With the discovery of a new class of high-temperature superconductors by reseachers in Japan (see page 922), history seems to be repeating itself. In 1986, Georg Bednorz and Alex Müller of IBM's Zurich research laboratories discovered that a complex oxide of barium, lanthanum and copper became superconducting at 35 K. This sparked an orgy of research that led to the discovery of a related compound (yttrium barium copper oxide) with a superconducting transition temperature of 90 K: high enough to be attained with relatively cheap liquid-nitrogen cooling. It also won Bednorz and Müller a Nobel prize just a year later.

The excitement stemmed from the prospect of exploiting low-cost superconductivity for loss-free electrical transmission, magnetic levitation and other dazzling applications. In superconductors, currents flow essentially without electrical resistance, the source of energy loss through heating. Before 1986, most superconductors were metals and alloys, with generally paltry transition temperatures that no one had managed to push above 23 K.

Now Hideo Hosono of the Tokyo Institute of Technology and his colleagues have shown that another complex oxide, containing lanthanum, iron, arsenic and a little fluorine, will superconduct at 43 K when squeezed by around 40,000 atmospheres pressure: a higher temperature than anything bar the copper oxides (see H. Takahashi et al. advance online publication: doi:10.1038/nature06972). Like them, the new material has a sandwich structure of alternate conducting and insulating layers. And like them, doping (replacing some oxygen with fluorine) injects electrons into the conducting layer that contribute to the supercurrent.

The copper-oxide materials have found some uses, but nothing to match the expectations heaped on them in the late 1980s — levitating trains and so forth. It has proved hard to fashion these brittle materials into wires and progress is slow. So cynics might grumble that the new breakthrough will merely renew the same unfulfilled promises.

But the new compound already offers more. For one thing, it reveals how much remains to be discovered about complex solid-state compounds. The combinatorial possibilities for four or more elements are so vast that we have barely scratched the surface, despite efforts to automate the search. And as before, the discovery followed from sound chemical intuition. Bednorz and Müller were led to the copper oxides from the apparently unpromising strontium titanium oxide, a superconductor at a mere 0.3 K, by reasoning what kind of crystal chemistry might boost the requisite interactions between electrons. Hosono and his colleagues similarly picked a systematic path from their initial discovery, in 2006, of superconductivity at about 4 K in a related material, a temperature so low that it attracted little interest. They raised this to 26 K by the start of 2008, and rightly figured that squeezing would take it further.

Most crucially, the 'iron oxypnictides' show that high-temperature conductivity is not the sole preserve of copper oxides. As in that case, superconductivity in the new materials seems to be related to magnetic behaviour. But quite how this works has remained a mystery. With an entirely new family of compounds to play with, the mechanism might be persuaded to start giving up some secrets. With a theory to hand, 'designer superconductors' with much higher transition temperatures might not look like a fool's quest. There could be another Nobel prize in that. For now, it is enough that the oxy-pnictides have set the community buzzing in a way that recalls the last heyday of superconductors two decades ago.