The question guide 1 left hanging
In the previous guide we met the four ways a pattern can map onto itself and look unchanged — the symmetry operations — and the geometric features they turn about, the symmetry elements. The workhorse of the set is the rotation axis: a line you spin the crystal around so that after a fraction of a full turn every atom lands exactly where an identical atom sat before. An n-fold axis repeats the pattern n times in 360 degrees, so its smallest turn is 360/n degrees — a 3-fold axis clicks every 120 degrees, a 6-fold every 60.
Now here is the catch. A lone object can carry any rotation it likes: a starfish wears a clean 5-fold axis, a maple seed a 2-fold, a snowflake a 6-fold, a car wheel with seven spokes a 7-fold. But a crystal is not a lone object. It is a space lattice — the same motif stamped at every point of an endlessly repeating grid, an infinite 3D wallpaper that must fill all of space with no gaps and no overlaps. That single demand — perfect, gap-free periodicity — quietly outlaws most rotation axes. Only five values of n survive.
The floor-tiler's picture: which tiles fill a floor
Forget crystals for a moment and think about tiling a bathroom floor with a single regular shape, corner to corner, no gaps. Equilateral triangles work: their corners are 60 degrees, and six of them (6 times 60 = 360) close neatly around every meeting point. Squares work: 90 degrees, four to a point (4 times 90 = 360). Regular hexagons work: 120 degrees, three to a point (3 times 120 = 360) — that is the honeycomb. These are the only three regular tiles that fill a plane, and they carry 3-fold, 4-fold and 6-fold symmetry respectively.
Now try regular pentagons. Each interior corner is 108 degrees. Fit two around a point and you have used 216 degrees; fit a third and you reach 324 degrees, leaving a wedge of 360 minus 324 = 36 degrees — a gap too narrow for a fourth pentagon (which needs 108). Pentagons cannot close up; they always leave holes. Heptagons (about 128.6 degrees) and everything larger overshoot instead, overlapping before they close. So among regular tiles, exactly 3, 4 and 6 fill the plane, while 5 and 7-and-up cannot — the very same n that a crystal keeps and rejects.
Be honest about what this picture proves. Tiling a floor with regular polygons is an intuition, not the full theorem — it shows 3, 4 and 6 directly, while 1 and 2 come for free (every lattice already has a 2-fold axis). To pin down all five values cleanly and rule out 5 for certain, we need the algebra below. The tiling picture is the feeling; the algebra is the proof.
The proof in one line of algebra
The rigorous argument uses nothing but the lattice translation vector — the step that carries you from one lattice point to an identical neighbour. Pick the shortest such vector t along a row of points, of length a. Suppose the crystal also has an n-fold axis standing on a lattice point, turning by angle alpha = 360/n. Because the axis is a symmetry, spinning t by +alpha lands on another lattice vector, and spinning it by -alpha lands on yet another. Add those two rotated copies together: by simple trigonometry their sum points straight along the original row and has length 2a times cos(alpha).
- That sum is a lattice vector along the row, so it must be a whole-number multiple of the shortest step a: 2a times cos(alpha) = m times a, for some integer m.
- Cancel a from both sides and you are left with a startlingly tight rule: 2 times cos(alpha) = m, an integer.
- Since cosine of any angle lies between -1 and +1, the integer m can only be -2, -1, 0, +1 or +2 — just five choices.
- Solve each choice back for alpha, then for n = 360/alpha, and the five allowed axes drop out — with no room for a sixth.
m 2cos(a)=m -> cos(a) alpha=360/n n verdict -----+--------------------+-----------+-----+--------- +2 cos = +1.0 0 deg 1 allowed +1 cos = +0.5 60 deg 6 allowed 0 cos = 0.0 90 deg 4 allowed -1 cos = -0.5 120 deg 3 allowed -2 cos = -1.0 180 deg 2 allowed -----+--------------------+-----------+-----+--------- ? n = 5: 2cos72 = 0.618 (not an integer) FORBIDDEN ? n = 7: 2cos51 = 1.247 (not an integer) FORBIDDEN
That is the whole of it. Five-fold fails for a reason you can check on a calculator: the turn it demands, 72 degrees, gives 2 times cos(72) = 0.618, which is not a whole number of lattice steps, so the rotated vectors never land back on the grid. This is exactly what the name forbidden symmetry refers to — 5-fold, 7-fold, 8-fold and beyond are geometrically impossible for anything built on a repeating lattice. Notice the theorem never mentioned atoms, bonds or chemistry; it followed from periodicity and nothing else.
Why this rule is the backbone of the whole rung
This little restriction is not a curiosity tucked in a corner — it is the reason the next three guides even have finite lists to teach. Because a crystal may use only the 1, 2, 3, 4 and 6-fold axes (plus the mirror and inversion elements from guide 1), there are only so many self-consistent ways to combine them about a point. Do the combining carefully and you get exactly 32 distinct symmetry recipes, the 32 crystallographic point groups, which sort without remainder into the seven crystal systems. Add lattice centring and you land on exactly 14 Bravais lattices.
The shock: quasicrystals break the spell
For seventy years 'no 5-fold in crystals' was iron law — which is why April 1982 was an earthquake. Dan Shechtman shot electrons through a rapidly cooled aluminium-manganese alloy and got a diffraction pattern with sharp, bright spots arranged in a perfect 10-fold star. Sharp spots mean long-range order; a 10-fold star means a forbidden symmetry. Both at once should have been impossible. The reaction was brutal: Linus Pauling scoffed that 'there is no such thing as quasicrystals, only quasi-scientists.' Shechtman was right, and in 2011 he won the Nobel Prize in Chemistry for it.
So was the theorem wrong? No — and this is the subtle, beautiful part. The theorem forbids 5-fold in a periodic lattice. Shechtman's alloy is a quasicrystal: it has long-range order but it is not periodic. Its structure tiles space the way a Penrose tiling does — two tile shapes, fat and thin rhombi (or kites and darts), covering the plane completely with 5-fold flavour, yet never repeating, never quite random. Because there is no repeating lattice, the theorem's one premise fails, so its conclusion simply does not apply. The rule was never violated; it was side-stepped by a material that refused to be periodic.
The lasting lesson sharpens a distinction from earlier rungs: order and periodicity are not the same thing. A crystal is order-plus-periodicity; a glass is short-range order with no long-range order; a quasicrystal is the third door nobody expected — genuine long-range order with sharp diffraction, but no periodic lattice at all. It is an aperiodic crystal, and its discovery forced the very definition of 'crystal' to be rewritten around what diffracts sharply rather than what repeats. Every rule in this guide still stands — for periodic crystals, which is almost everything you will ever meet.