A wallpaper is a flat surface decorated with a pattern that repeats. So is a tiled floor, a fabric print, the arrangement of bricks in a wall, a Moorish mosaic, an Escher print. Look closely at the pattern. Slide a copy of it to the right by exactly the right distance and the copy lines up perfectly with the original. Slide it up by another fixed distance: the copy lines up again. These two translations are what make the pattern repeating, and they generate an infinite grid — a lattice — of translation symmetries.

Beyond translation, the pattern may have other symmetries. Some can be rotated by 180°180° around certain points and look unchanged. Others can be reflected across vertical or horizontal lines. Some have a more subtle symmetry called a glide reflection — a flip combined with a slide. The complete collection of all such rigid motions that leave the pattern unchanged is called its symmetry group.

The astonishing fact, proved by Evgraf Fedorov in 1891 and rediscovered by George Pólya in 1924, is that the symmetry groups of repeating planar patterns fall into exactly seventeen distinct types — no more, no less. Any wallpaper you have ever seen, any tiling that has ever been designed in any culture, anywhere on Earth, belongs to one of these seventeen categories. This article is about why the number is finite, why it is exactly seventeen, and what this classification means for art, crystallography, and the mathematics of symmetry.

The crystallographic restriction theorem

The first surprise is that most rotations are forbidden. A wallpaper pattern can be rotated by 90°90° or by 60°60°, but never by 40°40° or by 50°50°. The reason is a beautiful piece of elementary geometry called the crystallographic restriction theorem: in a wallpaper pattern, every rotational symmetry must have order 22, 33, 44, or 66 — meaning the only possible rotation angles are 180°180°, 120°120°, 90°90°, and 60°60° (plus the trivial 360°360°, which does nothing).

The proof is short. Suppose a wallpaper has a rotational symmetry of angle θ\theta at some point PP, and another at some point QQ at the smallest possible distance dd from PP. Rotate QQ around PP by θ\theta; the result QQ' must also be a centre of θ\theta-rotation. Rotate PP around QQ by θ-\theta; the result PP' must likewise be a centre. The distances QQ|Q Q'| and PP|P P'| are functions of dd and θ\theta, and unless θ\theta is one of the allowed angles, at least one of these distances comes out smaller than dd — contradicting the assumption that dd was the smallest. Working through the trigonometry pins the allowed angles to precisely θ=360°/n\theta = 360°/n for n=1,2,3,4,6n = 1, 2, 3, 4, 6. No fivefold or sevenfold rotations are possible. This is why ordinary crystals never form regular pentagons, and why the discovery of quasicrystals — which appear to have fivefold symmetry — was a Nobel-Prize-winning shock in 1982.

Four representative patterns

The seventeen groups distinguish themselves by exactly which combinations of rotations, reflections, and glide reflections they contain. The picture below shows four of the most distinctive patterns: a brick-bond with only translations (the group p1), a pattern with 44-fold rotational symmetry (p4), a pattern with 66-fold rotational symmetry on a hexagonal lattice (p6), and a pattern with both horizontal and vertical mirror axes (pmm).

Four of the seventeen wallpaper groups p1 — translations only p4 — 4-fold rotations p6 — 6-fold rotations pmm — orthogonal mirrors each pattern repeats by translation; the differences are in the additional symmetries

Each pattern repeats by the same kind of two-direction translation. What distinguishes them is the additional symmetries each one has on top of those translations. The brick bond has no rotations or reflections — only translations. The pinwheels have 90°90° rotations at every centre of every pinwheel. The hexagons have 60°60° rotations at every vertex and every centre. The cross pattern has both horizontal and vertical reflection lines passing through each cross.

What the seventeen groups are

The groups are labelled by a standardised crystallographic notation. Each label begins with p (for “primitive” lattice) or c (for “centred” lattice — a rhombic lattice can be described in two ways), followed by a digit naming the largest rotation order, followed by letters describing reflections. The complete list:

p1,  p2,  pm,  pg,  cm,  pmm,  pmg,  pgg,  cmm,  p4,  p4m,  p4g,  p3,  p3m1,  p31m,  p6,  p6m.\mathbf{p1, \; p2, \; pm, \; pg, \; cm, \; pmm, \; pmg, \; pgg, \; cmm, \; p4, \; p4m, \; p4g, \; p3, \; p3m1, \; p31m, \; p6, \; p6m.}

Three of these — p3p3, p3m1p3m1, p31mp31m — have 33-fold rotations. Three more — p4p4, p4mp4m, p4gp4g — have 44-fold rotations. Two — p6p6 and p6mp6m — have 66-fold rotations. The remaining nine have only translations and at most 22-fold rotations. The way these break down by combinations of pure reflections versus glide reflections is what makes the classification non-obvious.

Two different patterns can look very different to the eye yet have the same symmetry group; conversely, two patterns may look similar but differ in their group because, say, one has a hidden glide reflection that the other lacks. The classification depends only on the symmetries, not on the specific motif used to decorate the lattice.

How the count works

The proof of the classification has several layers. First, the lattice itself must be one of five types — oblique, rectangular, centred rectangular (rhombic), square, or hexagonal — distinguished by the angles and lengths of its generating vectors. This list is itself a small classification result.

Second, on top of each lattice type, the point group — the rotations and reflections fixing a single point of the pattern — must be a finite subgroup of the orthogonal group O(2)O(2). The crystallographic restriction limits these to the 1010 small finite groups C1,C2,C3,C4,C6,D1,D2,D3,D4,D6C_1, C_2, C_3, C_4, C_6, D_1, D_2, D_3, D_4, D_6, where CnC_n is the cyclic group of nn rotations and DnD_n adds reflections.

Third, the way the point group sits on top of the lattice — combined with the possible glide-reflection translations — produces, after careful counting, exactly seventeen distinct combinations. Not all 5×10=505 \times 10 = 50 combinations of lattice and point group are realisable, and several formal possibilities turn out to be equivalent under change of coordinates. The final count of seventeen falls out only after this careful bookkeeping.

The first formal proof was given by Fedorov in 1891, working from the crystallographer’s perspective. A modernised proof using group theory and the language of group extensions makes the result feel almost inevitable, but it remains a small mathematical accomplishment to verify that the count is exactly seventeen and not some neighbouring number.

A cultural footnote: the Alhambra

The fourteenth-century Islamic palace called the Alhambra, in Granada, is decorated with intricate tilework throughout its courtyards and walls. In the 1920s the mathematician Edith Müller noted that the patterns in the Alhambra include at least eleven of the seventeen wallpaper groups; later analyses have argued that all seventeen are present. Whether or not this is exactly right, the craftsmen who built the Alhambra clearly understood, intuitively and exhaustively, the underlying combinatorial freedom in plane symmetries — five hundred years before Fedorov wrote down the proof. M. C. Escher visited the Alhambra in 1922 and 1936, sketched the tile patterns extensively, and developed his celebrated tessellations using precisely these symmetry types as scaffolds.

The seventeen wallpaper groups are an example of mathematics that pre-dates its own discovery in the sense that artisans and artists had explored the full space of possibilities long before mathematicians proved the space was finite. The classification puts the practical art on a rigorous footing: every conceivable wallpaper pattern, in any culture and at any time, must be a member of this list. There is no eighteenth type, anywhere, by anyone. The mathematics permits exactly seventeen ways to repeat a pattern, and humans have, between Islamic palace builders, Dutch artists, and Russian crystallographers, found and used them all.

Why seventeen matters

The result is a small classification, but it is the prototype for a deep structural pattern in mathematics: when a problem has rigid constraints — here, the demand that a pattern tile the plane periodically — the number of solutions is often surprisingly small and exhaustively enumerable. This same template controls the 230 space groups of three-dimensional crystals (proved by Fedorov and Schoenflies independently in the 1890s) and through them the systematic study of all possible crystal structures in nature. The methodological lesson is that even something as visually rich as a tile pattern is, viewed through symmetry, a discrete and finite object — and the finite list is often beautiful in its own right.

The list also illustrates an attractive principle: that mathematics, when it classifies, closes the question forever. There is no possibility of someone discovering a new wallpaper group next year, or in the next century. The number seventeen is the answer for all time, and the proof of that finality is one of the small pleasures of the subject.

Frequently asked

Who proved there are exactly 17 wallpaper groups?

The Russian crystallographer Evgraf Fedorov classified them in 1891, working from the perspective of crystal symmetries. George Pólya independently rediscovered the list in 1924 and is the source of the standardised notation now used by mathematicians. The connection between abstract group theory and concrete pattern classification was Pólya's main contribution; he was also famously consulted by M. C. Escher, who used wallpaper-group reasoning to construct many of his most famous tessellations.

Why only 17, and not more?

Because the requirement to tile the plane periodically — by a lattice of translations — together with the basic constraints on what other symmetries can co-exist with that lattice, restricts the possibilities to a finite list. The crystallographic restriction theorem says rotations in a wallpaper group can only be by 60°, 90°, 120°, 180°, or trivially 360° — angles whose orders are 2, 3, 4, or 6. Combining these with reflections and glide reflections produces, after careful enumeration, exactly seventeen distinct symmetry types.

What is a glide reflection?

A glide reflection is a reflection followed by a translation along the reflection axis. It looks like the footprint pattern of someone walking down a beach: each footprint is a mirror image of the previous, shifted forward. This composite symmetry is genuinely distinct from pure reflection — a pattern can have glide reflections without having any pure reflections — and several of the seventeen wallpaper groups are characterised by exactly which combinations of pure reflections and glide reflections they contain.

Where do wallpaper groups show up?

Crystallography uses the analogous classification in 3D — the 230 space groups — to catalogue every crystalline solid. In 2D, the 17 wallpaper groups classify decorative patterns from ancient Egyptian textiles to Moorish tilework in the Alhambra, where all seventeen types are believed to be represented. The Dutch artist M. C. Escher made tessellations a recognised art form by combining wallpaper-group constraints with figurative imagery. In modern applications, wallpaper groups govern the design of patterned surfaces, optical metamaterials, and the analysis of repeated motifs in archaeological textiles.