The Zero'th Law of Thermodynamics states that : If two systems are in thermal equilibrium with a third system, then they are in thermal equilibrium with each other.

Clearly, this introduces an equivalence relation between all thermodynamic systems and as a result, it is frequently stated that the purpose of the zeroth law is to recognize this empirical fact and introduce the notion of temperature $T$ to characterize the equivalence classes. My first issue is that if a zeroth law introducing temperature ($T$) as characterizing an equivalence class is necessary for the logic of thermodynamics, then why do we not have 4th and 5th law's introducing chemical potential and pressure as characterizing equivalence classes? For example, we could right the 4th law as : If two systems are in diffusive equilibrium with a third system, then they are in diffusive equilibrium with each other.

Why is the zero'th law and the notion of temperature needed for the internal logic of thermodynamics to make sense when my supposed 4th law is not needed? What makes temperature so difference from pressure or chemical potential in this regard?

Secondly, from my understanding, fundamental laws in physics should meet two criteria in order to be regarded as fundamental laws. Firstly, they should be statements that comport with empirical evidence to such an extent that we regard them as accepted truths. And secondly, they should not be derivable from any other statements or fundamental laws. If they are derivable from some other statement, then they can be discarded and the statement from which they are derived can be classified as fundamental instead.

If we look through the perspective of the microcanonical ensemble though, temperature is simply a derived quantity defined by $\frac{1}{T}\equiv\left(\frac{\partial S}{\partial U}\right)_{V,N}$. In this ensemble, we absolutely need the first and the second law's of thermodynamics. That is, we need to assume energy conservation, and we need to assume that the entropy of an isolated system never decreases. But once we've made these two assumptions, we get the notion of temperature for free. Similarly, if we allow 2 subsystems of an isolated system to alter their volume, from the first and second laws alone, we get the notion of pressure for free $\left(\frac{P}{T}\equiv\left(\frac{\partial S}{\partial V}\right)_{U,N}\right)$. The same goes for diffusion and chemical potential. It seems to me that the introduction of temperature as a state variable through the zeroth is not on equal footing with the introduction of energy as a state variable (via the first law) or entropy as a state variable (via the second). Its importance is merely equivalent to that of pressure or chemical potential? What about this am I missing? Why is temperature and the zeroth law so crucial to thermodynamics even though it seems to be derivable from the 1st and 2nd laws whilst pressure and chemical potential are not deemed to be so crucial?

  • 2
    $\begingroup$ In the formulation you give the law does not specifically mention temperature, but thermodynamic equilibrium, so the proposed 4th and 5th law are already covered by it. What really matters is the transitivity of thermodynamic equilibrium. $\endgroup$
    – Roger V.
    Commented Oct 13, 2022 at 9:47
  • 1
    $\begingroup$ see also this answer: physics.stackexchange.com/a/464589/226902 Regarding the derivation from the microcanonical in the title: physics.stackexchange.com/q/623364/226902 Regarding the existence of temperature from the 0-law: physics.stackexchange.com/q/229794/226902 $\endgroup$
    – Quillo
    Commented Oct 13, 2022 at 10:30
  • $\begingroup$ @RogerVadim Thanks for the response. Okay so then am I correct in thinking that pressure and chemical potential are entirely on equal footing with temperature? Peter Atkins states that "each law provides an experimental foundation for the introduction of a thermodynamic property". He then states that the zeroth law introduces the property we call Temperature. So based off Atkins' statements, I find it hard to believe that there isn't something that raises the importance of temperature above that of pressure or chem potential etc. $\endgroup$ Commented Oct 14, 2022 at 7:20
  • $\begingroup$ @Quillo thanks for the links. Looking through them though, none of them seem to explain why if we require the zeroth law to introduce the notion of temperature, why do we not need 4th and 5th laws to introduce the notions of pressure and chemical potential as well. Also, they do not explain why the zeroth law is required when temperature is simply a derived quantity in the microcanonical ensemble. $\endgroup$ Commented Oct 14, 2022 at 7:25
  • $\begingroup$ Pressure was known well before temperature and beyond thermodynamic context - so it is not introduced by the zeroth law. But it is an intensive thermodynamic variable, appearing on equal footing with temperature (see, e.g., the expression for the first law, $dU=TdS - PdV$). Chemical potential perhaps seems less important only because it is introduced later in stat phys texts... or because it appears only in grand canonical ensemble. $\endgroup$
    – Roger V.
    Commented Oct 14, 2022 at 15:11

2 Answers 2


There is a school of thought that attempts to formulate classical thermodynamics in terms of a minimum set of axioms so as to make all other results of thermodynamics inevitable. This approach strives for logical consistency though not for pedagogy, let alone for physical insight. And when it comes to "axioms", one is left to decide what constitutes an axiom versus what constitutes knowledge that exists prior to the axiom. Since the axiomatic approach is divorced from physics, it needs to express in axioms what the physicist considers "obvious" based in empirical observation.

One could say that the "zeroth" law, which was added to the list of axioms long after it had been routinely and unquestionably accepted as obvious, to express the fact that if a system is in equilibrium, then every one of its internal parts is in equilibrium with all others. Thus we can speak of a uniform temperature, pressure and chemical potential across the entire system.

At the end of the day one does not need axioms to describe nature. One needs observations and a quantitative hypothesis (theory) that ties the observations together. Statistical mechanics does that beautifully.

Statistical mechanics is built not on axioms, but on a physical hypothesis: that all microstates with the same energy, volume and number of particles contribute equally to the mechanical properties of a system in equilibrium. This is and remains a physical hypothesis because there is no way to measure the probability of microstates experimentally. The validity of stat mech is contingent upon evidence: as long as its predictions agree with measurement, we accept the theory as valid.


I think you inverted the logic a bit when showing how $T$ can be derived from entropy. In fact it is the opposite. In classical thermodynamics you define entropy as $d S=\delta Q/T$.

You may argue that entropy can be defined statistically (with no regards for thermal state) when you go down into microscopics but in fact here it is even more complicated and confusing. The existing dynamical theories (classical or quantum mechanics) are reversible. They do not accept probabilities. In the absence of a ${\it thermal}$ bath the system is always in a pure state, and hence its entropy is zero. Not a very useful quantity.

But when there is a bath there is an entropy one may say. True, but what is a bath if not another system in a thermal equilibrium? How did ${\it that}$ system enter a ${\it thermal}$ state and what does it even mean?

So to me, the statement of the zeroth Law of Thermodynamics (since it is not a canonic law different people formulate it differently, and I admit my understanding might not match what you have in your mind) is that ${\it there\, exists}$ a thermal state with the properties that you have described.

How obvious it is? Well, I believe it is not. For instance, there is a very active sub-field in condensed matter physics that works what they call "many body localization". What this means is that under certain circumstances the interactions in a many-body system "break ergodicity" and thus prohibit it from reaching a thermal state.

P.S. As for pressure and other quantities such as chemical potential, to me, they are not on equal footing with temperature, as all of them have a very transparent meaning that can be derived from pure mechanics. Whereas temperature is different. As I tried to articulate it above, it (or the very idea of a thermal state) doesn't really fit the rest of physics, so its existence has to be postulated.

  • 1
    $\begingroup$ "you may argue that entropy can be defined statistically"? This is the only way to define temperature. There is no definition of temperature in classical thermodynamics that does not require a "thermometer" device, i.e. "the height of a liquid in a capillary in contact with boiling water". $\endgroup$
    – Themis
    Commented Oct 14, 2022 at 15:47
  • $\begingroup$ What you describe is a temperature ${\it scale}$ (btw you agree a liquid capillary is not a good thermometer for a universal thermodynamic scale). But yet it is not a definition of ${\it temperature}$. In textbooks they just postulate there is a function of state $\Theta$ that has to be same for bodies in thermal equilibrium. For example take Kardar (Statistical Physics of Particles, 1ed, page 3): "...there are many potential choices of $\Theta$, the key point is the existence of a function that constrains the parameters of each system in thermal equilibrium". $\endgroup$
    – John
    Commented Oct 14, 2022 at 16:04
  • $\begingroup$ Kardar does not define temperature, only temperature scale. The ideal gas temperature scale on p4 is based on a different thermometer, the ideal gas. It is as unsatisfying as the liquid capillary because it does not tell us what temperature is, only how to measure it. $\endgroup$
    – Themis
    Commented Oct 14, 2022 at 19:09
  • $\begingroup$ That is what I mean. Temperature is postulated to exist as some unique function of state which tells which way heat flows and monotonically grows upon heat absorption. Only then in order to have a quantitative measure of this quantity one defines a scale. $\endgroup$
    – John
    Commented Oct 14, 2022 at 20:40
  • $\begingroup$ By the way ideal gas scale is not as unsatisfactory as liquid capillary. For one thing it can be used all the down absolute zero and does not depend on the geometry of the vessel. $\endgroup$
    – John
    Commented Oct 14, 2022 at 20:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.