Operation and Element: the Verb and the Noun
In the last rung you built the scaffolding of periodic order — the space lattice, the unit cell, and the fourteen Bravais lattices that exhaust every way to tile space with points. This rung asks a different question about the very same crystal: not where the pattern repeats, but what shape of symmetry it has about a single point. Hold a snowflake still and turn it by 60 degrees — it looks exactly as it did. Face a starfish and turn it by 72 degrees — unchanged. Look at your two hands — each is the mirror image of the other. Every one of these is symmetry: an object looks unchanged after some definite move.
Crystallography splits that idea into two crisp halves that beginners constantly blur together. A symmetry operation is the move itself — the act of rotating, reflecting, or inverting the object so that it lands back exactly on top of where it started. A symmetry element is the geometric thing the move is performed with respect to — the line you rotate about (an axis), the plane you reflect across (a mirror), or the point you invert through (a centre). Operation is the verb; element is the noun. "Rotate by 90 degrees" is an operation; the four-fold axis it turns about is the element.
Rotation Axes — and the Numbers a Crystal Allows
The most familiar element is the rotation axis. An axis is called n-fold if turning the object by 360/n degrees about it leaves it unchanged — so the object comes back to itself n times in a full circle. A snowflake sits on a 6-fold axis (60-degree steps); a starfish on a 5-fold axis (72 degrees); a square floor tile on a 4-fold axis (90 degrees); an equilateral triangle on a 3-fold axis (120 degrees); a plain rectangle on a 2-fold axis (180 degrees). The larger n is, the richer the rotational symmetry.
Here comes a genuine surprise. In a periodic crystal — one built by tiling identical unit cells edge to edge with no gaps — only five values of n are possible: 1, 2, 3, 4, and 6. Five-fold symmetry is flatly forbidden, and so is every value of 7 or more. The reason is the same one that says you can tile a bathroom floor with squares, triangles, or hexagons but never with regular pentagons: 5-fold rotation is simply incompatible with filling space by repetition. This ban is the crystallographic restriction, and the next guide, "Why Five-Fold Symmetry Is Forbidden," proves it properly. It is exactly why quasicrystals — which show clean 5-fold diffraction spots — were such a scandal when they turned up in 1982: they are ordered, but not periodic.
One more thing about pure rotations: they never change an object's handedness. Spin a right glove any way you like and it stays a right glove — a rotation just moves it, it never turns it into a left glove. Operations of this kind are called proper. To flip handedness you need a different sort of move altogether, and that is where mirrors and inversion come in.
Mirrors, the Centre, and Handedness
A mirror plane, written m, does exactly what a real mirror does: it reflects every point of the object straight across the plane to an equal distance on the far side. The vertical line down the middle of a capital A, or the six mirror lines threading a snowflake, are mirror planes. Reflection is the operation; the plane is the element. Crucially, a mirror swaps left for right — your reflection raises its left hand when you raise your right — so unlike a rotation it does reverse handedness.
The centre of inversion, written as a one with a bar over it (1-bar), is subtler. Pick a special point inside the object and push every atom in a straight line through that point and out the same distance on the opposite side: the atom at (x, y, z) is carried to (-x, -y, -z). It is the three-dimensional cousin of a point reflection, and it works like a pinhole camera, which flips an image top-to-bottom and left-to-right at once. An object has a centre of inversion when every feature has an identical partner directly opposite through the middle — and, like a mirror, inversion reverses handedness.
Mirrors and inversion belong together as improper operations: both reverse handedness, turning a right-handed motif into a left-handed one. This single fact underlies chirality. An object whose only symmetries are proper rotations — with no mirror, no centre, no improper axis at all — cannot be laid on top of its own mirror image, exactly like a left and a right hand. Such chiral crystals come in two mirror-related forms called enantiomorphs; quartz is the classic case, growing as distinct left- and right-handed crystals. Guide 5 in this rung returns to chirality in earnest; for now just hold the split: rotations preserve handedness, mirrors and inversion reverse it.
Rotoinversion: Two Moves Fused Into One
The last kind of element is the sneakiest: the rotoinversion axis, written n-bar (an n with a bar). It is a compound operation — rotate by 360/n degrees and then immediately invert through a point on the axis, the whole thing counting as one indivisible move. The catch is that the object need not be symmetric under the rotation alone, nor under the inversion alone; only the two performed back-to-back bring it into coincidence. A 4-bar axis, for instance, contains a plain 2-fold rotation (do it twice and the two inversions cancel) yet is genuinely different from a 4-fold axis, and it turns up in tetrahedral molecules and crystals.
This family quietly swallows the two operations we just met. Rotoinversion with n = 1 is nothing but a rotation by a full turn followed by an inversion — that is, plain inversion, so 1-bar is the centre of symmetry. And 2-bar (a half-turn plus inversion) works out to be exactly a mirror plane perpendicular to the axis, so 2-bar is the mirror m. That leaves 3-bar, 4-bar, and 6-bar as the only genuinely new rotoinversion axes. So the complete toolkit of point symmetry is smaller than it first looks: rotations, and rotoinversions — with mirror and centre falling out as special cases of the second.
A heads-up before the table. Everything above is the international, or Hermann-Mauguin, way of naming things, which builds improper moves out of rotation-then-inversion. There is a rival system, Schoenflies, favoured by chemists and spectroscopists, which instead uses rotation-then-reflection (a rotoreflection, written S with a subscript) and labels rotations C, mirrors sigma, and the centre i. The two describe identical physical symmetry; they just carve it up differently. Guide 4 lays both out side by side — for now, do not be alarmed to see the same crystal wearing two different name-tags.
operation symmetry element H-M Schoenflies ---------------- --------------------- -------- ----------- identity nothing (do nothing) 1 E rotation n-fold axis 2 3 4 6 Cn reflection mirror plane m sigma inversion centre of symmetry 1-bar i rotoinversion inversion axis 3 4 6-bar Sn
From a Handful of Moves to the 32 Crystal Classes
Now the pay-off. Take those point operations — rotations and rotoinversions about a shared point — and ask how many self-consistent combinations exist that are also compatible with a repeating lattice (so restricted to 1-, 2-, 3-, 4-, and 6-fold axes). The answer is astonishing in its finality: exactly 32. These are the 32 crystallographic point groups, also called the crystal classes, and they sort tidily into the seven crystal systems from last rung. Each is named in both languages you will meet — a Hermann-Mauguin symbol like 4/mmm that reads off the elements present, or the Schoenflies label D4h for the very same class. Like the 14 Bravais lattices and the 230 space groups, the number 32 is a proven, complete enumeration — not "32 so far," but 32, full stop.
Why sweat over which class a crystal belongs to? Because whole physical properties switch on or off with symmetry. The sharpest divide is centrosymmetric versus non-centrosymmetric — whether the class contains a centre of inversion. A crystal with a centre cannot be piezoelectric, cannot be pyroelectric, and shows no second-harmonic generation; squeezing a centrosymmetric crystal produces no voltage because for every push there is an equal and opposite push through the centre that cancels it. Piezoelectricity therefore requires a class with no inversion centre — 20 of the 21 non-centrosymmetric classes qualify. Handedness matters here too: only certain non-centrosymmetric classes permit chiral, enantiomorphic crystals like quartz.
There is one honest catch that ties straight back to how we see structure. Ordinary diffraction cannot tell a non-centrosymmetric crystal from a centrosymmetric one, because diffraction intensities obey Friedel's law and effectively add a centre of symmetry the crystal may not really have. As a result X-ray diffraction sorts the 32 point groups into only 11 Laue classes — it simply cannot resolve the finer distinction. This is a cousin of the phase problem from the last rung: diffraction is powerful, but it quietly throws certain information away, and knowing exactly what it discards is part of using it honestly.