Operations that must close on themselves
Guide 1 of this rung handed you the two halves of point symmetry: the symmetry operation (the move that leaves the object looking unchanged) and its symmetry element (the geometric thing — a rotation axis, a mirror plane, a point — that the move acts about). Guide 2 fenced off the allowed rotations: a periodic crystal may carry 1-, 2-, 3-, 4-, or 6-fold axes and never 5-fold, the crystallographic restriction. Add to those axes the center of inversion and the rotoinversion axis, and you now hold the complete alphabet of point symmetry. This guide asks the natural next question: how many self-consistent words can you spell from that alphabet?
The new rule that answers it is closure. You cannot toss operations into a bag at random. If a crystal has a 4-fold axis AND a mirror, then doing one right after the other is itself a symmetry of the crystal — so that combined move must ALSO already be in the bag. A set that is closed in this way — every combination already inside it, plus the do-nothing identity, plus an inverse that undoes each move — is exactly what mathematicians call a group. A point group is such a group whose operations all leave at least one point fixed: nothing ever slides.
Why the count stops at exactly 32
Closure is a surprisingly tight leash. Try to build a group from a single 2-fold axis and one mirror that contains it: the two together FORCE a second mirror at right angles to the first — you simply cannot have the pair without the trio. That closed group is written mm2. Try instead two 2-fold axes crossing at 90 degrees: closure immediately conjures a third 2-fold perpendicular to both, giving the group 222. Every legal group is trapped like this — you keep adding whatever operations closure demands until the set stops growing on its own.
Now feed in the crystallographic restriction. Only 1-, 2-, 3-, 4-, and 6-fold axes are on the menu, and you may decorate them with mirrors, a center, and rotoinversion axes — but ONLY in combinations that both close AND stay compatible with a periodic lattice. Grind patiently through every possibility and the self-consistent groups number exactly 32. These are the 32 crystallographic point groups, also called the 32 crystal classes. There is nothing to add and nothing to remove; the list is airtight.
Be clear on the honesty here: 32 is not 'thirty-two so far'. Like the 14 Bravais lattices and the 230 space groups, it is a complete, proven enumeration — the mathematics is finished, in the same way there are exactly five Platonic solids. Drop the demand for periodicity and let 5-fold in, and the point groups become infinite (icosahedral symmetry among them): that is precisely the door quasicrystals walked through. Inside the crystalline world, though, 32 is a closed and final count.
Sorting the 32 into seven systems
The 32 groups do not scatter at random — each one demands a lattice of a particular shape to host it, and that sorts them into the seven crystal systems you met at the end of the lattice rung. The bridge is the characteristic (highest-symmetry) axis of each group. A single 4-fold axis needs a cell with a square cross-section, so it belongs to the tetragonal system; four 3-fold axes running along the body diagonals force full cubic symmetry. The lesson to keep: it is the symmetry that names the system, not the accidental cell dimensions.
CRYSTAL SYSTEM DEFINING SYMMETRY POINT GROUPS (crystal classes) COUNT
-------------- ---------------------------- ------------------------------ -----
triclinic none beyond 1 or -1 1 -1 2
monoclinic one 2-fold axis (or mirror) 2 m 2/m 3
orthorhombic three 2-folds / mirrors 222 mm2 mmm 3
tetragonal one 4-fold axis 4 -4 4/m 422 4mm -42m 4/mmm 7
trigonal one 3-fold axis 3 -3 32 3m -3m 5
hexagonal one 6-fold axis 6 -6 6/m 622 6mm -6m2 6/mmm 7
cubic four 3-folds (body diagonals) 23 m-3 432 -43m m-3m 5
------------
TOTAL 32
( -n means an n-fold rotoinversion axis; n/m means a mirror perpendicular to n )One honest subtlety the table can hide: the system is fixed by the symmetry a crystal actually carries, never by numbers that happen to line up. A crystal whose three edges come out equal by sheer numerical coincidence is still not cubic unless it genuinely possesses those four 3-fold axes down the body diagonals. Shape follows symmetry, and not the other way round — which is why we sort classes by their axes first and only then notice what cell they require.
Two names for one group
Each of the 32 groups carries two standard names, and you will meet both in the wild. Hermann-Mauguin (the 'international' notation) lists the symmetry elements found along the crystal's important directions — mm2, 4/mmm, -43m — and it is the one crystallographers and the International Tables use, because it extends cleanly to the space groups. Schoenflies uses compact letter-and-subscript labels — C2v, D4h, Td — long favoured by spectroscopists and chemists.
A quick Rosetta stone: the tetragonal group written 4/mmm in Hermann-Mauguin is the very same object Schoenflies calls D4h; cubic 23 is T, and -43m is Td. Reading these symbols fluently — knowing that '4/m' means a mirror perpendicular to the 4-fold axis, or that a bar over a number signals rotoinversion — is the entire job of the next guide, so we will not drill them here. For now just hold the fact firmly: one group, two names, and they refer to exactly the same set of symmetry operations.
Does it have a center? Why that one question matters
Of the 32 groups, exactly 11 contain a center of inversion — a point through which every feature maps onto an identical feature straight opposite. These are the centrosymmetric classes; the other 21 are non-centrosymmetric. This single yes/no split governs some of the most useful properties a crystal can have, through a deep rule called Neumann's principle: any physical property of a crystal must be AT LEAST as symmetric as its point group.
Take piezoelectricity — squeeze the crystal and it develops a voltage across its faces. A center of inversion forbids it outright: inversion would flip the electric polarization into its own negative, and the only quantity equal to minus itself is zero. So piezoelectricity requires a non-centrosymmetric class. Twenty of the 21 qualify (the lone exception, cubic 432, has other symmetry that still cancels the effect). Quartz, class 32, is the textbook example — the trembling, voltage-tapping heart of a quartz watch.
- Look first for a center of inversion. If it is present, the class is centrosymmetric — no piezoelectricity, no bulk second-harmonic generation. You can stop here.
- No center? The class is non-centrosymmetric and (except for cubic 432) piezoelectric — squeeze it and read a voltage.
- Now look for a unique polar direction left unmoved by every operation. If one exists, the class is a polar class — it can be pyroelectric and is a candidate for ferroelectricity. The 10 polar classes are 1, 2, m, mm2, 4, 4mm, 3, 3m, 6, 6mm.
- Finally, if the group contains ONLY proper rotations — no mirror, no center, no rotoinversion — it is chiral: it comes in left- and right-handed forms. There are 11 such enantiomorphic classes, and quartz (class 32) is one of them.
That last step is why quartz grows as left- and right-handed crystals that are mirror images you can never superimpose, like your two hands — the property called chirality, with the two forms known as enantiomorphs. Guide 5 develops chirality, the polar classes, and the drawing tool in full; here just register the payoff. The mere presence or absence of a center and of mirrors quietly decides whether a crystal can be piezoelectric, pyroelectric, or optically handed — properties worth billions, all read straight off the point-group symbol.
What diffraction can never see: the 11 Laue classes
Here is a humbling limit. When you shine X-rays through a crystal, the diffraction pattern always looks centrosymmetric — even when the crystal itself is not. The reason reaches back to the phase problem you met earlier: diffraction measures intensities, the squared magnitude |F|^2 of the structure factor, and throws the phase away. That loss makes the intensity of reflection (hkl) equal to that of (-h,-k,-l). This is Friedel's law, and it means diffraction quietly adds a phantom center of inversion to whatever it looks at.
So a diffraction pattern cannot report all 32 point groups — only the symmetry left after forcing a center onto each one. Merge the 32 groups that become identical once inversion is added, and they collapse into exactly 11 Laue classes. Every crystal's diffraction symmetry is one of these 11. In practice that means ordinary diffraction cannot tell a non-centrosymmetric class from its centrosymmetric parent in the same Laue class, and — famously — it gives left- and right-handed quartz identical patterns. Telling the enantiomorphs apart needs a subtler trick, anomalous scattering, that gently breaks Friedel's law.