Penrose Tilings
Two tiles that never repeat
Take a floor and cover it with copies of a single shape — square tiles, or hexagons — and the
pattern repeats: slide the whole thing over by one tile and it lands back on
itself. That sliding-repeat is called being periodic, and almost every tiling
you have ever seen is periodic.
In 1974 the mathematician (and later Nobel laureate) Roger
Penrose found something astonishing: a way to tile the whole infinite plane using just
two tile shapes, with no gaps and no overlaps, such that the pattern
never repeats — there is no slide at all that maps the tiling back onto itself.
These are the Penrose tilings, and they are the most famous
aperiodic tilings in mathematics.
The secret is not the tiles alone but the matching rules — markings on the
edges that say which edge may touch which. With those rules the two tiles can still
cover the plane, but only non-periodically. Hidden inside every Penrose tiling
you find five-fold symmetry (the symmetry of a five-pointed star) and the
golden ratio
\varphi — the very fingerprints a repeating tiling is forbidden to
have.
The two shapes: kite & dart, or two rhombs
Penrose tilings come in two standard tile sets, and both are carved straight out of the
regular pentagon, whose
corner angle is 108^\circ:
-
the kite and dart — two four-sided pieces you get by cutting up a pentagon;
and
-
the two rhombs — a thick rhombus and a
thin rhombus. This is the set we will focus on, because its angles are the
easiest to read off.
A rhombus is a "pushed-over square": all four sides equal, opposite angles equal. So each rhomb
is described by just one number — its sharp (acute) corner angle — and the other corner is
whatever is left to make a straight 180^\circ along the side.
Worked example 1 — the angles of the two rhombs
Every angle in a Penrose rhomb is a multiple of 36^\circ — one fifth
of a half-turn — because 36^\circ = 180^\circ / 5 is the pentagon's
natural unit.
The thick rhombus. Its sharp corner is
72^\circ (that is 2 \times 36^\circ). The
other corner completes the straight line beside it:
180^\circ - 72^\circ = 108^\circ.
So the thick rhombus has angles 72^\circ,\,108^\circ,\,72^\circ,\,108^\circ
— and 108^\circ is exactly the interior angle of the regular pentagon.
The thin rhombus. Its sharp corner is
36^\circ (a single 36^\circ), so its blunt
corner is
180^\circ - 36^\circ = 144^\circ.
The thin rhombus therefore has angles
36^\circ,\,144^\circ,\,36^\circ,\,144^\circ. Notice every one of the
eight angles — 36, 72, 108, 144 — is a whole number of
36^\circ steps. That is what lets so many of them fit snugly around a
point.
Worked example 2 — the tile ratio is the golden ratio (and why that forbids repeating)
Count the tiles in a large Penrose patch. As the patch grows, the ratio of
thick rhombs to thin rhombs settles down to a single fixed
number — the
golden ratio:
\frac{\text{number of thick rhombs}}{\text{number of thin rhombs}} \;\longrightarrow\; \varphi = \frac{1 + \sqrt{5}}{2} \approx 1.618\ldots
(The kite-and-dart set does the same: the ratio of kites to darts also tends to
\varphi.) Now here is the punchline. The golden ratio is
irrational — it can never be written as one whole number divided by
another.
Suppose, for a moment, the tiling were periodic. Then it would have a repeating
unit cell — a fixed block that copies across the plane — and that block would
contain some whole number of thick rhombs, say a, and some whole
number of thin rhombs, say b. The overall ratio would then have to be
the fraction
\frac{a}{b} \quad\text{(a ratio of two whole numbers — i.e. rational).}
But we just said the ratio is \varphi, which is
irrational. A number cannot be both a whole-number fraction and irrational, so
the assumption breaks. There is no unit cell — the tiling
cannot be periodic. The golden ratio quietly proves the aperiodicity all by
itself.
It gets stranger still. There are uncountably many genuinely different
Penrose tilings — infinitely more of them than there are whole numbers. Yet they are
locally indistinguishable: take any finite patch out of one Penrose
tiling, however large, and that exact patch appears somewhere in every other Penrose
tiling too. You could never tell which tiling you were standing on by looking at a finite area
— you would have to see the whole infinite plane. Two tiles, and out of them tumbles endless,
bottomless variety.
See the two rhombs and a five-fold star
Step through the figure. First meet the thick rhombus with its
72^\circ / 108^\circ angles, then the thin rhombus
with its 36^\circ / 144^\circ angles. The coloured corner dots stand
for the matching rules — only like-coloured corners are allowed to meet.
Finally, watch five thick rhombs slot around a single point:
5 \times 72^\circ = 360^\circ closes the circle exactly, and a
five-fold star — forbidden to any repeating tiling — springs into being.
Why five-fold symmetry forces aperiodicity
A repeating (periodic) tiling can only have rotational symmetry of order
2, 3, 4 or 6 — that is the
crystallographic
restriction. Five-fold symmetry is precisely the one it forbids: you simply cannot
make a repeating pattern that looks the same after a 72^\circ turn.
But Penrose tilings are full of local five-fold (and ten-fold) symmetry — the
star you just built is only the simplest example. A pattern that shows a symmetry no repeating
pattern is allowed to have must not be repeating. So five-fold symmetry is a
second, independent proof that Penrose tilings are aperiodic — arriving at the same verdict as
the golden-ratio argument.
-
it uses just two tile shapes — the kite and dart, or the
thick (72^\circ/108^\circ) and
thin (36^\circ/144^\circ) rhombs;
-
the edge matching rules are essential — they let the tiles cover the plane
but only aperiodically (no slide maps the tiling onto itself);
-
it shows local five-fold symmetry, which the
crystallographic restriction
forbids to any periodic tiling;
-
the ratio of the two tiles tends to the golden ratio
\varphi \approx 1.618 — an irrational number, which itself rules
out any repeating unit cell.
A little history, and where this leads
Penrose tilings began as a beautiful mathematical curiosity, but in 1982 Dan
Shechtman discovered real crystals — quasicrystals — whose atoms are arranged in
exactly this kind of ordered-but-non-repeating way, with the "impossible" five-fold symmetry.
The discovery was so contrary to textbook crystallography that it took years to be accepted; it
won Shechtman the Nobel Prize in Chemistry in 2011.
The ideas here open onto a whole subject: how a Penrose tiling can be grown by
substitution and inflation
(repeatedly subdividing each tile), how it can be produced as a slice of a higher-dimensional
lattice by the
cut-and-project method,
the modern hunt for
aperiodic monotiles
(a single tile that tiles only aperiodically), and the physics of
quasicrystals.
-
Aperiodic does not mean random. A Penrose tiling is completely
ordered and deterministic — every tile is placed by the matching rules (or,
equivalently, by substitution). "Aperiodic" means only that there is no
translational repeat: no slide maps the whole tiling onto itself. It is
structured, not haphazard — nothing like scattering tiles at random.
-
The matching rules are essential. Without them, those very same two rhombs
will also tile the plane periodically — you can lazily line them up
into a boring repeating pattern. It is the edge markings that forbid the periodic
arrangements and leave only the aperiodic ones. Two tiles alone are not enough; two tiles
plus the rules are the whole trick.