Soap Bubbles & Minimal Surfaces: Physics Doing Calculus of Variations

Surface tension is an area-minimizer, not a shape-drawer

A soap film's energy is directly proportional to its surface area — every extra square millimetre of film costs a fixed amount of surface-tension energy, regardless of its shape. Left alone, any physical system relaxes toward lower energy, so a soap film spanning a fixed boundary (a bent wire loop, say) settles into whichever shape has the least possible area among all surfaces that share that boundary. This is why soap films look effortlessly elegant: they are not decorative, they are the physical embodiment of a solved calculus-of-variations problem, found by nature in real time through nothing more complicated than molecules pulling on their neighbours.

Zero mean curvature

The mathematical condition for a surface to be area-minimizing (more precisely, area-critical — a local minimum, maximum or saddle of area) is that its mean curvature H, the average of the two principal curvatures at every point, must be exactly zero everywhere:

H = (κ₁ + κ₂) / 2 = 0        (minimal surface condition, at every point)

κ₁, κ₂ = the two principal curvatures at a point (max and min curvature, in
         perpendicular directions) — for a minimal surface they are equal
         and opposite: the surface curves up one way and down the other by
         exactly the same amount, like a saddle

A flat plane trivially satisfies H = 0 everywhere, which is why a flat soap film across a flat loop is a valid (if boring) minimal surface. More interesting boundaries force the film into genuinely curved shapes that still balance out to zero mean curvature everywhere — the mathematics of Jean Baptiste Meusnier, who first derived this condition in 1776, connecting an intuitive physical observation to differential geometry more than a century before soap films were studied experimentally in earnest.

Classical examples: catenoid, helicoid, saddle

The simplest non-planar minimal surface is the catenoid, the shape a soap film forms between two parallel coaxial rings — a waist that narrows toward the middle, generated by rotating a catenary (the hanging-chain curve) around an axis. Pull the two rings too far apart, though, and the catenoid becomes unstable and collapses into two separate flat discs, one per ring — a striking demonstration that "minimal" only means locally minimal; beyond a critical separation, the disconnected two-disc configuration has less total area. The helicoid, generated by spiralling a straight line as it moves along an axis (the shape a spiral staircase's ramp approximates), is the other classical example, discovered by Jean Baptiste Meusnier alongside the catenoid, and it is famously the only known minimal surface (besides the plane) that is also a ruled surface — swept out entirely by straight lines. A saddle-shaped surface between four alternating high and low boundary posts is the simplest way to see H = 0 by eye: it curves upward along one direction and downward along the perpendicular direction, in exactly matching amounts.

Plateau's laws

Joseph Plateau, a blind Belgian physicist, spent decades in the 19th century systematically studying soap films and distilled his observations into rules that any configuration of connected films must obey — later proven mathematically by Jean Taylor in 1976:

1. Films are smooth surfaces of constant mean curvature between junctions.
2. Exactly THREE film sheets meet along any edge, always at 120° to
   each other (never 2, never 4+) — force balance between equal
   surface tensions demands it.
3. Edges (where three sheets meet) themselves meet in FOURS at a
   point, always at the tetrahedral angle ≈ 109.47°.

These rules are why a cluster of soap bubbles never shows a random tangle of film angles — every junction you can find in a real bubble cluster obeys the 120°/109.47° rules exactly, because any other configuration has an unbalanced tension force that immediately reshapes the junction until it settles into the stable angle.

Numerically finding a minimal surface

For small deviations from flat, the minimal surface equation linearises into the ordinary Laplace equation, ∇²h ≈ 0, where h(x, y) is the film's height above a reference plane — the same equation that governs steady-state heat conduction and electrostatic potential. Solving it numerically on a mesh, given fixed boundary heights, is a classic relaxation problem: repeatedly replace each interior point's height with the average of its neighbours until the whole grid stops changing.

// Jacobi relaxation toward ∇²h = 0 on an interior grid point
for (let iter = 0; iter < maxIterations; iter++) {
  for each interior point (x, y):
    hNext[x][y] = 0.25 * (h[x-1][y] + h[x+1][y] + h[x][y-1] + h[x][y+1]);
  // boundary points (the wire frame) stay FIXED every iteration
  [h, hNext] = [hNext, h];
}

For surfaces with genuinely large curvature (a true catenoid, not a small perturbation of flat), the full nonlinear minimal-surface equation is needed instead, but the underlying idea is the same: iteratively nudge every interior point in the direction that locally reduces total area, and stop when the surface stops moving — a discretised, computational version of exactly what a real soap film does physically, continuously, and in an instant.

Frequently asked questions

Try it live

Everything above runs in your browser — open Soap Bubble and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.