1. Can norms themselves be the contents of norms? Can for example a prohibition be itself commanded or permitted or prohibited?

If for some reason it be thought that norms cannot be the contents of norms the question may be raised whether norm-propositions can be so. Could it for example not be permitted that a certain thing is prohibited?

It appears more plausible to think that norm-propositions could be the contents of norms than that norms themselves could be so.

That a norm-proposition is true means that a certain norm exists. That a norm-proposition is the content of a norm would consequently mean that a certain state of affairs viz. the existence of a certain norm ought to or may or must not be (Ch. V Sect. 2). A norm with this content would be a norm of the type which we have called ideal rules (Ch. I Sect. 9).

It is doubtful whether there can be ideal rules about the existence of (other) norms. We shall not here inquire into this possibility.

Instead let us ask: How does the state of affairs which is the existence of a norm come to be? If the norm is a positive prescription i.e. a prescription with an empirical agent as its authority the answer is that it comes into being as a result of human action. Someone has given or issued the norm. Issuing norms is human action too. For action of this type we have previously coined the term normative action.

Even if it were not the case that the existence of norms could be meaningfully subject to norm it seems obvious that the human acts through which norms come into existence may themselves be obligatory or permitted or forbidden. And it may be suggested that this is what is really meant by the idea that norm-propositions sometimes are the contents of norms and also by the even more obscure idea that norms themselves may function as such contents.

By norms of higher order I shall here exclusively understand norms whose contents are normative acts. Since normative action is the giving of prescriptions norms of higher order are in a characteristic sense ‘about’ prescriptions. They may be prescriptions themselves. But they need not be so.

It may be thought that legislation in a state or the giving of prescriptions generally could be subject to some norms which are not themselves prescriptions but norms of a ‘moral’ or kindred nature. The idea that the laws of the state have (or should have) a ‘foundation’ in a Law of Nature postulates the existence of higher-order norms which are not themselves prescriptions but ‘govern’ (the giving of) prescriptions.

Norms of higher order which are prescriptions are of great importance to the legal order of a state and to other ‘hierarchies of commanding power’ such as e.g. an army.

Here we shall only be dealing with such norms of higher order which are (positive) prescriptions.

The interesting peculiarities of a logic of higher-order prescriptions have to do with relationships between norm-authorities and between norm-authorities and norm-subjects. These peculiarities cannot be treated within a theory of norm-kernels. For this theory takes into account only the character content and conditions of application of norms—omitting from consideration authority and subject.

The formal apparatus for dealing with norm-kernels which we developed in Chapters VIII and IX is therefore inadequate for dealing (interestingly) with norms of higher order. We shall not in this work however extend the scope of the strictly formal theory beyond the theory of the kernels. Our comments on the logic of higher-order norms will be ‘informal’. I hope they will invite formal treatment by making it plain that here is another virgin land of logical inquiry awaiting exploration.

2. Besides the acts of giving prescriptions there is another type of act which I shall also call normative viz. acts of cancelling (voiding withdrawing) prescriptions.

The act of issuing a norm transforms a world of which the negation of a certain norm-proposition is true into a world of which this norm-proposition itself is true.

The act of cancelling a norm again introduces a change into a world in which a certain norm exists i.e. of which a certain norm-proposition holds true. But which state of affairs does the cancellation produce? Two answers seem possible.

According to the first answer the state produced by the cancellation is simply one of which a formerly true (generic) norm-proposition is no longer true.

According to the second answer cancellation—like issuing—brings a new norm into existence. This new norm is the negation-norm of the cancelled norm.

On the second view of cancellation to cancel a command to do a certain thing thus entails issuing a permission to forbear this same thing and vice versa; and to cancel a prohibition to do i.e. command to forbear a certain thing entails issuing a permission to do this same thing and vice versa.

The question which is the ‘right’ concept of cancellation: the idea that cancellation simply dissolves or annihilates an existing normative relationship or the idea that cancellation creates a new norm is as such pointless. These are just two concepts of cancellation. But whether within normative orders of certain types such as e.g. the laws of the state cancellation actually is of the one kind or the other and what further consequences a choice between the alternative possibilities has can I think be problems of considerable interest. We shall not however discuss these problems here.

3. Norms whose contents are acts other than normative acts I shall call norms of the first order.

A norm of higher order is a norm of the second order if the normative act which is its content is the act of issuing or cancelling certain norms (prescriptions) of the first order. A norm is of the third order if the normative act which is its content is the act of issuing or cancelling certain norms (prescriptions) of the second order. In an analogous manner we define norms of the fourth fifth… nth order.

The subjects of norms of higher order i.e. the agents whom those norms address are themselves authorities of norms of lower order. We may call the authority of a norm of the first order an authority of the first order the authority of a norm of the second order an authority of the second order etc.

The authority who issues one norm may also issue another norm and the norms may be of different order. When we say of an authority that he is of order n we are therefore speaking of him as authority of one or several norms of this order n and not as authority of norms of some higher or lower order.

If the normative act which is performed by an agent a₁ when he issues a norm is itself the content of a norm which has been given by an agent a₂ to this agent a₁ then a₁ in issuing this norm will be said to be acting as sub-authority relative to or under a₂.

If the normative act of issuing a certain norm is not itself the content of any higher order norm then the agent who performs this act (issues this norm) will be said to act as sovereign or supreme authority of the norm in question.

4. It is probably right to say that among norms of the first order commands and prohibitions hold the most prominent position. Among norms of higher order the relative prominence of the various types of norm appears to be different. It is probably right to say that higher-order permissions are of peculiar interest and importance.

A higher-order permission is to the effect that a certain authority may issue norms of a certain content. It is we could say a norm concerning the competence of a certain authority of norms. I shall call permissive norms of higher order competence norms.

In the act of issuing a competence norm i.e. a permissive norm of higher order the superior authority of higher order may be said to delegate power to a sub-authority of lower order. ‘Power’ here means ‘competence by virtue of norm to act as an authority of norms’. I shall also speak of it as normative competence or power.

An important aspect of the study of norms of higher order is therefore the study of the logical mechanism of the phenomenon known in legal and political philosophy as the delegation of power.

It is essential to what I here call ‘the delegation of power’ that the norm delegating power should be permissive. If an authority commands or prohibits an agent to issue norms of such and such a content we shall not say that he is delegating power to the sub-authority. For an aspect of what we call the sub-authority's power is that he should be free to issue or not to issue the norms which it is within his competence to issue.

Are the permissions whereby power is delegated tolerations or rights? This question will for the time being be left open. (We shall presently decide that they are rights.)

It is however a noteworthy fact that the delegation of power to a sub-authority is often combined with an order to this authority to issue norms about certain types of act. The city magistrates for example may have the right to issue specific traffic-regulations about say speed limits and parking and the use of the horn when driving in the street but at the same time be ordered to issue some regulations about these things i.e. be ordered not to leave these things unregulated. Then it is not within the competence of the magistrate to decide whether there are going to be traffic-regulations or not but only to decide which these regulations will be.

It is easy to see what could be the raison d'être for this combination of a higher-order permission with a higher-order command. The supreme authority wants to have certain things subject to regulation perhaps for the sake of that which is also called the common good. But he leaves the details of the legislation to a lesser authority which has a better insight into what are the specific requirements of this end the common good in the particular case.

The limits of delegated power are often set by certain prohibitions. The authority may issue norms of a certain kind but must not issue norms of certain other kinds. It may be argued that norms the issuing of which is not expressly permitted to the authority are in fact forbidden to him to issue. This however cannot be deduced from the nature of permission as such. The prohibition if there is such a prohibition is a norm in its own right.

In Section 14 of Chapter V we briefly discussed the principle nullum crimen sine lege. We said that it could not be regarded as a logical principle to the effect that whatever is not prohibited is thereby ipso facto permitted. But it may be regarded as being itself a permissive norm to this effect. Such a norm which confers a normative status upon all human acts which are not already subject to norm may be said to close the system of norms to which it belongs.

As far as norms of the first order are concerned it seems natural to take the view that everything which is not forbidden is permitted but not very natural to take the converse view that everything which is not permitted is forbidden. It may in fact easily be shown that this latter view involves a contradiction unless either the doing or forbearing of every conceivable human act has already been individually permitted. For otherwise both the doing and the forbearing of acts which are not expressly permitted would be forbidden. This as we know is an impossibility (cf. Ch. V Sect. 14).

As far as norms of higher order i.e. norms regulating normative activity are concerned it appears much more natural to think that ‘whatever is not permitted is forbidden’ than to think that ‘whatever is not forbidden is permitted’. If we understand the ‘whatever’ as referring only to the doing and not to the forbearing of normative acts there is no contradiction in this idea (cf. Ch. V Sect. 14). There is however another logical difficulty to be noted:

A prohibition to the effect that no norm-authority must issue norms which he has not an explicit permission to issue would prohibit all normative activity whatsoever including the act through which it itself came into existence unless some permissive norms had first been issued. If a contradiction is to be avoided the sovereign authority who delegates power to a sub-authority must be exempt from the prohibition in question. Only the sub-authorities can be its subjects. Their normative acts may consistently be regarded as forming a normatively closed field of acts in the sense that these authorities are allowed to exercise only such normative power as has been delegated to them and no other. Whether this is an altogether reasonable view of the competence of subordinate norm-authorities I shall not discuss here. But it is a logically possible view.

5. Within a theory of higher-order norms we can illuminate one of the most controversial and debated notions of a theory of norms viz. the notion of validity.

What is meant by the ‘validity’ of a norm? There are at least two different relevant meanings of the words ‘valid’ and ‘validity’ in connexion with norms. Several controversies in the theory of norms will be seen to be futile when we realize that apparently opposed opinions really pertain to different notions of validity.

One sense in which a norm can be said to be valid is that it exists. A person comes across something which he interprets as a norm-formulation say on a notice-board or in a statute-book. He concludes that such and such a norm has been issued and thus also has existed at least for some time in the past. But he may be curious to know whether it still exists or whether it has been cancelled or has passed out of existence by what in jurisprudence is called desuetudo. The question ‘Does this norm still exist?’ is often couched in the words ‘Is this rule still valid?’ and the answer ‘It still exists’ in the words ‘It is still valid’. Since validity here means existence it would perhaps be better not to use the word ‘validity’ at all. For this word is also used with a quite different meaning.

Under this other meaning the validity of a norm means that the norm exists and that in addition there exists another norm which permitted the authority of the first norm to issue it. If we decide to call the act of issuing a norm legal (or lawful) when there is a norm permitting this act then we may also say that the validity’ in the sense now contemplated of a norm means the legality of the act of issuing this norm.

The words ‘valid’ and ‘validity’ when applied to a norm thus sometimes refer to the existence as such of the norm and sometimes to the legality of the act as a result of which this norm came to be. In English a norm or law which is valid in the sense that it exists is also said to be in force (cf. Ch. VII Sect. 8). To say of a norm that it ‘exists’ is not ordinary usage but philosophic jargon invented for special purposes. The question ‘Is this law valid?’ can often be rendered more unambiguously by ‘Is this law in force?’ In German however one would nearly always use the same word ‘gültig’ (i.e. ‘valid’) for the two cases. And in the Swedish language ‘gällande rätt’ which literally means ‘valid law’ is the technical term for law which is in force and thus in our philosophic terminology for existing law or law in existence. These peculiarities of various languages may offer a partial explanation of the fact that philosophers of law at least in Germany and Scandinavia have found it hard to see that there are two utterly different concepts covered by the same word ‘valid’ and have often thought that an account of one meaning of the word could cover both concepts. Some philosophers such as Hans Kelsen in his early period have tended to identify validity with the legality of norm-giving acts and to ignore or underemphasize the factual aspect of law as the efficacy of a commanding will. Others such as Axel Hägerström have put all the emphasis on efficacy on ‘law as fact’ ignoring the normative notion of validity as legality.

For the sake of avoiding ambiguities I shall here always understand ‘validity’ in the normative sense of ‘legality’ and never in the factual sense of existence or being ‘in force’.

Some authors have thought of validity as a parallel attribute to truth. Statements of fact (propositions) are true or false; norms it is said are not true or false but valid or invalid. What truth-value is in the world of propositions validity is in the world of norms.

The analogy between validity and truth is a bad one and should therefore not be used. Validity is neither a ‘substitute’ for nor a ‘parallel’ to truth in the realm of norms.

The notion of validity which we are discussing is a relative notion. A norm is valid if at all relative to another norm permitting its issuing or coming into existence.

This relativity of the notion of validity however must not be misinterpreted. It does not mean that the issued norm is valid if the norm permitting its issuing is valid. The first norm does not ‘get’ its validity from the validity of the second. The validity of a norm in the sense now under discussion is not validity relative to the validity of another norm. It is validity relative to the existence of another norm hierarchically related to the first in a certain way.

In this respect validity is unlike truth. By saying that a proposition is true ‘relative to’ another proposition one could hardly mean anything else but that if the second proposition is true then the first proposition is true also. The first proposition ‘gets’ its truth from the truth if it be true of the second proposition.

If we do not see clearly the difference between validity and truth but believe that they are analogous concepts we are easily led to the following mistaken idea: If validity of a norm is validity relative to the validity of another norm of higher order the validity of this higher-order norm will in its turn mean validity relative to a third norm of still higher order and so forth. If this chain is infinite the concept of validity would seem to lose all meaning or be hanging in the air. If again the chain is not infinite then the validity of the norm in which the chain terminates cannot mean ‘validity relative to some other valid norm’ since there are no other norms to refer to. It must mean validity ‘absolutely’ or ‘in itself’. The relative notion of validity is thus thought to require or presuppose an absolute notion in much the same sense in which a notion of relative truth can rightly be said to presuppose a notion of absolute truth.

But this argument is fallacious. The notion of relative validity which we have been explaining does not by logical argument force upon us a notion of absolute validity. The relative notion is self-sufficient so to speak. But as we shall see later the notion can be supplemented in a way which may be said to create an analogue to an absolute notion.

To the notion of validity which we have here explained corresponds a notion of invalidity. We shall say that a norm is invalid if the issuing of that norm by a certain authority is forbidden to this authority by virtue of some higher-order norm. If we decide to call the act of issuing a norm illegal when there is a norm prohibiting this act then we may also say that the invalidity (in the sense now under discussion) of a norm means the illegality of the act of issuing such a norm.

It should again be observed that the standard of invalidity of a norm is the existence and not the validity of a certain other norm hierarchically related to the first in a certain way.

It is clear that a norm need be neither valid nor invalid in the senses here defined. A sovereign norm for example cannot be valid or invalid.

It may happen that a norm which is valid relative to one norm of higher order is invalid relative to another. A norm can thus be both valid and invalid.

There is nothing illogical (contradictory) about this. One and the same norm n is both valid and invalid when there is one higher-order norm which permits and another higher-order norm which prohibits the authority of the norm n to issue the norm n. If however one and the same norm happened to be both valid and invalid then the permissive norm validating it and the prohibiting norm invalidating it must emanate from different authorities. For one and the same authority cannot both permit and prohibit the same act to the same agent on the same occasion. It is a straightforward application of this rule that one and the same superior authority cannot both permit and prohibit the same sub-authority to issue a certain norm. But one superior authority may permit and another superior authority prohibit the same inferior authority to issue a certain norm. And if then the inferior authority issues this norm the norm which thus comes into existence is both a valid norm and an invalid norm.

6. Assume that x orders or permits y to order or permit z to issue some norms. Assume further that y actually orders or permits z to issue these norms and that z does this.

On these assumptions we shall say that y in giving the norm to z acts as immediate subordinate to x and that z in giving norms to some further agent or agents acts as immediate subordinate to y and as remote subordinate to x.

Conversely we may also say that y acts as z's immediate superior and that x acts as y's immediate but as z's remote superior. And what we say of the agents in these respects we may also say of their acts and since these acts are normative acts also of the norms in which they result.

Of these three acts of x and y and z and of the norms in which they result we shall say that they form a chain of subordinated acts and norms or simply a chain of subordination. We shall do this notwithstanding the fact that the first act in the chain is not subordinate to any other act in the chain.

We shall call the acts of x and y and z and the corresponding norms links in the chain in that order. The act of x constitutes the first the act of y the second the act of z the third link. The first act and norm we shall say is linked to the third act and norm thanks to the intermediary of the second act and norm.

A chain of subordination may of course contain more than two links. Links may be omitted from either end of the chain and what remains—if the remainder is at least two links—is still a chain of subordination. But one cannot omit links from other places in the chain but the ends without ‘breaking’ the chain.

It is essential to the notion of a chain of subordination as I have explained it here that each link in the chain—with the exception of the first link—is a valid norm (and normative act) relative to the next superior link in the chain. A norm is valid when the act of issuing this norm is permitted. It is a theorem of deontic logic that if an act is commanded then it is also permitted. Therefore an order to issue norms entails that the norms issued under that order are also valid i.e. their issuing is permitted because commanded. We can also say that each inferior link in the chain is by transitivity valid relative to every superior link in the chain and that each inferior link derives its validity immediately from the next superior link remotely from those superior to this and ultimately from the first link in the chain.

By saying that one norm (and normative act) can be (‘normatively’) traced back to another norm (and normative act) we shall understand that there exists a chain of subordination of which the first norm is an inferior and the second a superior (relative to the first) member.

A norm which cannot be traced back to any other norm cannot by definition be valid relatively to any other norm. It will be either invalid relatively to some norm of next higher degree or it will be neither valid nor invalid i.e. sovereign.

If the number of individual norms which have been issued is finite the process of tracing back norms will always in a finite number of steps take us to a norm which can no longer be traced back to yet another norm. This assumption of finitude we can I think safely make. Thus we are entitled to say that any finite chain of subordination terminates in or originates from a norm which is either sovereign or invalid.

All the norms which are links in at least one chain which originates from the sovereign norms issued by one and the same authority will be said to belong to one and the same normative hierarchy or order or system. The sovereign norms themselves we shall include by definition in the system. A system of norms is thus a class of one or several sovereign norms which are issued by one and the same authority and norms which may through chains of subordination be traced back to these sovereign norms.

We can make use of the notion of a system of norms for defining a new concept of validity of a norm. This new concept will be called validity in a system or absolute validity. That a norm is absolutely valid will mean simply that it can be traced back to a sovereign norm. This as we have already seen is not trivially the case with every norm. For the process of tracing back may terminate in an invalid norm.

Since the notion of ‘tracing back’ a norm is defined by means of the notion of relative validity it follows that the notion of ‘absolute validity’ as defined by us presupposes or is secondary to the notion of relative validity. The opposite is the case with the notions of absolute and relative truth. That a proposition is true relatively to another proposition means that the first is true absolutely if the second is true absolutely. The notion of relative truth is secondary to the notion of absolute truth since it is defined in terms of absolute truth.

The question may be raised: Do the laws of the state constitute a normative system (hierarchy order) in the sense here defined? If they do who is the sovereign authority in a state? These are no doubt extremely interesting questions of political and legal philosophy. The questions can be raised empirically for the law of a given country. They can also be discussed as purely conceptual questions. The answers to the conceptual questions will depend on how we mould our concept of the state. The answers to the empirical questions again will depend on how well those empirical phenomena of an enormously complex structure which we know as so-called sovereign states conform to the concept as moulded by the political philosopher. I shall not however discuss these problems in the present work.

7. Suppose that a chain of subordination terminates in an invalid norm. This means that there exists some norm which prohibits the authority of the invalid norm to issue it. The act of the sub-authority was therefore an act of insubordination relative to this higher-order norm. In issuing the invalid norm he transgressed the limits of his normative competence as set by the superior authority. He seized or usurped a power which had not only not been delegated to him but which had been expressly denied to him. Invalid normative acts might therefore also be called acts of usurpation.

It should be noted that on the definition which we have given the invalid norm and the norm relative to which it is invalid are both in force (exist). The authorities who issue them succeed in establishing normative relationships between themselves and the subjects of their norms. The authority of the invalid inferior norm is the subject of the superior norm. That the superior norm is in force and that the authority of the invalid norm is its subject entails that the authority of the superior norm tries to make the authority of the inferior norm forbear such illegal acts. He may for example order him to be prosecuted for disobedience and punished. He will also probably take steps to dissolve the relationships under norm which the usurper had succeeded in establishing. There might be a whole chain of such ‘illegal’ relationships. If the superior authority is successful the illegal norm and its possible repercussions in the form of norms subordinate to it will vanish cease to exist.

The outcome of the struggle of authorities may however also be the reverse. The usurper of power is successful. The normative relationships which he has established remain acquire relative permanence. The authority who was superior to the usurper resigns in his efforts to make the usurper obey. This means that the superior norm relative to which the usurper's act was invalid passes out of existence—perhaps dies as a consequence of an act of cancellation. If this happens the usurper's norm ceases to be invalid. It is now neither valid nor invalid relatively to any other norm. It has become a sovereign norm. The chain or chains of subordination to which it has given origin will then together with possible further normative acts of the same ex-usurper constitute a normative system in its own right. And the norms which may be traced back to this once invalid norm will not only be valid relatively but valid absolutely in the sense that they are valid within a system.

One may distinguish two kinds of acts of usurpation.

Assume that x has prohibited to y the issuing of commands to z. nevertheless issues a command to z to do something. This act on y's part is an act of usurpation.

Assume further that in addition to the prohibition from x to y and the command from y to z there is also a prohibition from x to z to do the very same act which y has ordered z to do. A positive and a negative command with different authorities but identical content subject and occasion we have previously called conflicting commands. In the case under discussion we thus have an invalid order from y to z which conflicts with a (sovereign or) valid order from x to z. Then we shall say that y's invalid normative act was not only usurpatory but also revolutionary. y in issuing the invalid order did not only himself violate a prohibition by transgressing the limits of his normative competence as assigned to him by x. y also urged another agent z who takes orders from x to disobey orders from that quarter. This is the ‘logic of revolution’: seizing illegal normative power and urging the citizens to disobey existing regulations. ‘Revolution’ is very much the mot juste to describe the case. For if the usurper is successful in the sense that his illegal commands become effective i.e. generally obeyed by those to whom they are addressed then since his commands conflict with existing valid commands these latter will have to become ineffective cease to be generally obeyed by the citizens. Two conflicting commands as we know can coexist and ‘contend’ with each other at least for some time. But it is logically impossible that they should both become effective in the sense of being generally obeyed by their subjects. A revolutionary usurpation of norm-giving power which is successful will therefore necessarily overthrow an existing effective legal order or a part of it and institute a new effective order in its place.

8. No normative act can be both sovereign and subordinate. But one and the same authority of norms may perform both sovereign and subordinate normative acts. When Mr. X who is a judge sentences a thief in court he performs a subordinate normative act. But when he orders his children to go to bed he acts as sovereign (unless there is a norm to the effect that parents are entitled to give orders to their children).

The example also shows that the normative acts of one and the same authority of norms may belong to different systems of norms. This is a fairly trivial observation. Of more interest is it to observe that one and the same normative act may belong to two or more different systems of norms. It is conceivable that two agents x and y whose normative acts cannot be traced back to the same sovereign act-e.g. because they both act as sovereigns themselves—authorize (permit) a third agent z to issue norms to w. If z makes use of the power delegated to him i.e. if he actually issues a norm to w then the normative act of z can be traced back both to the normative act of x whereby x gave him this power and to the normative act of y whereby y gave him this same competence. Since the two normative acts of x and y respectively belong to different systems and therefore also the norms in which they result the act of z will be a common member of two systems of normative acts. And the norm which z issued to w will belong to at least two systems of norms.

If two systems of norms and normative acts have common members the two systems will be said to intersects. If they do not intersect they are independent.

9. A command which belongs to one system S may conflict with a command which belongs to another system S¹. That the commands conflict means that they demand incompatible modes of conduct of the same subject on some occasion. A special case of conflict is when the one command requires the subject to do and the other requires him to forbear the same thing on the same occasion. In this case the two conflicting prescriptions are related to one another as command and prohibition with identical content.

When two systems contain conflicting commands we shall say that there is a conflict between the systems. For example: x and y are two sovereign commanders. x orders z to do a certain thing. y prohibits z to do this same thing. Then there is conflict between the system emanating from x and the system emanating from y.

Conflict between systems of norms is a special case of that which we have previously (Ch. VIII Sect. 7) called conflict of wills. We have just studied another case of conflict of wills in the realm of norm viz. the case when a revolutionary usurpation of power takes place. The normative concept of a revolution necessarily entails a conflict of commanding wills.

Can conflicting commands coexist within one system of norms? Revolution it should be observed is not an example of conflict between norms belonging to the same system. Revolution entails conflict between norms but it also presupposes the occurrence of an invalid act of usurpation of power. And the norm which is the result of the invalid normative act by definition does not belong to the system but marks a recession from the system.

In order to find out whether conflicting norms can coexist within a system we must first make it clear what such a conflict would mean.

10. That a conflict of commanding wills occurs within a system of norms means the following: Some agent w receives from an authority y an order to do something on a certain occasion and from another authority z an order to do something else on that same occasion. Both orders can be traced back to sovereign norms issued by an authority x. But the contents of the two orders are incompatible modes of conduct.

Let us assume for the sake of argument that x has permitted y to command w to do a certain thing and that x has also permitted z to prohibit w to do this same thing. The question may now be raised whether there is anything ‘illogical’ about the case which makes its factual occurrence impossible. Can such cases happen?

That w received the conflicting orders from y and z is certainly possible. This is just as possible as any conflict of will is possible. If there is anything ‘illogical’ about the case it can only be because the conflicting orders were both valid relatively to norms of the same supreme authority. By permitting y to order w to do a certain thing and z to order w to forbear this same thing x as it were ‘endorses’ a possible conflict of commanding wills within the system. The conflict need not arise. The competent sub-authorities need not make use of their power. But the conflict may validly arise. This must be the ‘illogicality’ if there is one.

Thus the problem before us is this: Can x issue the two permissions to y and z respectively without somehow ‘contradicting himself’? I find this question very puzzling. One cannot settle the difficulty by saying that since x on our assumption has permitted those acts to y and z this shows that this can happen and therefore is logically possible. This is no answer since the question is whether x can do anything which can be truly described as giving two permissions of the kind now under discussion. He can of course say to y ‘I permit you to command w to produce the state p’ and to z ‘I permit you to command w to forbear the production of p’. But this is not to say that x can permit y to command w to do a certain thing and z to command w to forbear this same thing.

In order to answer our question we must therefore first become clear about what x is supposed to do when he gives the two permissions. It is the problem of the nature of permissions recurring.

If we take the view that a permission is a ‘toleration’ then x's two permissions are two declarations of intention or two promises to the effect roughly speaking that x will leave y in peace should he choose to give a certain command to w and that x will leave z in peace should he choose to give a certain other command to w (cf. Ch. V Sect. 16). And since these two acts by y and z are simultaneously possible albeit conflicting it is difficult to see that there could be any logical inconsistency concealed in x's two permissive normative acts.

If on the other hand the two permissions (or at least one of the two) amounted to rights the situation would be different. A right we have suggested (Ch. V Sect. 15) entails a prohibition to others to do that which one has resolved or promised not to do oneself viz. hinder the holder of the right from availing himself of his permission. Now assume that x grants a right to y to command w to do a certain thing. This entails that x prohibits z to interfere with y's action should y command w to do this thing i.e. should he take certain steps to make w do it. If z has this prohibition from x he cannot at the same time hold a permission from x to prohibit w to do the thing in question i.e. a permission to make or try to make w forbear it. For an attempt on the part of z to make w forbear this thing falls under the prohibition issued to him by x to interfere with y's attempts should he make such attempts to make w do it. Hence a permission to z to prohibit w to do a certain thing would conflict with the prohibition to z to interfere with y's attempts to make w do this thing and consequently with y's right to command w to do it. It follows finally that x cannot consistently (without inconsistency) permit y to command w to do a certain thing and permit z to prohibit w to do this same thing if one of the permissions (or both) are rights.

The upshot of the discussion is thus as follows:

It is logically possible for a sovereign agent to endorse a conflict of will within a system of norms if endorsing the conflict means to permit in the weak sense of tolerate the issuing of conflicting commands by two sub-authorities. But it is not logically possible for a sovereign to endorse a conflict of commanding wills within a system of norms if endorsing the conflict means to permit in the stronger sense of granting a right the issuing of conflicting commands. By granting such rights the authority is contradicting his own will.

The answer to the question whether a conflict of commanding wills is logically possible within a system of norms thus depends upon how we understand the permission whereby superior authorities in the system delegate power to inferior authorities. If these permissions amount merely to declarations or promises that the superior authority is going to tolerate certain normative actions on the part of the inferior authorities then there may occur a conflict of commanding wills in the system. But if the power-delegating permissions amount to rights to issue certain norms i.e. if the superior authority undertakes to protect the normative actions of the subordinate authorities by prohibiting other agents to interfere with such actions then conflicts of will are logically impossible within the system.

A system of norms which is in the sense explained logically immune to conflict possesses the same coherence and unity which is characteristic of that which we have called a corpus of norms. A corpus is a class of norms which have the same authority (see Ch. VIII Sect. 7). Within a corpus a conflict between prescriptions is excluded as being contrary to the nature of a rational will. In a system of norms there are (normally) several authorities. But in a system which is logically immune to conflict and thus has the coherence of a corpus the sub-authorities cannot contradict the will of the sovereign but only ‘transmit’ it. In a sense therefore there is only one commanding and permitting will within such a system viz. the sovereign will.

We could sharpen our definition of validity in such a way that to say that a norm is valid shall mean that the authority who issues it has a permission amounting to a right to issue the norm. Normative competence or power would then mean permissions in the stronger sense of rights to perform certain normative acts. Such redefinitions of the notions of competence and validity would give to the notion of a normative system the coherence of a corpus. I think that this reshaping of our definitions should take place. The higher-order permissions of which we have here been talking should be regarded as rights.

I would not however myself say that these findings support the view that (all) permissions ‘essentially’ are rights. Permissions as ‘mere’ tolerations have a normative status of their own. But it is most illuminating I think in regard to the logical nature of this much-debated and controversial idea of permissive norms to see clearly that only permissions which are rights may serve the purpose of giving to a normative system the concord of commanding wills which is characteristic of rational willing and which it is at least highly reasonable to think that a class of norms such as say the laws of a state should possess. This also makes the idea—entertained by so many philosophers—that legal permissions are rights more understandable.