Pure Theory of Law

Pure Theory of Law


Hans Kelsen was an Austrian legal theorist, who worked in Germany until the rise of the Nazi Party, and then in the USA. He published the first edition of The Pure Theory of Law in 1934, and a second, expanded edition (which I read) in 1960.

The theory is ‘pure’ because it separates jurisprudence from other disciplines like ethics, politics and psychology. This is important because different disciplines have different methodology, and so it’s hard to analyse law when its all mixed up with other things. Kelsen’s pure theory allows for a pure ‘legal science.’

The word ‘science’ is a translation of the German Wissenschaft, which is normally translated as ‘knowledge’ or ‘study,’ and can apply to disciplines like literary theory. The English word ‘science’ is translated into German as Natur Wissenschaft i.e. ‘natural science.’ We should therefore not jump to conclusions about the relationship between legal science and natural science. Nevertheless, a comparison seems at least plausible.

Descriptive and Normative Statements

We begin the idea from Kant’s philosophy that reality can only be understood by humans through different formal categories of statements. Here we are concerned with two categories in particular, which give two different ways of talking about reality.

The first category is ‘Descriptive Statements’ (descriptions), which deal with existence, or Kelsen’s words: “A statement that something is.” An example is ‘the door is closed.’

The second category is ‘Normative Statements’ (norms), which deal with preference, “A statement that something ought to be.” An example is ‘the door should be closed.’

Kelsen’s description of these categories highlights the words ‘is’ and ‘ought.’ I don’t like this, because it confuses the meaning of a sentence with the particular language used to express it. To decide whether a statement is descriptive or normative, you need to look at its meaning. For example:

  • A statement does not require the word ‘is’ to be a description: the sentences ‘the man runs away’ or ‘the apple will fall’ are descriptions.
  • Not all statements of the form ‘X is Y’ are in fact descriptions. For example, the statement ‘Murder is bad’ is in fact the norm ‘Murder should not be done.’

The two classes of statements are logically separate. There are no correct statements of the form ‘X is, therefore Y should be’: one cannot derive an ‘ought’ from an ‘is’. The classes also cannot contradict, i.e. the statement ‘the door should be closed’ does not contradict with ‘the door is open’ or even ‘the door does not exist.’

The two classes are related in asymmetrical ways. A descriptive statement can have a normative statement as its ‘meaning’. For example, ‘a robber says ‘Stop!’’ has the normative meaning ‘You should stop.’ We cannot conclude that the normative meaning is valid, but it is still attached to the descriptive statement. For this reason, we consider norms to be subordinate to descriptions.

It is important to keep in mind the distinction between norms and descriptions. The norm ‘Theft should be punished’ is different from ‘Theft will be punished,’ or ‘The policeman thinks theft should be punished,’ or ‘The Criminal Code says theft should be punished,’ all of which are descriptions.


We can now define legal laws as a subset of normative statements, and scientific laws as a subset of descriptive statements.

Kelsen introduces a third category that sits above normative statements. This category contains descriptive statements about norms, as opposed to about reality. The work of a legal academic contains such descriptive statements. If we care only about law, we can call these statements ‘rules of law.’ This is a horrible name – do not confuse a rule of law with ‘The Rule of Law’ meaning that the government should obey the law.

Descriptions, about reality or law, can be true or false, while norms can be valid or invalid.

Descriptions about reality contain statements about causality, i.e. statements of the form ‘If X is, then Y is.’ Rules of law contain an equivalent process called imputation, i.e. ‘If X ought to be then Y ought to be.’ You can use causality or imputation to derive valid statements from other valid statements. For example, you could derive the norm ‘Do not kill your neighbour’ from ‘Love your neighbour.’ Imputations can also use a description as the minor premise, e.g:

Major Premise (norm): You should do what your father says.
Minor Premise (description): Your father says “Go to bed.”
Conclusion (norm): You should go to bed.

Using this method, we can derive the validity of a norm using a description and an already valid norm.

Under Kelsen’s theory, no norm is objectively valid. If we assume one valid norm then we can derive the validity of others, but any such derivation is based on this first presupposed norm.

Natural science, i.e. the study of nature, what we understand in English to be ‘science’, is the study of reality and descriptions about reality. The study of legal norms and rules of law is Kelsen’s legal science.

Law as a System of Sanction-Prescribing Norms

Kelsen’s definition of a legal system, which I shall call ‘Kelsenian law’ is a system of norms that prescribe sanctions. A sanction is a punishment or reward that follows from a condition; an example of a sanction-prescribing norm is ‘A thief should be imprisoned.’

A ‘delict’ is an action that is prohibited by the law. An action is a delict if there is a sanction in place to discourage people from committing that action. Not all conditions for sanctions are delicts: for example, the sanction of forced quarantine may be imposed after illness. It follows that delicts are never directly prohibited, but rather are made the conditions of sanctions – there is no law that says “Do not steal,” only a law that says “Thieves should be imprisoned.”

Be aware of the distinction between ‘Thieves should be imprisoned’ and ‘Thieves will be imprisoned’ – the latter is a not a norm but a description, and therefore is not Kelsenian law. It is instead a description of what actually happens. It could be false, and the law would still be a law (e.g. if the thief was not caught).

A legal system includes general norms such as “Thieves should be imprisoned” , but also specific norms such as ‘This man should be imprisoned.’ A norm with specific application is still a Kelsenian law.

Dependent Norms

An independent legal norm is a norm that prescribes a sanction, e.g. ‘This man should be imprisoned.’ A dependent legal norm is a legal norm that does not itself prescribe a sanction, but is connected to a legal norm that does. For example, the norm ‘This judge is authorised to hear cases’ is valid only insofar as that judge can prescribe/dismiss sanctions, or make decisions that ultimately lead to sanctions

If a norm is not connected to a sanction-prescribing norm, then it is not a Kelsenian law. Such a norm is instead a moral or social norm posited by a legal organ who is not exercising their legal capacity.

All information about dependent norms can be described as a “If X then Y should be” statement (a rule of law), where X is all the conditions stipulated by the dependent norms, and Y is a sanction.

  • A licence to drive can be described as “If (1) a person drives, and (2) they have no licence, then they should be punished.”
  • A law that authorises a judge to preside over a theft case: “If (1) a person is found to have stolen by a judge, and (2) that judge was authorised to preside over the case, then that person should be punished.”
  • A contract to pay $100 in return for a service: “If (1) a person does not pay $100, and (2) that person was required to do so by a legal contract, then that person should be punished.”
  • The Constitution, as applied to theft: “If (1) a person has stolen, and (2) all laws relating to this case were created in accordance with the Constitution, then that person should be punished.”


A taxman, backed up by the State, and a robber backed up by a gun, both demand your money. Both men’s actions have the normative meaning “Give this man your money,” and both of these norms are backed by sanctions. We want a definition of law such that the former is law, and the latter is not.

We can do this by asking which of these two norms is valid. The norm “Give the taxman your money” can be derived from a series of higher norms:

Obey the constitution
Obey laws passed by the legislature
Obey tax law
Obey the tax office and its decisions
Obey the taxman’s authority
Give the taxman your money

If a norm is valid, any norm derived from that norm by imputation will also be valid. The taxman’s norm will be valid under this system, while the robber’s will not. A legal system is therefore a system of coercive, valid norms.

Source of Validity – The Grundnorm

Under Kelsen’s theory, no norm is objectively valid. The ‘source’ of validity therefore must be a norm that is assumed to be valid. Kelsen formalises this assumption through a new object: the basic norm or Grundnorm. The Grundnorm is not a legal norm, it is instead a norm that is assumed to be valid by the legal scientist examining the legal system. A legal norm is then any norm derived from this Grundnorm.

A Grundnorm refers to a specific constitution or other source of law. For example, the Australian legal system’s Grundnorm is “One ought to obey the Australian Constitution.”

An analogy can be made with science: if we want to say any valid statements about reality, we need to start with something we assume to be valid. Such assumed statements could “experience corresponds to reality” or “things will continue to behave the same way as they have in the past.” Such an assumption is the ‘source’ of the validity of a scientific system.


In order to qualify as a valid legal system, a norm must also be ‘by and large effective’, i.e. the norm must be obeyed most of the time. This means that a legal theorist’s choice of Grundnorm is not arbitrary – we have to choose the Grundnorm that gives validity to the actually effective legal system. We can’t assume that the Grundnorm is ‘Obey Hugh’ because society does not follow this. This is especially important during a revolution – if the old legal order ceases to be effective, either because of a hostile takeover, or just because people stop following the law, then there will be a new legal system and a new Grundnorm.

Legal Decisions

A law is valid if it was derived from a higher norm. But how does a legal scientist determine whether this derivation was legitimate?

Under Kelsen’s system, there is no objectively right or wrong way to derive a norm from a higher norm, just as there are no objectively valid norms in the first place. All we care about is a legal norm’s validity, and here, the source of validity is whether the person who derived the legal norm was authorised to do so. Therefore, every derivation made by an authorised legal organ is automatically valid. A legal scientist can describe the possible interpretations of a particular law, or the relationships between laws, but they cannot decide which derivation is ‘correct’ – that decision is entirely up to the judge.

It seems to me that there is some wiggle room here, to prevent extreme cases. For example, what if a corrupt judge claims to derive the norm ‘give me money’ from tax law? In this case, the legal theorist could argue that the decision is so unreasonable that it is not derived from the higher norm at all. The corrupt judge’s decision is either invalid and no different from a robbery, or derived from a more general norm of the form ‘This legal organ may posit new norms.’

If a case is successfully appealed, it doesn’t mean that the decision was wrong from the start – it just means that a higher legal organ chose to overturn the decision. For example, if a judge finds a person guilty of murder who was in fact innocent, the judge’s decision is still valid because the judge was authorised to make that decision. Similarly, if the legislature passes an unconstitutional statute, the statute is still valid unless and until it is repealed.

This means that if a decision cannot be appealed, the organ which makes that decision has ultimate legal authority. They can flout the constitution or any other law they choose, and their decision will still be valid. The only considerations that prevent such an organ from disrupting the law this way are non-legal ones, such as the organ’s morality or the organ’s fear of losing office.

This also means that a legal organ’s thinking process, reasoning and arguments behind a decision are not part of legal science. Kelsen admits that these factors are legal in nature, but they instead belong to a different discipline called ‘legal politics’. [1]

Fitting Other Definitions of Law into Kelsen’s Categories

Kelsenian law is then a subset of the set of norms, specifically, norms given validity by a Grundnorm that are effective and prescribe sanctions.

Scientific Law and Physical Law

In his book The Concept of Physical Law, Norman Swartz makes a distinction between ‘scientific law’ and ‘physical law.’ Scientific Laws are laws like Newton’s Laws and Ohm’s Law; they are posited by humans and can be changed or created at will. Physical Laws exist ‘out there’ and actually correspond to reality – they are the true laws, and are independent of our understanding. One could argue then, that the goal of science is to successfully approximate the physical laws with the right set of scientific laws. It is an open philosophical question as to whether the physical laws actually exist. I won’t go into this now, but it may be the topic of another post.

These sit inside ‘Descriptive Statements,’ and physical laws are a subset of ‘scientific laws’ that are true. The question of what ‘true’ means is a discipline in philosophy, but I’ll see if I can learn at least something about it.

Natural Law

The theory of natural law claims that within the set of norms there is a subset of objectively correct laws. A society’s positive law is an approximation of the natural law, and improvements to positive law bring it closer to natural law. Natural law is a bit like physical law then, in that our positive law (scientific or Kelsenian) is an approximation for the ‘true’ law (physical or natural).

Natural law then fits into Kelsen’s scheme as a special subset of normative statements that are objectively valid.

The Pure Theory argues that by definition natural law cannot exist – a law can only derive validity from a presupposed Grundnorm, therefore and no law can be objectively correct. I don’t know how a natural lawyer would get around that argument, so that’s another thing to look into.

Existing Law

So far we have talked about physical law as objectively true rules, or natural law as objectively valid norms. You can go one step further, and argue that these laws actually exist in reality somewhere. This would give you statements like “The law of gravity causes the mass to accelerate.” I will call these laws that exist in reality ‘Existing Law.’ They just go into the ‘Reality’ spot in Kelsen’s scheme.

This theory sits comfortably with the idea of a God, who exists in reality and creates the physical and natural laws. Note that this of course does not need to be the Christian God, or any other particular God – for example, Einstein’s ‘cosmic spirituality’ fits this description.

A more secular interpretation of this could nonetheless describe laws as objects that exist, and can affect matter. A natural lawyer could describe norms as arising from ‘human nature,’ which fits into the language of ‘fundamental human rights’ and other such things. [2]

Impurities in Kelsen’s Pure Theory

1) Effectiveness

According to Kelsen, all legal norms are either valid or invalid, and one of the conditions for validity is that the legal norm be ‘by and large effective.’ You need to take a middle path in terms of effectiveness: effectiveness must be taken into account, so that you don’t get absurd legal systems, but you also don’t need all the laws to be effective all of the time.

You definitely need effectiveness, because otherwise any hypothetical system of coercive norms can be law, and the legal scientist has to arbitrarily choose the ‘correct one.’ But effectiveness has problems.

Effectiveness is both imprecise and impure (in terms of the Pure theory). Kelsen does not explain how to determine whether a norm is effective, and such a process would need to answer various non-legal questions, such as:

  • Are some laws more ‘serious’ than others, i.e. is it more important that the law of murder be obeyed than the law of theft?
  • If we want to determine whether the law against speeding is effective with respect to a particular citizen, do we need to take into account how often she drives? What about her moral attitude towards speeding?
  • Effectiveness lies on a spectrum – where do we draw the line?

Effectiveness is looking like a necessary impurity in Kelsen’s theory.

This is also a problem in science: the various scientific disciplines describe one reality, but use different methodology. So they often make assumptions that seal each other off from each other – physics assumes lots of simplifying mathematical properties that might not be well defined, and biology assumes lots of physical properties that might not be well defined. This is a ‘flaw,’ but a flaw with practical benefits, because you get useful answers. People acknowledge, when they make such assumptions, that their model/theory will be no good if the assumptions turn out to be false.

In terms of Kelsen, maybe you could argue that the only way to make effectiveness precise would be to compromise the Pure Theory and bring in all the sociological and moral considerations. By refusing to elaborate on ‘by and large effective,’ Kelsen could be sealing off those considerations and keeping the theory Pure.

To seal off questions about effectiveness, you would need to make an assumption like ‘it is always clear whether a law is by and large effective’. In situations where effectiveness is not clear, such as when a society is unstable, Kelsen would have to concede that his theory had major problems.


A legal scientist can describe the possible interpretations of a legal norm, but only an authorised legal organ can choose which interpretation to actually implement. However, if a legal organ chooses an interpretation that the scientist thought was ‘impossible’, the scientist may not consider the resulting norm to be valid. Legal interpretation can therefore affect the content of the Pure theory.

The theory definitely needs to allow legal scientists to interpret norms. Otherwise, a legal norm would be meaningless to a legal scientist, and they could only understand it by looking at its effects. However, legal interpretation uses things like ‘purpose’ that are beyond legal science. So these other disciplines have another way to bleed into the pure theory.

Kelsen sets a low standard here though: the scientist doesn’t need to choose the correct interpretation, they just need to identify all the ‘possible’ interpretations. This gives us an assumption that ‘seals off’ the politics of interpretation from the pure theory: it will always be clear whether an interpretation is possible.

You can easily think of hypothetical scenarios where this assumption is no good, for example, if a judge tries to prosecute a murderer with a law ‘Thieves should be punished,’ on the grounds that the murderer ‘stole’ life. I think it is not clear whether this interpretation is ‘possible.’ But in reality, such situations seem implausible, which makes the assumption seem pretty good. Kelsen would still have to admit there are some situations (such as a corrupt judge abusing their power) where this assumption could cause problems.


Kelsen’s desire for rigour and precision is risky because the theory might begin to deviate from a reasonable understanding of law. The two fudge factors, effectiveness and interpretation, work well to keep Kelsen’s theory grounded in reality, but they also compromise its purity. If the questions ‘is this law effective?’ or ‘is this a possible interpretation?’ are difficult questions, then the Pure theory does not have the right machinery to analyse the law.



  1. By Hugh McCarthy
  2. Id.

See Also

  • Legal Positivism
  • Natural Law Theory
  • Interpretivist Theories of Law
  • Legal Realism
  • Nature of Law

Further Reading

  • C. Fred Alford (2010). Narrative, Nature, and the Natural Law: From Aquinas to International Human Rights. Palgrave Macmillan.
  • The Stanford Encyclopedia of Philosophy (Fall 2011 Edition), Edward N. Zalta (ed.)
  • Robert C. Baker (2011). Natural Law, Human Sexuality, and Forde’s “Acid Test”. In Robert C.
  • Baker & Roland Cap Ehlke (eds.), Natural Law: A Lutheran Reappraisal. Concordia Pub. House
  • Finnis, J. (2011, first published 1980). Natural law and natural rights. Oxford: Oxford University Press
  • Robert C. Baker & Roland Cap Ehlke (eds.) (2011). Natural Law: A Lutheran Reappraisal. Concordia Pub. House.
  • Gail Belaief (1971). Spinoza’s Philosophy of Law. The Hague,Mouton.
  • Stephen Buckle (1991). Natural Law and the Theory of Property: Grotius to Hume. Oxford University Press.






Leave a Reply

Your email address will not be published. Required fields are marked *