Shortly after 9 o'clock on a Sunday morning in January 2007, Raymond Rodriguez was in his driveway, changing a tyre on his car. Parked close by in the quiet cul-de-sac in Tampa, Florida, was a cherry red Ford Mustang. It belonged to Andre Javage, who had come to visit the night before. Out of nowhere, Rodriguez heard a high-pitched whistling. Then a 13-kilogram chunk of ice landed smack on the roof of Javage's Ford. The windshield blew out and the car bounced, Rodriguez said, a metre into the air. The roof was squashed flat.

Rodriguez lives under the flight path to Tampa international airport. So when he and Javage got over the shock, they figured the ice bomb had fallen from an aircraft. But incident investigator Fred Kaiser of the Federal Aviation Administration wasn't so sure. Ice does fall from aircraft, but it usually comes from a damaged toilet drain tube, and it's blue because of the chemicals added to the waste tank to deodorise and break down solid waste.

Usually this blue ice has melted before it hits the ground. Any chunks that survive are usually only large enough to crack roof tiles, and sometimes the briefest check will confirm their origin. One resident of north Kensington in London reported a nappy (diaper) encased in blue ice sailing onto their balcony.

The ice that totalled Javage's Mustang, however, was white, pure and huge. "I've never seen that big a piece of ice fall off an airplane," Kaiser said as he began his investigation. But Kaiser has discovered that falls of large masses of white ice, though rare, are not as uncommon as he first thought. In the space of a week in April 2006, there were two such falls in California. The second involved a chunk of ice the size of a microwave oven that dropped from a cloudless sky, punching a hole 75 centimetres wide through the metal roof of a recreation centre in Loma Linda.

Jesús Martínez-Frías, a planetary geologist at the Centre for Astrobiology in Madrid, Spain, has dubbed these prodigious lumps of falling ice megacryometeors. His interest was piqued in January 2000 when a football-sized chunk smashed though the windshield of a car in south-east Spain, heralding a week of unusual falls.

Martínez-Frías and his team quickly dismissed plumbing in planes as the source. Not only was the ice not contaminated, but there were no reports of any aircraft over the locations where two of the ice balls fell. A hoax perhaps? While a few bogus ice balls did show up in Spain after the initial flurry, Martínez-Frías was able to discount them by analysing their chemical and isotopic composition, which also told him the megacryometeors were not comet fragments from outer space. These ice balls were made of Spanish rainwater, pure and simple. Mostly, Martínez-Frías concluded, they resembled giant hailstones.

In a paper in the Journal of Atmospheric Chemistry , Martínez-Frías and his co-authors argued that the megacryometeors must have formed from atmospheric water vapour. After all, they had the same layering, air bubbles and silica particles (common in the atmosphere) as those found in conventional hailstones. There was just one problem, which has made many meteorologists sceptical of Martínez-Frías's ideas. Many of the 50-plus megacryometeors Martínez-Frías has catalogued, including the ones which bombarded south-east Spain, fell out of a clear blue sky, and regular hail just does not do that.

No storm required

Hail forms inside the cumulonimbus clouds of a thunderstorm, where supercooled water droplets freeze around condensation nuclei such as dust particles or ice crystals. Strong updrafts repeatedly drive the hailstones up through the clouds, and ice collects around them like the layers of an onion. And though conventional hailstones can grow to lethal size - in 2002, a hailstorm in China left 25 dead - the largest hailstones don't come much bigger than a baseball, altogether much smaller than the basketball-sized chunk that crushed Javage's Mustang. And even that was pea-sized compared to specimens found in Brazil and Spain, which weighed in at hundreds of kilos. It would be hard to miss the kind of thunderclouds that might produce such monsters, if such clouds existed.

So how could gigantic hailstones form on a clear day? Conditions over south-east Spain in early January 2000 were certainly anomalous. Three days before the first megacryometeors hit, ozone levels in the lower stratosphere, between 10 and 50 kilometres above ground, were substantially below normal, allowing more radiation than usual to pass through to the troposphere below. With less radiation being absorbed in the lower stratosphere, it became cooler. Furthermore, satellite data showed that the boundary between the troposphere and stratosphere, known as the tropopause, dropped by 2.5 kilometres over Spain in the days before the strikes.

The consequence, says Martínez-Frías, was strong wind shear at high altitude. Though no clouds formed, the lower stratosphere and tropopause were unusually moist, according to a study by meteorologist Millán Millán of the Mediterranean Centre for Environmental Studies in Valencia, Spain. Martínez-Frías speculates that in these conditions charged ions, possibly released in the contrails of cruising jet aircraft that had flown by sometime earlier, could have acted as seeds for the formation of ice crystals. Once formed, they could have been kept aloft by the unusually turbulent wind conditions until a handful had grown to massive sizes.

The number of megacryometeors reported has mushroomed in the last few decades, and Martínez-Frías says this may be due to climate change. Global warming might be making the tropopause colder, moister and more turbulent - ideal for producing huge ice balls.

However, when Martínez-Frías mooted his ideas on how a megacryometeor might form without a thunderstorm, meteorologists weren't impressed. Among them was hail expert Charles Knight at the University Corporation for Atmospheric Research in Boulder, Colorado. "I don't like to claim that anything is absolutely impossible," he said in 2002, "but this comes awfully close." The intervening years haven't changed his opinion: "Growth of ice in that environment to mega sizes would probably take weeks, if not months or years," he says. "Meanwhile, of course, the ice would be falling."

Knight still believes the most likely explanation is that ice is falling off aircraft. It's a view David Travis, a climatologist at the University of Wisconsin at Whitewater, who has collaborated with Martínez-Frías, leans towards. "The theory that they are forming on their own in the atmosphere just doesn't make physical sense," says Travis. He points out that only a thunderstorm can provide the huge vertical thrusts needed for hailstones to form. "For megacryometeors to form there is a need for equal or greater vertical motion, and we don't understand how this could occur anywhere outside of thunderstorms."

Travis agrees that unique atmospheric conditions coincide with some falls, and that this points to enhanced turbulence in the atmospheric layer where Martínez-Frías says the megacryometeors form. But he is not convinced that these conditions are extreme enough to create huge masses of ice. "The fact that most events happen in winter [and] early spring suggests there is a seasonality to their occurrence, which could support either theory," he says. That time of year, he notes, is when aircraft most often have icing problems. However, Travis struggles with the aircraft theory too. "It's quite hard to find a place on an aircraft that could support such a large build-up of ice," he says. But what about ice in a protected spot, such as a plane's wheel well?

Initially Kaiser speculated that an aircraft taking off somewhere snowy might accumulate ice on a wheel. As the wheel-well doors open for landing, the ice might fall out. But he remained unconvinced. After take-off aircraft wheels are braked as they retract, and this creates a lot of heat. Kaiser believed this would melt any ice brought up into the wheel wells. What's more, Martínez-Frías has documented a few megacryometeors that fell before the age of aviation.

Kaiser's latest theory for the ice block that wrecked Javage's Mustang suggests it came from a faulty water heater leaking from one of 15 aircraft in the vicinity of Tampa at the time. Unfortunately, none of those aircraft reported a problem with a water heater, leaving him stuck. "Unless a 'suspect' aircraft [reported] a heater failure, it will be impossible to discover an actual cause," he admits.

Whatever the cause, the fact remains that large chunks of pure ice are falling from a clear sky and nobody is sure where they are coming from. Even with blue ice, it is rare for investigators to be able to connect incidents with particular aircraft. Among 52 cases of large ice lumps falling on the UK in the last two years, just one has had its origin identified - a faulty toilet - by the Civil Aviation Authority. "It's just a matter of time before someone gets hit on the head with one of these things," says Travis, "and then it will finally generate some serious attention."

Ed Douglas already drives a dented car. He blames his wife.