This paper is a sequel of an article I wrote twenty years ago that I now think can be put more tightly and clearly. That early paper was born of the irritation I felt, and continue to feel, at much of the classical liberal discourse about limited government. At least since Locke, that discourse sets out a normative ideal of government: the protector of “rights” its citizens are in some fashion endowed with, and the guarantor of liberty that ranks above rival values. Such government uses coercion only to enforce the rules of just conduct. This ideal is attractive enough to the liberal mind. The reason why it nevertheless irritates is that it makes it seem that the writing of a constitution of liberty is a plausible means for transforming the normative ideal into positive reality. The message is that “we” can have limited government in the above sense if only “we” understand why we ought to wish it. The “we” is crucial, for it suppresses the essence of collective choice. Collective choice starts where unanimity ends, and involves some deciding for all, where the “some” control the apparatus of government. It is the potential for some to benefit morally and materially at the expense of others that creates the bone of contention and that limits on government are meant to move out of reach. It is odd that little or no awareness is shown of the “incentive-incompatibility” (if we may use ugly but handy jargon) of limits that would exert real rather than illusory restraint.
Similar confusion between “ought” and “is” is inherent in another major strand of political thought centered on the idea of the fictitious social contract. The contractors give the government a limited mandate to help execute and enforce the contract. There is an obvious potential gain to the government or, to be pedantic, the persons in charge of it, from exceeding this mandate and the means are available for doing so. The incentive-incompatibility of the limits upon the government is nevertheless ignored and not dealt with in mainstream theory.
When Ulysses was sailing close to the sirens, he had his shipmates tie him to the mast lest he should be lured to jeopardy by the sirens’ song. This metaphor is used to bring home the rationally self-interested nature of self-imposed restraint. However, the metaphor is an ill-fitting one for governmental self-restraint. Ulysses obviously would not tie his own knots, for he would know that he could and would untie them again when temptation beckoned. He had shipmates; it is not evident that government has any capable to tie it to the mast. If there are any, they might be just as eager to untie him and to yield to the temptation as was Ulysses himself. For the metaphor to work, some explanation is needed why their incentives differed from those of Ulysses, why Ulysses took on such uncongenial shipmates and why, being of sterner stuff than their weak-willed skipper, they did not take over the ship rather than leave him in command.
A different metaphor I proposed in my The State is a perhaps more apt illustration of the use of a mechanical device to serve as a substitute for reliable commitment to self-restraint. A lady accepts to wear a chastity belt to reassure her lord during his prolonged absence from home. (A government may likewise adopt constitutional limits to gain the confidence of its subjects). However, the key to the belt’s padlock is within reach, and thus the belt can occasion delay, but cannot stop Nature from ultimately having its way. A padlock whose key has been thrown away is a metaphor for a constitution that cannot legally be amended — a problematical legal rule I intend to touch upon in Section I.3. Meanwhile, we must not forget that if there is no key, there is always the locksmith.
My early paper on the subject concluded, if that is the right word, that the possibility of limited government lies beyond the scope of rational choice theory. I still think that there is something to be said for this view, but that the argument leading to it should be formulated somewhat differently and before winding it up more account must be taken of real-life contingencies that may or may not occur and persist.
My present argument will be set out in what I hope to be a tolerably organised pattern of nutshells. Section 1 will deal with rule-bound social orders, 1.1 sketching the conventional rule system of ordered anarchy, 1.2 the rule system made for and by government, 1.3 considering some problems raised by the nature of vows, self-referring rules, and enforcement, and 1.4 some consequences of a winning coalition seeking both to maintain itself and maximize the benefit from winning. Section II is mainly devoted to contingencies, 2.1 speculates about what happens when money buys elections, 2.2 points to limits of government arising from taxable capacity, 2.3 deals with how panic about a dysfunctional spread of government may bring about a reversal of the usual time preference for the present over the future, while 2.4 evokes taboos as factors that may limit government in ways the usual conception of rational choice would not do.
1. Rule-Bound Conduct
Social coexistence may be chaotic or rule-bound. Here I am only seeking to deal with the latter. Rule systems can be fruitfully divided into two types, conventional rules and rules by and for government.
1.1. Conventional Rules of Ordered Anarchy
Conventional rules can be best understood in game-theoretic terms. They are equilibria of coordinated behavior that are payoff-enhancing (without necessarily being payoff-maximizing) for those who adopt the behavior in question provided enough others adopt it, too. Standard examples include the use of a common language, the rule of the road that drivers will adhere to if many travel on the same road, or the orderly passage of two persons through a narrow door where if one chooses “after you” and the other chooses “before you.”
These are examples of equilibria that are self-enforcing like all equilibria. However, most of the conventional rules of coordinated behavior that have the greatest significance for the social order are equilibria only because deviation from them by one individual provokes sufficient probability of the deviant being punished by adjacent individuals, so that the probability-weighted payoff from deviation is lower than that from adherence to the convention. Formally, some or all of the “players” pursue a contingent “strategy” that kicks in to respond to another’s harmful deviation and sanctions it by one of a variety of more or less heavy punishments. The best examples relate to property and contract. Here, the deviant strategy of trespass, theft, usurpation, and default would yield vastly higher payoffs than respect for property and fulfilment of contracts, were it not for an adequate probability of sanctions. Administering a sanction involves some cost, but the conventions of property and contract function as equilibria no one can expect profitably to deviate from if a sufficient proportion of “players” deem it that the strengthening of the convention due to their own action in administering punishment is worth more to them than the cost of doing so. Bearing the cost of punishing deviation is a classic public goods problem. The public good is the convention. Not punishing the deviation when one does not know whether enough others are going to do so augments the risk of weakening the convention. There is an incentive to diminish the risk by assuming the tasks of punishment.
In the Hobbesian tradition of political thought, the likelihood that conventions of property and contract will spontaneously emerge and be protected by the voluntary defensive action of those benefiting from the conventions is never envisaged, and the task is entrusted to Leviathan, despite ample evidence that such conventions have since time immemorial been deeply anchored in people’s consciousness and conduct. Hume, I believe, was the first to recognise that conventions, including those regarding property and the keeping of reciprocal promises (i.e. contracts), exist and are the outcome of spontaneous rational conduct. He implicitly but clearly scotches the Hobbesian idea of a need for Leviathan when he says “…the stability of possession, its translation by consent and the performance of promises. These are…antecedent to government.” 
The complete set of conventional rules banning torts against life, limb and property, nuisances, and incivilities is neither imposed nor sponsored by authority. Nor is it the outcome of bargaining. It constitutes ordered anarchy. There are plausible explanations why ordered anarchy, despite its moral attraction and its nature as an equilibrium, does not subsist through history, but is overlaid by a different rule system involving government. Surveying these explanations is outside the purpose of my paper. However, the concept of ordered anarchy is a standard of reference that must always serve as the background in considerations of the nature and role of government.
1.2. The Rule System by and for Government, and the Rule of Submission
In a complete rule-bound system, every rule except at least one is the product of a rule. There must be at least one that is not rule-made, for if there were none, there could be no first rule, hence there could be no rule. The exception must be the product of something outside the rule-bound system and once created by that thing, can in its turn create rule-made rules. We shall call it the rule of rule-making (which includes rule-change).
The origin of the rule of rule-making may be divine command, moral authority, material power, or the unanimous will of all concerned by it and its consequences. What it is or what it is believed to be is an important question for its legitimacy, but not strictly relevant to our purpose. What matters here is that it is a valid rule if it functions to make rules that turn out to be binding.
The rule of rule-making must define the person or persons, for short the government, entitled to make rules and the conditions it must fulfill in doing so. For instance, it may lay down that government is carried out by the king, that he must be anointed and crowned, and the rules he makes must be decided by him in his council. The general form of the rules is to say what may, shall, or shall not be done by whom. Accordingly, it may be addressed to individuals, or to the government itself. The latter type is a collective choice rule (in the literature inspired by the work of Kenneth Arrow it is called, perhaps a little suggestively, a “social choice rule”), or constitution.
It is apparent that in laying down a constitution, the rule of rule-making refers to itself, which would vitiate its logic and make rule-bound constitution-making nonsensical. However, as it refers to other things as well, it is only partially self-referring and is saved from being nonsensical. This difficulty will be briefly faced, though not really resolved, in 1.3.
A necessary complement of the rule of rule-making is the rule of submission which requires that all shall obey the rules, including this rule, regardless of whether they approve or disapprove of them. The enforcement problem of the rule-bound rule system impinges entirely on this rule.
The crucial aspect of the rule of rule-making seems to me the frontier it draws between the “zones” of individual and collective choice. There is a zone of the feasible universe where choices are reserved for individuals and another where power is given to collective choice to override individual choices. To the extent that the collective choice rule can define or by a rule-change amend the frontier, its zone may encroach on the zone of individual choices, while the reverse is of course not the case; individuals have no rule-making power enabling them to encroach on the collective zone. Much of the problem of limited government is summed up by this sentence.
What follows immediately below is a leap from the wholly abstract to the mundane and almost comic. As far as I am aware, no constitution at present in force imposes on the government a limit to taxation. Collective choice, in other words, is not restrained by its rules from encroaching on individual choice in the matter of pre-empting material resources. Germany’s post-war constitution is no exception. However, its Article 14 guarantees the ownership of property (though it is not made clear against what, and whether income qualifies for the guarantee). The article adds (gratuitously, it seems, for a document that should lay down what shall be done) that private property also serves the common weal. The presiding judge of one of the two chambers of the Constitutional Court chose to interpret Article 14 to mean that half of a person’s owned resources may be taken from him to serve the common weal, which in turn implied that taxation of all kinds must not exceed 50 per cent of anyone’s resources. The other chamber did not support this interpretation and the ensuing debate among constitutional lawyers about the “Half-and-Half Principle” seems now to be petering out.
Equally quaint and droll, but in a way also rather ominous, is a remark Hayek makes about taxation in one of his essays, quoted verbatim in my 1989 article “Is Limited Government Possible”. The gist of Hayek‘s astonishing statement is that while government may use coercion to enforce the rules of just conduct, it may also provide many useful services that do not involve “coercion except for raising the necessary means.” Is one to take it that the usefulness of the services it renders is a warrant for the government to “raise the means”? If this were a clause in a constitution, government would only be limited by physical feasibility, for its useful services are potentially no doubt unlimited.
1.3. Self-Referring Rules and Inseparable “Powers”
It is difficult to quarrel with John Austin’s view that every legal rule must be backed by a legal rule sanctioning the breach of the rule, though one may argue that the sanction need not always be a material one. However, the requirement of sanction by some punishment creates a logical difficulty. If the initial rule is breached, the sanctioning rule orders punishment, but if that rule is also breached, it must order to sanction itself for its own breach. Such self-reference is nonsensical. To avoid it, there must be a further rule ordering to sanction a breach of the sanctioning rule, and so on in an infinite regress.
In a magisterial essay, Herbert Hart reviews Hans Kelsen’s criticism of Austin. Though Hart with characteristic courtesy omits to say so, Kelsen’s solution seems to be to call the infinite regress to a stop, which is clearly one way of resolving the mater. Hart argues that the logical defect of self-reference can be overcome if the rule refers to itself and also to another rule, i.e., if it is only “partly self-referring.” He shows that Article V of the American constitution and other rules of the same structure refer to themselves but to other rules as well, and on this ground he refutes Alf Ross’s widely remarked attack on such clauses in his Law and Justice.
It is probably unwise to fault Hart on logic and the present author intends to do nothing of the kind. However, while admitting that partly self-referring rules are technically well-formed, they may still leave some unease. Hart himself cites the example of a South African constitutional clause that was impeccably “entrenched” by invoking itself and another clause, and that successfully withstood challenge in the courts, but was got round by the government. One might say that the clause, like the chastity belt, merely occasioned delay before Nature took its course.
It is also the case that quite momentous changes in rules can be effected in ways that many consider breaches of the spirit if not of the letter of the constitution, but that have no recourse to the rule-changing rule. The post-war evolution of American constitutional law at the hands of the Warren Court is an example.
However, the fundamental problem of any structure of a rule of rule-making is not its logico-juridical impregnability, but the role of power in enforcing the rule of submission. If the rule of rule-making states that the king shall not make a certain type of rule and the king breaches that rule, how does the king enforce the rule of submission against himself? He may, of course, submit voluntarily, contrite at his own offence and withdraw the offending rule. (The same voluntary submission may be shown by ordinary subjects to rules demanding their obedience). But if this is not the case, ultimate recourse to force must be known to be possible, however distant and rare actual violence may be. The potential for violence is under the king’s control; the state is supposed to have a monopoly of it. If so, we are reduced to the entirely and purely self-referring reality that the king enforces the king’s obedience to his own rule.
One of the dangerously misleading phrases in this context that has penetrated political thought is the “separation of powers.” It is dangerous because it tacitly suggests that such separation can resolve the paradoxical feature of every constitution which the king enforces against himself (or a government against the mandate of its own majority). Montesquieu uses words that do not illuminate the distinction between separate functions of a government and separate repositories of power under separate control that may act independently of one another or even against each other. The latter kind of separation of what ultimately boils down to armed formations and firepower is difficult to conceive of within a single government. It is fairly obvious that Montesquieu did not mean it, and if we mean it when we use the phrase, it is that we do not really think of what the words could mean. The result is a blind belief that the separation of functions among legislature, executive, and judiciary contains within itself a solution to the constitutional paradox of real, though perhaps not logico-legal, self-reference.
1.4. Maximization and the Size of Coalitions
The person or persons seeking control of the government, and retention of control once gained, must generally form coalitions to achieve this. This is true even of ostensible one-man rule, where the absolute monarch or dictator must rely on concentric circles of supporters, from the innermost circle of the most committed and most highly rewarded to the outermost one of consenting masses. It is more explicitly true of governments that can only gain control by fulfilling some formal requirement such as ballot-counting giving them simple or qualified majority support. Let us call the government and its supporters the “winning coalition.” It is formed in an auction where rival bidders make election promises to attract the desired number of supporters.
In very plain terms, a government must reconcile two objectives that are diametrically opposed. One is to maximize the total of all gains, religious and lay, moral and material, cultural and economic, that it can obtain for itself and remunerate its supporters at the cost of the losing coalition. The larger and richer the losing coalition, the greater is the loss that can potentially be imposed upon it. The obvious corollary is that the smaller is the winning coalition, the larger is its potential gain both in absolute total and per head.
No less obviously, the more the winning coalition is reduced in size to maximize its gain, the greater is the probability that it will fail to get control of the government or lose it even upon some minor shift in public attitudes. The safe dose of support is in excess of that just necessary for gaining and retaining power. To maximize the excess is to minimize the gain, and of course vice versa.
There has been an irregular, often interrupted and reversed, movement throughout medieval and modern history from smaller to larger and from informal to formally defined winning coalitions. From the conquering war band and its leader we have moved to a looser feudal structure, to absolute monarchy, to constitutional monarchy, representative government (either unelected or elected on a restricted franchise), and finally to simple majority rule and universal suffrage with constitutional limits or with absolute sovereignty. The latter may well be a rule-of-thumb approximation of some optimum compromise between the two contradictory objectives of maximum gain for the winning coalition and maximum security of its tenure. With a touch of good-natured irony, we might say that if this is the optimum combination, it is the “end of history.”
2. Contingent Limits of Government
The interim conclusion that might be drawn from Section I is that self-imposed rules attempting to limit the scope of collective choices, such as constitutions, are not strong and though they may be observed if they are innocuous and only forbid government to do what it is not strongly interested in doing, they could hardly be expected to restrain government from doing what it is anxious to do or must do to preserve its tenure of power. The general absence from constitutions of restrictions of taxation lends some verisimilitude to this conclusion, though it would still have to be regarded as tentative.
However, circumstances and events may occur that are contingent, not constant and integral features of the state of affairs but sufficiently frequent to be reckoned with. They do seem to place limits on collective choice when they occur.
2.1 Campaign Finance as a Limit of Government
There is an incomplete, partial indicator of the extent to which collective choice overrules individual choice, namely the share of central and local government spending (including expenditure by compulsory social insurance schemes) in GDP. It is the best quantitative measure for the good reason that it is the only one we have.
It turns out to be the case that in some European countries such as Scandinavia and France this share is just under or just above 55 percent, in others in the mid- to high 40s, while in the United States it is in the mid-30s. Popular wisdom has it that this large difference is due to European nations being left-leaning and falling in readily with collectivist policies, while the American people are rugged individualists, preferring opportunity to security and mistrusting big government. There is perhaps a little truth in this belief, but such as it is, it is probably getting less true as time passes. It might be interesting to look at a less heroic hypothesis.
Let us suppose that campaign expenditure has a significant probabilistic effect on election results. In Europe, campaign finance is in many countries provided more to parties than to individual candidates, and is provided more or less even-handedly to each party out of general taxation. A party will make campaign promises that may succeed in recruiting enough votes to form the winning coalition. Its promises and subsequent policies are directed at forming and maintaining the winning coalition and have no or scant influence on how much campaign finance it will obtain.
In the United States, it is still largely individuals and not parties that get elected. Party discipline is loose compared to Europe and candidates raise their campaign expenditure to a large extent by personal effort for their personal purposes. To the extent that campaign donations are sought from higher income donors, a candidate’s program must be more “conservative” and less redistributive than if donations came from all income groups in proportion to their income. If elected, a legislator has both a debt of honor to pay to his high-income donors and must establish a record that will help him gather donations on future occasions if there are any such.
This structural difference in campaign finance and the allegiance of legislators between Europe and America may or may not be the real cause of the more limited pre-emption of GDP for collective purposes in the United States than in Europe, but it is hard to think of a more likely or much stronger cause.
2.2. Taxable Capacity
The maximum gain a winning coalition derives from winning depends solely on the size and wealth or income of the losing coalition if, but only if, the proportion of the losing coalition’s wealth or income extracted from it to the benefit of the winners is a constant.
Obviously, this may but need not be the case. The tax rates the winners choose may be higher or lower depending on what they think the total tax yield will be at various rates. They may not choose to impose rates that produce the maximum yield because of the adverse effect this may have on the level of support the winning coalition needs to retain its tenure with some margin of safety.
However, a further limitation is liable to arise. Tax rates that fall short of maximizing the immediate tax yield may yet be too high to maximise the discounted present value of the multi-year tax yield over the expected tenure of the winning coalition. The reason is that rising taxes trigger two streams of exit: a “brain drain” typically composed of the most enterprising, ablest, and best-trained members of society, as well as a flight of capital in the form of money, and of going concerns that “de-localize” to low-tax jurisdictions.
Tolerating the brain drain and the capital flight reduces the present value of the expected tax yield. Not tolerating it, but imposing emigration restrictions and harsh controls on capital movements also reduces it, and probably more drastically, too. For restrictions of a moderate sort would very likely accelerate the brain drain and the capital flight as those concerned would want to flee before escape routes are closed. Really strict controls severely enforced, on the other hand, would be liable to transform the country to a replica of pre-1989 East Germany, with the disastrous consequences that are well known and among which impoverishment is not the greatest. Modern governments have in the last decade generally taken the opposite line of defense, reducing the corporate tax rate and moderating the progressivity of personal taxation.
There is, then, in modern economies, a taxable capacity that governments are well advised to respect, which is lower than the physically feasible one and also lower than the one that would maximize the immediate tax yield. When modern economies in fact react as sketched under the present heading, taxable capacity that seems sub-optimal in the short run but is probably optimal in the longer run, does seem to constitute a limit a government may rationally choose to observe as a drinking man might limit his daily intake of whisky in order to preserve his liver.
Government in Great Britain from 1945 to 1979 was alternately controlled by Labour, the Conservatives and then again by Labour. From the perspective of three decades later, the government can be seen as left-leaning to a greater or lesser extent. Three main features characterize the policy of this epoch. The first was the development of the welfare state, particularly the free health service, as well as the liquidation of the pre-war education system of selective schools and its replacement by a unified non-selective system. The second feature was the helping hand extended to the labor unions in their ascendancy, the securing of their legal immunity and the complete lack of controls over picketing. Labor unions were treated as a second-tier government with influence on policy that bordered on veto rights. The third defining feature of post-war British policy was budgetary softness that greatly contributed to chronic weakness of the external accounts, a spectacular decline of the pound sterling, and humiliating recourse to the IMF.
The winning coalition, regardless of periodic shifts in its composition from Labour to Tory and back to Labour, gave the impression of helpless drift, maintaining itself by developing ever more elaborate welfare entitlements. It consistently responded to strikes in the public sector by capitulating, which led to the recognition that strikes bring easy glory to union hierarchies, and that of course bred more strikes. Meanwhile, the country’s financial outlook was becoming truly alarming.
By 1979, the electorate had passed through the phases of egalitarian enthusiasm, complacency, dissatisfaction, and disquiet about the symptoms of decline and disorder. It had reached a state of virtual panic about the future.
The normal state of mind of a typical electorate is to discount the future at a high and apparently higher rate than do its individual members. In its role as collective decision-maker, the wining coalition fairly consistently favors more public goods and services paid for at the margin by public dissaving, i.e. a fairly chronic budget deficit (whereas individual members of the coalition save).
The 1979 election victory of Margaret Thatcher, who in the popular imagination embodied austerity, welfare cuts, budgetary discipline, and “union-bashing” (though this image was in large part kitsch and fabrication) was certainly contrary to the sort of result one would expect from an electorate apparently addicted to a constant expansion of welfare entitlements and redistribution. It can be analytically explained by a reversal of the usual strong collective preference for the present over the future. It was as if instead of applying a discount on future goods, the collective political will turned around and applied a premium. This may be the rational reaction to a drastic, panicky lowering of the expected stream of future goods, which in turn is consistent with a sense of panic about order unraveling and a bleak and poor future becoming a real threat. (It is not absurd to suppose that a similar, though much milder, panic was partly responsible for the replacement of Jimmy Carter by Ronald Reagan in 1980 and the French electorate’s turning its back on 26 years of the equally welfarist and equally soft Mitterrand and Chirac presidencies in 2007).
Panic, then, may be the limit a government inadvertently runs into when it expands recklessly, yields promptly to blackmail, and shows weakness before partial interests.
2.4. Beyond Interest
Individual choices obey conventional rules that arise spontaneously and are self-enforcing. They also obey collective choices which are enforced upon them. Collective choice involves an exogenous rule-making rule which collective choice itself may endogenously amend over the course of history to suit the maximizing conduct of winning coalitions. Formally, such adaptation need not involve a breach of the rule-making rule thanks to its self-referring nature. Metaphorically, the lady holds the key to her own chastity belt. Thus, collective choice is as good as omnipotent, potentially antagonistic to individual choice and dominating it within rules of its own making.
Yet it is a truism that collective choice is in some manner composed of individual choices. No matter how indirect and involved the manner in which the individual will is transmitted to the collective one, it is the minds of persons that ultimately decide what will be chosen.
If we steer clear of the concepts of utility and preference as irredeemably tautological, having no meaning apart from the choice they are claimed to inspire, we are left with a simple account of human motivation that has the merit of having a descriptive content other than as the description of a choice.
Human motivation, then, is twofold, identifiable interest and everything else or mixtures of the two. The present paper has throughout been running in terms of identifiable interest and the choices such interest would help to explain. Limits on government were examined in the light of their “incentive-compatibility.” As a last stage in the argument, it is now necessary to look at the “everything else” that is not interest, and in particular to its element that is basic to political philosophy, unreasoning conduct according to standards.
Standards differ from rules because we follow rules at least in part because of their consequences and the consequences of breaching them. Calculus plays its part and even if rule-obedience has become a habit, the habit was first formed under the influence of calculus. Standards may have consequences, but if we follow them, it is not because we reason about them. We do it reflexively. However, whether we follow standards, and which ones, is a purely contingent fact of the social state of affairs; there are possible worlds where every human choice is motivated by reasoned interest and nobody observes any standard, though the idea of such a world might give not be a reassuring one.
Unreasoning, even plainly unreasonable standards can effectively limit government if they are widely followed. For about a century and a half before Keynes’s General Theory became common currency for the literate and the semi-literate, it was widely believed that repeated deficits in the state household were mortally dangerous, liable to lead to the country’s ruin and to be countenanced only in desperate circumstances. Balancing expenditure and revenue was an unreasoned standard that did to a large extent tie the hands of governments. Now, of course, deficits are not believed to act as magic poisons, and they are decided by the interplay of interests.
Perhaps the purest and strongest type of limit on government is the standard, non-interest motive arising from superstition and taboo. To say so is not to denigrate such standards. Nor is it meant to belittle religious ethics or a lay moral code. But it expresses the apprehension that religion and lay morals may not be sufficiently impervious to utilitarian enlightenment, the reasoning that is driven by interest.
 Jasay, A. de, “Is Limited Government Possible?” Critical Review, 1989, Vol.3, No.2, repr. in Radnitzky, G. and H.Bouillon, eds., Government: Servant or Master?,1993, Amsterdam and Atlanta, Rodopi, repr. In Jasay, A. de, Against Politics, 1997, London, Routledge.
 Hume, D., A Treatise On Human Nature, 2nd ed., 1749/1978. Oxford, OUP. P.541.
 In Hayek, F.A., New Studies in Philosophy, Economics and the History of Ideas, 1978. Chicago, Univ. of Chicago Press, p.144.
 H.L.A. Hart, Self-Referring Laws, in Hart, Essays in Jurisprudence 1983, Oxford, OUP.