometimes insisted upon (M, obiter in Michael) and
sometimes not (Barrett, Morcom). A future Supreme Court will need to decide
whether, and the extent to which, duties of care for pure omissions on the basis of an
assumption of responsibility extend beyond Reliance.
Here are some candidate non-reliance-based principles:
Promise. If A promises B that A will do an act, A will be under a tortious duty

of care to do that act.
Taking on a task. If A takes on a task for B the careful performance of which

A knows or ought to know will affect whether physical harm to B is
prevented, A will be under a duty of care to perform that task carefully.
Role or Relationship duty. If A accepts a socially recognised role or enters into

a socially recognised relationship to which positive obligations to take care to
improve B’s physical safety are conventionally attached, then A will be under
a positive duty of care in the performance of that role or relationship.
Promise is supported by Morcom: D promised to hold the pole in place and was held
to be under a duty to take care to hold the pole in place. It may also explain the duties
of care of hospital doctors: in accepting a patient, or in operating in a hospital at all,
they implicitly promise to take reasonable care of the patient’s health. It is also
68
Michael v Chief Constable of South Wales [2015] UKSC 2, at [67], see generally
[66]-[69].
69
Michael v Chief Constable of South Wales [2015] UKSC 2, at [138].
70
Kent v Griffiths [2001] Q.B. 36.
71
72
For an example of a case where the content of the communications did amount to a
sufficient assurance to generate an assumption of responsibility on the part of the
police, see Sherratt v Chief Constable of Manchester Police [2018] EWHC 1746, at
[80].
16
supported by Kent, where the assurance that the ambulance service would come itself
created the duty of care. 73 But this analysis is artificial in the case of Barrett: putting
the unconscious officer in bed is not an implicit promise of any kind: a reasonable
person would not understand this conduct as communicating an intention to become
obligated to take reasonable care. Taking on a task, which merely requires the actual
undertaking of an act, rather than any promise to do the act, explains the result in
Morcom, Barrett, and the duty of care of doctors accepting patients. In each case, the
duty-bearer could be said to have taken on a task – holding a pole, caring for a person
– careful performance of which will affect whether physical harm to another is
prevented. 74 However, Taking on a task is inconsistent with other decisions. For
instance, fire brigades who take on the task of putting out a fire at a particular
person’s property are not under a duty of care to perform the task carefully in such a
way as to prevent harm to the property. 75 Role or relationship duty differs from both
principles: it is both broader than Promise and narrower than Taking on a task. It is
broader than Promise because the duty of care that arises is one that is attached to the
particular role or relationship by social convention, it need not be one promised by the
defendant, albeit that by entering into the role, one will often be understood to be
promising to comply with its associated obligations. It is narrower than Taking on a
task because it only applies to a subset of ‘tasks’ – those which amount to socially
recognised roles or relationships of care. Role or relationship duty would explain the
duties of doctors, but not those in Morcom or Barrett.
It is submitted that Promise and Role or relationship duty are consistent with
the underlying rationale of the general rule. Under both principles, the freedom-based
objection underpinning the general rule lacks force. This is because, in both cases, the
defendant chooses to engage in a specific act or activity which attracts liability. So
long as these acts or activities are relatively narrow in scope the defendant has some
choice whether to enter into the activity. This is not to say that the positive
justification of the defendant’s liability is their choice to engage in the activity. It is
rather that the defendant’s choice removes, or weakens, their freedom-based moral
73
Kent v Griffiths [2001] Q.B. 36, at 54; [2000] 2 All E.R. 474: “The acceptance of
the call in this case established the duty of care”.
74
This principle, or a variant of it, has some support in the United States authorities
and literature: “It is ancient learning that one who assumes to act, even though
gratuitously, may thereby become subject to the duty of acting carefully, if he acts at
all”: Glanzer v Shepherd 135 N.E. 275, 276 (1922, New York) per Cardozo J. D.
Waisman, “Negligence, Responsibility, and the Clumsy Samaritan: Is There a
Fairness Rationale for Good Samaritan Immunity” (2013) 29 Georgia State
University Law Review 609, n 140 cites United States v. Lawter, 219 F.2d 559, 562
n.1 (5th Cir. 1955), where the Coast Guard were held liable when a shipwrecked
person fell to their death due to negligent operation of a helicopter cable during the
rescue attempt when there were “no boats or vessels nearby to rescue”. See also
Restatement (Second) of Torts § 323 comment e (1965): “Clear authority is lacking,
but it is possible that a court may hold that one who has thrown rope to a drowning
man, pulled him half way to shore, and then unreasonably abandoned the effort and
left him to drown, is liable even though there were no other possible sources of aid,
and the situation is made no worse than it was.”
75
Capital & Counties v Hampshire County Council [1997] Q.B. 1004, 1035. See,
similarly, East Suffolk Rivers Catchment Board v Kent [1941] A.C. 74, 102, [1940] 4
All E.R. 527; Horsley v Maclaren (The Ogopogo) [1972] S.C.R. 441.
17
objection to being under a duty to protect the claimant’s welfare. It is true that, under
both principles, the defendant is held liable for failing to confer a benefit. Part of the
justification of the general rule is simply that failing to confer a benefit is easier to
justify than harming a person. But where one promises to take care to confer a benefit,
or chooses to occupy a role which is conventionally understood to involve the
incurring of obligations to confer benefits, this objection is substantially weakened.
The defendant did not merely fail to improve the world; they failed to improve the
world in breach of a promise or in contravention of an assumed role obligation. While
Taking on a task meets the freedom-based objection, since the defendant chooses to
undertake the activity to which the duty is attached, it is not clear why merely taking
on the task, without any promissory undertaking, makes the failure to confer the
benefit worse than it would otherwise be. It is not plausible to contend that, other
things being equal, an unreasonable failure to complete a rescue is worse than an
unreasonable failure to conduct a rescue at all. Promise and Role or relationship duty
are consistent, then, with the general rule, but Taking on a task is only partially so. If,
however, Taking on a task were limited to actual performance of a promised task,
then it would fully meet the objection.
Since Promise, Taking on a task, and Role or relationship duty primarily meet
the objections underpinning the general rule, the law could, in principle, adopt them
while coherently adhering to the general rule. However, it does not follow that there is
a positive case for their adoption. Should, then, any of these principles be endorsed?
Since the primary aim of this article is to consider the consistency of exceptions to the
law on omissions with the general rule, the analysis here is necessarily tentative.
Consider, first, Promise. This principle contradicts the rule that not all
gratuitous promises are binding. 76 This counts against Promise: the fact that the law
normally requires consideration, reliance, or a deed to enforce a promise is evidence
of the need to restrict the category of legally enforceable promises. 77 This objection
does not rule out, however, a narrower version of Promise. One possibility is that
gratuitous promises to protect a person’s physical safety ought prima facie to generate
a tortious duty of care, or perhaps only promises to protect a person’s physical safety
which are made by employees of public authorities in the course of exercising their
functions. However, it is not at all obvious why the enforceability of a gratuitous
promise without reliance or a deed should cut across the divide between public and
private persons in this way. The difficulty of carving out a coherent and clear
exception to the general rule in relation to gratuitous promises, and the slender
76
This objection is already found in Coggs v Bernard (1703) 2 Ld. Raym. 909, where
Holt CJ, at X, rejected the idea that a bare promise to do something founded a claim
in damages: only the actual undertaking of the promised act imposed a duty to be
careful in doing it. It may be responded that Promise only imposes a duty to take care
to perform a promise, not a duty to perform the promise. It does not, therefore, strictly
enforce a gratuitous promise. However, if a duty to take care to perform a promise
requires one to perform a promise unless there is significant justification for not doing
so, in substance it involves the enforcement of a gratuitous promise. Further, if the
duty to take care to perform the promise is merely a logical implication of the duty to
perform, then it could also be said to be a gratuitous promise.
77
N.J. McBride and R. Bagshaw, Tort Law, 6th edn, (Harlow: Pearson Publishing,
2015), x.
18
support in the authorities for such a principle, tends to suggest that Promise or a
variant thereof should not be adopted. 78
Consider, now, Taking on a task. The main objection to Taking on a task is
that its rationale is unclear. If one is under no duty to confer a benefit, why does
taking on the task of conferring that benefit impose a duty to take care to confer it?
The mere fact that a person indicates by action an intention to confer a benefit seems
a weak reason to impose an obligation to take care to confer it. Two possible
rationales may be considered.
First, in taking on a task, one typically increases the probability that others will
be dissuaded from taking on the task. One’s taking on the task may therefore increase
the probability that the person assisted will be rendered worse off than had one not
intervened. This justification fails, however, in cases where it is entitrely clear that no
one else would have intervened had the duty-bearer not taken on the task.
Second, it could be argued that by taking on the task of providing a benefit,
one chooses to forego one’s freedom-based objection to the imposition of a positive
obligation. Generally, the law does not enforce a person’s moral duty to provide
assistance to a person at risk of imminent physical danger when they could provide
this assistance at minimal cost. But, it may be argued, where a person chooses to go to
another’s assistance, they cannot say that a positive obligation is being foisted upon
them. This justification has some merit, but it would support a much narrower version
of Taking on a task – one which is limited to cases where a person is taking on a task
which would be required by a moral duty of easy recue. This is likely to be limited to
cases involving persons at significant risk of imminent physical harm that can be
avoided at minimal cost to the duty-bearer. However, other things being equal, it does
seem unfair to impose a duty to take care upon careless rescuers (who are not
otherwise under a duty to act) but not on those who deliberately fail to provide
assistance that they could provide at minimal cost. This treats those who choose to
aim to protect a person from harm unfairly. 79
Finally, consider Role or relationship duty. It is submitted that this is the better
explanation for a doctor’s positive duty towards an accepted patient. The role of
doctor is a socially constructed package of duties, duties which can be modified by
the law, and which include a positive duty of reasonable care. 80 By entering into the
role, one undertakes to adhere to those duties. Upon acceptance of a patient, the role
duty becomes focussed upon the particular individual patient. 81
78
For a recent re-iteration of the proposition that a bare promise does not generate a
duty in negligence, absent the actual undertaking of the task itself: Lejonvarn v
Burgess [2017] EWCA Civ 254, [2017] P.N.L.R. 25, at [67]-[68]. Furthermore, even
Morcom can be rationalised as a case where the duty arose only because of the
performance of the undertaking.
79
N.J. McBride and R. Bagshaw, Tort Law, 6th edn, (Harlow: Pearson Publishing,
2015), x. Futhermore, the acceptance of Taking on a task could discourage persons
from assisting others in need due to the threat of legal liability if they are careless in
the course of doing so: D. Nolan “The Liability of Public Authorities for Failing to
Confer Benefits” (2011) 127 L.Q.R. 260, 282.
80
See further M. Hardimon, “Role Obligations” (1994) 91 Journal of Philosophy 333.
81
Some positive duties to non-patients can also be coherently attached to the role if
there is policy justification for doing so: see, e.g., ABC v St George’s Healthcare NHS
Trust [2017] EWCA Civ 336, [2017] P.I.Q.R. P15. Once imposed, this duty can be
understood as chosen by virtue of acceptance of the role.
19
In summary, this section has demonstrated that each possible version of the
assumption of responsibility version idea: Reliance, Promise, Taking on a Task, Role
or relationship duty is consistent with normative rationale of the general rule. It has
further shown that the law has not adopted a clear position on when reliance is a
necessary condition for liability. It has tentatively rejected two non-reliance-based
principles – Promise and Taking on a task – and endorsed one: Role or relationship
duty. However, further analysis will be required as to precisely which roles or
relationships merit legal enforcement of their positive obligations.
2. Creation of risk
In Mitchell v Glasgow City Council, Lord Scott explained that: “[i]f a

defendant has played some causative part in the train of events that have led
to the risk of injury, a duty to take reasonable steps to avert or lessen the
risk may arise”. 82 Lord Rodger gave this example: “If I negligently collide with a
cyclist who is knocked unconscious, I must surely take reasonable care to move her
from the path of oncoming vehicles”. 83 Two points may be made as to the scope of
this principle. First, the defendant’s conduct must be an act: it is not possible to
contribute to the risk of harm through a pure omission. 84 This is simply part of the
meaning of ‘creation of risk’. Second, the principle extends to innocently creating a
risk or causing damage, thereby increasing the risk of a person suffering further
harm. 85 In Lord Rodger’s example, then, it is not essential that the collision is
negligent. The principle can be formulated thus:
Creation of risk. If A creates a risk of physical harm to B, and A knows or

ought to know of the existence of this risk, then A is under a duty of care to
protect A from the consequences of that risk.
Is this principle consistent with the rationale of the general rule? In truth, Creation of
risk is not an exception to the general rule at all: it is merely an instance of the type of
conduct that, according to the general rule, normally attracts a duty of care without
special reason being required. This is especially clear if Creation of risk is limited to
risks which were reasonably foreseeable prior to their creation. If so, Creation of risk
simply describes the basic situation in which a person will be under a duty of care in
the tort of negligence: where they create the risk of foreseeable of physical harm by
their act.
However, Creation of risk appears to extend to risks which were unforeseeable ex
ante, of which the defendant later becomes aware or of which they ought later to be
82
[2009] 1 A.C. 874, 893. N.J. McBride and R. Bagshaw, Tort Law, 5th edn, (Harlow:
Pearson Publishing, 2015) §7.3; see also Restatement Third of Torts, §39.
83
[2009] 1 A.C. 874, 899.
84
See the reference to ‘positive act’ in this context by Lord Hoffmann in Gorringe v
Calderdale [2004] 1 WLR 1057, at [13].
85
Hobbs v Baxenden Chemicals [1992] 1 Lloyd’s Rep. 54, 65: “a manufacturer’s duty
of care does not end when the goods are sold. A manufacturer who realises that
omitting to warn past customers about something which might result in injury to them
must take reasonable steps to attempt to warn them, however lacking in negligence he
may have been at the time the goods were sold”. See also Mitchell v Glasgow City
Council [2009] 1 A.C. 874, 894.
20
aware. For instance, suppose that a manufacturer later becomes aware of a dangerous
defect in a circulated product, but the defect was not reasonably foreseeable prior to
distribution in the market. Here, Creation of risk generates a duty of care in relation to
a risk which was unforeseeable at the time of its creation. However, it is doubtful
whether a distinct ‘exception’ is needed to accommodate this. When a person is held
to be under a duty of care under Creation of risk, their duty is to take care to prevent it
being the case that the person is worse off as a result of their own conduct. That is
merely an instantiation of the general rule.
3. Control
The fact that A has control over a person or object has been recognised to justify a
duty upon A to take care to protect B from foreseeable risks of physical harm
connected with the person or object. Thus in Dorset Yacht v Home Office, Borstal
officers were held to be under a duty of care to surrounding property owners to take
care to prevent the Borstal boys in their custody from escaping and causing damage to
surrounding property. 86 Lord Morris emphasised the fact that the officers had “a right
to exercise control over the boys” and Lord Pearson observed that “control imports
responsibility”. 87
Is the recognition of a duty of care in such situations consistent with the
underpinning of the general rule? A natural suggestion is that control justifies a duty
of care because the fact of control indicates that A is well placed to prevent the harm
risked by the person or object. The existence of control singles out the controller as
particularly able to prevent or minimise the occurrence of harm. This explanation is
difficult to reconcile, however, with the general rule. Even if a stranger is uniquely
well placed to rescue a drowning child from a shallow pool of water, it is clear they
are under no legal obligation to do so by virtue of the general rule.
Ripstein offers another explanation. He argues that the cases in which control
is invoked can be explained simply as cases where one person imposes a foreseeable
risk of physical harm upon another. 88 In essence, this collapses the ‘control’ category
into the creation of risk category: the claim is that the defendant creates, or at least,
imposes, a risk of physical harm, and this generates a duty of care. This idea can
explain some of the cases. For instance, in Lewis v Carmarthernshire, a nursery
school teacher was held to owe a duty of care to a driver not to allow a three year old
child in her charge to escape unattended from the school onto a nearby road where the
latter posed a danger to the driver, who was killed when swerving to avoid the child. 89
The House of Lords drew attention to the fact that the presence of children near a road
posed significant risks of harm both to the child and drivers. 90 Running a school near
such a road, then, could reasonably be considered to involve the imposition of a risk
of physical harm on others.
This argument is convincing, but it needs refinement and supplementation in two
respects. First, the current law draws a distinction in this context between risks over
which one has control which one chose to create and those which one did not choose
to create. Thus, in cases where a source of danger to B’s land arises on A’s land
86
[1970] A.C. 1004, [1970] 2 All E.R. 294.
87
[1970] A.C. 1004, at 1038 and 1055 respectively.
88
A. Ripstein, Private Wrongs, p.94.
89
Carmarthenshire County Council v Lewis [1955] A.C. 549, [1955] 1 All E.R. 565.
90
Carmarthenshire County Counvil v Lewis [1955] A.C. 549, at 564 per Lord Reid.
21
without A’s responsibility for the existence of that risk, and A knew or ought to have
known of its existence, A comes under a duty of care, but that is a measured duty of
care. That is to say: the duty imposed demands reasonable care, but the demands of
reasonable care are tailored to the defendant’s physical and mental capacities. 91
Second, not all cases where a duty is justified on the basis of control are
straightforwardly analysed as cases where A has created or imposed a risk. Consider
Goldman v Hargrave, where the defendant was under a duty to take reasonable care
in relation to a tree that caught fire on his land as a result of a lightning strike. 92 It
would be incorrect to say that the defendant created the risk of harm to the claimant in
Goldman: the risk was created by the lightning strike. It is true that the defendant’s
property posed a risk to the claimant and this is indeed sufficient justification for
imposing a duty of some kind upon the defendant. But the Goldman v Hargrave duty
seems to arise even in relation to risks on one’s land that one neither created nor
which are posed by part of the land itself. For instance, B could be under a measured
duty to take steps to eject trespassers who are using the land as a base from which to
damage A’s land. 93
A duty in such situations can sometimes be justified on the basis that the
defendant’s having a right in the land creates a risk that a failure to act will render the
claimant worse off. The fact that one has a property right over a space itself gives
people reasons not to enter that space without one’s consent. So it might be argued
that your having property in the land is likely to put people off from intervening if
they become aware of a danger on the land. But the duty arises without the need to
prove that any specific individual has been put off from intervening to quell the
danger. This explanation is therefore insufficiently general.
Stevens argues that the duty is justified as an incident of the defendant’s right to
exclude. 94 In return for the benefit of the right to exclude from a space, one is
legitimately burdened with an obligation to take care that objects on or parts of that
space do not harm neighbouring land. This explanation is incomplete: why is this
particular obligation the quid pro quo of the right to exclude? If the reason is because
the existence of the right creates a risk that others will not intervene, this explanation
faces the generality objection. A second problem with Stevens’ explanation is that it
generalises to bodily rights. Why not say that in return for legal protection of one’s
bodily rights through the criminal law and tortious regulation of one’s rights, one has
an enforceable obligation to assist others in peril? The answer cannot be that one
freely chooses to occupy land: not occupying land is not a feasible alternative.
It is submitted that a duty of care to protect others from risks on one’s land or other
property can, nonetheless, be justified as consistent with the general rule. This is
because the duty has limited scope: it only arises in relation to dangers imposed by
one’s land or which arise on one’s land. By contrast, a duty to assist endangered
people in general would apply across all aspects of a person’s life. It thus raises a
more significant liberty-based challenge than a limited property-based obligation.
In summary, then, some cases of control are consistent with general rule on the
basis that they are simply cases where one creates or imposes a risk through a person
91
Goldman v Hargrave [1967] 1 A.C. 645, at 662-663, [1966] 2 All E.R. 989. For a
recent summary: Vernon Knight Associates v Cornwall Council [2013] EWCA Civ
950, [2014] Env. L.R. 6, at [49].
92
Goldman v Hargrave [1967] 1 A.C. 645.
93
Lippiatt v South Gloucestershire Council [2000] Q.B. 51.
94
R. Stevens, Torts and Rights (Oxford: Oxford University Press) at p.16.
22
or object. Other cases are justified on the basis that the positive obligation imposed is
not a general one spanning all aspects of a person’s life.
4. Special Relationships
The existence of a special relationship between the injured person and the duty-bearer
is recognised as a source of positive tortious duties. A special relationship is a
relationship which exists between people which does not obtain between all people. In
Michael, Lord Toulson listed the following as “relationships in which a duty to take
positive action typically arises: contract, fiduciary relationships, employer and
employee, school and pupil, health professional and patient”. 95 Others typically listed
in this category include: parent and child, 96 landlord and tenant, and occupier and
visitor. 97 This section examines how, in principle, a relationship could justify the
imposition of a positive duty of care and whether such a justification is consistent
with the general rule.
We can distinguish, following John Gardner, two possible ways in which a
relationship could justify a duty. 98 The first, let us call it the reductive view, holds that
a relationship itself does not justify the duty, but rather the relationship is associated
with other facts which justify the duty. The second, let us call it the relational value
view, holds that the value of the relationship in which the duty-bearer and victim stand
itself justifies the duty. Consider, for instance, the relationship between a school and a
pupil. We have seen that a schoolteacher will owe a duty to a driver to take care not to
allow a child onto a nearby road. 99 As we have also seen, this duty could be justified
simply on the ground that having the child near a busy road creates a risk of harm to
drivers. If a stranger in control of a child brings the child near a busy road, they too
would have a duty of care to the road user. 100 This account of the schoolteacher’s duty
is consistent with the reductive view: the value of the schoolteacher and pupil
relationship has no role in the justification of the duty. 101 What justifies the duty is
merely the fact that one person is imposing a significant risk upon another person
through their activity. If all special relationship duties in negligence can be reduced to
duties which arise in virtue of an assumption of responsibility, control, or creation of
risk, then their consistency with the general rule follows from the analysis of these
categories already given. 102
Are there any positive duties of care which arise because of the value of the
relationship itself between the parties? One might think that parenthood is clear
95
96
AJ Allan (Blairnyle) Ltd v Strathclyde Fire Board [2016] CSIH 3, [62].
97
N.J. McBride and R. Bagshaw, Tort Law, 5th edn, (Harlow, Pearson Publishing,
2015), 245-248.
98
J. Gardner, From Personal Life to Private Law (Oxford: Oxford University Press,
2018), at pp.25-26.
99
Above, x.
100
McCallion v Dodd [1966] 1 N.Z.L.R. 710, at 724 per Turner J, at 729 per
McCarthy J.
101
This is consistent, however, with thinking that the value of the relationship affects
other features of the duty, such as its stringency.
102
Lord Toulson in Michael treats special relationship as an instance of the
assumption of responsibility category: [2015] UKSC 2, at [101].
23
example. Although there is a paucity of authority in English law, it is relatively clear

that a parent may have positive obligations to their child in negligence. 103 It is
unclear, however, whether the fact that a person is a parent has any independent
justificatory significance in the law on omissions. 104 In many cases, the control
exercised over the child by the parents (biological or otherwise) will justify a duty to
the child in the following way. By exercising parental authority over the child, the
parents have acted in such a way as to prevent the child being cared for by others:
others will (to some extent) defer to their authority over the child. A failure to take
care of the child will thus often render the child worse off than it otherwise would
have been. Similarly, in any case where the parent brings the child near a source of
physical danger that the child is ill equipped to take steps to avoid by exercising its
own capacities, the parent could be said to have created a risk of physical harm to the
child. 105
The issue of whether the relationship as such has justificatory significance has
been addressed in other jurisdictions. In Hahn v Conley, the High Court of Australia
considered whether a grandfather owed a duty of care to the claimant, his
granddaughter. The child was staying at her grandparents’ home for the weekend
when she ran across the road to the grandfather after he called out to her from the
other side of the road. She was injured by a vehicle when crossing. Barwick CJ,
McTiernan J, Windeyer J, and Menzies J each agreed that the mere fact that a person
is connected by a blood relationship to another person is insufficient to generate a
positive duty to act. 106 A parental duty is triggered by facts, such as creation of risk,
and taking the child into one’s charge, that would also trigger a duty in a stranger: the
mere fact of parenthood is not the source of tortious obligation. 107 Similarly, in
McCallion v Dodd, McCarthy J and Turner J both explained parental duties of care as
grounded in facts that would generate duties of care in similarly situated strangers to
the child. 108 Both considered that it is either the creation of danger or an undertaking
to care for the child – an undertaking that could be given by a stranger who takes care
103
Consider the duty of care in supervising a child using a bouncy castle organised by
the parent recognised in Harris v Perry [2008] EWCA Civ 907. The obiter doubt
expressed in XA v XY [2011] P.I.Q.R. P1, at [143], as to whether there is a parental
duty of care to protect a child from abuse by another parent was not based on the fact
that the putative duty would be a positive one.
104
It should be noted that whether parental positive duties can be justified by
reference to other legal categories, such as control, is logically a separate question
from whether the value of the relationship of parenthood is itself part of the legally
recognised justification of these duties. The claim here is that both are unclear.
105
This justifies the parental duty acknowledged (but held not to have been breached
in the case of a foster parent) in Surtees v Kingston upon Thames Royal Borough
Council [1992] P.I.Q.R. P101.
106
Hahn v Conley (1971) 126 C.L.R. 276, at 284 per Barwick CJ, at 287 per
McTiernan J, at 288 per Menzies J, and at 294 per Windeyer J. Barwick CJ,
McTiernan and Windeyer J, the majority, held that no duty arose at all. Menzies J
held that a duty arose but was not breached.
107
Hahn v Conley (1971) 126 C.L.R. 276, at 284 per Barwick CJ: “parenthood is not
itself the source of the duty”.
108
McCallion v Dodd [1966] 1 N.Z.L.R. 710, at 724 per Turner J, at 729 per
McCarthy J.
24
of a lost child – that generates the duty. 109 Even, then, the duties attached to a parental
relationship have been considered to be reducible to other sources of duty.
Would it be consistent with the general rule for the law to recognise that the value
of a special relationship itself generates special duties? There are two reasons to
answer in the affirmative. First, in almost all cases, special relationships are
voluntary: one chooses to enter or remain in them. 110 Of course, the freedom with
which persons enter special relationships is not always symmetric. Parents typically
choose to have children; children do not choose to be born. This asymmetry may
justifiably affect the incidence of positive duties in these relationships. Second,
special relationships are, by definition, limited in scope: one is only in a special
relationship with a limited category of persons. This helps to meet the general
freedom-based objection to positive duties.
CONCLUSION
The first part of the article demonstrated that the act-omissions distinction does not
fully capture the distinction between situations where a duty of care generally arises
in relation to physical damage simply upon the basis of reasonable foreseeability of
that damage, and those where a duty of care in respect of physical damage does not
generally arise, in virtue of the nature of the defendant’s conduct. It proposed a
revised principle that centres around whether the defendant’s conduct makes the
victim worse off than they would have been independently of the defendant’s body or
resources at the time of the careless conduct.
The second part of the article argued that the law enjoys a large measure of
overall coherence in the sense that the currently recognised exceptions to the general
rule – assumption of responsibility, control, creation of risk, and special relationships
– can all be understood as consistent with the normative underpinnings of the general
rule. The second part of the article also demonstrated, however, that the law on
omissions is subject to criticism in four main respects. First, the concept of an
assumption of responsibility refers to several importantly distinct ideas which are
rarely disambiguated. Second, the law has adopted an inconsistent position on the
requirement of reliance in relation to assumption of responsibility. The article argued
against the acceptance of two non-reliance based principles, but tentatively endorsed
one – the Role or relationship duty principle. Third, the creation of risk ‘exception’ is
not truly to be classified as an exception. Fourth, the justificatory basis of the ‘special
relationship’ exception is unclear: it is not clear whether special relationships can be
reduced into the other exceptional categories.
Uploaded by : Ahmed I H
Lecturer - Newcastle Law Academy
Mob : +8801911080106
109
Ibid.
110
Robinson v Chief Constable of West Yorkshire Police [2018] UKSC 4, [69.4] per
Lord Reed.
25

Sandy-Steel-Rationalising-Liability-for-Omissions-in-Negligence (1) - 1

Uploaded by

Copyright:

Available Formats

Sandy-Steel-Rationalising-Liability-for-Omissions-in-Negligence (1) - 1

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sandy-Steel-Rationalising-Liability-for-Omissions-in-Negligence (1) - 1

Uploaded by

Copyright:

Available Formats

Forthcoming in 2019 Law Quarterly Review. Please cite the published version.

RATIONALISING OMISSIONS LIABILITY IN NEGLIGENCE

II. The General Rule

The general rule is conventionally formulated in this way: 5

An omission is a failure to act. In Robinson v Chief Constable of West Yorkshire

Acts. A is generally under a duty of care to B in the tort of negligence in

Beyond acts and omissions

The function of the general rule is to identify a category of conduct in respect of

a. Making worse and failing to make better

No benefit: A is not generally liable to B in negligence for failing to confer a

This formulation is appealing in that it tracks an intuitively important moral

No Benefit 1: A is not generally liable to B in tort for failing to make B better

No Benefit 2: A is not generally liable to B in tort for failing to make B better

No Benefit 3: A is not generally liable to B in tort for failing to make B better

This suggestion is uninformative. This proposal says: a person is entitled to be made

No Benefit 4: A is generally only liable in negligence to B if A’s conduct

The problem of overinclusiveness

classified. 18 The objection, then, is that No Benefit 4, in combination with a revised

The problem of underinclusiveness (1): harming and benefitting

No rescue. A walks past B, an unconscious stranger, who is drowning. B dies.

Aborted rescue. A sees B, an unconscious stranger, who is drowning. A begins

Damaging rescue. A sees B, a lightweight, drowning, unconscious, stranger.

No Benefit 5. A is generally only liable in negligence to B if A’s conduct

The problem of underinclusiveness (2): overdetermination

The problem of underinclusiveness (3): removal of one’s own property

In summary, No Benefit 5 provides an appealing formulation of the general

Creation of risk: A is not generally liable to B in negligence in respect of a

III. EXCEPTIONS TO THE GENERAL RULE

It is well established that an ‘assumption of responsibility’ can generate a duty to take

Welsh v Chief Constable of Yorkshire Police is illustrative. 46 A Crown Prosecution

Furthermore, Lord Justice McCombe interpreted other authorities as suggesting that

through the call operator. Lord Toulson’s discussion of assumption of responsibility

Promise. If A promises B that A will do an act, A will be under a tortious duty

Taking on a task. If A takes on a task for B the careful performance of which

Role or Relationship duty. If A accepts a socially recognised role or enters into

In Mitchell v Glasgow City Council, Lord Scott explained that: “[i]f a

Creation of risk. If A creates a risk of physical harm to B, and A knows or

example. Although there is a paucity of authority in English law, it is relatively clear

You might also like