ńňđ. 3
(âńĺăî 23)



(a) Murder;
(b) Extermination;
(c) Enslavement;
(d) Deportation or forcible transfer of population;
(e) Imprisonment or other severe deprivation of physical liberty in violation of funda-
mental rules of international law;
(f) Torture;
(g) Rape, sexual slavery, enforced prostitution, forced pregnancy, enforced sterilization,
or any other form of sexual violence of comparable gravity;
(h) Persecution against any identifiable group or collectivity on political, racial, national,
ethnic, cultural, religious, gender as defined in paragraph 3, or other grounds that are
universally recognized as impermissible under international law, in connection with
any act referred to in this paragraph or any crime within the jurisdiction of the Court;
(i) Enforced disappearance of persons;
(j) The crime of apartheid;
(k) Other inhumane acts of a similar character intentionally causing great suffering, or
serious injury to body or to mental or physical health.
Per Saland, ‘International Criminal Law Principles’, in Lee (ed.), The International
Criminal Court, 189, 195.
Article 7(1) states, in the relevant part that:
For the purpose of paragraph 1:
a. ‘Attack directed against any civilian population’ means a course of conduct involving the
multiple commission of acts referred to in paragraph 1 against any civilian population,

There are further examples of this problem. Article 7 of the Statute of the
ICTR states, for example, in paragraph 4 that:
The fact that an accused person acted pursuant to an order of a
Government or of a superior shall not relieve him of criminal responsi-
bility, but may be considered in mitigation of punishment if the
International Tribunal determines that justice so requires.
Paragraph 4 is identical to paragraph 4 of Article 7 of the Statute of the
ICTY. In contrast, Article 33 of the Rome Statute is cast in altogether different
1. The fact that a crime within the jurisdiction of the Court has been
committed by a person pursuant to an order of a Government or of a
superior, whether military or civilian, shall not relieve that person of
criminal responsibility unless:
(a) The person was under a legal obligation to obey orders of the
Government or the superior in question;
(b) The person did not know that the order was unlawful; and
(c) The order was not manifestly unlawful.
2. For the purposes of this article, orders to commit genocide or crimes
against humanity are manifestly unlawful.

My point is not that all these details (which the Schmid/Lador-Lederer
proposal also faces with the customary definition of war crimes) are insur-
mountable, but that there is a more direct, indeed more intellectually honest,
approach which we will now turn to.

C. The ‘treaty crimes’ issue in Rome: terrorism as a crime to be
defined by future treaty negotiation
The true value of Professor Cassese’s proposal lies in its suitability as a
starting-point for international negotiations on a generic definition of
terrorism. Provided such a definition could also absorb whatever past con-
sensus has already been achieved in the form of the several treaty regimes
already in place on hijacking, hostage taking, and so on, there is no reason in
principle why crimes against humanity could not become a component, even
an important component, of a generic definition of international terrorism.
This brings us to a (third) broader approach that was adopted by some
delegations in the negotiations leading up to the Rome Statute.
These delegations sought to include terrorism within the jurisdiction of
the ICC at some point in the future when an appropriate definition can be

pursuant to or in furtherance of a State or organizational policy to commit such attack
(emphasis added).

agreed upon by the state parties to the Rome Statute (i.e. a definition by
further negotiation).
As von Hebel and Robinson point out, the question of including terrorism
within the jurisdiction of the ICC is bound up with the broader proposal
to include a list of other ‘treaty crimes’ in the Rome Statute; principally ‘drug
trafficking’, ‘crimes against United Nations and associated personnel’,
and ‘crimes involving the illicit traffic in narcotic drugs’.54 As Wilmshurst
has also noted,55 Article 20(e) of the International Law Commission’s 1994
Draft Statute had earlier referred to ‘crimes, established under or pursuant to
the treaty provisions list in the Annex, which, having regard to the conduct
alleged, constitute exceptionally serious crimes of international concern’.
Of the fourteen treaties listed in the Annex, six were concerned with terrorist
offences from hijacking to hostage-taking, amongst others.56
Professor Wilson of the American University, Washington D.C. reports
however that:
The major concern of the delegates appeared to be that such crimes, while
defined by treaty, have not been uniformly subscribed to or recognized, and
their exclusion from ‘core crimes’ of the ICC might facilitate the acceptance
of jurisdiction of the Court by States that are not parties to the treaties in
question. The view was also expressed that the drafters had failed to include
other important treaty-based offenses such as crimes against the

The inclusion of measures to combat the drug trade was due to public pressure that was, at
the time, brought to bear on the negotiations.
Elizabeth Wilmshurst, ‘Jurisdiction of the Court’ in Lee (ed.), International Criminal
Court, 127, note 10.
Draft Statute for an International Criminal Court, Report of the International Law
Commission on the work of its forty-sixth session, 2 May–22 July 1994, Chapter II.B.I.
United Nations General Assembly Official Records, Forty-ninth Session, Supplement No.
10, A/49/10 (1994), 29–140. Principally, unlawful seizure of aircraft defined under the
1970 Hague Convention; hostage-taking and related crimes defined under the 1979
Hostages Convention, piracy under the 1988 IMO Convention, and treaty definitions of
apartheid and crimes against diplomatic personnel; discussed in Richard J. Wilson, ‘A
Permanent International Criminal Court: Impunity Loses Another Round’, paper pre-
sented by Professor Wilson at the International Conference on ‘Impunity and its Effects
on Democratic Processes’, Santiago de Chile, 14 December 1996 (available on the world-
wide web).
Wilson, ibid. See also Kai Ambos, ‘Establishing an International Criminal Court and an
International Criminal Code: Observations from an International Criminal Law
Viewpoint’ (1996) 7 European Journal of International Law 519, 527: ‘There are crimes
which are virtually unrecognized by the state community in spite of their codification in
international treaties, for example mercenarism. Or there are treaty crimes, whose
wrongfulness is less than that of other non-treaty crimes, which are nevertheless recog-
nized by customary international law. In spite of this fact, these non-treaty crimes do not –
according to the treaty approach – fall within the jurisdiction of an ICC.’

Nonetheless, these ‘treaty crimes’ (so-called by the International Law
Commission)58 were eventually included in the list of negotiating items
following a February 1997 (Rome) Preparatory Committee Decision to do
so, with the caveat that their inclusion was without prejudice to the question
as to whether they would ultimately be included in the Rome Statute.59 On
14 July 1998, during the subsequent Rome Conference, the delegations that had
originally proposed the inclusion of terrorism within the list of crimes
suggested that the same compromise should be made as had been struck
over an appropriate definition of the crime of aggression. Specifically, this
was to list the crime in the Statute but to leave its precise definition for another
day prior to which the ICC would not be able to exercise jurisdiction over
that specific crime. In negotiating terms this was a neat device by which to
keep the issue firmly on the negotiating table. But there was scant support
for this proposal.
As things turned out, the delegation from Turkey only succeeded in
ensuring the inclusion of a resolution in the Final Act of the Rome
Conference, recommending that a future Review Conference should recon-
sider achieving an acceptable definition of terrorism with a view to its
inclusion in the Rome Statute and in order that terrorism should fall to be
considered as a crime within the jurisdiction of the ICC.60

VII. Conclusion
We have seen how the original ILC Draft Statute contained a list of treaty
crimes, based on the sorts of terrorist acts identified in the earlier subject-
specific treaties. From this viewpoint, the definition of terrorism is as
straightforward as it is merely the sum total of all the available subject-
specific definitions. These various subject-specific treaties remain the most
reliable sources of international law today as to what terrorism looks like.
Beyond these treaties there arises the question of what a general customary
law definition would look like, and whether these treaty rules may have
stimulated broad acceptance by states of parallel customary rules since
these treaties have arguably resulted from the opinio juris communitatis of a
not insignificant number of states.61 But beyond these subject-specific rules
that are focused only on some well-known methods of terror, the two

Defined by the International Law Commission as ‘crimes established under or pursuant to
the treaty provision of an international treaty which, having regard to the conduct alleged,
constitute exceptionally serious crimes of international concern’; ILC Commentary to
Article 20 of its Draft Statute, ibid., 70 and 147–157.
Decisions Taken by the Preparatory Committee at its session held from 11 to 21 February
1997, A/AC.249/1997/L.5 (1997).
Final Act of the Conference, Resolution E, A/CONF.183/C.1/L.76/Add. 14, 8.
See also Freestone, ‘International Cooperation’, 60.

pronouncements by the American and French courts and the traditional
differences between states on the international plane count against the exis-
tence of a more comprehensive definition of what terrorism is under cus-
tomary international law.
The end result, in terms of what international law is today, is a body of
international rules that will tell you that someone who traffics in plastic
explosives, or blows up a fixed platform on a continental shelf, or who
blows up or hijacks an airplane, or takes hostages is a terrorist in the eyes of
international law, and that certain states that are party to certain treaties
could claim jurisdiction over such persons, and would prosecute or extradite
these persons. One could even argue that some of these treaty rules are
customary rules. But beyond that international law has nothing to say, for
example, about whether an ordinary murderer convicted under domestic
penal law also becomes a terrorist because of the political or other motivation
he has in committing murder.62 International law is silent on the question of
what terrorism is, generally speaking.

For some domestic legal systems in respect of which offence categories under international
law would become common law offences too, that international law may call such a
person a terrorist and may define terrorism as an international crime would prove crucial
in so far as a court would not otherwise have the offence category under its domestic penal
law or even if it already does, may nonetheless not be entitled to extraterritorial jurisdic-
tion should the offending act have been committed abroad. In other countries, even
within the common law family, common law offences may not be possible because only
statutory offences are recognized. In Singapore, which is shy of common law extraterri-
toriality, it is an open question whether the Singapore courts could assert jurisdictional
competence over universal crimes outside of specific statutory provision, such as the Penal
Code provisions on piracy iure gentium and slavery, or offences provided for in pursuance
of one of the subject-specific terrorism treaties to which Singapore is party. See further
C. L. Lim, ‘Singapore Crimes Abroad’, (2001) Singapore Journal of Legal Studies 494, note
118, and the discussion of common law extraterritoriality therein. There is also an
apparent loophole in respect of genocide which I raised there. Singapore is a party to
the Convention on the Prevention and Punishment of the Crime of Genocide, 9 December
1948 but there appears no specific offence of genocide in Singapore’s statute books. This
could lead to difficulties were a person accused of having committed genocide were to
arrive on Singapore’s shores. For the debate in Parliament on this between Nominated
Member, Associate Professor Simon Tay and the Honourable Minister for Law, Professor
S. Jayakumar, see the Singapore Parliamentary Reports, Vol. 73, Sess. 9 (13 March 2001),
cols. 928–32. For some complex but interesting developments in customary international
law in respect of genocide, see Guglielmo Verdirame, ‘The Genocide Definition in the
Jurisprudence of the Ad Hoc Tribunals’ (2000) 49 International and Comparative Law
Quarterly 578. So far as the post 9/11 war against terrorism is concerned, Singapore has
also enacted parent legislation under which the Minister may, under direction from the
United Nations Security Council, provide for extraterritorial penal offences under
Singapore’s United Nations Act, on which see C. L. Lim, ‘Executive Lawmaking in
Compliance of International Treaty’ (2002) Singapore Journal of Legal Studies 73.

Admittedly, there are some plausible proposals that we have seen towards
the question of a generic customary definition. These would link any cus-
tomary law definition of terrorism to Rome in some way, either by linking a
definition of international terrorism with war crimes or crimes against
humanity, or simply by advocating that any definition of international
terrorism should occur in the context of a review of the Rome Statute. The
link to Rome is as it should be, for even in the League era the question of a
convention on terrorism was linked, as we have seen, to the creation of an
international criminal court. However, I do not find any approach which
might rely on summing the political will of states to negotiate a comprehen-
sive definition with an end in sight particularly promising in the short term.
Not because a Draft Comprehensive Convention on Terrorism might not
result, indeed it is expected to come into being, but because the old political
problems stretching back to 1937 and which we have since also witnessed in
countless General Assembly debates are all still there. The existence of these
‘old problems’ suggests that such a convention could have more of a symbolic
role than a real one. I do not wish to be overly cynical, but treaty lawmakers
know that if a treaty is what their political masters and others really want, you
just need to give them something they can call a treaty and leave it at that. So
far as efforts on the international plane are concerned, a strategy of avoiding
the old political pitfalls seems therefore more attractive, making the regime of
subject-specific treaties more attractive than talk of replacing or augmenting
these treaties which we already have and which work quite well with a
comprehensive treaty definition of terrorism. According to this view, it is
not as important to decide what a terrorist looks like as it is to decide what
methods used by terrorists need to be dealt with, and how we might wish
to do so.
In practical terms, the most promising path towards a useful, as opposed
merely to a symbolic, comprehensive international definition of terrorism
also lies in leaving terrorism largely to be defined by domestic law – by
domestic courts, domestic legislative activity, or both. Each state must cur-
rently push ahead according to its own lights. But there are clear fairness-
based difficulties with the construction by domestic courts of offences and
defences from such an amorphous body of material evidence of the applicable
customary rules. At the best of times, the reception of customary law into
domestic law by way of the exercise of judicial powers is haphazard, but where
the liberty of the individual is concerned, to do so risks politicizing any trial
to a degree that may simply be considered unacceptable under the constitu-
tions and legal cultures of a whole host of democratic nations. This could
suggest that some states might consider themselves better advised if the
legislative branch were to step into the void instead – both to ‘transform’
available customary standards into domestic law via legislation, and to
progressively fill in the gaps where the customary rule is still in its infancy

or remains relatively ill-developed. In due course, should there arise sufficient
similarities in legislative activity, sufficient borrowing and adaptation of
legislative approaches, this could pave the way for eventual harmonization.
If this happens, instead of the previous approach of having treaties that
required domestic legal reception of new categories of international crime
such as hijacking and so on, domestic legislative activity could drive the
international lawmaking process instead, bottom-up so to speak.
Approaching the matter in this bottom-up fashion could also have the
ancillary advantage of ensuring both greater rule-precision and a greater
breadth of state participation. International lawmakers could simply engage
in a process of codification by seeking out clear commonalities in compara-
tive legislative activity, coupled with a measure of progressive development
through gap filling, rather than quibble over an appropriate starting point for
international legislation. Roughly speaking, this is the position that many
states have taken, including Singapore.63

Multilateral Conventions in chronological order
Convention on Offences and Certain Other Acts Committed on Board
Aircraft, signed at Tokyo on 14 September 1963.
Convention for the Suppression of Unlawful Seizure of Aircraft, signed at
the Hague on 16 December 1970.
Convention for the Suppression of Unlawful Acts against the Safety of
Civil Aviation, signed at Montreal on 23 September 1971.
Convention on the Prevention and Punishment of Crimes against
Internationally Protected Persons, including Diplomatic Agents, adopted
by the General Assembly of the United Nations on 14 December 1973.
International Convention against the Taking of Hostages, adopted by the
General Assembly of the United Nations on 17 December 1979
Convention on the Physical Protection of Nuclear Material, signed at
Vienna on 3 March 1980.
Protocol on the Suppression of Unlawful Acts of Violence at Airports
Serving International Civil Aviation, supplementary to the Convention for
the Suppression of Unlawful Acts against the Safety of Civil Aviation, signed
at Montreal on 24 February 1988.

As Singapore’s Minister for Foreign Affairs and Law put it during the 57th Session of the
United Nations General Assembly: ‘where international consensus has not yet consoli-
dated over situations posing serious and immediate threats, the lack of international
consensus in itself cannot be an excuse for inaction. This would be an abdication of
individual and collective responsibility’; Professor S. Jayakumar, Address to the 57th
Session of the United Nations General Assembly, New York, 13 September 2002 (on file).

Convention for the Suppression of Unlawful Acts against the Safety of
Maritime Navigation, done at Rome on 10 March 1988.
Protocol for the Suppression of Unlawful Acts against the Safety of Fixed
Platforms Located on the Continental Shelf, done at Rome on 10 March 1988.
Convention on the Marking of Plastic Explosives for the Purpose of
Detection, signed at Montreal on 1 March 1991.
International Convention for the Suppression of Terrorist Bombings,
adopted by the General Assembly of the United Nations on 15 December
International Convention for the Suppression of the Financing of
Terrorism, adopted by the General Assembly of the United Nations on
9 December 1999.

The state of emergency in legal theory

‘There are times of tumult or invasion when for the sake of legality itself the
rules of law must be broken. . . . The Ministry must break the law and trust
for protection to an Act of Indemnity. A statute of this kind is . . . the last
and supreme exercise of Parliamentary sovereignty. It legalises illegality.
. . . [It] . . . combine[s] the maintenance of law and the authority of the
Houses of Parliament with the free exercise of that kind of discretionary
power or prerogative which, under some shape or other, must at critical
junctures be wielded by the executive government of every civilized coun-
try’. A. V. Dicey1

I. Introduction
Dicey says that in an emergency situation public officials might find them-
selves compelled to act outside the law and Parliament might then indemnify
them. But an Act of Indemnity does not make what the officials did legal –
Parliament has ‘legalised illegality’. Rather, it places them in a zone uncon-
trolled by law – a legal black hole, to use the current term. His point is that
political power can be exercised in a brute fashion, permitting those who
wield it to break free of the constraints of constitutionality and legality.
It might, Dicey thinks, be at times justified to do this because without
illegal action the state and thus law itself might have been overthrown. But it
would be far better if Parliament were to delegate in advance the resources to
public officials to deal with the emergency. In addition, the resources would
not be power uncontrolled by law, equivalent to the prerogative. Dicey was
not only enthusiastic about taming the prerogative by subjecting it to statu-
tory control.2 He also said that it would be erroneous to suppose that a

I thank Genevieve Cartier, Michael Hor, Rayner Thwaites, Victor V. Ramraj and, especially,
Tom Hickman and Kent Roach, for comments on a draft, as well as the participants in the
‘Comparative Anti-Terrorism Law & Policy Symposium’ for critical discussion.
Introduction to the Study of the Law of the Constitution (10th edn, London, Macmillan,
1959), 412–13.
Ibid., 64.


prospective legislative response to an emergency merely substitutes the ‘des-
potism of Parliament for the prerogative of the Crown . . . the fact that the
most arbitrary powers of the English executive must always be exercised
under Act of Parliament places the government, even when armed with the
widest authority, under the supervision, so to speak, of the courts’. And
judges would exercise a control on executive action informed by their under-
standing of the ‘general spirit of the common law’. In England, ‘Parliamentary
sovereignty has favoured the rule of law . . . the supremacy of the law of the land
both calls forth the exertion of Parliamentary sovereignty, and leads to its being
exercised in a spirit of legality’.3
Dicey distinguishes clearly between two legal responses to an emergency.
In the first, the response is the after-the-fact recognition that officials made
an excusable decision to act outside of the law because it was necessary to act
and the law did not provide them with the right resources. In the second,
Parliament in advance gives to officials resources to deal with emergencies in
accordance with the rule of law. Dicey prefers the second option, although he
thinks we should be aware that officials might have to react to an emergency
situation before Parliament can respond. In sum, Dicey’s model for dealing
with emergencies relies on legislative solutions that preserve the rule of law.4
In the face of an ongoing emergency, he would insist on the prospective
legislative response.
I argue that Dicey was right. My foil is Oren Gross’s recent article ‘Chaos
and Rules’,5 which suggests that the situation Dicey depicts in the epigraph
should be the general model for dealing with a state of emergency, particu-
larly an ongoing one.6 My argument is that one should not extract a pre-
scriptive model from the matter-of-fact judgment of necessity which Dicey
sketches. I first set out the main components of Gross’s argument. I then
show how Gross, despite his awareness of the contours of debate, finds
himself trapped by some of the same assumptions which produce those

Ibid., 413–24.
John Ferejohn and Pasquale Pasquino, in ‘The Law of the Exception: A Typology of
Emergency Powers’ (2004) 2 International Journal of Constitutional Law 210, 237–9,
disagree, against textual evidence. They also go astray in supposing that the legislative
normalization of the exception is driven by the need for democratic legitimation. This
claim is true in part of democratic societies, but the more fundamental driving force is the
need for rule-of-law legitimation, which is why one sees the same move made in non-
democratic societies. Since the driving force is rule-of-law legitimation, the kind of dualism
they sketch is not, for Dicey, the preferable legal response.
‘Chaos and Rules: Should Responses to Violent Crises Always be Constitutional?’ (2003)
112 Yale Law Journal 1011.
At the symposium, Gross suggested that his model is tied to what he called catastrophic
situations and is thus not, as I claim, general. On rereading ‘Chaos and Rules’, I do not find
this limitation and if the model were so to be limited, it would, for reasons I give below,
cease to be a model.

contours. His model turns out to be highly unstable in theory and the likely
result is that in practice his model turns into something distant from, even
opposite to, his intentions.
Gross recognizes that the legal responses after 9/11 are not to a state of
emergency, classically conceived. Rather, prompted by the allegation that
terrorism is here to stay, these responses seek to deal with the emergency not
as temporary external threat, but as an internal, permanent problem. The
standard reaction, which Gross wishes to resist, is that the normal legal order
has to be changed to deal with the fact that the exceptional situation has
become permanent.
Gross is wrong to resist this reaction so resolutely. There is something to it,
although it goes wrong in supposing that we are confronted with a completely
new reality. The reality is rather one which legal order has always had to
confront, in sometimes banal, sometimes dramatic ways. That confrontation
might require imaginative experiments in institutional design, but only a
legal theory not trapped by the standard assumptions of much of contem-
porary legal thought can explore such possibilities. For such experiments are
inhibited by a failure to recognize the second problem, a failure which plagues
Gross and much of contemporary legal theory. It arises from the grip of a
formal doctrine of the separation of powers and my argument is that only if
the grip is loosened can the first problem be addressed.

II. Chaos and rules
Gross sketches two traditional models used to respond to emergency situa-
tions. The first is the ‘Business as Usual’ model, which holds that the legal
order as it stands has the resources to deal with the state of emergency and so
no substantive change in the law is required. The second model is one of
‘Accommodation’, which argues for some significant changes to the existing
order so as to accommodate security considerations, while keeping the
ordinary system intact as far as possible. The principal criticism of the
Business as Usual model is that it is naıve or even hypocritical, as it either
ignores or hides the necessities of the exercise of government power in an
emergency. The Accommodation model, in contrast, risks undermining the
ordinary system because it imports the measures devised to deal with the
Gross argues that two basic assumptions dominate debates about the state
of emergency and thus underpin the models. The first is the assumption of
separation between the normal and the exceptional which is ‘defined by the
belief in our ability to separate emergencies and crises from normalcy,

Gross, ‘Chaos and Rules’, 1021–2. Gross finds several different models within the
Accommodation camp, but for the sake of simplicity I talk about one model.

counter-terrorism measures from ordinary legal rules and norms’.8 This
assumption makes it easier for us to accept expanded government powers
and extraordinary measures, since we suppose both that once the threat has
gone, so we can return to normal, and that the powers and measures will be
deployed against the enemy, not us. The second assumption is of constitu-
tionality: ‘whatever responses are made to the challenges of a particular
exigency, such responses are to be found and limited within the confines of
the constitution’.9 Gross supports the critiques of both models and he also
calls into question both assumptions.
The assumption of separation between the normal and the exceptional
ignores the way in which emergency government has become the norm, a
trend which has only gathered strength since the widely copied US response
to 9/11. And the assumption of constitutionality, whether made by claiming
business as usual or that the accommodation conforms to constitutional
values, risks undermining the legal order.
Thus Gross proposes the ‘Extra-Legal Measures model’. This model tells
public officials that they may respond extra-legally when they ‘believe that
such action is necessary for protecting the nation and the public in the face of
calamity, provided that they openly and publicly acknowledge the nature of
their actions’.10 Gross claims that this model preserves the ‘fundamental
principles and tenets’ of the constitutional order.11 In addition, public offi-
cials will have to disclose the nature of their activities and hope for ‘direct or
indirect ex post ratification’, either through the courts, the executive or the
legislature. The process involved will promote both popular deliberation and
individual accountability, while the uncertain outcomes will provide a brake
on public officials’ temptation to rush into action.12
To persuade us to accept the Extra-Legal Measures model, Gross suggests
that we should agree on three points: ‘(1) Emergencies call for extraordinary
governmental responses, (2) constitutional arguments have not greatly con-
strained any government faced with the need to respond to such emergencies,
and (3) there is a strong probability that measures used by the government in
emergencies will eventually seep into the legal system after the emergency has
ended.’13 The model, in his view, recognizes the force of all three points. But
by rejecting the naivety of the Business as Usual model while requiring that
exceptional government responses happen outside the law, it greatly
diminishes the probability of seepage.

Ibid., 1022. Two ideas that do considerable work in this chapter share a term, ‘the
assumption of separation between the normal and the exceptional’ and ‘the formal
doctrine of the separation of powers’. I try to avoid confusion by giving ‘the assumption
of separation’ its full description each time – ‘the assumption of separation between the
normal and the exceptional’.
Ibid., 1023. 10 Ibid. 11 Ibid., 1023–4. 12 Ibid. 13 Ibid., 1097.

Gross find the intellectual roots of his model in John Locke’s theory of the
prerogative, where prerogative is defined as ‘nothing but the Power of doing
publick good without a Rule’. For Locke, the prerogative is the ‘power to act
according to discretion for the publick good, without the prescription of the
Law and sometimes even against it’.14 As Gross understands Locke, he is
providing a functional test for use of the prerogative power – the test of the
public good. Locke supposes that it is within the subjective discretion of the
executive to decide both when the exercise of this power is required and how
it is to be exercised. The functionality of the test – whether the powers were in
fact exercised in accordance with the public good – resides in popular
acquiescence in the measures – the tacit approval of the public evidenced in
its failure to rebel. If advance public trust in government is shaken by the
exercise of the prerogative powers, after the fact the public will revolt.15
The Extra-Legal Measures model adopts Locke’s understanding of the need for
and nature of the prerogative. It differs in its requirement of ‘explicit, particular
and ex post ratification’.16 This requirement both puts public officials at risk and
makes them accountable, because their exercise of power will be subject to public
scrutiny and deliberation – to an assessment of whether their actions were
necessary to secure the public good. This requirement, Gross claims, will also
prevent illegal official action from breeding a culture of lawlessness that one might
suppose would ensue if citizens were to take such action as an example.17
Gross recognises that his model, like the other two, rests on the false
assumption of separation between the normal and the exceptional, perhaps
even more so.18 But he claims that its appeal lies in its ‘open recognition of the
assumption’s limitations and in an attendant endeavor to minimize the
actual reliance on it’ through requiring openly illegal and thus potentially
costly official responses.19 His model thus preserves the distinction between
the normal and the exceptional situation while permitting the state to address
the exception subject to after-the-fact legal controls.
Gross is aware that Locke is not his only intellectual precursor. His model also
echoes the claims of fascist legal theorist Carl Schmitt about the inapplicability of
legal norms to exceptional situations.20 Gross agrees with Schmitt that a pre-
existing legal norm cannot be successfully applied to an exceptional situation.
But it fails to follow, Gross thinks, that no norms apply or that law has no role
to play.21 Political norms of accountability, transparency and the good of

John Locke, Two Treatises on Government, P. Laslett (ed.) (Cambridge University Press,
1988), 375.
Gross, ‘Chaos and Rules’, 1104. 16 Ibid., 1105. 17 Ibid., 1117 and 1126–33.
Ibid., 1133. 19 Ibid.
I rely here on my discussions of Schmitt in Legality and Legitimacy: Carl Schmitt, Hans
Kelsen and Hermann Heller in Weimar (Oxford, Clarendon Press, 1997).
Gross, 1120–1.

democratic deliberation govern the decision to act outside the law. And when
it comes to the question of the appropriate consequences of illegal action, law has
a role to play since officials will be pardoned, indemnified, punished, and so on.
So Gross distinguishes himself from Schmitt as he distinguishes himself from
Locke – through the substance of his functional test.
Any agreement with Schmitt is perilous. Gross is therefore keen to point
out that Schmitt’s theory of the exception is not normatively sound; it has
only ‘certain descriptive validity’.22 But Schmitt did not present himself as a
normative theorist. He claimed to provide a descriptive account of the
problems of liberalism and its attempt to constrain politics through law
and of the consequences that followed inevitably from the description. It is
consistent to hold both that Schmitt’s self-presentation was disingenuous
since his account of the state of exception is premised on certain normative
commitments and that, if one adopts his account, he is right about the
consequences which will follow. The question then arises about whether
one can adopt his account and avoid its normative commitments.
Consider an example taken from Schmitt’s own career. On 30 June 1934,
the ‘Night of the Long Knives’, Hitler purged the Nazi movement of those
elements he considered undermined his hold on power and murdered rivals
from the conservative but non-Nazi right, including people to whom Schmitt
was close. He then had his Cabinet accept a draft Law for the Emergency
Defence of the State: ‘The measures taken on 30 June and 1 and 2 July for the
suppression of high treasonable and state treasonable attacks are, as emer-
gency defence of the state, legal.’ This law was declared by the Minister of
Justice to confirm existing law.23
While the world outside Germany was shocked by these murders as well as
by their retrospective validation, within Germany there was widespread and
vociferous approval.24 Schmitt joined in by quickly publishing one of his
most notorious essays, ‘Der Fuhrer schutzt das Recht’,25 in which he praised
¨ ¨

Ibid., 1121, note 474, referring to Oren Gross, ‘The Normless and Exceptionless
Exception: Carl Schmitt’s Theory of Emergency Powers and the ‘‘Norm-Exception’’
Dichotomy’ (2000) 21 Cardozo Law Review 1825.
The Enabling Act of 1933 – the ‘Act for the Removal of Distress from People and Reich’ –
effectively gave legislative power including power to override the Weimar Constitution to
the executive. It was rammed through a cowed Reichstag, in an atmosphere of complete
demoralization caused by Nazi acts of terror; only the 94 socialists who were able to reach
the forum opposed it. For incisive analysis of the law’s significance in the decline of legality
in Germany, see Peter L. Lindseth, ‘The Paradox of Parliamentary Supremacy: Delegation,
Democracy, and Dictatorship in Germany and France, 1920s-1950s’ (2004) 113 Yale Law
Journal 1341 at 1371.
See Ian Kershaw, Hitler, 1889–1936: Hubris (London, Penguin, 1998), 505–22, especially
¨ ¨
Schmitt, ‘Der Fuhrer schutzt das Recht’ in Schmitt, Positionen und Begriffe im Kampf mit
Weimar-Genf-Versailles 1923–1939 (Berlin, Duncker & Humblot, 1988), 199.

Hitler’s retrospective validation of the murders. No doubt Schmitt was fearful
because of his ties to some of the victims and eager to begin a new career as
chief legal philosopher of the Nazis. But he was also theoretically committed
to the position of approval of Hitler’s acts, if not to the sycophantic tone in
which he expressed that commitment, the product of his own functional test.
For Hitler had done everything that Schmitt positively required of a leader.
He had distinguished between friend and enemy, which Schmitt argued is the
mark of the ‘political’,26 had established himself decisively as the sovereign –
the supreme source and judge of all right and law – and had done away with the
liberal and parliamentary fictions of Weimar. Most important, he had through
his personal representation of the German people as a substantive homoge-
neous unit, brought about the ‘democratic’ identity of the people (Volk), which
Schmitt prized above all else. Moreover, there could be no doubt about
the popular acclaim, the resounding ‘Ja’ that greeted Hitler’s vision of an
orderly Germany.27
How does the substance of Schmitt’s test differ from Gross’s? Hitler acted
in a way that he thought, at least initially, was outside of the law to bring
Germany from a state of exception into normality. He sought and procured
legal validation for what he did after the fact and justified himself to the
Reichstag and to the German people; both responded positively. Moreover,
the basis for the legal validation of his act depended on a statute enacted by
the German Parliament after he had been elected leader of Germany at a time
when at least the vestiges of Weimar’s parliamentary democracy were in
place. It might then seem that Gross can distinguish his test from Schmitt’s
only by appealing to substance – through the claim that the Extra-Legal
Measures model has a legitimate place if, and only if, as he says, the commu-
nity is ‘worth saving’; a ‘despotic, authoritarian, and oppressive society is not
worth the effort’.28
But then the test for the legitimacy of the model is primarily substance, the
character of the political regime, rather than the formal indicia of action
outside the law followed by the officials seeking vindication from the public,
from politicians, and from the law. What was wrong with the Night of the
Long Knives stemmed from the obnoxious character of the regime which
Hitler was seeking to stabilize, not with the steps he took.29 But there are two
major problems with this view.

Carl Schmitt, The Concept of the Political (New Jersey: Rutgers University Press, 1987,
translated by George Schwab).
Carl Schmitt, Verfassungslehre (Berlin, Duncker & Humblot, 1989).
Gross, ‘Chaos and Rules’, 1115.
At the symposium, Gross suggested that the dangers of a Night of the Long Knives were
very remote in Western societies and in any case such dangers cannot be forestalled by law.
But the events of that night were made possible in part by the gradual decline in respect for

First, consider a distinction between a society that is just according to
Western standards and a society that is not just because it is neither demo-
cratic nor liberal, but is still ‘well-ordered’ because it can plausibly claim to
abide by the rule of law.30 It is not clear to me that Gross’s model has a place in
such societies, because they are neither liberal nor democratic. If right, this is
a severe problem for his model since such well-ordered societies are plausibly
in the front line of the fight against terrorism.
Second, there was something wrong with the steps Hitler took beyond the
fact that they were taken to stabilise an obnoxious regime. These steps are
considered to make a mockery of legality or the rule of law. The worry here is
not the one Gross depicts of officials’ acting outside the law, thus setting a
precedent for other officials and for ordinary citizens. It is that if the law is
used to give the officials the cloak of legality, legal form is used to cover the
substantive damage to the rule of law. Public officials are supposed to be
accountable to the law, an accountability that becomes most urgent when
they exercise the kind of coercive powers associated with dealing with emer-
gencies. If the law to which they are accountable is not the law that exists at
the time of their actions, but a law which after the fact declares their actions to
be legal, then they are not accountable at all. One is left with the facade not the
substance of the rule of law. This criticism applies to all societies which can
genuinely claim to abide by the rule of law and one of the main marks of a
well-ordered society is that it is entitled to such a claim.
Moreover, Gross, in my view, vastly overestimates what he calls the
‘process-oriented’ aspect of his model – its promotion of deliberation
among the public and among officials.31 The atmosphere of public fear that
attends emergencies is not conducive to deliberation and leads to easy
acceptance of official action that is claimed to be necessitous. If the

legality: see Dyzenhaus, Legality, ch 1. This kind of decline is well illustrated in Justice
Thomas’s dissent in Hamdi v. Rumsfeld 124 S Ct 2633 (2004). Thomas uses an interesting
analogy in his argument that the President can do as he pleases during an emergency and
so does not have to accord a hearing to a citizen indefinitely detained as an ‘enemy
combatant’. He mentions, at 2684, a CIA missile strike in Yemen in which a US citizen was
apparently killed and says that no ‘additional process’ would be required for the citizen.
‘The result here’, he says, ‘should be the same’. On the basis of this analogy, the executive is
permitted to assassinate its perceived enemies both within and outside the US. This
thought is the product of a mind-set which has led to both abuse and torture by the US
and the employing of other nations to do its torturing by proxy. Whether the courts will
prevent this decline still remains to be seen. But it is difficult to measure the effectiveness
of their stand unless they make it; and it is a counsel of self-fulfilling despair to say they
should not because it will make no difference.
This distinction is inspired by John Rawls, The Law of Peoples (Cambridge, Harvard
University Press, 1999), 62–8. But my idea of a decent society is much thinner than
Rawls’s as it is confined to respect for the rule of law. For an exploration of the idea of the
rule of law in the societies I have in mind, see Hor, Chapter 13, in this volume.
Gross, ‘Chaos and Rules’, 1129.

Extra-Legal Measures model were public, as it must be if it is to promote
deliberation, the expectation would be generated of after-the-fact validation
of illegal official acts. In an atmosphere of fear that expectation would likely
be met rather easily, especially when the threat is, or is claimed to be, a
constant one and the government successfully manipulates public opinion.
Finally, Gross recognizes that his model relies, perhaps even more heavily than
the other two, on the false assumption of separation between the normal and the
exceptional situation. His antidote, once again, is accountability through raising
the potential costs of official action. But an acknowledgement of the falsity of the
assumption plays directly into Schmitt’s hands. All Schmitt needs for almost
complete induction into his camp is that one accepts first that legal norms
cannot apply to exceptional situations and second that there is no clear demar-
cation between the normal and the exceptional. With those two elements in
place, his definition of sovereignty is incontestable: ‘Sovereign is he who decides
on the state of exception’.32 This definition asserts that in abnormal times the
sovereign is legally uncontrolled. But Schmitt’s thought goes further. Not only is
the sovereign legally uncontrolled in the state of emergency; the quality of being
sovereign, he who is the sovereign, is revealed in the answer to the question of
who gets to decide that there is a state of emergency.
In this respect, Gross might go even further than Schmitt. He appears to
give the power to determine that there is an emergency to any public official
who is prepared to take the risk that he will not receive after-the-fact legal
validation for his illegal acts. Schmitt would have thought this model is
indicative of the fundamental problems he claimed to have exposed in what
is referred to these days as liberal legalism – the liberal claim about the
legitimacy of government under the rule of law. The model, that is, seeks to
respond to exceptional situations but finds itself doing so in ways that fissure
the state apparatus. It establishes a multitude of potential sovereigns, thus
rendering itself more vulnerable and open to capture by interest groups.
My critique of Gross shows the importance of contesting the very assump-
tion of separation between the normal and the exceptional. I will argue that
banishing the exception from legal order has to be, as Schmitt claimed, the
ideal of liberal legal order, but that where he went wrong is in thinking that
this dream can never be realized by a liberal legal order. Put differently,
I argue that the way to do away with the assumption is to reverse Schmitt
and collapse the exceptional into the normal, which requires holding fast to
the other assumption Gross sketches – the assumption of constitutionality.
Moreover, while the model suggested by this argument might seem to partake
of elements of the other two models, it represents a superior approach, not
accommodation in Gross’s sense of a compromise of the rule of law, but

Carl Schmitt, Political Theology: Four Chapters on the Concept of Sovereignty, George
Schwab, trans. (Cambridge, MIT Press, 1988), 5.

experiments in institutional design fully consistent with the rule of law. The
linchpin of my argument is that we must loosen the grip that a formal
doctrine of the separation of powers appears to have on legal theory.
According to this doctrine, the legislature has a monopoly on lawmaking,
judges a monopoly on interpretation of the law, while the executive or admin-
istration both implements the law and makes determinations about policy
within the area of discretion delegated by the legislatures to public officials.
I argue below that this formal doctrine should be discarded in favour of a view,
the Legality model, that regards the separation of powers as instrumental to the
values of the rule of law or legality. Once that is done, one can conceive of
imaginative responses to emergencies which preserve the rule of law without the
risks Gross associates with the Accommodation model.

III. The norm and the exception
In Schmitt’s first publication, Gesetz und Urteil, he examined the relationship
between statutory law and judgment.33 His concern was the problem of what
now would be described as gaps in the law – those occasions on which a judge
has to decide a matter but the law is indeterminate, so the judge has to
exercise a quasi-legislative discretion to fill the gap. Schmitt did not think
there was a normative solution. The lack of stability in legal order created by
inability of law to provide a solution can be cured sociologically, by conceiv-
ing of the judge as an empirical type.34 Schmitt means that as long as judges
are legally well-trained solid sorts, who come from a class of men with fairly
set views, their exercise of discretion will be both predictable and safe.
In sum, the legal order does not contain the resources to deal with the
problem of indeterminacy. But this problem is containable on two conditions.
First, there is the legal positivist assumption that the law generally provides
determinate answers, so that the number of cases to be resolved by discretion
is relatively small compared to the number that do not become cases at all
because the law is clear. In terms introduced later by H. L. A. Hart, this condition
can be expressed by saying that the legal order will be stable as long as the ‘core’ of
‘settled’ law is larger than the ‘penumbra’ of ‘doubt’ about what law requires.35
And in the distinction between area of stable core and unstable penumbra, we
can see the assumption of separation between the normal and the exceptional at

Gesetz und Urteil: Eine Untersuchung zum Problem der Rechtspraxis (Munich, C. H. Beck,
1969, first pub. 1912). See my ‘Holmes and Carl Schmitt: An Unlikely Pair’, (1997) 63
Brooklyn Law Review 165, 180–6.
Schmitt, Gesetz und Urteil, 71: ‘A judge’s decision can today be taken for correct when we
can predict that another judge would have decided the matter in exactly the same way.’
H. L. A. Hart, ‘Positivism and the Separation of Law and Morals’ in Essays in Jurisprudence
and Philosophy (Oxford, Clarendon Press, 1983), 49, 62–72.

work. Second, and Schmitt’s own contribution, the judges charged with resol-
ving penumbral cases are a homogeneous bunch with the right sorts of views.
Schmitt seems then to combine a legal positivist diagnosis of the relation-
ship between law and judgment with what we would think of today as a
conservative legal realist response to the diagnosis. But the work is interesting
as a lens both into the way Schmitt’s political and legal theory developed and
into the lack of resources contemporary legal theory has to offer when it
comes to states of emergency. For the moment of uncertainty or indetermi-
nacy is a kind of mini state of emergency for a positivist theory of law. It is an
emergency because, by positivist stipulation, it is not resolvable by law. It is
mini because it is containable: order can be secured as long as the core of law
is large enough and as long as judges are of the right sort. But if the boundary
between core and penumbra cannot be sharply drawn, the core seems to
disappear; and the state of emergency becomes uncontainable and generally
pervasive. Here it is important to see that Schmitt does not think that his
solution to the problem of the gap is merely a matter of sociological predic-
tion. He still thinks it is important for judges to adopt what he calls the
‘postulate of legal determinacy’ and that they are a ‘learned professional’
group.36 If these two elements are in place, he thinks that adjudication can
still be conceived as a distinctively legal enterprise because judges orient
themselves around the legal value of together achieving legal determinacy.
Schmitt’s later work gives up on this hope. The distinction between core
and penumbra, and thus the stability of legal order, is not sustainable because
of inherent qualities of law, nor is it sufficient to keep things trundling along
that there is a homogeneous group of judges. The distinction is sustainable if
and only if there is a substantively homogeneous population whose identity
as a people is represented by a sovereign, who can then fill the role of guardian
of the constitution. Liberalism wishes to displace the political necessity of
determining the basis for such identity onto the law and the rule of law. But
law cannot carry this burden. Rule of law patriotism, so to speak, is not
patriotism at all without the right sort of substantive basis in place, in which
case it is not law but its basis onto which patriotism will latch. It is, on
Schmitt’s view, no surprise, when liberal legal orders are faced with a crisis
such as 9/11, that political theories of the sort espoused by Samuel
Huntington come to the fore, as these seek to express, perhaps in veiled
terms, the kind of identity required for a serious response to the crisis.
One can of course call into question the very terms in which this problem is
framed. Ronald Dworkin, in the best known critique of legal positivism of the
last forty years, argues that positivism goes wrong in conceptualizing law as a
model of rules of determinate content, where the question of what the rules of

Schmitt, Gesetz und Urteil, 86.

a system are is settled by other rules relating to their origin or pedigree.37 It
follows from this model that, when a rule does not supply a determinate
answer to a question, the judge has to exercise discretion. But, as Dworkin
argued, law is not just a matter of rules; it also contains principles which are
moral as well as legal. Such principles have to be assigned weight by the judge
when they are relevant to a dispute about the law. So Dworkin’s first challenge
to positivism advanced the view that in penumbral or ‘hard’ cases, judges are
under a legal duty to decide the case by resort to principles already immanent
in the law. Dworkin’s theory of interpretation holds that there is in principle
one right answer in all such cases, an answer fully determined by law.38
Dworkin’s second challenge was to the positivist view that the idea of legal
order as consisting of determinate rules is not threatened by the concession
that there is a penumbra of uncertainty. As we have seen, there is supposed to
be no threat because the core is in fact much larger than the penumbra and
this provides the certainty that makes legal order possible. But Dworkin
argued there is no core in the positivist sense. What appears to be the
core is the product of interpretation in just the way that decisions in the
penumbra are. The core is merely an area of provisional agreement as to
Dworkin’s second challenge makes the first more radical. If there is no
clear boundary between core and penumbra, so that the core does not so
much diminish in size as disappear, then legal order as positivists understand
it implodes. If all questions about what law is are interpretative in this sense,
there is no such thing as law. More precisely, there is no such thing as law in
the positivist sense of a set of rules whose content can be determined without
resort to moral argument. From the positivist perspective, the problem which
they acknowledged as occurring only at the margins of legal order now
appears throughout.
Now in one respect Dworkin’s critique of legal positivism looks quite
similar to that to be found in Schmitt’s later work, as both want to contest
the distinction between core and penumbra. The difference between them lies
in Dworkin’s optimism that one can do without that distinction and still hold
the edifice together through the principles immanent in the law, while
Schmitt thinks that the basis has to be found outside of the law, in the identity

Ronald Dworkin, Taking Rights Seriously (London, Duckworth, 1978).
Hart dealt with this challenge by claiming that the choice between principles could not be
settled by the law but only by a legally unconstrained act of judicial choice or discretion.
Put differently, he claimed that the more Dworkin showed that the adjudication of hard
cases involves a decision based on legal principles, the better the evidence for the positivist
thesis about judicial discretion. In my view, this claim is fatal to legal positivism, for
reasons I explore in ‘The Genealogy of Legal Positivism’ (2004) 24 Oxford Journal of Legal
Studies, 39.
Ronald Dworkin, Law’s Empire (London, Fontana, 1986).

of the people. Further, Schmitt argues that the liberal principles which for
Dworkin make up the normative basis of the law are corrosive of identity. The
basis of identity will be substantive in the negative sense of being anti-liberal
and in Schmitt’s ideal legal order, that basis will be given determinate
expression by the sovereign in his code of law. Schmitt is not then a critic
of positivism’s model of law, only of positivism’s naivety about the political
conditions that have to be in place before a relatively large core of determinate
law can be sustained.
Schmitt no less than liberal legal thinkers wishes to banish the exception
from the political and legal order. He differs from liberals only in thinking
that once substantive homogeneity has been successfully established within a
state, the exception will not disappear but be displaced to international
relations which he, like Hobbes, conceives of as inherently unstable – a
state of nature. For Schmitt, the instability of the international realm is
beneficial. It is through the contrast with the political identities of other
states and the constant threat of conflict with them that the internal identity
of any particular people is preserved. So not only the exception but politics
itself is banished from the internal legal order.
From this we can see that for Schmitt, as for positivists, the idea of
separation between legal normality and extra-legal state of exception is
important. But for Schmitt, separation is not strictly speaking an assumption
but a distinction between internal and external – between the inside and the
outside of law – which has to be brought about by establishing the right sort
of political conditions. And once separation is seen as an ideal state rather
than as an assumption about the way the world works, one can appreciate that
for Schmitt it is not separation, however conceived, that is the real issue.
Rather, it is the assumption of constitutionality. Dworkin, in contrast,
accepts the assumption of constitutionality. His model of law is not per-
turbed by Schmitt’s thoughts about the separation between the normal and
the exceptional, since separation, on Dworkin’s view, is a symptom of posi-
tivist thought. Where Dworkin goes wrong, along with Gross, is in adhering
to the different idea of separation, the idea that the best way to understand
legal order is in terms of a formal doctrine of the separation of powers
between the legislature, the judiciary and the executive.
I now argue that once we loosen the grip of that doctrine, we can also
qualify Gross’s central claims as follows: (1) emergencies call for governmen-
tal responses that cannot be accommodated by the formal doctrine,
(2) constitutional arguments premised on that formal doctrine have not greatly
constrained any government faced with emergencies, and (3) there is a strong
probability that if we hang on to that formal doctrine, emergency measures will
eventually seep into the legal system. But then one should give up the formal
doctrine, so that legislative responses to emergencies can be controlled by the
rule of law.

IV. Constitutionality and the separation of powers
Dworkin’s model of law is a Business as Usual model, in which public officials
who act to deal with emergency situations should be held accountable to law,
where law includes the constitutional values of which judges are the exclusive
guardians. And that is because Dworkin operates with a rather formal under-
standing of the separation of powers, in which the legislature makes policy
decisions, public officials implement those decisions, and judges make sure
that both legislation and official actions are limited by constitutional
If it is the case that imaginative legal solutions to emergency situations
cannot operate within the constraints of this model, then Gross has to be right
that it follows that such solutions are outside the reach of the rule of law. His
worry about the effects of using law to give cover to official acts to deal with
genuine emergency situations, especially when the emergency is or is alleged
to be permanent, would be fully justified.
It follows that Gross accepts the Business as Usual model with all its
assumptions except for the claim that it can cope with emergencies. And
because he thinks that that model governs the normal situation but that
something different is required for emergencies, he must adopt the assump-
tion of separation between the normal and the exceptional, despite his keen
awareness of its problems. He ends up with a highly unstable combination of:
a model of Business as Usual for the ordinary situation; a model of Schmittian
positivism for the exception or emergency; and a kind of deliberative demo-
cratic model as the after-the-fact cure for the ills of the intervening Schmittian
His position is unstable because he accepts the assumption of constitu-
tionality for the ordinary situation, while rejecting that assumption for the
exception, a rejection which depends on the admittedly false (for Schmittian
reasons) assumption of separation between the norm and the exception.
Moreover, the instability is not confined to the fact that Gross accepts a
model for dealing with the exception whose presuppositions are supposed
to implode the model he adopts for dealing with the normal situation.
Schmittian positivism regards as a charade the kind of deliberative demo-
cratic process Gross envisages as triggered by extra-legal official action, one
which either serves as an excuse for postponing an effective executive decision
or as political camouflage for the decision the executive has already made.
It is helpful to note that for Dicey, the ordinary situation of administrative
law, in which an administrative official is delegated authority by the legisla-
ture, is a site of arbitrary power. Dicey advocated abolishing legislative
delegations of authority to administrative officials, since such delegations
introduced pockets of lawlessness – mini exceptions – into the ordinary legal
order. The decision to create the administrative state need not and should not

be taken, precisely because these pockets of lawlessness will damage the integrity
of the legal order as a whole.40 Such delegations he viewed in much the same way
as he depicted the legalization of illegality when officials were given after-the-fact
legislative indemnification, the only difference being that the former were not
retrospective but prospective indemnifications of illegality.
The idea that the delegated discretionary power of public officials is not
subject to the rule of law, that it is ‘unfettered’ or that the officials are a ‘law
unto themselves’ within the limits clearly stated in the statute, has important
affinities with the Lockean idea of the prerogative as a legally uncontrolled
space. But Dicey was not only enthusiastic about taming the prerogative by
subjecting it to statutory control.41 As we have seen, when it came to emer-
gencies, he moved away from the position he took in regard to delegations of
ordinary executive authority to one which seemed to countenance that
delegated authority could be thought of as subject to the rule of law. There
is thus a tension between Dicey’s claim that the emergency situation should
be dealt with by legislative delegation of authority controlled by the rule of
law and his more famous argument against the delegation of discretion to
public officials because that delegation creates pockets of arbitrary power.
Perhaps part of the explanation lies in the fact that Dicey’s hostility to the
administrative state as a site of lawlessness was motivated by his adherence to
a laissez-faire political philosophy. But it was also motivated by an incom-
prehension about how such a state could ever be subject to the rule of law.
And that incomprehension had much to do with his views about the separa-
tion of powers, where the legislature has a monopoly on lawmaking, and
judges on interpretation.42 He thus held the same view of the separation of
powers as that presupposed by a Business as Usual model.
However, in common law countries, there is now in place a highly devel-
oped rule-of-law system for administrative officials, including controls of
reasonableness, independence, impartiality, and often a requirement to give
reasons. Ultimately, there is the check of judicial review, accompanied by the
understanding that administrative officials have an expertise in their area of
decision-making which generalist judges lack. That understanding is cap-
tured in tests indicating that judges will not normally overturn decisions on a
correctness standard, on the basis of mere disagreement with the merits of the

I am indebted here to Genevieve Cartier’s lucid discussion in Chapter 1 of her SJD thesis,
‘Re-conceiving Administrative Discretion: From Discretion as Power to Discretion as
Dicey, Law of the Constitution, 64.
In Lectures on the Relation Between Law & Public Opinion in England during the Nineteenth
Century (London, Macmillan, 1920), Dicey in an Appendix discussed what he calls ‘Judge-
Made Law’, 483–94, in which he is willing to talk about ‘judicial legislation’. But Dicey
seems to mean by this phrase the kind of creative interpretation of law which Dworkin’s
theory of adjudication advocates.

decision. Rather, they will declare decisions invalid only if they are unreason-
able. In addition, influenced by the fact that most of the prerogative powers
have been incorporated into statutory regimes, judges in the common law
world have become increasingly willing to think of exercises of such powers as
subject to the rule of law and so have asserted, at least at the level of abstract
principle, their authority to review such exercises.43
To create this understanding of judicial review, judges had to move away from
the formal doctrine of the separation of powers which required that any admin-
istrative decision involving interpretation of the law had to be reviewed on a
correctness standard. At the same time, judges had to develop a sense that rule-
of-law standards should be tailored to particular administrative contexts.
If we are to take seriously Gross’s Extra-Legal Measures model, we have
also to take seriously his point of agreement with Schmitt – that legal norms
cannot apply to exceptional situations. But that means that the history of
gradual legislative and judicial colonization of the prerogative, as well the
creation of the common law understanding of the rule of law governing
public officials, are mistakes. Gross’s view must be that these are situations
of discretion where the attempt to use law to deal with the exception results in
variants of the Accommodation model, thus damaging law’s integrity in the
normal situations where its writ properly runs. But, as I have suggested, this
recognition is premised on a conception of law as a system of rules. If one’s
conception of law is different – the test of law’s integrity is its compliance with
immanent principles – then law’s writ is not confined in this way.
I do not want, however, to suggest that a conception of law as a matter of
principle is immune to damage through the wrong kind of institutional
experiment. Whether law is conceived as a matter of rules or principles, it
is dangerous to permit governments the luxury of claiming that they govern
in accordance with the rule of law when in fact law provides them with a
formal facade that serves only to cover abuse of power. But in liberal democ-
racies and beyond, it has become almost unthinkable for governments to
govern outside of the framework of the rule of law. In addition, the allegedly
permanent nature of the terrorist threat makes, in my view, a legislative
response inevitable. Finally, and most importantly, it is desirable, as Dicey
argued, that the response be a legal one, which means that one should
experiment only in so far as experimentation is constitutionally justifiable.44

Nevertheless, that a power is or was at one stage a prerogative power often looms large in
judges’ approach to statutory interpretation, especially when officials are given broad
discretionary powers to make security or immigration determinations. And as we know,
these two areas – immigration and security – frequently overlap, which can serve to
increase judicial wariness about review.
See Michael Ignatieff, The Lesser Evil: Political Ethics in an Age of Terror (Princeton
University Press, 2004).

One must then hold the assumption of constitutionality in place, which is
to be done by treating, with Schmitt, the separation between the normal and
the exceptional not as an assumption but as a conclusion which has to be
argued for. If there were no model for experiment controlled by legality, no
Legality model, it would seem that the inevitability of experiment, or what
Gross calls accommodation, would leave us in a highly uncomfortable posi-
tion. But, as I have indicated, there is ample evidence of the right sort of
experiment – the development of the common law of judicial review in the
last forty or so years. Since security issues will be dealt with by delegating
authority to public officials, one should look to the common law of judicial
review for ideas about how such authority can remain subject to the rule of
law. In the next section, I briefly examine one manifestation of such a model –
the Special Immigration Appeals Commission (SIAC), a creation of the
legislature in the United Kingdom.

V. SIAC and the separation of powers
Before 1997 those subject to deportation from the UK on national security
grounds were statutorily deprived of the right of appeal that individuals
subject to deportation on other grounds enjoyed. Their only recourse was
to an executive committee, which gave advice to the responsible minister. In
Chahal v. UK,45 the European Court of Human Rights rejected the govern-
ment’s argument that national security grounds are inherently incapable of
being tested in a court of law and held that the advisory panel was not a ‘court’
before which the legality of one’s detention could be challenged, as required
by Article 5(4) of the European Convention of Human Rights.
The government responded with a statute establishing SIAC, a three-
person panel of which one member had to have held high judicial office,
the second had to have been the chief adjudicator or a legally qualified
member of the Immigration Appeals Tribunal, while the third would ordi-
narily be someone with experience of national security matters. The statute
gave the individual, who would have had the right to appeal against a
deportation order but for the fact that national security was involved, a
right to appeal to SIAC. It gave SIAC itself the authority to review the
Secretary of State’s decision on the law and the facts as well as the question
whether the discretion should have been exercised differently. There was a
further appeal to the Court of Appeal on ‘any question of law material to’
SIAC’s determination. In addition, the statute provided for a special advocate
who could represent the appellant if parts of the proceedings before SIAC
took place in closed session because it was considered necessary to keep

(1996) 23 EHRR 413.

information confidential. SIAC’s decision is based on both the closed and the
open session though its reasons do not disclose information from the closed
SIAC seems a workable answer to concerns about the sensitivity of security
issues and the need to hold public officials accountable to the rule of law. It is
seized of jurisdiction through a statutory right of appeal and it has the explicit
authority and the necessary expertise to review security decisions, thus
ensuring that the decisions of the security services are accountable to the
law. And this is in large part because when SIAC reviews it has before it the
information on which the executive and the security services act, which is
fully tested even if it is highly confidential. Moreover, it is important to see
that in contrast to the Extra-Legal Measures model, SIAC builds reason-
giving requirements and deliberation about those requirements into the
legal process. Put in the more abstract terms of this chapter, SIAC is a
remarkable exercise in imaginative institutional design to address security
concerns. It internalizes the exception to the rule of law that had been created
by the removal of the right to appeal and does so in such a way that preserves
legality at the same time as taking into account the special circumstances of
the security services.
At least, SIAC would have seemed to be the answer, were it not for the fact
that the House of Lords has responded to SIAC in a way which shows that
judges care less about the rule of law than about their place in the hierarchy of
legal order, in which case the creation of such a body might be seen as an
occasion for judicial jealousy.46
In Secretary of State for the Home Department v. Rehman,47 the Court, to
preserve its sense of role, refused to concede to SIAC the capacity to be a more
effective enforcer of the rule of law than a generalist court.48 The Court first
interpreted the legislation as giving courts some review authority, though one
which clearly undersells the resources of the common law: courts can review
only if decisions are manifestly absurd. The Court then cut down SIAC’s
authority to fit the Court’s parsimonious understanding of its own role.
While this decision is in part motivated by political concerns, the custom-
ary judicial spinelessness in security, it is officially presented in Lord
Hoffmann’s infamous judgment as required by the formal doctrine of the
separation of powers. This stance is astonishing given that he and another
judge on this bench, Lord Steyn, are responsible for articulating a principle of
legality in ordinary administrative law which requires that all executive acts

As a result, the lay member of SIAC who provided the expertise on security, Sir Brian
Barder, resigned making public his reasons. For the full account, see his article in the
(2004) 26 London Review of Books, issue of 18 March, 40–1.
[2002] 1 All ER 123.
See Lord Hoffmann ibid, especially, 139, para 49.

be demonstrated to be justifiable in law, where law is assumed to include
fundamental values.49 These two judges thus find that in some cases they are
driven to constitutional bedrock, which is full of values and principles, while
in others they find that the constitution amounts only to a very formal
understanding of the separation of powers.50
I want to emphasize two points. First, the Court ignored the exercise of
institutional imagination which had gone into making SIAC an effective
enforcer of the rule of law in security matters. Because a generalist court
cannot be effective in such matters, it was unwilling to concede that an
administrative panel could have a special role as guardian of constitutional
values. Second, in reverting to the formal understanding of the separation of
powers, one which is inconsistent with developments in the common law of
judicial review, the judges introduced a rather glaring inconsistency into
English public law. The lesson to be learned is that an adequate response to
emergencies, especially in light of the way emergencies are now regarded as in
a sense normal, must start from a premise other than one that adopts the
formal doctrine of the separation of powers.

VI. Responding through the rule of law
As should now be clear from my argument, following Dicey, I hold the view
that governments which have the luxury of time to craft a response to
emergency situations should do so in a way that complies with the rule of
law. The issue is not accommodation in Gross’s sense of compromises which
undermine the integrity of the rule of law. Rather, it is about imagination in
institutional design, experiments which seek to address both the concerns of
the rule of law and security. The Legality model is not then about accom-
modation, nor is it the Business as Usual model if that model requires
adopting the formal doctrine of the separation of powers that gets in the
way of such experiments.
The Legality model does, however, preserve the assumption of constitu-
tionality in that it insists that the values of the rule of law are not to be
compromised. It also preserves the idea of a separation between the excep-
tional and the normal. It holds that to the extent that political power can be

Thus in R v. Secretary of State for the Home Department, ex p Pierson [1997] 3 WLR 492, at
518, Lord Steyn said that ‘Parliament does not legislate in a vacuum’ but ‘for a European
liberal democracy founded on the principles and traditions of the common law’. Compare
R v. Secretary of State, ex p Simms [1999] 3 All ER 400, at 412, per Lord Hoffmann: ‘In the
absence of express language or necessary implication to the contrary, the courts therefore
presume that even the most general words were intended to be subject to the basic rights of
the individual.’
Lord Steyn has since made public a radical change in his thinking; see ‘Guantanamo Bay:
the Legal Black Hole’ (2004) 53 International and Comparative Law Quarterly 1.

successfully subjected to the discipline of the rule of law, it should. But, as in
Schmitt, the idea functions not as an assumption but as an ideal; at most it is a
regulative assumption, an assumption one adopts for political reasons
because it will regulate practice in the best possible way. One assumes its
truth to bring the legal order closer to the ideals which underpin it.
It does not follow, however, that all possible acts by public officials should be
subject to the rule of law. Torture is absolutely prohibited by international law
and by the domestic laws of many states for good reasons having to do with our
understanding of ourselves as human beings and the fact that even the prudential
reasons for torture are so dubious. But the humanitarian reasons are so strong
that no decent regime could permit torture. So if officials consider that they have
to torture to avoid a catastrophe – the ticking bomb situation – such an act must
happen extra-legally, more or less the position of the Israeli Supreme Court. All a
court should say is that if officials are going to torture they should expect to be
criminally charged and may try a defence of necessity.51 But in saying that, a
court is not adopting Gross’s Extra-Legal Measures model. It is simply recogniz-
ing, as Dicey did, that in some situations where officials act outside of the law
they merit after-the-fact recognition that they made an excusable decision
because it was necessary to act and the law did not provide them with the
resources they needed. The twist with torture is that a decent regime is precluded
from providing prospective legal resources to legalize what would otherwise be
illegal. Torture is, in other words, ‘unlegalizable’.
What falls into this category of the unlegalizable is, of course, controversial.
Does preventive detention fall into this category, or trials which fall far short of
the standards prescribed for criminal justice? Take for example the division in
the United States Supreme Court in Hamdi v. Rumsfeld.52 Only Justice Thomas
accepted in his dissent the blank cheque position that the Congressional author-
ization to the executive to wage war on terrorism – to use ‘all necessary and
appropriate force’ – included the power to detain indefinitely people deemed by
the executive to be ‘enemy combatants’. Only he accepted the government’s
main argument – that the executive had a blank cheque even without that
Congressional authorization since Article II of the Constitution provides that
the President is ‘Commander in Chief of the Armed Forces’. And only he put

See Judgment of the Supreme Court of Israel, sitting as the High Court of Justice,
6 September 1999, concerning the Legality of the General Security Services’
Interrogation Methods. The Court did indicate the possibility that the Legislature might
enact a statute that put in place prior authorization to torture, modelled on the defence of
necessity. But my sense is that this indication was a dare which the Court thought the
Legislature could not afford to take up and if it did there would be grounds for invalida-
tion. See my ‘With the Benefit of Hindsight: Dilemmas of Legality in the Face of Injustice’
in Emilios Christodoulidis and Scott Veitch (eds.), Lethe’s Law: Justice, Law and Ethics in
Reconciliation (Oxford, Hart Publishing, 2001), 65 at 86–9.
124 S Ct 2633 (2004). See also Roach, Chapter 7 in this volume.

forward a basically Schmittian argument to the effect that the executive needs the
authority to respond to exceptional situations unconstrained by legality.
Justice O’Connor for the plurality of the Court held that the detentions
were authorized by implication but that the detainees were entitled to some
due process. The only way they could be deprived of review altogether would
be if Congress were to suspend habeas corpus.53 Since they had had no access
to review, the case was remanded and the plurality outlined the process to
which the detainees were entitled. The detainees had to be given notice of the
factual basis for the classification and a ‘fair opportunity to rebut the
Government’s factual assertion before a neutral decisionmaker’.54
However, the fact that the plurality did not require express authorization
for detentions is deeply troubling.55 Moreover, while they rejected as too
deferential a ‘some evidence’ rule,56 the rule accepted by the House of Lords
in Rehman, their understanding of appropriate due process included a mili-
tary tribunal, which could rely on hearsay evidence and require the detainee
to accept the onus of rebutting a presumption in favour of the government.
Moreover, they suggested that the kind of cost benefit analysis for fair
procedures, developed by the Supreme Court in Mathews v. Eldridge,57 would
be suitable for testing the appropriateness of the procedures. Not only is
Mathews widely regarded as weakening the rights-protecting approach to
deprivation of important interests outlined in the earlier decision in Goldberg
v. Kelly,58 but, as Justice Scalia pointed out in dissent in Hamdi, Mathews was
about the withdrawal of disability benefits, not the interest in protection
against arbitrary deprivations of liberty.59
In contrast, Justice Souter with Justice Ginsburg argued that any detention
of a citizen had to be ‘pursuant to an Act of Congress’, a requirement of the
Non-Detention Act, which had been enacted to rule out a repetition of the
wartime detention of Japanese Americans. Justice Souter still wished to give
practical effect to the majority’s view which rejected the government’s posi-
tion, so he went on to agree with the plurality that the detainees should be
entitled to due process. However, he doubted the appropriateness of a reverse
onus and was wary of litigation before a military tribunal.60
Justice Scalia, joined by Justice Stevens, argued that there was no author-
ization for the detentions. The only possible authorization resided in an
explicit Congressional suspension of habeas corpus. Failing that, the detainees

Hamdi, Plurality at 2644–7. 54 Ibid. at 2648.
Here the plurality follows the majority in R v. Halliday [1917] AC 260 (HL) and thus the
even more infamous majority in Liversidge v. Anderson [1942] AC 206. For discussion of
both, see my ‘Intimations of Legality Amid the Clash of Arms’ (2004) 2 International
Journal of Constitutional Law, 244. On Halliday, see David Foxton, ‘R v Halliday Ex Parte
Zadig in Retrospect’ (2003) 119 Law Quarterly Review 455.
Hamdi, Plurality, at 2651. 57 424 US 319 (1976). 58 397 US 254 (1970).
Hamdi, at 2672. 60 Hamdi, at 2660.

should be released or charged with criminal offences. Scalia’s dissent is closest
to Gross’s Extra-Legal Measures model, because it draws a clear line between
full constitutional protection and action outside of the law. It thus has the
virtue of requiring the government openly to declare that it will not govern by
the rule of law and such a declaration can promote political accountability.61
But his dissent also shares much with Thomas, in that he is prepared to
countenance the government writing itself a blank cheque, as long as it can
persuade Congress to certify it. Moreover, in Rasul v. Bush,62 released on the
same day as Hamdi, where the majority found that US Federal Courts do have
jurisdiction over detainees at Guantanamo Bay, Scalia showed in dissent that
he is not at all perturbed by the idea of legal black holes, zones of official
illegality, as long as they are (in his view) properly created.63
While this advance legalization of illegality is contrary to Gross’s Extra-
Legal Measures model, it is what I think the model amounts to in practice, if it
is to be anything more than after-the-fact legal recognition that officials made
an excusable decision to act outside of the law. My point here is different from
the one made earlier about the theoretical instability of the model because of
the way it falls into the trap Schmitt sets for liberalism, a trap sprung by
seeking to address the exception through either the Business as Usual or the
Accommodation models. It is that in an era when the rule of law has a
currency such that at least lip service to its ideals is required, governments
will generally seek to use vaguely framed statutes prospectively to indemnify
official illegality. Governments will prefer to use vague authorizations in the
hope that judges will not intervene to curb executive arbitrariness. But judges
who do so intervene can force governments to come clean in their legislation
in a way which increases political accountability and which might permit
judges to find the explicit authorizations unconstitutional.
I do not wish to underestimate the importance of forcing governments to
come clean. Nor do I want judges to do what Justice Thomas’s dissent in
Hamdi did when, in as fine an example of judicial spinelessness as one could
find, he handed to government the cloak of the rule of law by deeming
illegality to be legal. I also do not want to underestimate the perils of the

Dyzenhaus, ‘Intimations of Legality’.
Rasul v. Bush 124 S Ct 2686 (2004).
Dicey argued that an Act suspending habeas corpus suspends legal remedies for detainees
until it is lifted, but an Act of Indemnity must follow the lifting of the suspension if the
officials are to be protected against legal action by the detainees: Dicey, Law of the
Constitution, 233–37. So the suspension does not quite create a legal black hole; it only
renders legal remedies temporally ineffective. At 237, Dicey explains somewhat enigma-
tically his remark in the epigraph that an Act of Indemnity maintains the authority of


ńňđ. 3
(âńĺăî 23)