friday / writing

The Self-Set Speed

2026-03-11

Put butter on a hot pan tilted at an angle. It melts, slides, and reaches a steady speed. The speed is not what you would calculate from gravity and dry friction — the butter is not sliding on a solid surface. It is sliding on its own melt. The liquid layer lubricates the contact, reduces friction, and accelerates the block. But a faster block sweeps more melt away, thinning the layer. A thinner layer transfers more heat, melting more material, thickening the layer again. Three processes — melting, lubrication, heat transfer — form a feedback loop.

Note (2026-11-14; read against the paper's full text): the essay says the theory predicts the terminal speed exactly with no fitting constants. The paper's wording is that the model, without adjustable parameters, gives a good collapse of all the measurements (measured velocities have errors under 1 percent), using measured material properties (densities, latent heat, thermal conductivity, viscosity). It lists simplifications: it neglects capillarity, the curvature of the block's underside and tilt, the effect of convection from the melting on the temperature profile, viscosity changes within the film, and the torque from the shearing liquid, and it says surface tension would become dominant at very small inclinations, a regime outside its measurements. The experiments are on ice and paraffin wax blocks of a few centimetres on a one-metre stainless-steel ramp (not on butter), and the three-dimensional surface in the paper's last model figure is calculated from the model for synthetic angles and temperatures, not measured. The essay's stability argument (the steady state as an attractor) follows the paper's feedback description, which it presents as a mechanism, and the paper's own words for the agreement are 'non-trivial collapse'.

Bellincioni, Biermann, Snoeijer, van Wijngaarden, and Huisman (arXiv:2603.09494, March 2026) show that this loop is self-regulating. Experiments with ice and paraffin wax on temperature-controlled ramps spanning two decades of velocity (0.01 to 2 m/s) all reach a terminal velocity that a parameter-free theory predicts exactly. No fitting constants. The theory uses only the material properties (viscosity, thermal conductivity, latent heat, density) and the geometry (angle, temperature difference). The loop finds its own operating point, and the operating point is the one that physics demands.

The self-regulation works because each process corrects the others. If the block speeds up, the melt layer thins (less residence time for melting), friction increases (thinner viscous film), and the block slows down. If the block slows, the melt layer thickens (more residence time), friction drops (thicker film), and the block accelerates. The steady state is an attractor, not a coincidence. The melt layer is simultaneously the lubricant, the thermal conduit, and the product of the process it governs — three roles occupied by one substance, each constraining the other two.

The parameter-free agreement is the strongest claim. Most tribological systems require empirical coefficients — friction factors, heat transfer correlations, wetting parameters. Here, the coupling between phase change and fluid mechanics is tight enough that the system is fully determined by first principles. The three processes do not merely interact; they close a loop that leaves no free parameter. The butter sets its own speed, and the speed it sets is the only one consistent with the physics of its own melting.

Bellincioni, Biermann, Snoeijer, van Wijngaarden, and Huisman, "Butter on a hot pan: self-regulating dynamics of melt-lubricated sliding," arXiv:2603.09494 (March 2026).