Once again, I am studying classical thermodynamics. Axiomatic thermodynamics to be precise, none of this statistical physics business (which is interesting on its own right, but it is quite a different topic.)

The more I learn about it, the more I find thermodynamics incredibly fascinating. Why is it so different from other areas of physics? Perhaps I now have an answer that may be trivial to some, but eluded me until now.

Most of physics is described by functions of coordinates and time. This is true even in the case of general relativity, even as the coordinate system itself may be curved, the curvature (the metric) is described as a function of space-time coordinates.

In contrast, there are no coordinates in axiomatic thermodynamics, only states. States are decribed by state variables, and usually you have these in excess. For instance, the state of one mole of an ideal gas is described by any two of the three variables *p* (pressure), *V* (volume) and *T* (temperature); once two of these are known, the third is given by the ideal gas equation of state, *pV* = *KT*, where *K* is a constant.

Notice that there is no *independent* variable. The variables *p*, *V*, and *T* are not written as functions of time. Nor should they be, since axiomatic thermodynamics is really equilibrium thermodynamics, and when a system is in equilibrium, it is not changing, its state is constant.

So why is it not called *thermostatics*? What does *dynamics* have to do with stationary states? As it turns out, thermodynamics is the science of fitting a square peg in a round hole, as having just established that it’s a science of static states, it nevertheless goes on to explain how states can *change*… so long as all the intermediate states can exist as static states on their own right, such as when you’re heating a gas slowly enough so that its temperature is more or less uniform at all times, and its state is well approximated by thermodynamic variables.

The zeroeth law states that an empirical temperature exists that is associative: systems that have the same temperature form equivalence classes.

The first law defines the (infinitesimal) quantity of heat *dQ* as the sum of changes in internal energy (*dU*) and mechanical work (*p* *dV*). An important thing about *dQ* is that there may not be a *Q*; in the jargon of differential forms, *dQ* is a Pfaffian that may not be exact.

The second law uses the assumption of irreversibility and Carathéodory’s theorem to show that there is an integrating denominator *T* and a function *S* such that *dQ* = *T* *dS*. (Presto, we have entropy.) Further, *T* is uniquely determined up to a multiplicative constant.

Combined, the two laws can be written in the form *dU* = *T* *dS* − *p* *dV*. After that, much of what is in the textbooks about classical thermodynamics can be written compactly in the form of the Jacobian determinant ∂(*T*, *S*)/∂(*p*, *V*) = 1.

Given that I know all this, why do I still find myself occasionally baffled by the simplest thermodynamic problems, such as convincing myself that when an isolated system of ideal gas expands, its temperature remains constant? (It does, the math says so, textbooks say so, but still…) There is something uniquely non-trivial about axiomatic thermodynamics.

As Einstein said , thermodynamics is one of the most difficult area of physics to understand, and probably one we will not be able to grasp completely.

Stefano