1. The ontological problem of norms is essentially the question what it means to say that there is (exists) a norm to such and such effect.

It is reasonable to think that the logical nature of the facts which make norm-propositions true will be different for the different types (kinds species) of norm which there are. For most types of norm however these facts are contingent (empirical). It is a contingent fact that there are such and such customs in a community or laws of a state. In one sense it is contingent that chess is played according to such and such rules. For it is a contingent fact that there should exist the game which we call ‘chess’. But in another sense it is necessary that chess is played according to these rules. For a game with different rules would not be chess.

Are there norms which have necessary existence? The question is complicated by the fact that the very notion of necessary existence is problematic. Some may think that moral norms have necessary existence if they are theonomous i.e. the commandments (law) of God. Others may hold that moral norms have necessary existence as a ‘law of nature’. That moral norms are not contingent in the same way (sense) as customs and prescriptions exist contingently seems fairly obvious. But it does not follow from this that we must attribute necessary existence to them. The question will not be discussed further in the present work.

We shall here limit the discussion of the ontological problem of norms principally to prescriptions. Our main question will thus be: What does it mean to say that a prescription (command permission prohibition) to such and such effect exists?

We shall attack this problem in a somewhat roundabout way. The point of departure of the discussion will be the idea associated chiefly with the ethics of Kant that Ought entails Can. The justification of this procedure will be plain I hope from the answer which we are going to propose to our main problem under discussion.

2. The idea that Ought entails Can has been the subject matter of much discussion in recent times also. We may raise the question what Kant meant by it. This we shall not discuss at all. We may also ask what different things could be meant by it. And one may discuss whether the idea when understood in a certain way is true or not.

In our discussion of the principle we shall in turn focus attention on each of the three words contained in its formulation viz. ‘ought’ ‘entails’ and ‘can’.

Since ‘ought to’ and ‘must not’ are interdefinable (Ch. V Sect. 12) it is fairly obvious that the principle must be regarded as applying to norms which prohibit action just as much as to norms which enjoin action. What does the principle say when formulated explicitly for prohibitions? This is not quite clear. One suggestion would be this: If there is something which one must not do then one can forbear this. If however ‘can’ here refers to the generic ability (Ch. III Sects. 9 and 11) the above suggestion would be equivalent to: If there is something which one must not do then one can do this. The formulation of the principle for norms which prohibit thus challenges the question how the ‘can’ should be understood. We shall return to this question presently.

Does the principle apply to permissive norms? Does May too entail Can?

It is obvious that the answer to this question depends upon what we think of the nature of the permissive norm-character and of its relation to obligation (cf. Ch. V Sects. 13–16).

If we accept the view that permission is mere absence or lack of prohibition then it is clear that there are any number of things which one is permitted to do but which one cannot do. In fact anything which an agent cannot do he would then be permitted to do.

If we define permission in terms of prohibitions to a third party then it would follow from the principle that Ought entails Can that if something is permitted to an agent then other agents can prevent him from doing this thing. But since one can prevent an agent from doing only such things as he can (in the generic sense) do it would follow a fortiori that if something is permitted to an agent then this agent can do this thing. May too would then entail Can.

If finally we regard permission as an independent norm-character we cannot deduce from Ought entails Can that May entails Can. A decision is called for. We make this decision as follows: In the same sense of ‘entail’ and ‘can’ that we accept the principle that Ought entails Can we shall accept the principle that May entails Can.

We can formulate the principle for prescriptions in the following way: That something is the content of a prescription entails that the subject of the prescription can do this thing.

3. What is the meaning of ‘entails’ or ‘implies’ in the principle under discussion? Is the alleged connexion between norm and ability a logical (conceptual) or a physical (causal) connexion?

This last question we shall answer by saying that the connexion is logical. The tie between norm and ability which the principle envisages is a conceptual tie.

‘Entails’ will thus mean ‘logically entails’ and ‘implies’ will mean ‘logically implies’. The question may be raised however whether there is not a better name for the logical relation in question than either ‘entails’ or ‘implies’. We shall return to this point presently.

The idea that Ought entails Can has sometimes been thought to constitute a counter-argument against the well-known view associated chiefly with the name of Hume that there is a sharp distinction between norm and fact between Ought and Is. Those who wish to maintain a sharp distinction between the two it is said may be right in thinking that one cannot from the fact that this or that is the case conclude that something or other ought to be the case. But if it is admitted that duty entails ability then one may modo tollente from the fact that something can not be done conclude that there is not a duty to do this thing either. And although the duty to do a certain thing does not entail that this thing is done it nevertheless entails another factual conclusion viz. that this thing can be done.

Instead of using Kant's principle as an argument against Hume's view we may wish to make the assumed sharp distinction between Is and Ought a ground for refuting the view that a norm could entail factual consequences about human ability.

I think that both ways of arguing here—with Kant against Hume and with Hume against Kant—are wrong and that the conflict between the Kantian and the Humean viewpoints is apparent only. Those who think that the viewpoints conflict are guilty of a confusion between norms and norm-propositions. If I am right this shows the importance of keeping this distinction clear.

The principle that Ought entails Can as I understand it does not affirm a relation of entailment between a norm and a proposition. The entailment is between (true or false) norm-propositions on the one hand and propositions about human ability on the other hand. The antecedent (premiss) is to the effect that there is a norm of such and such character and content. The consequent (conclusion) is to the effect that the enjoined or permitted thing which is the content of the norm can be done. On this interpretation the Kantian principle that Ought entails Can is in no conflict with the Humean idea of the logical independence of Ought and Is.

There is a sense in which facts about human ability can be said to be prior to facts about the existence of norms. Whether a man can or cannot do certain things can normally be decided independently of considerations as to whether the acts or forbearances in question are subject to norm. But on our interpretation of the principle that Ought entails Can whether there is or is not a norm to such and such effect cannot be decided without first consulting facts about human ability. The existence of a norm depends logically on facts about ability. This is how we here understand the principle that Ought entails Can.

Considering what has been said about logical priority it would seem more to the point to replace the words ‘(logically) entails’ in our formulation of the principle under discussion by ‘logically presupposes’. Ability to act is a presupposition of norms. Norms cannot exist or better: cannot come into existence unless certain conditions about human ability are (already) satisfied.

For norms which are prescriptions we now get the following formulation of the Kantian principle:

That there is a prescription which enjoins or permits a certain thing presupposes that the subject(s) of the prescription can do the enjoined or permitted thing.

4. In Ch. III Sect. 9 we distinguished two meanings of ‘can do’. We called them the ‘can do’ of ability and the ‘can do’ of success. The distinction is connected with that between generic and individual acts events and states of affairs.

The question may be raised: When ‘can’ in the shorthand formulation ‘Ought entails Can’ means ‘can do’ does it then refer to ability or to success? Does ‘can’ in other words mean that the agent or agents in question can do the kind of thing which the norm enjoins or permits—or does it mean that the agent or agents in question can on such and such occasions do the thing enjoined or permitted?

If the principle that Ought entails Can is interpreted as laying down a (logical) condition for the existence of norms then it seems fairly obvious that the ‘can’ which is involved in it must be the ‘can’ of ability i.e. of generic acts. If we accepted the alternative interpretation of ‘can’ we should run into the following ‘paradox’:

Consider a person who has been commanded to do a certain thing on a certain occasion. He tries to do this thing but fails. We should then since he could not do the thing in question have to say that strictly speaking he was not even commanded to do it. Whenever a person unsuccessfully tried to follow a prescription there would be no prescription (for him). Failure to obey the norm would annihilate the norm. But this is certainly not how we wish to shape our notion of a prescription or norm. Therefore if we wish to make the principle that Ought entails Can an ingredient of our concept of a norm we must understand its ‘can’ in a sense which is compatible with the ‘cannot’ of failure. That is to say we must understand ‘can do’ to imply ability but not to imply success in each individual case.

Compelling and preventing is an annihilation of power to do or forbear. The power thus annihilated however is the ‘can do’ which refers to act-individuals and not the ‘can do’ which refers to act-categories (cf. Ch. III Sect. 12). This observation has the following consequence for the principle that Ought entails Can:

When we say that only such things can be commanded or permitted or prohibited to an agent as this agent can do we need not qualify this by the phrase ‘unless he is prevented from doing them’. For the ability to which the principle that Ought entails Can refers as understood by us is the generic ability and this does not pass out of existence when the agent is prevented from exercising it.

5. In Ch. I Sect. 9 we distinguished between norms concerning that which ought to may or must not be and norms concerning that which ought to may or must not be done. The first we also call ideals (ideal rules).

The question may be raised whether the principle that Ought entails Can applies to ideals also. Ideal rules as was observed are largely concerned with so-called states of character. They say that a man ought to be brave temperate truthful etc. Would an application of the principle that Ought entails Can to ideal rules mean for example that if a man ought to be brave then he can be brave? And would it then follow that if a man is a notorious coward and incapable of showing bravery the ideal rule does not apply to his case?

I think that the answer to the last question is in the negative. It does not follow however that the principle that Ought entails Can does not apply to ideal rules. But it follows that it cannot be interpreted as saying strictly that what ought to be also can be. My suggestion is that when applied to ideals the principle should be understood to mean that if a man ought to be such and such then he can become such and such-unless he already is this.

A man as he is now may not be able to live up to the ideal. His character may be undeveloped or corrupt. But the ideal may nevertheless apply to his case too. It does this if or as long as his case is not ‘hopeless’ i.e. if or as long as it is true to say that he may become like the ideal.

This application of the principle that Ought entails Can to ideal rules raises interesting problems of moral philosophy. How shall the ‘may (can) become’ be understood? Does it refer only to that which a man may become as a consequence of his own efforts and training? Does it include that which may befall him as a consequence of natural causality e.g. processes affecting his bodily and mental development? Or is the suggestion that whether a man can become like the ideal or not depends neither on his own efforts nor on causality in nature alone but also on the grace of God?

Since this is not a treatise on ethics we shall not discuss these questions. But it may be useful to see their connexion with the more elementary problems which occupy us here.

6. I hope that the discussion in the preceding sections has made clear the sense in which we here understand the principle that Ought entails Can. The question may now be raised what grounds there are for thinking that the principle as understood by us is true.

This question of truth must not be misunderstood. It is a question neither of empirical verification nor of logical proof. The adoption of the principle is rather a matter of decision. The purpose of the principle as I see it is to help to mould or shape the concept of a norm. The question of the truth of the principle is essentially a question of how well it serves the philosopher's purposes. It should perhaps rather be called a question of ‘acceptability’ or of ‘plausibility’ than a question of ‘truth’.

It must not be taken for granted that the adoption of the principle is equally plausible for every kind or type of norm.

Shall we for example regard the principle as being valid for norms of the kind we call rules? We shall not attempt to answer the question. The first reaction to it is I think that it is not quite clear how the principle applies to rules. Consider e.g. the rules of a game. Obviously the existence of the rules of a game is independent of whether individual men master the moves of the game. But what shall we say of the case when there is a contradiction in the rules so that a situation may arise when no player could possibly comply with the demands of the rules? One thing which could be said is that the game ‘collapses’ if its rules put contradictory demands on the players. It ceases to be a ‘proper’ game. It is a logical requirement of rules of a game that it must not be impossible to satisfy the requirements which the rules make on the players. This would be a way of applying the principle that Ought entails Can to rules (of a game).

Of more interest to the discussion of norms in the present work is the application of the principle that Ought entails Can to technical norms.

Let the norm be that if I want to attain a certain end e I ought to do a certain act a. Can I not want to attain this end independently of whether I can or cannot do any act which is necessary for its attainment? The answer to this question is not as obvious as may at first appear.

That e is something I want can mean several things. It can mean for example that e is something which I would ‘welcome’ if it happened to me—as a grace of fate or thanks to the action of some other agent. In this sense e can be a thing wanted by me even though I cannot do that which is necessary for its attainment. Or that e is something I want can mean that I wish that e would happen to me. This too I can do without being able to use the necessary means for the attainment of e. But to want something can also mean to pursue it as an end of action. This is neither the same as to wish for it nor as to welcome it if it happens. It may be argued that pursuit of something as an end of action is not independent of my abilities but that on the contrary it requires or presupposes that I know how to attain the end can do the things which are necessary for its attainment. I may of course fail to attain an end which I pursue although this requirement on my ability is satisfied. For as we know ability to do something is no infallible guarantee of success in the individual case.

I shall accept the view that pursuing something as an end of action presupposes ability to do the things which are necessary for the attainment of the end. This conceptual connexion between the pursuit of ends and ability to do things will turn out to have important consequences for the application of Kant's principle to norms which are prescriptions.

7. When applied to prescriptions Kant's principle as already observed (Sections 3 and 4) states that the existence of a prescription enjoining or permitting a certain thing presupposes ability on the part of the norm-subject(s) to do the kind of thing enjoined or permitted. Is this an acceptable view of the relation between prescription and ability? We shall consider the question in the light of an example.

An officer orders a soldier to swim across a river. The soldier refuses to plunge into the water. He gives as an excuse that he cannot cross the river swimming. Must we not however say that he was commanded to swim across the river irrespective of whether his excuse is truthful or not? How can we say that he refused to do something if he cannot truly be said to have been asked to do this?

Assume that our soldier is court-martialled and charged with disobedience. If he cannot substantiate his claim not to be able to swim across the river then clearly he can be sentenced and punished for disobedience. But if he can substantiate his claim can he then not be sentenced and punished? The soldier can of course be treated in the way which is characteristic of punishment and which involves the infliction on him of some kind of pain or disagreeable thing. This treatment may even rightly be described as punishment. It would be punishment for the manner in which he answered the officer or punishment because he did not on the spot prove that he could not perform the required act e.g. by plunging into the water and letting the officer thus test his ability. Or he may be punished because he had not learnt to swim although he was supposed to have learnt to do so in the course of his training. But whatever he is being punished for must—if his lot is to be called punishment as distinct from mere maltreatment—be something which he could have done but neglected to do. And since on our assumption the soldier cannot do that thing which the order to swim across the river requires he cannot be punished for having disobeyed this order. He cannot have disobeyed it for there is ‘room’ for disobedience only where obedience is possible. And obedience is possible only when there is ability to do the required thing.

An attempt to describe the case of the officer and the ‘disobedient’ soldier reveals conflicting conceptual tendencies. On the one hand there is an inclination to say that since he could not do the required thing he could not even be commanded to do it. On the other hand there is an inclination to say that there was a command since he obviously was required to do something. How shall these two inclinations be reconciled?

One possibility of reconciliation would be by means of a distinction between the giving of prescriptions and the receiving (taking) of prescriptions. One can give an order to somebody it might be argued irrespective of whether that person can carry it into effect or not—but one cannot take an order from anyone unless one has the ability to comply with it. Similarly it may be argued that a permission can be given to an agent irrespective of his abilities but that one cannot have (‘enjoy’) a permission unless one can do the thing permitted.

How does this splitting up of prescriptions into a giving- and a receiving-aspect affect the question of the existence of prescriptions? It would be tempting to say that this existence depends upon the giving-aspect alone. Then it would appear that ability to do the prescribed things is not a logical precondition of the existence of prescriptions.

I shall try to show however that even if the existence of a prescription depends upon its giving alone the conclusion that this existence is independent of the abilities of its receiver does not necessarily follow.

8. Prescriptions originate come into existence through a peculiar mode of human action. For this mode of action the giving of prescriptions (orders permissions prohibitions) we have earlier coined the name normative action (see Ch. V Sect. 7).

We have distinguished between act and activity and between the result and the consequences of action (see Ch. III Sects. 5 and 6). The questions may now be raised whether the giving of a prescription is an act or an activity and whether the existence of a prescription is the result or a consequence of normative action. I propose to answer these questions as follows:

The giving of a prescription is an act the successful performance of which results in the existence of a prescription. The consequences of normative acts broadly speaking are the effects which (the giving of) prescriptions may have on the conduct of those to whom the prescriptions are given.

In acts we have said (Ch. III Sect. 6) activity is usually involved e.g. in the form of muscular activity and movements of limbs. The activity which is characteristic of normative acts is verbal activity. It consists in the use of norm-formulations to enunciate or as we also say promulgate the norm (prescription) to the appropriate subjects.

We thus distinguish between the act of giving a prescription and the verbal activity which is involved in the act. The point of making this distinction can perhaps best be illustrated through an analogy between the giving of prescriptions and the giving of promises:

To promising as to prescribing the use of language is essential. The giver of a promise usually utters a certain form of words ‘I promise to…’. The uttering of these words is activity.

The mere fact that somebody utters a promise-sentence does not entail that a promise has been given. If a small child says to me ‘I promise to give you a thousand pounds to-morrow’ or if I say to a friend ‘I promise to make you Emperor of China’ or if an actor says on the stage ‘I promise to revenge my father’ nothing has been promised. The child was talking unwittingly I was joking the actor was acting a role. This is trivial—but it shows that whether the uttering of a promise-sentence ‘constitutes’ an act of promising depends upon other factors beside the verbal activity which is essential to the act. The same holds true of prescriptions. Mere uttering of imperative sentences and use of other forms of prescriptive language does not establish that a command permission or prohibition has been given does not by itself ‘constitute’ an act of commanding or permitting or prohibiting.

What then is required in addition to the verbal performance to constitute normative action? For answering this question also the comparison between promises and prescriptions is illuminating.

When the uttering of a promise-sentence ‘constitutes’ an act of promising or ‘results’ in a promise having been given there exists henceforth and for a time a relationship between the giver and the receiver of the promise the promiser and the promisee. The promiser is as we say under an obligation to fulfil his promise i.e. to do the thing which he has promised to do. It is natural to call this a ‘normative relationship’ between the two parties. It would not be quite right to ‘identify’ the promise with this normative relationship. But it is certainly right to say that when the uttering of a promise-sentence leads to or results in the establishment of this normative relationship then a promise has been given.

Similarly when the uttering of a command-sentence ‘constitutes’ an act of commanding there exists henceforth and for a time a relationship between the giver and the receiver of the command the commander and the commanded. We could call this too a ‘normative relationship’ between the two parties. I shall prefer to call it a ‘relationship under norm’ between them. Again as in the case of promises it would not be right to identify the command with this relationship under norm. But it is right to say that when use of prescriptive language leads to or results in the establishment of this relationship between a norm-authority and some norm-subject(s) then the prescription has been given the normative act successfully performed and the norm has come into existence.

Prescriptions do not only come into being; they also pass out of existence. Prescriptions cease to be when the relationships under norm which the giving of the prescriptions established dissolve. The life-span of a prescription is thus the duration of a relationship between a norm-authority and one or several norm-subjects. As long as this relationship lasts the prescription is said to be in force. The existence of a prescription is not the fact as such that it has been given but the fact that it is in force.

9. Let the question be raised: Why does a certain agent command (order) another agent to do or forbear a certain action?

Sometimes an order is given because the giver of the order has in his turn been ordered to give it.

When a prescription is given because there is an order to give it then the normative act is itself subject to i.e. the content of a norm. This is a common and important type of case. Some logical problems connected with it will be discussed later in Chapter X. For present purposes we can however ignore it. For it only removes the question ‘Why?’ in which we are here interested to the ‘second order’ normative act through which the order to perform the ‘first order’ normative act came into existence.

When the normative act of giving an order is not itself the content of a norm the common type of answer to our question appears to follow this general pattern:

The giver of the order wants the result of the prescribed act to happen. Therefore he wants the subject of the prescription to do the act in question i.e. to make the wanted change happen. By commanding the subject he may make him do the act. Therefore he gives the order. The normative act is a means to the norm-authority's ends. It is a means to making the norm-subject do something and this in turn is a means to making a certain thing happen. If we wish to say as we are I think free to do that wanting to attain an end entails wanting to use the means which are actually used for the sake of attaining this end then we may also say that the norm-authority wants to command the norm-subject and that he wants to make the subject do the prescribed act.

When we say that the norm-authority wants a certain thing to happen and therefore wants the norm-subject to do this thing (make it happen) we ground the second want on the first. One can distinguish between necessary and sufficient grounds. In the case under consideration the first want is a sufficient and not a necessary ground of the second want. This means: wanting an agent to do a certain thing does not (logically) presuppose that I want this thing to happen. I may for example want somebody to do a certain thing merely because I want to put him in motion and not because I am interested in the result of his act. But it is probably right to say that normally we order people to do things because we are anxious to have those things done.

Wanting an agent to do something is obviously a sufficient ground for wanting to make him do that thing. As far as I can see the first want is here also a necessary ground of the second. This means: One cannot (logically) want to make a person do a certain thing unless one wants him to do that thing.

Wanting to make an agent do a certain thing is a sufficient but certainly not a necessary ground for wanting to command him. Commanding is only one among several means of moving people to action.

These observations will suffice on the mutual relations of the four cases of ‘want’ which we distinguished in connexion with the normative act.

I do not wish to maintain that always when the normative act is not itself the content of a norm the question why it is done can be answered with a reference to wants according to the above pattern. Orders are sometimes given ‘for no particular reason’. This however does not necessarily mean that the giver of the order could not be truly said to want the receiver of the order to do a certain thing. It need only mean that there is no particular reason for his wanting this. Yet I shall not deny that an order could be given ‘for absolutely no reason’. This however would be a most uncommon case ‘conceptually alien’ to the institution of commanding. One could perhaps call it a ‘misuse’ or a ‘parasitic use’ of this institution.

10. What has been said in the last section of commands applies mutatis mutandis to prohibitions as well. The giver of a prohibition normally wants the receiver of the prohibition to forbear something and also wants to make him forbear this by prohibiting him.

The giver of a permission cannot normally be said to want the receiver of the permission to do the permitted action. To permit is to let somebody do something. The question can be raised: What does the agent do who lets another do a certain thing?

This is but to raise afresh the question of the nature of permissions which we discussed briefly in Sections 13–16 of Chapter V.—A person can be said to let another do a certain thing when he has not prohibited the doing of that thing to that person and is perhaps not even aware of his doing it. This use of ‘let’ would correspond to the view of permission as mere absence of prohibition. Letting another do a certain thing can however also mean that one tolerates this act by that person and is aware of the possibility that he will do it even if not of his actual doing of it. This use of ‘let’ answers to the view of permission as toleration. I shall call the first form of ‘letting do’ passive and the second active.

If we take the view that a permission is a toleration in combination with a prohibition of non-interference then to give a permission to somebody is actively to let this other person do a certain act and to want others to forbear a certain other act viz. the act of making the first act impossible to the permission-holder. If finally we take the view that only the prohibition of noninterference is essential to the permissive norm then giving permission is wanting others to behave in a certain way and wanting to make them behave thus by commanding them.

Passively to let another person do a certain thing does not involve any kind of wanting on the part of the letting agent. But if a person actively lets another do a certain thing i.e. is aware of the possibility that he will do it and tolerates this then the first agent can also be said to want to leave the second agent free to do this. Wanting to leave an agent free to do something corresponds in the case of permissions to wanting to make an agent do something in the case of commands.

11. With the remarks in the last two sections on the intention and reasons involved in normative action we have arrived in the neighbourhood of a well-known ‘classical’ theory of the nature of norms. We can call it the will-theory of norms. According to it approximately speaking norms are the expressions or manifestations of the will of some norm-authority with regard to the conduct of some norm-subject(s).

The will-theory of norms has a primary application only to norms which are prescriptions. For it is essential to this theory that norms should emanate from an authority. Historically the will-theory of norms is known above all as a theory of (the nature of) the law of the state. Laws on this view are sometimes said to express the will of the state. As a theory of legal philosophy the will-theory of norms may be said to challenge the question of the nature of the authority behind the legal norms and ultimately the question of the nature of the state.

As a theory of the ontological status of prescriptions generally the will-theory of norms appears to me substantially correct. As a theory of legal norms in particular its acceptance need not so far as I can see commit one to an anthropomorphic or theomorphic conception of the state as a being endowed with a will.

If one had to give a brief characterization of the will of which commands are manifestations one should call it I think a will to make agents (norm-subjects) do and/or forbear things. For short we could call it a will to make do or forbear. This will is seldom a will to make do or forbear ‘for its own sake’ but has some ulterior end in view. As observed in the last section the authority normally wants to make the subject do something because he wants him to do this. And he wants him to do this because he wants the thing done to happen. It is a major problem of political philosophy how these ulterior ends of the state as the authority of the legal norms are (or should be) related to the ends of the citizens of the state as the subjects of these norms.

The will which permissions manifest can be called a will to tolerate.

12. The art of commanding we could say consists in ability to make agents do or forbear things which we want them to do or forbear.

It is clear that ability to command does not presuppose that the giver of the order can make its receiver perform an individual act which results in the wanted thing. He may succeed with his order to make the subject do this act but he may also fail. When he succeeds we say that the subject has obeyed the order. When he fails we do not necessarily say that the subject has disobeyed. There are at least three different types of reason why commanding may fail of its aim on the individual occasion:

One reason is that the subject disobeys. That the subject disobeys will mean that he understands the order and can do the kind of thing ordered but forbears and does not even try to do it on the occasion in question.

Another reason why commanding may fail of its aim is that although the subject tries to do it and can do the kind of thing ordered he fails to accomplish the act. He could not do it on the occasion in question because prevented by ‘physical obstacles’ or the interference of other agents. This we do not call disobedience. But there is no sharp border in the individual case between disobedience and this type of failure to comply with an order.

A third reason finally is that the subject cannot do the kind of thing which he is ordered to do. Then he can neither obey nor disobey the order. In such circumstances it is natural to say that he cannot ‘receive’ the order at all. The subject is incapable of entering into the ‘normative relationship’ with the authority which the normative act of commanding aims at establishing. This incapacity of course lasts only as long as the subject has not learnt to perform acts of the category in question.

Does this third type of failure of a normative act mean that the authority cannot command the subject? Ability on the part of the authority to command we have said is an ability to make the subject do the kind of thing which is commanded. If the subject cannot do the kind of thing in question neither can he be made to do it by being commanded. (The subject may of course be taught to do it or learn to do it and then on some other occasion be made to do it by being commanded.) And if he cannot be made to do this kind of thing the authority does not possess the ability which on our view of the matter is logically required for commanding this subject to do that kind of thing. The answer to our question above is thus affirmative.

It follows from what has been said in this section that a necessary condition of the existence of a command from some authority to some subject to do or forbear a certain thing is that the subject of the command can do this kind of thing. It should now be clear in which sense and for which reasons the principle that Ought entails Can may be said to lay down a minimum condition of the existence of commands (and prohibitions).

To give permission we said is ‘actively to let’ an agent do or forbear a certain thing. If ‘active letting’ is defined as the toleration of some action in the power of some agent then it follows trivially that one can permit an agent to do or forbear only such things as that agent can do. On this view of permitting May entails Can also.

13. The ‘art of commanding’ admits of several degrees of generality so to speak. To say that an agent ‘can command’ may mean no more than that he can command somebody to do something some kind of thing. This is ability to command in the most general and extenuated sense. From it must be distinguished ability to command a certain agent to do something ability to command somebody to do a certain thing and ability to command a certain agent to do a certain thing.

Let there be a norm-authority a a norm-subject s and a norm-content c. We can then make a table of corresponding abilities of various degrees of generality to command and be commanded:

a can command somebody to do something	s can be commanded by somebody to do something
a can command somebody to do c	s can be commanded by somebody to do c
a can command s to do something	s can be commanded by a to do something
a can command s to do c	s can be commanded by a to do c

The two first pairs of abilities listed in the table consist of logically independent members. The two last pairs consist of logically identical members.

On the view which we take here there can exist a command from a to s to do c if and only if the ability of a to command and of s to be commanded match as in the fourth of the above pairs.

When the abilities of a and s match as in the first second or third of the above pairs it is possible but not certain that a can command s to do c. When a can command s to do something it is plausible to say that he can also try to command s to do c irrespective of whether he actually can command s to do c or not. Similarly when a can command somebody to do c it is plausible to think that he also can (at least) try to command s to do c irrespective of whether he actually can do this or not. It is more doubtful whether the mere fact that a can command somebody to do something should be said to entail that he can try to command s to do something try to command somebody to do c and/or try to command s to do c. The notion of trying to command is not in itself precise enough to make a decision possible. The notion has to be moulded. We could distinguish between several concepts (senses) of trying to command depending upon which of the above requirements as regards ability are satisfied.

Trying to command is compatible with but does not presuppose ability on the part of the agent whom we try to command to do the thing which we try to command him to do.

One must distinguish between trying to command and commanding to try. Commanding a person to do a certain thing presupposes I shall say that the commanded agent can try to do this kind of thing. As observed earlier (Ch. III Sect. 10) it is not the case that one can try to do just anything. One may even argue that one can try to do on an individual occasion only such things as one can do generically. But this requirement may appear too strong. Perhaps we should say that some things which one cannot do in the generic sense of ‘can do’ one can yet try to do. But this notion of ‘can try’ presupposes that one at least ‘has some idea’ of how to do the thing in question. When there is no such idea present one cannot even try. One does not know how to try.

Thus from the fact that a can try to command s to do c it does not follow that a can command s to try to do c. But accepting what was said above about trying to command from the fact that a can command s to try to do c it does follow logically that a can try to command s to do c.

The distinction between commanding and trying to command is of importance for the problem of the existence of commands and for the interpretation of the principle that Ought entails Can. Trying to command nearly always results at least in the production of the words or symbols which we called the norm-formulation. Now the norm-formulation is the perhaps most ‘conspicuous’ feature in which the existence of a norm shows itself. For this reason it is tempting to say that already when a person is trying to command another a command comes into existence. This is how we often and naturally express ourselves. It is not the philosopher's business to correct language here. His task is to note the conceptual differences between cases—even when the cases are such that ordinary language blurs the differences.

One chief reason for laying down the conditions of existence in a manner which presupposes the validity of the principle that Ought entails Can is that this keeps the distinction between commanding and trying to command clear.

14. Wherein does ability to make agents do or forbear things by commanding them consist? In order to get a firmer grasp of this question let us ask first: What does the agent who gives commands do?

With one aspect of what he does we are already familiar. This is the aspect which we called promulgation. It consists broadly speaking in making known to the norm-subjects by means of language or other symbols what the norm-authority wants them to do or forbear.

Promulgation is necessary but not by itself sufficient to the establishment of normative relationships among agents. Beside promulgation there is also a second component involved in normative action. I shall refer to it by using another term from legal philosophy viz. sanction.

Sanction may for present purposes be defined as an explicit or implicit threat of punishment for disobedience to the norm.

The existence of a threat of punishment is not by itself a motive for obedience. Fear of punishment however is. When threat of punishment constitutes fear of punishment I shall speak of an effective threat or sanction.

Fear of punishment need not be the sole motive for obedience to the norm. It may even be regarded as being of the essence of some types of prescriptions e.g. of laws of the state that there should be other motives beside fear for obeying them. It is probably right to say that normally when action conforms to prescriptions the motive is not fear of punishment or of other unpleasant consequences. The function of sanction is to constitute a motive for obedience to the norm in the absence of other motives for obedience and in the presence of motives for disobedience. When the subject is tempted to disobey fear of punishment is one of the things which may ‘call him to order’. In extreme cases it is the only thing with this appeal on him.

The existence of the motive for obedience which fear of punishment is does not entail that it is strong enough to overcome in the individual case motives for a contrary conduct. Effective sanction is compatible with disobedience to the norm. But disobedience must be occasional must be the exception and not the rule. If disobedience is habitual rather than exceptional sanction is ineffective punishment not (seriously) feared.

The meaning of ‘exceptional’ and ‘habitual’ (dis) obedience calls for a comment. If the command or prohibition is what we have (Ch. V Sect. 11) called eminently general disobedience to the norm is exceptional when most subjects on most occasions obey the norm. If the command or prohibition is general with regard to the occasion but addressed to a particular subject disobedience is exceptional when this subject on most occasions obeys the norm. Similarly if the prescription is general with regard to subject but for a particular occasion only disobedience is exceptional when most subjects on this occasion obey the norm.

If however the prescription is (completely) particular it does not make sense to speak of exceptional and/or habitual disobedience to this norm. Shall we then say that disobedience proves that sanction was ineffective? We could say this. But we could also in such cases make the question of the efficacy of sanction depend on the subject's reaction to repeated prescriptions of the same content by the same authority. The two tests answer to slightly different notions of an effective threat. For present purposes we need a notion of efficacy which relies upon a test of the second kind.

We can now answer the question what the agent who gives commands does as follows: He promulgates the norm and attaches to it an effective sanction or threat of punishment for disobedience. When this has been done a normative relationship between authority and subject has been established. The normative act has been successfully performed. As a result of its successful performance a prescription exists i.e. has been given and is in force.

15. It is by no means trivially the case that any man can effectively threaten any other man to visit him with evil. The mere use of threatening words does not constitute an effective threat.

Occasionally a threat can constitute a motive for obedience to an order even though the authority could not actually have carried his threat into effect. The subject may have overestimated the authority's power to make his threat effective.

A necessary condition that a threat shall be effective is that the person who is being threatened believes that the evil with which he is threatened will befall him if he disobeys. Instead of ‘believes’ we could also say ‘estimates that there is a considerable risk’.

The subject can of course be mistaken in this belief. He may later find out that he need not have feared punishment because the authority could not have punished him even if he had wanted to. But it is probably right to say that normally such a belief is not mistaken. It usually has some ‘ground’ or ‘foundation’ e.g. in what has happened in past cases of disobedience.

When the commander can actually punish (visit with evil) the commanded in case of disobedience I shall say that the first is in the relevant respect stronger than the second. Normally a threat of punishment will be effective only if the person who threatens can carry his threat into effect. Normally in other words commanding is possible only when the authority of the commands is in the relevant respect stronger than the subject(s) of the commands. Ability to command is thus logically founded on a superior strength of the commander over the commanded. Occasionally genuine commanding is possible even when this presupposition is not fulfilled. This happens when the subject mistakenly believes in the superior strength of the authority.

It is of course quite possible that a person who is well aware of the fact that another could not harm him with punitive measures yet does as that other person asks him to do. There may be plenty of motives for such conformity to the will of another person. But then he has not been effectively commanded and his conduct is not rightly called ‘obedience to a command’.

The superior strength on which ability to command is logically founded can be either accidental or essential.

A person may accidentally be in a position to make another person behave according to his orders. He knows e.g. of some ‘secret’ which if made public would damage that other person's reputation and social position. Blackmail is a species of commanding which is based upon accidental superiorities of strength in the relationships among persons.

Adults may be said to enjoy a natural superiority of strength over children. That is why adults can command children. When the children grow up and come of age there is a natural end to this superiority. When the superiority of strength vanishes commanding ceases too. Counsel and warning take the place of command and prohibition in the relations between adults and their offspring.

Adult people are among themselves approximate equals in strength i.e. they have roughly the same power to do (good and) bad to each other. This explains why adults do not under normal circumstances issue commands to each other.

Officers command soldiers and officers of superior rank command officers of inferior rank. Does this mean that the officers are stronger than the soldiers? ‘In a state of nature’ the individual officer need not be stronger than the individual soldier. The chances are that they are approximate equals. But as officer the officer is stronger. He can normally carry into effect the threats by which he threatens recalcitrant subordinates. That he can do this is a consequence of the fact that he can command other soldiers to punish the recalcitrants. That he can command these other men is in its turn founded on his powers to threaten them with punishment for disobedience. This fabric of commanding powers is in the last resort dependent upon the fact that men in the army on the whole obey orders. Occasionally the fabric collapses. Subordinates no longer fear punishment for insubordination. Orders are not obeyed. The officers ‘lose command’ of the army can no longer command.

The superior strength of the commander over the commanded is also the factual basis on which the legal order of the state is founded. The existence of a legal order is the existence of normative relationships between the authorities and the citizens. It is essential that the authorities should be able to back their prescriptions to the citizens with effective threats of punishment in case of disobedience. When this condition is not fulfilled the legal order collapses or dissolves as when there is a successful revolution.