Facts, values and norms
During the nineteen fifties and the sixties it was broadly accepted that there is a clear distinction between facts, on the one hand, and norms and values, on the other. The work of Brecht illustrates this for the field of political theory; the early work of Hare does the same for ethical theory, while Taylor’s book Normative Discourse makes the point in a more general fashion. That does not mean that the distinctions were not known earlier. Hume famously suggested that it is not possible to derive ought-sentences solely from is-sentences. Kant’s distinction between our knowledge of the external world which is structured by causality and therefore determinist and the freedom of the will also points in the direction of the distinction. Kelsen made the distinction between facts and norms one of the starting points of his pure theory of law.
Although the distinctions, or their extreme variants, the gaps, between Is and Ought and between fact and value have always been contested, arguments that they cannot be maintained have become increasingly popular since the second half of the 20th century. To mention a few examples: consider the contributions on moral realism, theological voluntarism and ethical naturalism in The Oxford Handbook of Ethical Theory, Searle’s work on the derivation of ‘Ought’ from ‘Is’ and Putnam’s work on the collapse of the fact/value dichotomy. 
The discussion about the distinctions, if not gaps, is important from a methodological perspective, because different methods seem to be in place depending on whether a question is one about facts, or about norms or values. For instance, it is often assumed that factual issues can be dealt with by empirical sciences, while normative and ev aluative issues require a non-empirical science, or cannot be dealt with in a scientific manner at all. A related line of thinking can be recognised in the positivist lawyers’ view that lawyers should deal with questions about the positive law and that evaluative matters belong to the realm of politics.
If the realms of Is and Ought are taken to be separated, it would be impossible for a phenomenon to belong to both realms. That would also hold for the law, which should therefore either belong to the realm of ought, or to the realm of is. The former view was famously defended by Kelsen, according to whom the law consists of norms where norms belong to the realm of ought. The latter view was defended by the Scandinavian legal realists, according to whom the law exists as a matter of fact.
Despite the importance of the distinctions and the methodological consequences attached to them, there seems to be insufficient clarity about the nature of facts, norms and values. For instance, are norms always normative in the sense that they prescribe behaviour? Can there be power conferring norms? Is it a fact that somebody is courageous, or that in the Netherlands the proper side of the road to drive on is the right-hand side? And can it be a fact that in the UK one ought to drive on the left? Is it, at least in theory, possible to establish scientifically (e.g. with the use of economics) what is the best way to deal with the Euro crisis? To answer these and similar questions, we need at least a good conceptual framework concerning facts, values and norms. The purpose of this chapter is to provide the outlines of such a framework. In the following sections, facts, rules, norms and values and their relations and differences will be discussed. At first sight these distinctions may seem confusing and needlessly complicated, but the reader who takes the trouble – and it will take some trouble – to master them will find that these distinctions, rather than creating confusion, create clarity in many a normative or evaluative discussion.
Perhaps it is useful to say a little about the method used to identify the distinctions and analyses provided in this contribution. The starting point for the analysis of concepts must be the use of these concepts in their natural habitat. For concepts used in daily life this will be common usage; for technical concepts, also when they are a refinement of ‘folk concepts,’ it will be the use in the discipline at issue. However, since it is the purpose of analyses like the present one to improve on the actual use of the concepts if they are obscure or ambiguous, their actual use cannot be the last word. Therefore, the following text sometimes offers analyses of concepts that deviate from common usage, or – more often – that try to impose clear and unambiguous meanings on a linguistic practice that is ambiguous and lacks this clarity. Any attempt to do this must be controversial. An analysis of concepts cannot be proven to be true, and its value can only be tested by means of more substantial theories in which the analysis is used. At a number of places the following text contains suggestions on how the suggested analyses prove their value.
Understanding the differences between facts, values and norms requires that it is at least clear what facts are.
As Strawson has pointed out, facts depend on language. A fact is always the fact that …, where the dots stand for a phrase expressed in some language. It is for instance a fact that it is not raining here and now. Facts also depend on the world, because it is the world, not language, that determines which facts obtain, for example whether it rains here and now. A language determines which facts can be expressed, the world determines which of the expressible facts actually obtain.
It is useful to distinguish between ‘facts’ that are merely expressible and facts that obtain in the real world. An expressible fact will be called a state of affairs. States of affairs are what is expressed by sentences with truth values. For instance, the sentence ‘It’s raining’ expresses the state of affairs that it is raining. States of affairs that obtain in the actual world will be called (actual) facts. A sentence that expresses a fact is true. False sentences express non-facts, states of affairs that do not obtain.
In most declarative sentences it is possible to distinguish one or more terms which denote entities in the world. Next to these terms there will be predicate expressions by means of which something is said about the denoted entities. For instance, in the sentence ‘John walks’ the word ‘John’ denotes John, while the predicate term ‘walks’ is used to say something about John. Logicians call the entities in the world individuals, and the expressions used to denote them terms.
Terms should be distinguished from full sentences. Declarative sentences are true or false; terms are not. So there is, from a logical point of view, a fundamental difference between sentences and terms. Sentences which are used to make statements have truth values; they do not denote. Terms, on the contrary, have no truth values, but denote. This distinction is important with respect to rules, because, arguably, rule formulations look like sentences but are terms. (See also Section 4.3). This implies, among other things, that there can be no logic of rules that is similar to traditional logic. (See Section 4.4).
In discussions about the difference between Is and Ought it is sometimes implicitly assumed that the ‘Is’ has to do with facts that obtain independently from the human mind and in particular from what we believe about them, while the ‘ought’ is not a matter of fact, because it depends on the human mind, in the form of rules and standards that humans have adopted or accepted. In that connection we often take some form of ontological realism for granted when it comes to facts. Ontological realism is the view that facts and things exist independent of the mind (including language), and more in particular independent of our beliefs about them. For many facts this is not adequate. Think for instance of the fact that the United Nations has its headquarters in New York City, the fact that tort-feasors are liable for damages, and – possibly more controversial – the fact that it is wrong to commit murder. In this section, the realist assumption with regard to facts will be relativized, thereby paving the path for the recognition of ‘ought-facts’ (deontic or normative facts; see Section 3) and ‘value-facts’ (Sections 3.1 and 5.4).
Taking Brian Leiter’s distinction between different kinds of objectivity as a point of departure, it is possible to distinguish theoretically between at least three kinds of facts: objective facts, intersubjective facts, and rule-based facts. It should be noted beforehand, however, that making this distinction does not commit you to the view that facts of all three kinds actually obtain.
The first category consists of facts of which it is assumed that they are completely mind-independent. We will call them ‘objective facts’. Objective facts would in the eyes of many include that the highest mountain on Earth is Mount Everest, that America was discovered (long) before 1500 AD, that there are Higgs particles and that the amount of solar systems in the universe equals some as yet unknown number.
Facts in social reality are those facts which exist because they are recognized or accepted as existing by sufficiently many and sufficiently relevant members of some social group. The precise conditions of existence of these facts are still object of discussion, but typical examples from the Netherlands are that Amsterdam is the capital of the Netherlands, and that legislation is a source of law. These facts will be called ‘intersubjective facts’.
Many facts are facts because they are the result of the application of some rule, principle, or evaluative standard, or the outcome of a real or best possible argument. These seemingly different categories of facts are taken together because arguments are based on inference rules, while standards, rules and principles need to be applied in arguments. In all three cases, the facts depend in some way on rule or rule-like entities. Examples of these ‘rule-based facts’ would be that in chess the person who has check-mated his opponent’s king has won the game, that nobody can chair the hockey club for more than two subsequent periods, that 3+5 equals 8, and that in the European Union, the Member States generally are not allowed to subsidize national industries.
Rule-based facts always presuppose some rule. But this does not mean or imply that they are merely relative facts in the sense that these facts themselves include reference to some standard. For example, the person who checkmated his opponent has won the chess game; he has not merely ‘won the game if and to the extent that the rules the rules of chess are such and so’. If this were different, the Earth would only be a planet if, and to the extent that, things like Earth are called ‘planets’.
The insight that rule-based facts are not for that reason relative facts has implications for ‘detached’ or ‘externalist’ views about morality and law. Such views hold that subscription to the moral or legal judgement that something ought to be done can very well go together with a complete lack of motivation to act accordingly. Moral and legal duties may presuppose moral principles and legal rules, but they are not facts relative to these principles and rules. If a particular murder was morally wrong, it was not merely ‘wrong according to the standard that committing a murder is morally wrong’. Non-relativized moral requirements are not, as Foot would have it, merely requirements in case one wants to act morally. They are requirements that follow from the principles or values of morality and presuppose commitment to, an internal perspective on, those principles and values. This is different for relativized moral judgments, such as the judgment that according to some Pre-Columbian cultures, child sacrifice is morally allowed. Then the relativity is built into the judgement, and such relativized moral judgments can be made without being committed. The fact that moral judgments are not relativized judgments implies that an externalist view of morality cannot be based on the alleged relative nature of these judgments.
The opposite holds for ‘detached legal judgments’ as mentioned by Raz. They do not commit the person giving these judgements to act accordingly, precisely because they are relativized judgments which inform us what would be required by, for instance, Islamic law. Alexy discusses these detached legal judgments under the heading of judgments from the observer’s perspective on law, which he opposes to the participant’s perspective. In his view, a participant is committed to the correctness or justice both of legal systems as a whole and of concrete legal judgments.
In connection with rule-based facts we can make a subdivision based on the issue whether the rules determine the facts or not. Mathematics would provide an example of rule-based facts in which the rules of for instance number theory determine the outcome of calculations. Such facts are determinate rule-based facts. Arguably, the law provides an example where facts – the legal outcomes of cases – depend on arguments, and therefore on standards that determine what are good arguments, but where in hard cases the rules of the argumentation game do not fully determine the right answer (pace Dworkin in his early days). Such facts are indeterminate rule-based facts.
Another subdivision of the category of rule-based facts is between facts based on rules that exist as a matter of intersubjective fact and facts based on rules that exist as a matter of rule-based fact (rule-based rules). The latter might be called ‘multiply rule-based facts’. The first category still has a ring of objectivity – they are at least independent of individual minds – but the multiply rule-based facts may seem to be completely subjective, in particular if the rules on which the rule-based rules depend are themselves based on rules which are based on rules, and so on …
It might seem that there exists even a fourth kind of facts, consisting of those ‘facts’ which are purely subjective, such as the fact that spinach tastes better than Brussels sprouts, and the fact that Deep Purple is a better rock band than Grand Funk Railroad was. Arguably, these ‘facts’ are not facts at all, but merely personal preferences, although it may be a fact that somebody has such a preference.
A common mistake in normative and evaluative theorising is to identify indeterminate and/or multiply rule-based facts and what is merely subjective. This mistake underlies projectivism in ethical theory, the view that value judgments are the result of projecting one’s subjective emotional response to a given set of facts onto these facts. That this is a mistake transpires from the phenomenon that we consider it as rational to argue about indeterminate rule-based facts such as the law in hard cases, and not to argue about purely subjective ‘facts’, such as the taste of Brussels sprouts. Apparently we make a difference between judgments of these two categories. The identification of the two categories only makes sense if we consider disputes in both cases as equally (un)reasonable.
It has often been noted that sentences that tell one what to do, can take the form of declarations. For example, the sentence ‘You ought to go to the supermarket’ looks very much like the sentence ‘You will go to the supermarket’. It has also been argued that this similarity is deceptive and that the former sentence is a kind of prescription and that it is not a fact if somebody ought to do something.
Several misunderstandings may underlie the view that there cannot be such a thing as ‘deontic’ or normative facts, facts of the type that somebody ought to do, or refrain from doing, some kind of action. One misunderstanding is that a sentence such as ‘You ought to go to the supermarket’ is ‘really’ a kind of order or exhortation, rather than the description of a fact, because it is, or can be, used to make somebody do something. This line of argument has been attacked by Geach and Searle, because, basically, the speech act which can be performed with a sentence does not determine the meaning of the sentence. The sentence ‘That is what I ought to do’ means the same if it stands on its own as when it is used in the conditional sentence ‘If that is what I ought to do, I will eat my hat’.
A second misunderstanding, which may have inspired the first one, is that a ‘real’ fact cannot depend on what we humans think, believe, project, accept or recognize. On the assumption that standards for goodness and for what should or ought to be done are mind dependent, this misunderstanding becomes that ‘real’ facts cannot depend on standards. Perhaps the clearest expression of this idea can be found in the work of Mackie, who claimed that facts involving an ‘Ought’ (and other normative ‘facts’) are ontologically ‘queer’. This misunderstanding is essentially that of applying an objectivist stance to domains in which facts are rule-based and therefore mind-dependent. If facts can depend on rules or standards, there is no good reason why there cannot be deontic facts. Moreover, for the same reason there is no problem with ‘evaluative facts’ or ‘value-facts’ either. It may be a fact that the decoration of this wall is terrible, just as it may be a fact that one should visit the dentist in case of a toothache. Both facts presuppose standards such as a standard for decent wall-decoration or a standard for prudence, but there is no good reason why the presupposition of a standard makes the existence of facts impossible.
In normative discussions, words like ‘ought’, ‘should’, ‘must’, ‘duty’, and ‘obligation’ are often used interchangeably. Although the meanings of these words in natural language are not fixed and also overlapping, there exist different categories of facts in the normative sphere. The differences between these categories are important, although the words used to denote them are not. In the following, I will point out some distinctions and propose to use particular words to denote the newly delineated categories. Let me start with some examples:
- Everybody has the duty not to steal, but normally there is no obligation to that effect.
- A and B are under obligations towards each other, because they entered into a sales contract, but these obligations are not duties.
- From the fact that P is under an obligation or a duty to do something, the consequence follows, pro tanto, (if only this reason is taken into account) that P ought to do it, but not the other way round.
The first distinction to be made is between what will be called duties and obligations. Both duties and obligations are reasons why somebody ought to do something, but neither the duty, nor the obligation coincides with the fact that this person ought to do it. A duty is connected to a role or status. It is for instance the duty of house owners to pay real estate tax, and the duty of a Mayor to maintain the public order in a municipality. All human beings are under a duty not to kill other human beings.
Whereas duties are connected to a particular status or role, an obligation is the outcome of an event and depends on that event having occurred. Typical examples of such obligation generating events are causing damage, making a promise, or contracting. Moreover, whereas a duty does not have to be a duty with regard to somebody in particular (e.g. the duty to stop for a traffic light, even if nobody is approaching), obligations are always ‘directed’, obligations towards somebody else. This directedness of obligation still holds if this ‘somebody else’ is (as yet) unknown, as when for instance a car was unlawfully damaged but the owner of the car is still unknown.
The term ‘obligated’ has no precise meaning, but it will be used here to denote the common denominator of duties and obligations: a person who is under a duty to do B is obligated to do B; a person who is under an obligation to do B is also obligated to do B.
By now we have encountered three normative concepts, ‘duty’, ‘obligation’, and ‘being obligated’. They all differ from the normative concept that is often used as a catch-all for all kinds of normativity, the concept of ‘Ought’. In connection with duties, obligations and being obligated, the more relevant notion is ought-to-do. The word ‘ought’ as I define it here stands for the outcome of the interplay of one or more reasons for acting, a kind of aggregate of these reasons. Examples are the legal ought, as the aggregate of legal reasons, and the moral ought as the aggregate of moral reasons.
An Ought itself is not a reason for acting, but merely the ‘aggregate’ of, or conclusion from, one or more reasons. So, where the facts that make that X is under a duty to pay real-estate tax are reasons why X ought to pay real-estate tax, the fact that X ought to pay the tax is not a reason for paying it, although it presupposes the existence of such a reason (the reasons underlying the duty, for example).
It is sometimes assumed that, as a matter of logic, a person only ought to do something if this person is capable of doing it: ‘ought implies can’. However, this slogan expresses a normative principle, rather than a logical truth. It is very well possible to be under an obligation to do what one cannot do. For instance if A contracts to pay B 100.000 euro and if A becomes insolvent before having paid this money, A is still under an obligation to pay B 100.000 euro.
It is also possible to have a duty to do something that is impossible under the actual circumstances. For instance, every house owner may be under a duty to keep the pavement before his house free from snow. This duty does not stop if one breaks ones leg, even though, let us assume, having a broken leg makes it impossible to keep the pavement free from snow. It may be argued, though, that the person with the broken leg may have the prima facie duty to clear away the snow, but that all things considered he is permitted no to do it. This argument would most likely be based on disregarding the duty because of the impossibility, but such disregard is only required on the substantive moral principle that having a broken leg exempts one from particular duties. It is not a matter of logic alone.
The step from having an obligation, or being under a duty, to owing to do something is, as became clear from the above, not an automatic one. There can be reasons for not taking it, and in the law this phenomenon is known as a justification of an exception. Such justifications come in two types. Sometimes they are reasons why an act ought not to be performed, even if there are colliding reasons why it ought to be performed. For instance, one ought to break the windows of the neighbouring house if this is necessary to save the children from the burning house. This is called ‘necessity’ or ‘force majeure’.
Sometimes justifications are merely reasons that make that other reasons do not count. Such reasons are called ‘exclusionary reasons’. An exclusionary reason is a reason, which makes that a fact that normally would count as a reason, does not count so anymore. An example of this is the permission of a patient that allows the dentist to hurt him in the course of a treatment. The earlier example of the person with a broken leg might be another.
The difference between colliding reasons and exclusionary reasons is important, because in case of colliding reasons it is necessary to balance the reasons. In case of exclusionary reasons the excluded reasons do not count anymore, and therefore there is no need for balancing. This distinction between colliding reasons and exclusionary reasons may be useful in discussions where human rights conflict with other human rights, or with mere interests, because arguably human rights may trump (exclude) mere interests, while they need to be balanced against other rights.
To understand the nature of deontic facts and rules, it is useful to distinguish the elements of deontic states of affairs. We will discuss these elements for states of affairs of the ought-type, but most of the discussion can mutatis mutandis be applied to duties and obligations as well.
Ought-states of affairs can be subdivided in states of affairs of the ought-to-be type and states of affairs of the ought-to-do type. An ought-to-be state of affairs involves some state of affairs that ought to be the case, without specifying anybody who ought-to-do something. An example is the state of affairs that every letter ought to be stamped. Ought-to-be states of affairs are useful for evaluating other states of affairs (they are not as they ought to be), but useless for guiding behaviour because they leave unspecified who should do what.
An ought-to-do state of affairs involves that somebody is either required or prohibited to do something, or to do something in a particular way, or at a particular time or place. These states of affairs can both be used to guide and to evaluate behaviour. An ought-to-do state of affairs consists of three or four elements, the deontic modality, the addressee, the act specification, and – occasionally – the specification of the act modality. Note, however, that the description of ought-to-do facts may not mention the actor of an ought explicitly, in particular if this actor can be identified by means of the context (e.g. ‘all houses ought to be well-painted’). Since it is an important function of law to guide behaviour, there is a strong presumption that sentences that tell us what legally ought to be the case belong to this category of implicit ought-to-do facts.
There is one basic deontic modality and two derived ones. The basic modality is that somebody ought to do something. One derived deontic modality is that somebody is forbidden to do something. This means that this person ought to refrain from doing something. The other derived deontic modality is that somebody is permitted to do something. This means that this person is not forbidden to do it.
A state of affairs of the ought-to-do type always specifies who is to do something, the addressee of the state of affairs. This addressee can be specific, particular or universal. If the addressee is specific, one or more specific persons ought to do something. For instance, John ought to repay his debt to Mary, or Derek and Jane ought to wash the dishes.
If the addressee is particular, all members of a set defined by some characteristic ought to do something. For instance, car drivers ought to carry a driver’s license. Notice that these particular ought-to-do states of affairs are not conditional ones. Conditions should be distinguished from specifications of the addressees. An example of a conditional state of affairs of the ought-to-do type is that car drivers ought to turn on the car lights if it is misty.
Many deontic states of affairs are universal: they specify what everybody ought to do or to refrain from doing. An example is that one should not kill human beings. This prohibition holds for everybody.
Deontic states of affairs of the ought-to-do type specify what ought to be done. That is, they mention the action type that ought to be done, or to be refrained from. Since ought-to-do facts are action guiding, they must refer to action types, not to individual concrete acts (act tokens). An individual act only exists if the act is in the process of being, or has already been, performed, and then an ought-to-do cannot guide the act anymore. However, an ought-to-do can be used to evaluate act tokens. (‘Well done. This was exactly what you ought to do!’) Action types can be abstract (refrain from killing) but also very concrete (post a hand-written letter before eight on a Sunday morning).
For some purposes (e.g. in connection with issues of causality) it is useful to distinguish between material and formal actions. A material action consists in bringing about a particular effect. Examples are committing a murder, harming the environment, and regulating a particular subject. A formal action is an action that is not material. Examples are walking and using a false name. Such acts can bring about a particular effect, but they are not defined in terms of bringing about this effect.
Duties and obligations can relate to the way in which things are to be done, including the time at, and place in, which they should be done, without necessarily stipulating that these things ought to be done themselves. For instance, Jane is permitted not to drive, but if she drives she ought to do so on the right hand side of the road. John ought to repay his debt to Mary; moreover he ought to do so before next year. If an ought relates to the modality of an act, rather than to the act itself, the deontic state of affairs must specify the act’s modality.
Notice that deontic states of affairs that deal with act modalities are not conditional deontic states of affairs.
An ought that concerns the modality of an action might be construed as a conditional ought: owing to drive on the right hand side of the road then becomes owing to drive on the right on the condition that one happens to drive. Such a construction confuses the ought of an act modality with a conditional ought. The failure to recognize that an ought may refer to an act modality rather than to an action type has caused quite some confusion within deontic logic research. Famous in this connection is the ‘paradox of the gentle murder’. A plausible assumption is that if one ought to do something then one also ought to do what is implied by this something. For instance, if A ought to post a letter, and this can only be done by walking to the letter box, A also ought to walk to the letter box. Suppose now that it is forbidden to commit murders, but that if one commits a murder one should do so gently. It has been argued that committing a gentle murder involves committing a murder and that if one ought to murder gently, it logically but paradoxically follows that one ought to murder. This inference characteristically makes the mistake of assuming that ‘P ought to do A in a particular way’ implies ‘P ought to do A’.
A common misunderstanding in dealing with rules is to see rules as a kind of sentences, be it not sentences that describe what is the case but as sentences that prescribe that something ought to be the case. However, if we recall the distinction (made in section 2.1) between descriptive sentences and terms, rules have more in common with ordinary objects or individuals (in the logical sense) that are denoted by terms than with the states of affairs that are expressed by descriptive sentences. Rules exist in time, they can be created and derogated, they can have characteristics such as effectiveness, etc, and in all these respects they resemble individuals.
Rules also seem to have a lot in common with descriptive sentences: they have a content and they can in some sense ‘correspond’ to facts. However, in this correspondence lies also a major difference with descriptive sentences. Descriptive sentences are ‘successful’ in the sense of ‘true’ if they match the facts. This is what Searle called the ‘word to world direction of fit’; the words aim to fit the facts in the world. Rules are successful in the sense of ‘valid’ if the facts match the rule, the ‘world-to-word direction of fit’. This match is not that the rule is obeyed, but that the content of the rule is imposed upon the world. So, the rule that theft is forbidden is imposed upon the world if (because of this rule) individual instances of theft count as forbidden acts. To that purpose it is not necessary that people refrain from committing theft. Existing rules have the world to word direction of fit because they constrain the world in the sense that not all combinations of facts are possible. As a consequence, rules bring about facts. Because rules impose themselves on facts rather than reflect them, they are not true or false.
Rules exist like other ‘things’. The precise conditions of their existence vary. The existence of a rule of positive law, for instance, is either an intersubjective fact (customary law) or a rule-based fact. The latter is the case if the rule has been created by legislation in accordance with valid rules for law making, or by a judicial verdict, or – less obviously – if it fits in the ‘system’ of the law as constituted by legislation, principles and case law.
Because rules are more like ‘things’ such as tables than like full sentences such as the sentence ‘The table is green’, they can, just like tables, not be derived from anything. Notice that this has nothing to do with the distinction between Is and Ought, and everything with the distinction between sentences and terms or – which boils down to the same thing - between states of affairs and individuals.
However, the existence of a rule, which is expressed by a full sentence, is a fact and is derivable from other facts. Often such a derivation takes the form of pointing out that a rule is to be found in a ‘source of law’ such as a statute, and by adducing the ‘rule of recognition’ that rules that stem from this source are valid legal rules. However, the existence of rules of customary law, which is a matter of intersubjective fact, needs to be derived from the existence of certain social practices.
Constitutive rules are usually contrasted with regulative rules. Arguably, however, all rules are constitutive, and regulative rules only regulate by constituting deontic facts such as duties and obligations. To see why this is the case, it will be necessary to distinguish between two main kinds of constitutive rules, namely between dynamic and static rules.
Dynamic rules attach new facts, or modify or take away existing facts as the consequence of an event. An event is a change in the set of all facts, any change in the world. Examples are that it starts to rain, that John promised Richard to give him 100 euro, that Eloise was appointed as chair of the French Parliament, and that Parliament passed a Bill. At least the last three of these events have consequences attached to them by dynamic rules. John’s promise has the consequence that from the moment of the promise on John has the obligation to pay Richard 100 euro. The appointment has the consequence that Eloise will be the chair of the French Parliament when its new term starts. Passing the Bill, finally has as a consequence that the rules contained in the Bill are from that moment on (or a little later) valid legal rules. As this last example illustrates, Hartian rules of change are dynamic rules.
Static rules come in at least two flavours. One kind of static rule attaches a fact to the presence of some other fact. We may call them fact-to-fact rules. An example is the rule which attaches to the fact that P owns O the fact that P is competent to alienate O. For example, if Smith owns Blackacre, she is competent to transfer her property right in this real estate to Jones. Another example is the conceptual rule which determines that if P is under an obligation towards Q to do A, then P is obligated to do A.
The second kind of static rules consists of the so-called counts-as rules. They have the following structure: Individuals of type 1 count as individuals of type 2. These ‘individuals’ may be human beings, such as in the rule that the parents of a minor count as the minor’s legal representatives. Often, however, the ‘individuals’ that count as another kind of individual are events. For instance, under particular circumstances, causing a car accident counts as committing a tort, or making a promise counts as undertaking an obligation.
Traditionally, constitutive rules are contrasted with regulative rules. It is not at all clear that this opposition makes sense. Take for instance the rule ‘If somebody unlawfully causes damage to somebody else, the former person has the obligation to compensate the latter person’s damage’. This rule may very well be interpreted as a dynamic rule which attaches to the occurrence of a tort the deontic fact that one person has an obligation towards some other person. All rules which attach a consequence to the occurrence of an event can be classified as constitutive rules, and for this purpose it does not matter whether the consequence is deontic or not.
The rule ‘Every citizen must declare his taxes annually’ might very well be interpreted as a fact-to-fact rule, which attaches the deontic fact (a duty) that a person must declare his taxes annually to the fact that he is a citizen. All rules that obligate persons who belong to a particular class can be analysed in this way and all these rules are constitutive because they create duties.
In general, all rules are constitutive, and some of them constitute deontic facts. Those are also called ‘regulative rules’. The only problematic regulative rules are those which unconditionally obligate everybody. An example would be the rule that it is forbidden to commit a murder. This rule does not seem to attach the prohibition to murder to anything. However, it is very well possible to treat such rules as constitutive rules which create deontic states of affairs ‘out of nothing’. For instance the rule ‘It is forbidden to commit a murder’ unconditionally creates the fact that it is forbidden to commit a murder. If one is prepared to take this step, one can abandon the distinction between constitutive and regulative rules because regulative rules would just be one kind of constitutive rule.
If all rules, including the regulative ones, are constitutive, the major difference between rules and declarative sentences is, as already discussed, that rules impose themselves upon the world, while descriptive sentences aim to reflect the word. The difference has nothing to do with the alleged normative nature of rules.
Suppose that there exists a rule to the effect that skateboards count as vehicles in the sense of some traffic regulation. Because this rule exists, skateboards count as vehicles in the sense of this traffic regulation. To state it differently: because the rule exists, it is a fact that skate boards count as vehicles. The fact that skate boards count as vehicles is not the same fact as the fact that the rule ‘Skate boards count as vehicles’ exists. The former fact is about skateboards; the latter is about a rule.
Apparently, a rule can constitute facts, which can be described by re-using the rule formulation. ‘Skateboards count as vehicles’ is both a descriptive sentence which expresses a state of affairs, and the formulation of a rule. Moreover, the existence of the rule tends to go hand in hand with the state of affairs being a fact. This phenomenon may be denoted by the expression ‘factual counterpart of a rule’. The factual counterpart of a rule is a fact which corresponds to the formulation of a valid rule and this fact exists because of the validity of this rule. For example, car drivers are obligated to turn on their car lights when it gets dark (a fact) because the rule exists that car drivers are obligated to turn on their car lights when it gets dark.
One of the most ambiguous terms in legal theory is ‘norm’. It may stand for rules in general, for deontic rules, for factual counterparts of deontic rules, for power-conferring rules, and for deontic facts in general. This is not the place to discuss all possible uses of ‘norm’, but it may be useful to point out that at least six categories should be distinguished:
- deontic and non-deontic facts;
- deontic and non-deontic sentences; and
- deontic and non-deontic rules.
It seems advisable to use the term ‘norm’, if at all, for only one of these categories, and more in particular for the category of deontic rules. That convention captures both the deontic nature and the rule-nature of norms. Then norms are not facts, not because they are deontic (facts can be deontic too), but because rules are not facts but individuals in the logical sense of the word. That would explain:
- why norms can exist (just like kings exist);
- that they can be created (just like organisations can be created);
- that they can be interpreted (just like texts); and
- that they can be evaluated (just like judicial decisions).
Norms themselves cannot be derived from facts or descriptions of facts, because derivation only applies to full sentences, not to logical individuals. Notice that this also applies to tables, kings and mountains; they cannot be derived either. The non-derivability of norms is the non-derivability of logical individuals and has nothing whatsoever to do with the fact that norms are deontic (see Section 2.1). Discussions of the logic of norms in terms of a logic of imperatives are therefore based on the wrong assumption that the behaviour guiding nature of (some!) rules is relevant for the possibility of a logic of norms.
Notice, however, that the sentence ‘Norm X exists’ expresses a state of affairs and that this sentence is amenable to derivation from other descriptive sentences. From which sentences it can be derived depends on whether the rule exists as an intersubjective fact or as a rule-based fact. If a norm exists as an intersubjective fact, its existence can be derived from data about the beliefs and attitudes within a certain population. If a norm exists as a rule-based fact, its existence can be derived from the fact that the rule can be found in a recognized source of law, or fits in the best possible reconstruction of existing legal materials.
If norms are more like – to take a weird but nevertheless correct example – screwdrivers, than like descriptive sentences, their application is better comparable to the use of tools than to logical derivation. That norms are in need of application explains why it is possible to make exceptions to them and that it is possible to discuss whether a norm should be applied analogically to a concrete case even if it is obvious that the case does not satisfy all the conditions of the norm. The logic of rule application is for this reason not that of the legal syllogism with which it is sometimes identified.
Beauty, truth, and friendship are values, but it is not very clear what this means. The status of values is more difficult to account for than the status of value judgments, such as that this is a beautiful piece of music, that this friendship is valuable, that this theory is better than its competitor, and that yesterday’s football match was bad.
In any case, it is obvious that values and norms are not the same. Norms, as individuals in the logical sense, exist as a matter of intersubjective or rule-based fact. Values are not ‘individuals’; they are abstractions of valuable things, such as friendship or paintings. Norms, in the sense of deontic rules, directly tell us what to do; values only have indirect impact on our behaviour on the normative assumption that we should promote values. Moreover, norms are ‘binary’ in a sense in which values are not. A norm is either violated or complied with; there is no in-between. Something may be more or less beautiful, with an infinite range of in-between possibilities depending on the extent to which the value of beauty is realized.
Norms provide us with direct reasons for acting, and the sum of these reasons can be expressed in ought-judgments. Values provide us with reasons why something is valuable, and the sum of these reasons can be expressed in judgments about goodness. If something is a value, then, pro tanto, the more it is realized, the better. More friendship and more truth are, pro tanto, better than less.
So it may seem that goodness is the ‘highest’ value. On second thought this is less obvious than it might seem. To begin with, it sounds a bit strange to say that goodness is valuable. In contrast to friendship, truth and even beauty, goodness is too indistinct to count as valuable. In fact, the very word ‘goodness’ is somewhat suspect. Whereas friendship, truth and beauty exist in reality (in the form of people being friends, a sentence being true, or a landscape being beautiful), this is not at all clear in the case of ‘goodness’.
Moreover, the evaluation ‘good’ can be used for rather different purposes, including purposes in which being good seems at first sight not to be valuable at all. ‘He is a good hired killer’ would be an example, although it may be argued that a good hired killer is valuable for somebody who happens to need one. In that case the proper standard to evaluate the value of hired killers is the measure in which they fulfil their function. Arguably, there is a kind of goodness, good in a function or good of a kind, where the appropriate standards are given with the function or kind. The goodness at issue is then relativized to the function (a good screwdriver), respectively the kind at stake (a good soccer match). Another kind of relativity of goodness is goodness from a point of view, such as moral goodness or goodness from the point of view of European integration. It may even be questioned whether there exists a non-relative kind of goodness. For instance, the judgment ‘It is good that the crime rates are decreasing’ evaluates a state of affairs as good, but the goodness of this state of affairs seems to be relative to the economic, moral, or legal point of view. Similarly, the judgment ‘It is good that she took her umbrella along’ made in connection with a lady who has to walk through the rain, seems to express goodness relative to the lady’s interests.
Norms and values have been lumped together as being both ‘subjective’ and non-derivable from facts. We have seen that norms cannot be derived from descriptive sentences, let alone facts, but they cannot be derived from other norms either, not even with the help of descriptive sentences. The reason is that norms are, from a logical perspective, individuals, ‘things’, and not facts or descriptive sentences. However, the sentences that describe the factual counterparts of norms are descriptive sentences and they can be used in derivations. If somebody is under a duty to post a letter, then the sentence that this person has a duty to post the letter is true. Moreover, if the only way to post the letter is to take the letter to the post office, this person is, pro tanto, obligated to take the letter to the post office. Something similar holds for values. The judgment that something is valuable is comparable to the factual counterpart of a norm, and it can be used to derive other judgments about what is valuable. If a high employment rate is valuable, and if a low corporation tax is the only way to achieve a high employment rate, then, pro tanto, a low corporation tax is valuable too.
The last line of argument explains the familiar distinction between intrinsic and instrumental values. Something is an instrumental value if its realisation is valuable because it contributes, either causally or constitutively, to the realisation of some other value. So a low corporation tax would be an instrumental value because its realisation causally contributes to the realisation of a high employment rate, which is considered to be valuable. The absence of torture is, arguably, an instrumental value in the sense that it contributes constitutively to human dignity. Intrinsic values are those values the realisation of which is valuable but not for the reason that this realisation contributes to some other value. Apparently the realisation of intrinsic values is valuable ‘in itself’, or- which means the same thing – intrinsically. Human dignity or happiness might be examples of intrinsic values.
The value of something supervenes upon its other characteristics. For instance, a car is good because it is fast, economical, spacey, and cheap. Another car, which is similar to the first one in all its good and bad making characteristics, must then be a good car too. In general, supervenience is the phenomenon that some of an object’s characteristics are not independent, but depend on other characteristics of the same object without being identical to them. As a consequence, two objects which agree in their subvening characteristics must then also agree in their supervening characteristics.
An immediate consequence of the fact that value is supervenient is that value judgments are universalizable. For example, if a car is good because it is fast, this can be generalized into that standard that fast cars are good. This works also the other way round: given the standard that fast cars are good, the fact that a car is fast goes, pro tanto, together with the fact that this car is good.
And this leads to a third observation, namely that value judgements express facts. Not objective facts in the sense given above (Section 2.2), because the value of something depends on both its subvening characteristics and on a standard. Value-facts if we may call them thus, are mind-dependent, but that is a characteristic that they have in common with all other rule-based facts, including the facts that 3+6 equals 9, that Amsterdam is the capital city of the Netherlands, and that legal parenthood can be created by means of adoption. So, although facts are not values, it can be a fact that something is valuable.
Because value judgements are supervenient, they can be derived from subvening characteristics of the ‘thing’ that is evaluated. What the relevant subvening characteristics are depends on the standards that are used for the evaluation. These standards are from a logical point of view individuals and their existence is a matter of either intersubjective, or of rule-based fact. For example, the standard that a soccer match with offensive, but fair, play is a good match exists as an intersubjective fact. (At least, let us assume that by way of example.) Then it is a rule-based fact that some particular soccer match, characterized by offensive but fair play, was a good match.
Depending on the mode of existence of the standard, the value judgments express respectively the way something tends to be evaluated, or the way something (rationally) should be evaluated. If the value judgment is a moral one, this distinction corresponds to the well-known distinction between positive or descriptive, and critical or normative morality. Judgments of positive morality are based on standards that exist as a matter of intersubjective fact, while judgments of critical morality are based on standards that exist as a matter of multiply rule-based facts (see Section 2.2).
If value judgments are to be rational, the standards on which they are based should be ‘good’ ones. The goodness of these standards is a supervenient characteristic which depends on subvening characteristics of the standard and a meta-standard that specifies what makes an evaluative standard into a good one. This meta-standard should obviously also be a good one, and its goodness is based on subvening characteristics and a meta-meta-standard, which …. and so on. This argument is the traditional one to show that there exists an insurmountable gap between fact and value. Value judgments depend on facts, but never on facts alone; they always presuppose other value judgments. And therefore the ‘worlds’ of fact and value appear essentially different.
However, this appearance is deceptive, because the proper comparison is not between facts and value judgements, but between value judgments and non-evaluative judgments. Value judgements express facts too, although most likely not objective facts. Both categories of judgments depend for their justification (an evaluative notion) on factual beliefs and standards which make these beliefs relevant for the judgments. These standards must be good ones, which makes that they depend on other standards, … and so on. As far as their dependence on standards is concerned, value judgments and non-evaluative judgments are in the same boat.
All sensible discussions about facts, values and norms, including methodological discussions about how to justify normative and value judgements and about how to develop a normative science presuppose clarity about the basic notions used in them. As long as it is unclear what we understand by facts, norms and values, such methodological discussions are prone to end in ambiguities, vagueness and confusion, for instance because we confuse the constitutive and the deontic aspect of norms, or because we do not see that what is deontic (an ought) can at the same time be a fact (not subjective). In this chapter I have attempted to provide at least the beginnings of a clear conceptual framework that can be used in the development of accounts of the nature of law and the proper methods for scientific legal research. It deals with a large number of distinctions between, among other things, facts and their descriptions, different kinds of objectivity of facts, different kinds of deontic facts, deontic facts and norms, different kinds of rules, and facts and values. I hope that many readers will make the effort to grasp the distinctions and the relations between the concepts, and that they will use the insights gained in their substantive theories. Because of the greater clarity, these theories will be more open to fruitful discussion, and therefore more amenable to progress.
 The author thanks Anne Ruth Mackor, Bart van Klink, Michal
Araszkiewicz, Wouter de Been,
Wibren van der Burg and Gustavo Arosemena for useful comments on draft versions of this paper.
 See Arnold Brecht, Political Theory: The Foundations of Twentieth-Century Political Thought (Princeton University Press 1959), RM Hare, The Language of Morals (OUP 1952) and Freedom and Reason (OUP 1963), and Paul W Taylor, Normative Discourse (Greenwood Press 1961).
 David Hume, A Treatise of Human Nature (OUP 1978; first published 1739/40), book III, part I, section I.
 Immanuel Kant, Grundlegung zur Metaphysik der Sitten (Felix Meiner Verlag 1906; first published 1785). The many interpretational issues that may be relevant here are ignored for the present purposes.
 See Hans Kelsen, Introduction to the Problems of Legal Theory (translation of the first (1934) edition of the Reine Rechtslehre by Bonnie Litschewsky Paulson and Stanley Paulson, Clarendon Press 1992), 92 and Reine Rechtslehre (2nd ed., Franz Deuticke 1960).
 See David Copp (ed.), The Oxford Handbook of Ethical Theory (OUP 2006), John Searle, ‘How to derive “ought” from “is”’, in Philosophical Review 73 , 43-58 and Hilary Putnam, ‘The Collapse of the Fact/Value Dichotomy’, in The Collapse of the Fact/Value Dichotomy and Other Essays (Harvard University Press 2002), 7-45.
 This last line of thinking is particularly clear in the work of the logical empiricists. See Moritz Schlick, ‘What Is the Aim of Ethics?’, Chapter I of M Schlick, Problems of Ethics (New Prentice Hall 1939), included in AJ Ayer (ed.), Logical Positivism (The Free Press 1959), 247-263 and AJ Ayer, Language, Truth and Logic (2nd ed. Penguin 1971; 1st edition 1936).
 Other interpretations of the distinction between law and politics are, of course, also possible.
 Kelsen, Reine Rechtslehre, 72 and 4.
 See Karl Olivecrona, Law as Fact (1st edition, Einar Munksgaard 1939).
 PF Strawson, ‘Truth’, in Proceedings of the Aristotelian Society, Supplementary Volume, 1950.
 Frege’s convention that sentences denote truth values is ignored here.
 Jaap C Hage, Studies in Legal Logic (Springer 2005), 192-200, and 'The deontic furniture of the world', in Jerzy Stelmach, Bartosz Brożek and Mateusz Hohol (eds.), The Normativity of Law, (Copernicus Press 2011), 73-114.
 Michael Devitt, Realism and Truth, (2nd ed. Blackwell 1991), 13-17.
 Brian Leiter, Naturalizing Jurisprudence. Essays on American Realism and Naturalism in Legal Philosophy (OUP 2007), 257-264.
 John R Searle, Making the social world, (OUP 2010), ch. 3.
 To the author’s knowledge, there exists no law that makes Amsterdam the capital, so the fact would not be rule-based. However, as Wibren van der Burg was so kind to point out to me, there is phrase in the Dutch constitution, which suggests that Amsterdam is the capital of the Netherlands.
 Arguably, some of these facts are (also) rule-based facts. This is the case because there is no hard difference between kinds of intersubjective facts that exist because they are broadly accepted and facts that are based on rules where the rules themselves exist by being broadly accepted (social rules in the Hartian sense).
 The difference in views between Jellinek and Kelsen, as described in the contribution by Van Klink and Lembcke to the present volume, may find its deep explanation in the unwillingness of Kelsen to recognize facts that are not rule-based as relevant for a ‘pure’ theory of law. Jellinek would be willing to also recognize intersubjective facts, as would Mackor, going by her contribution to this volume.
 See Philippa Foot, ‘Morality as a System of Hypothetical Imperatives’, in Virtues and Vices (Basil Blackwell 1978), 157-173.
 Joseph Raz, ‘Legal Validity’, in Archiv für Rechts- und Sozialphilosophie BdLXIII, 3 .
 R. Alexy, The Argument from Injustice. A Reply to legal Positivism (OUP 2002), ch. 3.
 Jaap Hage, 'Construction or reconstruction? On the function of argumentation in the law', in C Dahlman and E Feteris (eds.), Legal Argumentation Theory: Cross-Disciplinary Perspectives (Springer 2012), 125-144.
 Bart van Klink pointed out to me that this might also be considered an intersubjective fact, since Deep Purple has far more fans than Grand Funk Railroad. He might be right, but that depends on the precise criteria for the existence of intersubjective facts. For instance does the existence of an intersubjective fact require that people accept that something is a fact because they consider acceptance as the appropriate standard for being such a kind of fact? Should people consider Deep Purple as the better band for the reason that ‘everybody’ thinks so? This is an important topic, but falls outside the scope of this contribution.
 Richard Joyce, ‘Projectivism and quasi-realism’, The Stanford Encyclopedia of Philosophy (Summer 2009 Edition), Edward N. Zalta (ed.),
URL = <http://plato.stanford.edu/archives/sum2009/entries/moral-anti-realism/> (last visited on October 16, 2015).
 When we seriously discuss the taste of Brussels sprouts, this goes to show that we consider their taste not to be a subjective matter, but, for instance, a matter of multiply or indeterminate rule-based fact.
 See for example Hare, The language of morals, Ch. 11.
 The term ‘deontic’ in this connection is borrowed from deontic logic, the branch of logic that deals with what ought to be done or ought to be the case, and has its roots in the Greek language. The reason to prefer ‘deontic’ above the more common ‘normative’ is that the word ‘norm’ is very ambiguous.
 See PT Geach, ‘Good and Evil’, in Analysis 17 , 32-42 and John Searle, Speech Acts (CUP1969), 136-140.
 JL Mackie, Ethics (Penguin Books 1977), 38-42.
 See Hilary Putnam, Reason, Truth and History (CUP 1981), 147.
 Assuming that the aesthetic quality of wall-decoration is not merely a matter of taste.
 The proposal fits as well as possible with common usage, but is unavoidably somewhat stipulative. A more extensive discussion can be found in my ‘The deontic furniture of the world’.
 The term ‘obligation’ derives its technical meaning that is proposed here from the civil law tradition, according to which an obligation is particular kind of bond between a debtor and a creditor. In the English literature the difference between duties and obligations is not made so clearly, possibly under the influence of the common law.
 Being a human being might be the most abstract status to which duties are assigned.
 An interesting question in this connection is whether there can be obligations caused by undirected promises such as the promise to reward the person who returns my lost dog. (This was pointed out to me by Michal Araszkiewicz.)
 See my ‘The deontic furniture of the world’.
 The view is historically ascribed to Immanuel Kant Kritik der reinen Vernunft (Suhrkamp 1974; first edition 1781), A548/B576, but is – for different reasons - also a theorem in the standard system of deontic logic. See Brian F Chellas, Modal logic, an introduction (CUP 1980), 191.
 Joseph Raz, Practical Reason and Norms (Hutchinson 1975), 35-39.
 Jaap C Hage, Reasoning with Rules (Kluwer 1997), 110-118.
 Ronald Dworkin, ‘Is there a right to pornography?’ in Oxford Journal of Legal Studies 1 , 177-212.
 In the introductory section of their contribution ‘Exploring the Boundaries of Law: On the Is-Ought Distinction in Jellinek and Kelsen’ (present volume), Van Klink and Lembcke picture the law as dealing with what ought to be the case, rather than with what ought to be done.
 Since the three deontic modalities can be expressed in terms of each other in combination with negation, it does not matter from a logical point of view which modality is taken to be the basic one. Arguably, however permission is not a basic deontic modality, because it is the absence of such a modality.
 In series of articles, Westerman has paid ample attention to the phenomenon of ‘goal regulations’, the point of which is to guide actors to realize some goal, rather than specifying precisely which actions should be performed. For an overview, see http://www.paulinewesterman.nl/publications-dutch.php(last visited on October 16, 2015).
 The locus classicus of this approach is Georg Henrik von Wright, ‘A new system of deontic logic’, in R. Hilpinen (ed.), Deontic Logic: Introductory and Systematic Readings (Reidel 1971), 105-120.
 See JW Forrester, ‘Gentle murder, or the adverbial Samaritan’, in Journal of Philosophy 81 , 193-197, and Henry Prakken and Marek Sergot, ‘Dyadic Deontic Logic and Contrary-to-Duty Obligations’, in Donald Nute (ed.), Defeasible Deontic Logic (Kluwer 1997), 223-262.
 For a more elaborate discussion, see Jaap Hage, ‘Contrary to duty obligations’, in J. Breuker e.a. (eds.), Legal Knowledge and Information Systems. Jurix 2000: The Fourteenth Annual Conference (IOS-Press 2000), 89-102.
 John Searle, Expression and meaning. Studies in the theory of speech acts (CUP 1979), 12-14.
 About the world-to-word direction of fit of (constitutive) rules, but unnecessarily limited to ‘counts as rules’, see John R Searle, The construction of social reality (The Free Press 1995), 43-48. More elaborate about the world-to-word direction of fit of rules is Jaap Hage, ‘Separating Rules from Normativity’, in Michal Araszkiewicz, Pawel Banaś, Tomasz Gizbert-Studnicki and Krzystof Pleszka (eds.), Problems of Normativity, Rules and Rule-Following (Springer 2015), 13-30.
 The idea that rules are ‘things’, or – more precisely – ‘abstract objects’ plays a central role in Anne Ruth Mackor, ‘Legal Doctrine As a Non-Normative Discipline’, in Recht en methode in onderzoek en onderwijs 2 , 22-45, and in ‘Legal doctrine is a non-normative discipline. An argument from abstract object theory’, in the present volume.
 See Ronald Dworkin, ‘Hard Cases’, in Taking Rights Seriously, 2nd ed. (Duckworth 1978), 81-130, and Mackor, ‘Legal doctrine is a non-normative discipline. An argument from abstract object theory’.
 See Herbert L.A. Hart, The Concept of Law (3rd ed., OUP 2012; 1st ed. 1961), 94-96.
 The distinction between regulative and constitutive rules can be traced back to at least J. Rawls, ‘Two Concepts of Rules’, in Philosophical Review 64 , 3-32 and to Searle, Speech Acts , 33-42.
 For the philosophical implications of his simple observation, see Jaap Hage, ‘Juridical Acts and the Gap between Is and Ought’, Netherlands Journal of Legal Philosophy 42 , 50-66.
 Rules of recognition would then be counts-as rules, while primary rules of obligation would be either obligation-creating dynamic rules, or duty-imposing fact-to-fact rules. See section 4.2.
 Kelsen (Reine Rechtslehre 73) described this distinction as existing between Rechtsnormen and Rechtssätze. The distinction does not only exist in law, as Kelsen’s terminology suggests. Moreover, Kelsen opposed the rules (Normen) to descriptive sentences (Sätze), while it is more appropriate to keep the comparison either completely on the ontological level (as is done here) or on the linguistic level.
 As also becomes clear from the contribution of Van Klink and Lembcke to the present volume, Kelsen seems to mix up the rule-nature and the deontic nature of norms. The distinction between description and constitution (the word-to-world and the world-to-word direction of fit) is then confused with the altogether different distinction between is and ought.
 See the in this connection prototypical paper J Jörgensen, ‘Imperatives and Logic’, in Erkenntnis 7 [1937/8], 288-296.
 Raz, The Authority of Law, 37-52.
 See Ronald Dworkin, Law’s Empire (Fontana 1986), and Mackor, Legal Doctrine As a Non-Normative Discipline’, and ‘Legal doctrine is a non-normative discipline. An argument from abstract object theory’.
 See Jaap Hage ‘The Logic of Analogy in the Law’, in Argumentation 19 , 401-415.
 For instance in Neil MacCormick, Legal Reasoning and Legal Theory (OUP 1978), Ch. 2.
 Geach, ‘Good and evil’.
 See JO Urmson, The Emotive Theory of Ethics (Hutchinson 1968), Ch 9 and, specifically concerning what is morally good, Foot, ‘Are moral requirements hypothetical imperatives?’.
 In this respect, judgments about goodness are different from ought-judgments, which depend on a standard, but which are not for that reason relative judgments. See Section 2.2.
 Michael J Zimmerman, ‘Intrinsic vs. Extrinsic Value’, The Stanford Encyclopedia of Philosophy (Winter 2010 Edition), Edward N. Zalta (ed.),
<http://plato.stanford.edu/archives/win2010/entries/value-intrinsic-extrinsic/> (last visited on October 16, 2015).
 Paul Teller, ‘Supervenience’, lemma in Jaegwon Kim and Ernest Sosa (eds.), A Companion to Metaphysics, (Blackwell 1995), 484-486.
 This was emphasized in Hare Freedom and reason, Ch. 2.
 Notice that the formulation of the standard does not include the clause ‘pro tanto’. The standard is not a descriptive sentence and that fact that it may have exceptions does not influence the formulation of the standard, which can remain without exceptions, but it does influence the ‘logic’ of standard application.
 See Bernard Gert, ‘The Definition of Morality’, The Stanford Encyclopedia of Philosophy (Fall 2012 Edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/fall2012/entries/morality-definition/> (last visited on August 20, 2014).
 See Jaap Hage, ‘The Method of a Truly Normative Legal Science’, in M van Hoecke (ed.), Methodologies of Legal Research (Hart Publishing 2011), 19-44.