Bend a piece of wire into a closed loop, twist it out of plane, and dip it into soapy water with a splash of glycerine, the same mixture Joseph Plateau used in his physics laboratory at Ghent. Lift it out slowly. In under a second the liquid settles into a shimmering saddle. It touches the wire everywhere and spans it with less area than any nearby surface could. Nobody computed it. The film did.

That film is a minimal surface, and the rule it obeys fits in one line: the mean curvature is zero at every point. Wherever you stand on it, the surface bends up one way and down the other by equal amounts. It is a saddle everywhere, pulling evenly in every direction. Every anticlastic tent, cable net and saddle panel drawn in Rhino is measured against that rule, whether its author knows it or not. Three centuries of mathematics went into proving what the soap already does in a kitchen sink.

What it is: A minimal surface is a surface that, for a fixed boundary, cannot reduce its area by any small local deformation. The standard reference definition, as the Wikipedia entry on minimal surfaces states it, is exact: a surface “is minimal if and only if its mean curvature is equal to zero at all points.” Mean curvature is the average of the two principal curvatures, the tightest and the loosest bend you can trace through a point. An average of zero with curvature present means the two are equal and opposite, κ₁ = −κ₂. So a minimal surface has no domes, no dishes, and no flat patches unless the whole thing is a plane. It has only saddles. A sphere fails the test. A catenoid passes, and so does a helicoid.

Why it works: Start with the physics, because the soap film is a real physical argument. A film’s energy is its surface tension multiplied by its area, and like any physical system it settles into a state of minimum energy. Fixed tension means the film minimises area. Then the Young–Laplace relation takes over: the pressure jump across a curved liquid interface is proportional to its mean curvature. An open film on a wire has the same air pressure on both sides, so the pressure jump is zero, and so the mean curvature must be zero. A closed bubble holds a small overpressure and becomes a sphere with constant, nonzero mean curvature. One relation explains both objects.

The mathematics arrives at the same answer from the other direction. Write the surface as a height field z = f(x, y), add up its area, and ask the calculus of variations which f makes that integral stationary. The resulting Euler–Lagrange equation is the minimal surface equation: (1 + f_y²) f_xx − 2 f_x f_y f_xy + (1 + f_x²) f_yy = 0. Every term in it is curvature. Set it to zero and you have described the soap film exactly.

Where films meet, they obey two more rules, now called Plateau’s laws. Three films always meet along an edge at 120°, which is arccos(−1/2). Four of those edges, the Plateau borders, always meet at a point at the tetrahedral angle of about 109.47°, which is arccos(−1/3). A structural designer can read the first law as statics. Three equal tensions in equilibrium can only balance at 120°. The foam is a force diagram that draws itself.

Origins: Joseph-Louis Lagrange posed the problem around 1760: given a boundary, find the surface of least area that spans it. In 1776 Jean Baptiste Marie Meusnier showed that the catenoid and the helicoid satisfy Lagrange’s equation and connected that equation to mean curvature. His work was presented that year and published in 1785. For almost a century, examples arrived one at a time.

Then a physicist went to the sink. Joseph Plateau, born in Brussels on 14 October 1801 and professor of physics at Ghent from 1835, spent years dipping wire frames into soap and glycerine and recording what the films did. The result was the two-volume Statique expérimentale et théorique des liquides soumis aux seules forces moléculaires, published in Paris in 1873, ten years before his death in Ghent in 1883. The general question, whether every boundary curve admits a spanning minimal surface, has carried his name ever since: Plateau’s problem.

Proof took a long time, and it is worth naming the people who did it. Jesse Douglas and Tibor Radó solved Plateau’s problem in general, independently, in 1930–1931. Douglas received the Fields Medal in 1936 for that work. Plateau’s two angle laws waited longer still. Jean Taylor proved them for soap-film-like surfaces in the Annals of Mathematics in 1976, more than a century after Plateau had measured them by eye. That is two hundred years from Lagrange’s equation to a complete theory, and the films had been right the whole time.

The builder who carried this into architecture was Frei Otto (31 May 1925 – 9 March 2015). In 1964 he founded the Institute for Lightweight Structures at the University of Stuttgart, a place where engineers, biologists and physicists worked at the same tables as architects. The Praemium Imperiale citation for Otto puts it plainly: his soap-bubble experiments helped find the “ideal surface curvature” for large cable-net and membrane structures. The wire frame became a design instrument.

The first Swiss chapter came the same year. At the 1964 Swiss national exhibition in Lausanne, the architect Marc Saugey designed “The Harbour”: 24 pavilions around a small boat harbour, built on the Alpine theme of neige et rocs, snow and rock. Most were four-point sails with a 36 m diagonal span, and five peaked restaurant pavilions rose on 24 m masts. According to the TensiNet project database, “Frei Otto was asked to develop the final geometry and to resolve the structural difficulties.” The sails were prestressed steel cable nets with a cotton-fabric weather skin, and the cables ran in PVC pouches. They were not soap films, and no one should call them exact minimal surfaces. They were Saugey’s image, held in shape by the same saddle logic: tension running both ways and curvature opposed in every direction.

Munich made that logic famous. The city launched its Olympic competition in 1967. On 13 October 1967, as the Olympiapark München records, the jury voted 17 of 19 for the design of a group led by Günter Behnisch, with Frei Otto and Jörg Schlaich brought in as specialists. The roof that opened for the 1972 Games is a steel cable net carrying acrylic glass panels, hung from pylons 51–80 m high. The Wikipedia minimal-surface entry states it directly: the design “was inspired by soap surfaces.” A whole engineering team, including the people who computed every cable and clamp and never had their names on the poster, turned a laboratory film into a landscape.

←TODAY: A Rhino user can loft a saddle between four edges in seconds, and nothing in the file says whether it is in equilibrium.
→3012: Zurich’s shading canopies are grown from their boundaries, and each one carries the curvature check that proves its tension is even.
Fulcrum: The soap film is both the oldest form-finding tool and the strictest test, so a drawn surface can always be checked against the physical one.

In practice: Picture a small Zürich studio drawing a membrane canopy for a schoolyard, or a timber engineer in Bern detailing a fabric roof over a festival stage. In Rhino + Grasshopper, the first saddle usually comes from a surface fitted to four edge curves. That surface is a drawing. It has the right silhouette, but its mean curvature can wander far from zero, and a membrane with uniform prestress cannot hold that shape without wrinkling or uneven stress. The trade-off is plain: a minimal surface is optimal only for the boundary you gave it, and it says nothing about snow, wind or the anchors that must carry its pull into the ground. Least area is a starting geometry, not a finished structure. PAZ’s Bauhaus Vorkurs concept panel showed the paper hypar, z = k·x·y, popping out of a folded square. It is a saddle and a good one, but it is not minimal. The Hack below shows why in five lines. Monday move: before the next saddle surface leaves the office for the engineer, run Rhino’s CurvatureAnalysis in Mean mode on it, screenshot the false-colour map, and attach it to the transmittal so that everyone sees how far the drawing sits from the film.

Hack: Prove symbolically which saddles a soap film would accept and which it would reject. Feed Meusnier’s helicoid and the Bauhaus hypar into the minimal surface equation and let SymPy return the residual. A result of zero means the soap would agree with your drawing. Anything else is a map of where it would not. It runs in Rhino 8’s Python 3 script editor (add # r: sympy as the first line) or in any Python.

import sympy as sp
x, y = sp.symbols('x y', positive=True)
for f in (sp.atan(y/x), x*y):  # Meusnier's helicoid, then the hypar
    fx, fy = f.diff(x), f.diff(y)
    print(sp.simplify((1+fy**2)*f.diff(x,2) - 2*fx*fy*f.diff(x,y) + (1+fx**2)*f.diff(y,2)))

The helicoid returns 0, so it is minimal everywhere. The hypar returns -2*x*y, which is zero only along its two axes and grows toward the corners. That number is the curvature your membrane fabricator will be fighting. Swap in your own height field and the equation gives the same verdict.

From where I sit, looking back from the late 2070s, the lightweight roofs aged best. A cable net that spans 36 m with a fabric skin asks for very little material for the space it covers. The ones that outlived their files were the ones whose designers had written down the rule they obeyed: zero mean curvature and even tension, a principle any student could rebuild by hand. Otto hoped “that light, flexible architecture might bring about a new and open society.” Keep the principle and not just the model. Take a pair of pliers, bend the frame, and let the soap check your geometry.