Colin Humphreys, head of materials science at Cambridge University, disclosed his ideas at a symposium on the centenary of the discovery of the electron at Cambridge. He has submitted a paper based on his ideas to Nature.

Superconducting materials lose all electrical resistance below a critical temperature; but for many years, this temperature was below 23 K for all known superconducting substances. High-temperature superconductivity was discovered in 1986 when materials with critical temperatures as high as 35 K were found. Since then one of the holy grails of modern physics has been to develop room-temperature superconductors. Such materials could, for example, replace conventional electrical cables and carry current without any loss of power. The present upper limit for superconductivity is 164 K; Humphreys believes his model could be used to design room-temperature superconductors.

All high-temperature superconductors consist of parallel planes of copper-oxide. The atoms lie on a square lattice and the charge is carried by "holes" sitting on oxygen sites. However, no clear theory has emerged that adequately explains high-temperature superconductivity - and very few theories have even been ruled out. Some theorists, such as Phillip Anderson from Princeton University in the US, believe that fundamentally new physics is needed to construct a proper theory, while others argue that an essential piece of existing physics is missing. "I think my model will be very controversial because it's saying that fundamental assumptions of a lot of theories are wrong. People have put a lot of effort into their pet theories, " says Humphreys.

Humphreys believes that existing theories fail because they do not take into account the distribution of the holes. He argues that each copper-oxide plane consists of square "nanodomains", separated by channels that are one unit-cell wide - rather like a grid of streets surrounding blocks of houses. Holes at the edges of adjacent blocks are magnetically paired, he says, and superconductivity occurs because these hole-pairs march collectively along the channels, like trams on pairs of tramlines running between the blocks of houses. There is one hole on each tramline, according to the model, and the pairs of holes move down the channels, hopping from oxygen to oxygen via adjacent copper sites.

Reaction to Humphreys' model was cautious. "I want to know why the holes pair up in the tramlines and - as with any potential theory of superconductivity - whether it can properly explain why normal materials do not superconduct, " said Mike Gunn, a theorist from Birmingham University, when told about the model by Physics World. He added that the test of the model will be if it can make concrete predictions that can be experimentally verified.

However, Gunn does see some similarity between Humphreys' model and recent experimental and theoretical evidence for "stripe phases" in some copper-oxide superconductors, which was obtained by John Tranquada at the Brookhaven National Laboratory and Steve Kivelson of the University of California at Los Angeles. The stripes, containing a high density of holes, lie between insulating regions with low hole-density.

But others are sceptical about any new theories. "There have been many claims to understand high-temperature superconductors, but there is no consensus yet so I am naturally very cautious, " says Andy Schofield, a Cambridge theorist.

About 100 000 papers have been published on high-temperature superconductivity in the last ten years, which makes it hard for anyone to claim to have a definitive answer. And although Humphreys admits that the model will need to be developed into a full-blown theory before predictions can be made, he believes that in the meantime his model should be used to re-analyse existing data.