Chicken Little wasn't completely wrong. Some of the sky does fall, some of the time.

When a grit-sized particle traveling at many miles a second streaks into the Earth's atmosphere and burns up from friction with the air before reaching the ground, we call it a shooting star or a meteor. Some of us make a wish on it. We think of meteors as harmless and beautiful, especially when they come in large groups and provide spectacular displays such as the Leonid and Perseid meteor showers.

Meteors, however, have big brothers. These exist in all sizes from pebbles to basketballs to space-traveling mountains. If the speeding rock is large enough, it can remain intact all the way to the ground and it is then known as a meteorite. The reality of meteorites was denied for a long time - Thomas Jefferson said, "I could more easily believe that two Yankee professors would lie than that stones would fall from heaven" - but today the evidence is beyond dispute.

If one of these falling rocks is big enough, its great speed gives it a vast amount of energy, all of which is released on impact with the Earth. Even a modest-sized meteorite, 20 meters across, can do as much damage as a one-megaton hydrogen bomb. This sounds alarming, so let us ask three questions: How many rocks this size or larger are flying around in orbits that could bring them into collision with the Earth? How often can impact by a rock of any particular size be expected? And how does damage done vary with the size of the meteorite?

Direct evidence of past impacts with Earth is available only for large meteorites. For small ones, natural weathering by wind, air and water erases the evidence in a few years or centuries. However, we know that a meteorite, maybe 200 meters across, hit a remote region of Siberia called Tunguska, on June 30, 1908. It flattened 1,000 square kilometers of forest and put enough dust into the atmosphere to provide colorful sunsets half a continent away.

About 20,000 years ago, a much bigger impact created Meteor Crater in Arizona, more than a kilometer across. And 65 million years ago, a monster meteorite, maybe ten kilometers across, struck in the Gulf of Mexico. It caused global effects on weather, and is believed to have led to the demise of the dinosaurs and the largest land reptiles.

The danger of impact is real, and beyond argument. But is it big enough for us to worry about? After all, 65 million years is an awfully long time. How do we make an estimate of impact frequency?

The answer may seem odd: we look at the Moon. The Moon is close to us in space, and hit by roughly the same meteorite mix. However, the Moon is airless, waterless, and almost unchanging, so the history of impacts there can be discovered by counting craters of different sizes. Combining this with other evidence about the general size of objects in orbits likely to collide with Earth, we can calculate numbers for frequency and energy release. They are not totally accurate, but they are probably off by no more than a factor of three or four.

I will summarize the results by size of body, and translate that to the equivalent energy released as number of megatons of H-bombs. About once a century, a "small" space boulder about five meters across will hit us and produce a matching "small" energy equal to that released by the Hiroshima atomic bomb. It will probably burn up in the atmosphere and never reach the ground, but the energy release will be no less. Once every 2,000 years, on average, we will be hit by a 20-meter boulder, with effects a little bigger than a one-megaton H-bomb. Every two million years, a 500-meter giant will arrive, delivering as much energy as a full-scale nuclear war.

I found these numbers disturbing, so a few years ago I sent them to the late Gene Shoemaker, an expert on the bombardment of Earth by rocks from space. He replied, not reassuringly, that he thought my numbers were in the right ballpark, but too optimistic. We will be hit rather more often than I have said.

Even if I were exactly right, that leaves plenty of room for worry. Being hit "on average" every 100,000 years is all very well, but that's just a statistical statement. A big impact could happen any time. We have no way to predict it, or - despite what recent movies would have you believe - prevent it.

"Borderlands of Science" is syndicated by: