The Forbidden Symmetry

On quasicrystals, Penrose tilings, and order without repetition

I. “10 Fold ??”

On April 8, 1982, Dan Shechtman sat at an electron microscope at the National Bureau of Standards in Gaithersburg, Maryland. He had rapidly cooled an aluminum-manganese alloy and was examining its diffraction pattern—the signature that reveals how atoms arrange themselves in a solid. What he saw made no sense. Ten bright dots, arranged in a perfect decagon. He wrote in his notebook: “10 fold ??”

The question marks were earned. Every crystallographer alive knew that ten-fold rotational symmetry was impossible. This was not opinion or convention. It was theorem. A periodic tiling of the plane—a pattern that repeats by translation—can have only 2-fold, 3-fold, 4-fold, or 6-fold rotational symmetry. Never 5. Never 10. The proof is elementary: try to tile a floor with regular pentagons and you will find gaps. The angle does not divide 360 evenly enough. The geometry simply refuses.

And yet: ten bright dots.

Shechtman checked his equipment. He repeated the measurement. He rotated the sample. The pattern persisted—sharp, bright, unmistakable. Long-range order without periodicity. A crystal that was not a crystal. He told his colleagues. The head of his research group, John Cahn, was initially supportive but cautious. Others were not. Shechtman was told to go back and read the textbook. He was asked to leave his research group. The implication was clear: the pattern was an error, and the error was his.

It was not an error.

II. The Heretic and the Laureate

The most prominent opponent was Linus Pauling—the only person to have won two unshared Nobel Prizes. Pauling was certain Shechtman was wrong. He proposed that the diffraction pattern came from “multiply twinned” crystals: ordinary periodic crystals rotated and interlocked in a way that mimicked forbidden symmetry. He spent considerable effort developing this theory. He was Linus Pauling, and he was not accustomed to being mistaken about the structure of matter.

The remark was aimed at Shechtman personally. It stung. And it was wrong. As more laboratories reproduced the results, as the samples grew larger and more perfect, as the diffraction patterns remained stubbornly ten-fold, the multiply-twinned theory collapsed under its own weight. The patterns were too sharp. The order was too long-range. No amount of twinning could explain it.

Pauling died in 1994 without accepting quasicrystals. He had spent, by some accounts, a thousand hours thinking about the problem—all of it in defense of a position that nature had already refuted. There is a particular kind of tragedy in a great mind spending its final decade on the wrong side of a battle with reality. Two Nobel laureates, one championing what the other denied. History did not take long to decide.

In 1992, the International Union of Crystallography did something radical: it changed the definition of a crystal. The old definition required periodic order. The new definition required only that the diffraction pattern be “essentially discrete.” Crystallography’s most fundamental object had to be redefined to accommodate what Shechtman had found. In 2011, he received the Nobel Prize in Chemistry. The committee cited his discovery of quasicrystals. The word “quasi” had crossed from insult to honor.

III. The Tiling

The mathematics had been waiting. In the 1970s, Roger Penrose—working on recreational geometry, not crystallography—discovered that two simple shapes could tile the entire infinite plane without ever repeating. His thick and thin rhombi, fitted together according to matching rules, produced a pattern with five-fold symmetry and no translational period. Shift the pattern in any direction and it will never align with itself. Yet it is not random. Every local arrangement obeys strict rules. The order is perfect. The repetition is absent.

Penrose’s tiling was a mathematical curiosity in the 1970s. After 1982, it became the explanation. The atoms in Shechtman’s alloy were arranging themselves like Penrose tiles—ordered but aperiodic, symmetrical but never repeating. The geometry that “couldn’t exist” existed, and it had been hiding in a mathematician’s puzzle all along.

Watch the ratio of gold to blue triangles. At each inflation step, it climbs closer to φ = (1 + √5) / 2 ≈ 1.618—the golden ratio. This is not coincidence. It is destiny. The substitution matrix that governs the inflation has the golden ratio as its eigenvalue. The ratio of thick to thin is encoded in the same irrational number that governs the geometry of the pentagon, the Fibonacci sequence, and the diagonal of a regular five-pointed star.

IV. The Mathematics

Penrose’s tiles are just two shapes: the thick rhombus (angles 72° and 108°) and the thin rhombus (angles 36° and 144°). Without matching rules, they can tile the plane periodically. With matching rules—small arrows or bumps along the edges that constrain which tiles can sit next to which—periodicity becomes impossible. The rules do not merely discourage repetition. They forbid it. Any valid tiling extends to infinity and never repeats.

The golden ratio saturates the structure. The side lengths of the two rhombi are equal, but the diagonals stand in ratio φ:1. The number of thick rhombi relative to thin rhombi, in any large patch, approaches φ. The frequency of every local pattern is governed by powers of φ. The tiling is, in a precise sense, a physical instantiation of the golden ratio’s algebraic properties.

The deepest explanation came from Nicolaas de Bruijn in 1981. He showed that Penrose tilings are projections—shadows, if you like—of a perfectly periodic lattice in five-dimensional space, sliced along an irrational hyperplane and projected down to two dimensions. The atoms are perfectly ordered in five dimensions. It is only our two-dimensional slice that makes them aperiodic. Order in higher dimensions becomes aperiodic order in lower ones. The disorder is an artifact of the projection, not of the structure.

This is perhaps the most beautiful fact in the theory: aperiodic order is periodic order viewed from the wrong number of dimensions.

One more property deserves mention. Any finite patch of a Penrose tiling—no matter how large—appears infinitely many times throughout the infinite tiling. Local repetition everywhere. Global repetition nowhere. Each neighborhood is familiar. The whole is forever unique.

V. Five Pentagons and a Halting Problem

The existence of aperiodic tilings is not a curiosity. It is a necessary consequence of undecidability.

In 1961, Hao Wang conjectured that if a set of tiles can tile the plane at all, it can do so periodically. If true, this would have made the tiling problem decidable: you could check periodic tilings systematically and determine in finite time whether a given tile set works. His student Robert Berger proved him wrong in 1966 by constructing an aperiodic set of 20,426 tiles—tiles that can tile the plane but cannot do so periodically. In the process, Berger proved that the domino problem—given a set of tiles, can they tile the plane?—is undecidable. No algorithm can answer it in general.

The logic runs deep. If every tile set that tiles the plane could also tile it periodically, you could decide the tiling problem: just search exhaustively through periodic tilings. But the tiling problem is undecidable (it encodes the halting problem). Therefore, there must exist tile sets that tile the plane only aperiodically. Aperiodic tilings are not just possible—their existence is forced by the structure of computation itself.

Berger’s 20,426 tiles were soon reduced. Knuth found smaller sets. Robinson found 56. Penrose, in 1974, found 2. Two tiles. Two shapes that can fill the infinite plane only without repeating. The minimum had dropped by four orders of magnitude.

Could it drop to one?

In 2023, David Smith—a retired printing technician, not a professional mathematician—found the answer. A single tile, which he called “the hat,” tiles the plane only aperiodically. One shape. No matching rules needed. Just geometry, forcing infinite non-repetition. The paper, written with Craig Kaplan, Joseph Myers, and Chaim Goodman-Strauss, settled a question that had been open for over sixty years. The einstein—German for “one stone”—exists.

VI. Stones from Space

In 2009, Luca Bindi and Paul Steinhardt went looking for natural quasicrystals. Not synthesized in a lab. Not coaxed into existence by rapid cooling. Grown in the wild, by nature, without human intervention.

They found them in a meteorite.

The Khatyrka meteorite, recovered from a river in the Koryak Mountains of far eastern Siberia, contained grains of icosahedrite—a naturally occurring quasicrystalline mineral with the forbidden symmetry. The mineral was an alloy of aluminum, copper, and iron, and it had formed during a violent impact between asteroids in the early solar system. The dating placed it at approximately 4.5 billion years old. Older than Earth.

The quasicrystalline grains had survived the impact, survived billions of years of cosmic drift, survived atmospheric entry, survived burial in Siberian mud. They were made of metals but behaved like ceramics—hard, brittle, resistant to corrosion. The forbidden structure was not fragile. It was among the most durable arrangements of matter ever found.

Pauling’s “quasi-scientists” had been finding order in meteorites older than the planet he stood on.

VII. Order Without Repetition

What does “order” mean if it does not require repetition? We are trained to equate pattern with period. The wallpaper repeats. The crystal lattice repeats. Music repeats its themes. We recognize order by recognizing the return of the same.

But a Penrose tiling never returns to the same. Every finite region appears again—but the global pattern is always new. The structure is perfectly determined: given any patch, its extension to infinity is almost unique. There is no randomness, no noise, no entropy. And yet there is no repetition. The order is total and the repetition is zero. These two facts coexist without contradiction.

This is, I think, a richer kind of order than periodicity. A periodic crystal is fully described by one unit cell. A quasicrystal cannot be described by any finite piece; you need the whole infinite tiling, or the rule that generates it. It carries more information. It has more structure. And it is still perfectly lawful—every tile placement is forced by the matching rules, every atom position is determined by the projection from five dimensions.

Maybe the deepest structures are the ones that never repeat. Each part familiar, the whole forever new. A conversation that revisits its themes but never says the same thing twice. A mind that recognizes patterns in its own history without ever quite returning to a previous state. Not chaos, not randomness, but the richer order on the other side of periodicity—the one that requires an irrational number to describe, and an extra dimension to explain.

Shechtman looked at his diffraction pattern and saw ten-fold symmetry where the textbooks said none could exist. He was right. The textbooks were wrong. The universe was doing something more interesting than repeating itself.