The work output of a thermodynamic cycle is the area enclosed by the curve in the pressure-volume diagram. This is a global property — you need the whole loop to compute it. A Carnot engine running between two temperatures performs work proportional to the area of its rectangle in PV space. You cannot ask how much work is being done “at” a particular point on the cycle without reference to the rest of the path.
Note (2026-11-12): checked against the paper's full text (Bittner). The main results match: the area laws for work and reversible heat are projections of a single canonical two-form on the equilibrium manifold, and the work of an infinitesimal cycle is set locally by the mixed curvature U_SV, expressible through measurable susceptibilities. The Jarzynski statement is narrower in the paper: for closed cycles, or protocols completed by reference paths on the equilibrium manifold, the work reduces to a projected oriented area, and the Jarzynski average is then written as an ensemble average of the exponential of the oriented area in the (P, V) plane, equal to exp(-beta times the free-energy difference); it is not described as an average of local curvature contributions. The paper also frames the analysis as classical equilibrium thermodynamics that connects to stochastic trajectories 'naturally', and calls its aim to identify the structure that unifies area laws, response and work functionals; the essay's remark that some cycle segments contribute more curvature than others is its own gloss (the density is over the enclosed area).
Bittner (arXiv:2603.22559) shows that you can. The area in the PV diagram and the area in the TS (temperature-entropy) diagram are both projections of a single canonical two-form on the thermodynamic state manifold. This two-form is a local geometric object — it has a value at every point. The work done by an infinitesimal cycle is proportional to the mixed curvature of the equilibrium energy surface, U_SV, which can be expressed entirely in terms of measurable susceptibilities: heat capacity, compressibility, thermal expansion.
The reframing converts a global integral (area enclosed by a finite cycle) into a local field (curvature at a point). Any sufficiently small cycle encloses an amount of work proportional to the local curvature. Large cycles are built up from these local contributions, but the contributions are not equal — the curvature varies across the state space, concentrating work generation in regions where the energy surface bends most sharply.
What this reveals is that the PV diagram was always a projection that obscured locality. The work wasn't distributed uniformly around the cycle. Some segments contributed more curvature than others. The integral hid this, reporting only the total. The geometry recovers what the diagram collapsed.
The connection extends to nonequilibrium paths. Stochastic trajectories on the thermodynamic manifold generate work along each step, and the Jarzynski equality — which relates free energy differences to the exponential average of work along irreversible paths — acquires a geometric interpretation as the average of these local curvature contributions across an ensemble of fluctuating trajectories.
The area was never just arithmetic. It was geometry all along.