The box and the rule
In the last guide we met the big idea: a crystal is a repeating pattern, and the smallest repeating chunk is the unit cell. Now let us actually build with it. Picture a single small box of atoms held in your hand. The instruction for making a crystal out of it is almost laughably simple: place an exact copy of the box directly against each face of the first, then keep going, filling all of space with identical boxes packed edge to edge, no gaps and no overlaps. Repeat to infinity. That is the whole construction. A crystal is one cell plus the order "copy me forever, in every direction."
To pin the box down with numbers, you need just six measurements. Three of them are the lengths of the box's edges, and three are the angles between those edges. The edge lengths are the lattice constants — in a cube all three are equal, but in a squashed or stretched box they can differ. The angles tell you whether the corners are square right angles or leaning. Those six numbers — three lengths, three angles — completely fix the shape and size of the repeating box, and so, together with what sits inside, they completely fix the crystal.
The leanest possible cell
Here is a subtlety worth slowing down for. There is often more than one way to draw the repeating box. You could choose a small, cramped box that contains the bare minimum, or a roomier box that is easier to look at. When you choose the *smallest possible* cell — the one holding exactly one lattice point's worth of stuff, with nothing to spare — it has a special name: the primitive cell. It is the most economical description, the tile with no waste, the true atom of repetition.
So why would anyone ever choose a bigger box? Because beauty sometimes beats economy. A larger cell can show off the crystal's symmetry — its squareness, its right angles — far more clearly than a lean, slanted primitive cell that technically does the same job. Crystallographers happily trade minimalism for clarity, picking the cell whose shape reflects the crystal's true symmetry even when a smaller, uglier one would also tile space. This is why you will see, for instance, a cube drawn with extra atoms at its centre or face-centres: the cube is not the smallest possible cell, but it is by far the clearest.
Why there are only so many ways to fill space
Now comes the surprising and beautiful part. You might guess that since boxes can be any shape, there are endless kinds of repeating grid. There are not. Nature is sharply limited here, and the limit comes from one stubborn requirement: the copies must tile space perfectly, leaving no gaps. Tiling is fussy. You can tile a floor with squares, triangles, or hexagons, but never with regular pentagons — try it and unfillable slivers always appear. The same fussiness, lifted into three dimensions, slashes the number of genuinely different repeating grids down to a small, countable list.
When mathematicians worked it out carefully in the nineteenth century, the count came to exactly fourteen. There are precisely fourteen distinct ways to arrange points in a repeating, space-filling grid in three dimensions. These fourteen are the Bravais lattices, named after the French scientist who completed the catalogue. Every crystal that has ever been found, or ever will be, has its lattice somewhere on this list of fourteen. It is one of those rare moments where the messy real world turns out to obey a short, exact rulebook.
The fourteen, sorted into seven families
Fourteen is small, but you can organise it even more kindly. The fourteen lattices fall into seven larger families, sorted by the shape of the box — how its edge lengths and corner angles compare. It helps to think of the families as a series of boxes growing steadily less special, from the most symmetric to the least.
- Cubic — a perfect cube: all three edges equal, all corners square. The most symmetric, most familiar box (salt, copper, diamond all live here).
- Tetragonal — a cube stretched along one direction into a square-based box, like a tall matchbox. Two edges equal, one different; corners still square.
- Orthorhombic — a shoebox: all three edges different, but corners still square. Less symmetric again.
- Hexagonal — a box with a six-sided, honeycomb-like footprint, the basis for many metals and for ice.
- Rhombohedral (trigonal), Monoclinic, and Triclinic — progressively more leaning and lopsided boxes, where corner angles stop being right angles. Triclinic, with no two edges equal and no square corners, is the wonky cardboard box at the bottom of the pile: the least symmetric of all.
The reason there are fourteen lattices rather than just seven is that some of these box shapes can be decorated in more than one way and still tile space cleanly. A cubic box, for example, comes in three flavours: plain (points only at the corners), body-centred (one extra point dead in the middle), and face-centred (an extra point on each face). Each is a genuinely different repeating grid, so the seven families branch into fourteen lattices in total. You will meet the body-centred and face-centred cubes constantly — they are the homes of most common metals.
Putting it together
Step back and look at the whole machine you now own. To specify any crystal in the universe, you choose one of fourteen Bravais lattices (that is the grid), you write down a lattice constant or two (that is the size), and you state the basis — which atoms sit at each lattice point and where (that is the contents). Those three choices, and only those, describe every crystalline material there is. It is an astonishingly compact language for the entire solid world.
Two honest cautions before we move on. First, the basis can be a single atom or a complicated molecule with dozens of atoms — proteins crystallise too, and their basis is enormous, but the lattice idea still holds without a hitch. Second, the fourteen Bravais lattices cover crystals, the orderly solids. There exist strange materials with patterns that fill space and never quite repeat, breaking the fourteen-fold rule in a way that stunned scientists when it was discovered. Those exceptions are a story for much later; here, the fourteen are your reliable map.