Quasicrystals are freaks of the crystal world. Strictly speaking, they are not crystalline: although the five-fold symmetry some of them exhibit in two dimensions might be acceptable for, say, a starfish, it is not permitted in a perfect crystal. For example, a floor can be tiled using only triangles, or squares or hexagons; tiles with five-fold, eight-fold or twelve-fold symmetry, however, will leave gaping holes if flat, or a puckered surface if forced to fit.

But in this issue, Xiangbing Zeng et al. report their discovery of a highly ordered molecular quasicrystal that has the ‘forbidden’ twelve-fold, or dodecagonal, symmetry (Nature 428, 157–160; 2004). All but one of the quasicrystals known so far are metallic alloys (the notable exception is a liquid-crystal film with non-crystallographic helical symmetry). Instead of metallic atoms, the new quasicrystal is formed from organic dendritic (tree-like) structures that, Zeng et al. propose, have self-assembled into supramolecular spheres, or micelles.

This organic superstructure (shown here in simulation) was unexpected. Its stability may well have broader implications for cellular organization in tissues, for instance, or for a problem proposed by Lord Kelvin in 1887: in a foam, what shape of individual, equal-volume bubbles would yield the minimum surface area for the foam? Kelvin suggested a warped, 14-sided structure (a truncated octahedron) but was not able to prove it.

More than a hundred years later, Robert Phelan and Denis Weaire showed that a lower overall surface area would result if a foam contained two kinds of bubbles, one with 14 sides and the other with 12 (Phil. Mag. Lett. 69, 107–110; 1994). With their dodecagonal quasicrystal in mind, Zeng and colleagues suggest that this classic packing problem might now have a solution based on quasicrystalline structure.