Two New Zealand scientists think they can explain one of the great mysteries of the natural world - ball lightning.

These bright, hovering spheres of light seen during thunderstorms have been reported as far back as the Middle Ages. They can be as small as tennis balls or as big as beach balls.

They can be white, yellow, orange or blue and have been said to pass through windows and walls. Some people have even claimed to see ball lightning pass through aeroplanes.

But despite numerous theories and attempts to recreate the phenomenon in a laboratory, ball lightning has remained an enigma to modern science.

Now, John Abrahamson and James Dinniss, from the University of Canterbury, NZ, have put forward the theory that these ethereal objects are nothing more than burning particles of silicon.

Their experiments show that when ordinary forked lightning hits the ground, mineral grains in the soil can be converted into tiny particles of silicon and its compounds with oxygen and carbon.

Burning fluff balls

These particles, less than a tenth of a micrometre (millionth of a metre) in size, then link up into chains.

The filaments then cluster together into light, fluffy balls that are carried aloft by air currents. The silicon particles are very reactive and burn relatively slowly, emitting light as they do so.

"Lightning penetrates below the surface of the soil and heats a certain portion of it to quite high temperatures, so that it vaporises," Dr Abrahamson told the BBC.

"And then, when the lightning strike has finished, the vapour is free to erupt, to appear above the ground in the form of a ball.

"The jet of hot gas will be very much the same as the jet coming out of one's mouth when one blows a smoke ring - it forms a little re-circulating vortex and it's quite self-contained."

The scientists' model predicted that heating above a given starting temperature would lead to an explosive end for a fluffy ball, whereas with a lower starting temperature the ball would melt and fade away.

Because the ball would become visible only over the latter part of its lifetime, it would appear to materialise out of thin air after a lightning strike.

One theoretical umbrella

And Dr Abrahamson believes their theory will even explain how ball lightning passes through windows and walls.

"Most, especially old, houses have cracks around their windows and cracks near doors," he says.

"The network of silicon filaments should be very flexible and move wherever air moves. So if air can get through a crack as a draught, the ball should be able to squeeze through, and then rearrange itself on the other side."

Unfortunately, the researchers have not been able to generate ball lightning in the laboratory. But Dr Graham Hubler, of the United States Naval Research Laboratory, who has taken a keen interest in lightning balls, says the research has much promise.

"I have followed the theories that have come out over the years, and few can explain all of the features of ball lightning. This one, however, unifies an awful lot of the properties of ball lightning under one theoretical umbrella, so I think it stands a very good chance of perhaps being correct.

"It's not necessarily the whole story, or even the story, but the nice thing about it is that it can be tested experimentally. Then we'll know whether this theory has any merit in the future."

Both Dr Hubler (hubler@ccs.nrl.navy.mil) and Dr Abrahamson (j.abrahamson@cape.canterbury.ac.nz) are keen to hear from anyone who has pictures or video footage of ball lightning.