The natural world that scientists explore often seems pathologically arranged to thwart discovery. To peer inside the nucleus, we must build massive accelerators and detectors the size of office buildings; the mysterious beginnings of our Universe are shrouded by tremendous distances, requiring telescope mirrors as big as swimming pools to collect photons liberated billions of years ago. So there is something satisfying about turning nature to our own purposes, applying its peculiar properties to explore the edges of our knowledge and forcing nature itself to become the tool with which we explore the Universe. On page 294 of this issue, Brown et al.¹ describe how they have used a detector as big as the Earth itself — the planet's protective, blanketing atmosphere, which acts as a scintillator to detect the powerful but rare impacts of small asteroids or comets.

Meteors, those romance-inspiring flashes of light in the night sky, are created when small pieces of dirt called meteoroids slam into the Earth's atmosphere at speeds of tens of kilometres per second (Fig. 1). These fragments are the detritus of collisions between asteroids in the belt between Mars and Jupiter, or debris spawned from cometary nuclei, and their orbits gradually evolve onto paths that cause them to collide with the Earth. The burnt remains of these objects rain down on Earth and increase its mass by many tonnes every day — although the brightest meteor you may ever see might be the death beacon of a rock no bigger than a pea. As they are small and plentiful, it isn't particularly difficult to study the statistics of these impacts. They are found and measured regularly by meteor radar facilities² and automated wide-field camera searches³, and even by examining the surfaces of objects that have been in orbit about the Earth⁴.

Figure 1 Shooting stars. The Leonid meteor shower, which happens each year between 17 and 19 November, can be a breathtaking sight: small rocks, fragments of the comet Tempel–Tuttle, burn up or even explode as they hit the Earth's atmosphere. This meteor (inset) was photographed in the spectacular shower of 1966. Fortunately, impacts of much larger objects are rare. In 1908, a meteor as much as 50 m in diameter exploded in the Earth's atmosphere over Tunguska, Siberia. The resulting shock wave flattened trees in the surrounding area over hundreds of square kilometres (main image). Using satellite data, Brown et al.¹ have calculated the average rate of such devastating impacts to be about one per thousand years.

Tipping the scale in the other direction, massive mountains of rock and ice (asteroids and comets, respectively, in planetary jargon, though the distinction probably isn't as strict as was once thought), which may be many kilometres in diameter, also move in orbits that may one day intercept the Earth. The latest estimates for the number of these objects larger than a kilometre in diameter is about a thousand⁵, and there is a good chance that one of them will strike the Earth within the next million years. Theoretical predictions⁶ indicate that when objects this big collide with the Earth, there will be severe environmental consequences and massive continental-scale destruction. Fortunately for us, because these are relatively big objects, and because they are in orbits that swing them through the Earth's 'backyard' on occasion, the chances are good that planetary astronomers will discover the next big one before a collision takes place. Several groups around the world are surveying the night skies for these objects using modest-sized (about 1-m diameter) telescopes, and they are currently about halfway through cataloguing all of these most dangerous objects.

In between the harmless little meteors and the civilization-ending flying mountains lies an intermediate regime of objects, ranging from one metre to tens of metres in diameter. These objects are too small to be detected with telescopes and too uncommon in their rate of impact to be regularly visible from the planet's surface as they ignite in the upper atmosphere. Early extrapolations⁷ of the number distribution from the smallest to the largest size scales seemed to indicate a discontinuity (or more than one) between the two extremes — a condition ripe with opportunity for a scientific Goldilocks to identify the detector that is 'just right' for the task.

The detector in this case is the Earth's atmosphere. When a large meteoroid ploughs into its rarefied upper reaches, shock waves catastrophically disrupt the object, creating a blast equivalent to the explosion of many kilotonnes of TNT. Some of this energy is converted into visible light and other forms of radiation that can be detected by instruments on satellites operated by the US Departments of Defense and Energy, and these data have been used by Brown and colleagues¹. Although there are numerous uncertainties involved in correcting the observed signal intensity at the satellite to obtain the incoming energy and the size of the impactor, Brown et al. have succeeded in documenting this previously unexplored region of the size or energy distribution of objects in the neighbourhood of the Earth.

The most satisfying aspect of their measurement is the fact that the bridge between the larger and smaller objects fits the gap so well. It's as though a bridge were designed in Australia, built in France, transported to the United States and dropped seamlessly into a location in the Grand Canyon. Indeed, to within the errors of measurement, a single power-law seems to fit the size or energy distribution over ten orders of magnitude in energy (see Fig. 4 on page 296). This simple behaviour has been predicted analytically for self-similar collisional cascades⁸, in which a set of objects whose physical strength is independent of their size grind into, strike and disrupt one another.

In 1908 a tremendous explosion, estimated to be equivalent to about ten megatonnes of TNT, flattened trees over hundreds of square kilometres near Tunguska, Siberia (Fig. 1). If a similar event were to take place over a densely populated region of the world today, the death toll could easily be many millions of people. By extending the bridge they had measured by about 2.5 orders of magnitude in energy, Brown et al. estimate that the average time between impacts of this energy is about a thousand years — five times longer than was thought only ten or twenty years ago. Their results for a Tunguska-like impactor dovetail nicely with earlier and unpublished measurements by others, as shown in their Fig. 4. Because there are fewer of these objects than originally believed, we can all worry a little less about the risk of the next hazardous impact. Yet it is essential to emphasize that the job of cataloguing all objects down to the size of the Tunguska parent body is not currently achievable, and probably will not be completed for many decades.

The timely measurement by Brown et al.¹ of the size or energy distribution of objects in this range has linked the fields of meteor and comet/asteroid planetary astronomy in a manner that shows they are not merely distant relatives but kissing cousins. It seems that, as well as recognizing the Earth's atmosphere as a 'just right' detector system, Goldilocks has charmed the US government into releasing valuable scientific information obtained serendipitously while its satellites keep watch over the world's nuclear arsenals.