Early in the morning of January 23, 1999, a robotic telescope in New Mexico picked up a faint flash of light in the constellation Corona Borealis. Though just barely visible through binoculars, it turned out to be the most brilliant explosion ever witnessed by humanity. We could see it nine billion light-years away, more than halfway across the observable universe. If the event had instead taken place a few thousand light-years away, it would have been as bright as the midday sun, and it would have dosed Earth with enough radiation to kill off nearly every living thing.

The flash was another of the famous gamma-ray bursts, which in recent decades have been one of astronomy's most intriguing mysteries. The first sighting of a gamma-ray burst (GRB) came on July 2, 1967, from military satellites watching for nuclear tests in space. These cosmic explosions proved to be rather different from the man-made explosions that the satellites were designed to detect. For most of the next 30 years, each new burst had merely heightened the puzzlement. Whenever researchers thought they had the explanation, the evidence sent them back to square one.

The monumental discoveries of the past decade have brought astronomers closer to a definitive answer. Before 1997, most of what we knew about GRBs was based on observations from the Burst and Transient Source Experiment (BATSE) onboard the Compton Gamma Ray Observatory. BATSE revealed that two or three GRBs occur somewhere in the observable universe on a typical day. They outshine everything else in the gamma-ray sky. Although each is unique, the bursts fall into one of two rough categories. Bursts that last less than two seconds are short, and those that last longer--the majority--are long. The two categories differ spectroscopically, with short bursts having relatively more high-energy gamma rays than long bursts do. The January 1999 burst emitted gamma rays for a minute and a half.

Arguably the most important result from BATSE concerned the distribution of the bursts. They occur isotropically--that is, they are spread evenly over the entire sky. This finding cast doubt on the prevailing wisdom, which held that bursts came from sources within the Milky Way; if they did, the shape of our galaxy, or Earth's off-center position within it, should have caused them to bunch up in certain areas of the sky. The uniform distribution led most astronomers to conclude that the instruments were picking up some kind of event happening throughout the universe. Unfortunately, gamma rays alone did not provide enough information to settle the question for sure. Researchers would need to detect radiation from the bursts at other wavelengths. Visible light, for example, could reveal the galaxies in which the bursts took place, allowing their distances to be measured. Attempts were made to detect these burst counterparts, but they proved fruitless.

A Burst of Progress

THE FIELD TOOK a leap forward in 1996 with the advent of the x-ray spacecraft BeppoSAX, built and operated by the Italian Space Agency with the participation of the Netherlands Space Agency. BeppoSAX was the first satellite to localize GRBs precisely and to discover their x-ray afterglows. The afterglow appears when the gamma-ray signal disappears. It persists for days to months, diminishing with time and degrading from x-rays into less potent radiation, including visible light and radio waves. Although BeppoSAX detected afterglows for only long bursts--eight more years would pass before counterparts of short bursts would be discovered--it made follow-up observations possible at last. Given the positional information from BeppoSAX, optical and radio telescopes were able to identify the galaxies in which the GRBs took place. Nearly all lie billions of light-years away, meaning that the bursts must be enormously powerful. Extreme energies, in turn, call for extreme causes, and researchers began to associate GRBs with the most extreme objects they knew of: black holes.

Among the first GRBs pinpointed by BeppoSAX was GRB970508, so named because it occurred on May 8, 1997. Radio observations of its afterglow provided an essential clue. The glow varied erratically by roughly a factor of two during the first three weeks, after which it stabilized and then began to diminish. The large variations probably had nothing to do with the burst source itself; rather they involved the propagation of the afterglow light through space. Just as Earth's atmosphere causes visible starlight to twinkle, interstellar plasma causes radio waves to scintillate. For this process to be visible, the source must be so small and far away that it appears to us as a mere point. Planets do not twinkle, because, being fairly nearby, they look like disks, not points.

Therefore, if GRB970508 was scintillating at radio wavelengths and then stopped, its source must have grown from a mere point to a discernible disk. Discernible in this case means a few light-weeks across. To reach that size, the source must have been expanding at a considerable rate--close to the speed of light.

The BeppoSAX and follow-up observations have transformed astronomers view of GRBs. The old concept of a sudden release of energy concentrated in a few brief seconds has been discarded. Indeed, even the term afterglow is now recognized as misleading: the energy radiated during both phases is comparable. The spectrum of the afterglow is characteristic of electrons moving in a magnetic field at or very close to the speed of light.

GRB990123--the January 1999 burst--was instrumental in demonstrating the immense power of the bursts. If the burst radiated its energy equally in all directions, it must have had a luminosity of a few times 1045 watts, which is 1019 times as bright as our sun. In comparison, the photonic emissions of a supernova explosion are spread out over several weeks, with a luminosity that is only a tiny fraction of that of a GRB. Even quasars, which are famously brilliant, give off only about 1040 watts.

If the burst beamed its energy in particular directions rather than in all directions, however, the luminosity estimate would be lower. Evidence for beaming comes from the way the afterglow of GRB990123, among others, dimmed over time. Two days into the burst, the rate of dimming increased suddenly, which would happen naturally if the observed radiation came from a narrow jet of material moving at close to the speed of light. Because of a relativistic effect, the observer sees more and more of the jet as it slows down. At some point, there is no more to be seen, and the apparent brightness begins to fall off more rapidly [see box on next page].

For GRB990123 and several other bursts, the inferred jet-opening angle is a few degrees. Only if the jet is aimed along our line of sight do we see the burst. This beaming effect reduces the overall energy emitted by the burst approximately in proportion to the square of the jet angle. For example, if the jet subtends 10 degrees, it covers about one 500th of the sky, so the energy requirement goes down by a factor of 500; moreover, for every GRB that is observed, another 499 GRBs go unseen. Even after taking beaming into account, however, the luminosity of GRB990123 was still an impressive 1043 watts.

GRB-Supernova Connection

ONE OF THE MOST interesting discoveries has been the connection between GRBs and supernovae. When telescopes went to look at GRB980425, they also found a supernova, called SN1998bw, that had exploded at approximately the same time as the burst. The probability of a chance coincidence was one in 10,000. A firmer case is the association of GRB030329 with SN2003dh. This GRB was localized by NASA's second High Energy Transient Explorer satellite (HETE-2), launched in October 2000. Ground-based observations revealed the broad spectroscopic features of a supernova, basically identical to those of SN1998bw, 10 days after the GRB. The best case by far is GRB060218, which is tied quite nicely to SN2006aj. This GRB was discovered by NASA's Swift satellite, launched in November 2004. Ground-based telescopes were intensely scrutinizing the fading afterglow when the supernova appeared, three days after the GRB.

Of the three cases mentioned above, GRB030329 comes closest to being a normal long GRB; GRB980425 and GRB060218 are unusual in that they are underluminous, of long duration and predominantly x-ray events. Also, these two bursts occurred in, by GRB standards, relatively nearby galaxies. The two have long spectral lags, meaning that the high- and low-energy gamma-ray pulses arrive several seconds apart. These bursts may be best described as x-ray flashes, which will be explained later.

In addition to GRB030329, there is strong evidence that other normal long GRBs are associated with supernovae. GRB970228 was the first BeppoSAX GRB for which an optical afterglow was discovered. At 30 days after the burst, a bump in its optical light curve appeared that looked a lot like a supernova.

A link between GRBs and supernovae has also been suggested by the detection of metals, most notably iron, in the x-ray spectra of several bursts. Iron atoms are known to be synthesized and dumped into interstellar space by supernovae explosions. If these atoms are stripped of their electrons and later hook up with them again, they give off light at distinctive wavelengths, referred to as emission lines. Early detections of such lines by BeppoSAX and the Japanese x-ray satellite ASCA have been followed up with more solid measurements. Notably, NASA's Chandra X-ray Observatory detected iron lines in GRB991216, which yielded a direct distance measurement of the GRB. The figure agreed with the estimated distance of the burst's host galaxy. And in the shell of gas around GRB011211, the European Space Agency's X-ray Multi-Mirror Satellite found evidence of emission lines from silicon, sulfur, argon and other elements commonly released by supernovae.

A connection between GRBs and supernovae is now generally accepted by astronomers. Because GRBs are much rarer than supernovae--every day a couple of GRBs go off somewhere in the universe, as opposed to hundreds of thousands of supernovae--not every supernova can be associated with a burst. Perhaps jetting inside a supernova is common, and, in some small fraction of cases, relativistic jets escape from the supernova, and, in some small fraction of those cases, one of the jets is directed toward us, which allows us to observe a GRB. Also, if the jet is pointed just slightly away from us, then we may observe a lower-energy event, with more x-rays than gamma rays.

Great Balls of Fire

EVEN LEAVING ASIDE the question of how the energy in GRBs might be generated, their sheer brilliance poses a paradox. Rapid brightness variations suggest that the emission originates in a small region: a luminosity of 1019 suns comes from a volume the size of one sun. With so much radiation emanating from such a compact space, the photons must be so densely packed that they should interact and prevent one another from escaping. The situation is like a crowd of people who are running for the exit in such a panic that that nobody can get out. But if the gamma rays are unable to escape, how can we be seeing GRBs?

The resolution of this conundrum, developed over the past 10 years, is that the gammas are not emitted immediately. Instead the initial energy release of the explosion is stored in the kinetic energy of a shell of particles--a fireball--moving at close to the speed of light. The particles include electrons and their antimatter counterpart, positrons. This fireball expands to a diameter of 10 billion to 100 billion kilometers, by which point the density has dropped enough for the gamma rays to escape unhindered. The fireball then converts some of its kinetic energy into electromagnetic radiation, yielding a GRB.

The initial gamma-ray emission is most likely the result of internal shock waves within the expanding fireball. Those shocks are set up when faster blobs in the expanding material overtake slower blobs. Because the fireball is expanding so close to the speed of light, the timescale witnessed by an external observer in the path of the fireball is vastly compressed, according to the principles of relativity. So the observer in the path of the fireball sees a burst of gamma rays that lasts only a few seconds, even if it took a day to produce. The fireball continues to expand, and eventually it encounters and sweeps up surrounding gas. Another shock wave forms, this time at the boundary between the fireball and the external medium, and persists as the fireball slows down. This external shock nicely accounts for the GRB afterglow emission and the gradual degradation of this emission from gamma rays to x-rays to visible light and, finally, to radio waves.

Although the fireball can transform the explosive energy into the observed radiation, what generates the energy to begin with? That is a separate problem, and astronomers have yet to reach a consensus. One family of models, referred to as hypernovae or collapsars, involves stars born with masses greater than about 20 to 30 times that of our sun. Simulations show that the central core of such a star eventually collapses to form a rapidly rotating black hole encircled by a disk of leftover material.

A second family of models invokes binary systems that consist of two compact objects, such as a pair of neutron stars (which are ultradense stellar corpses) or a neutron star paired with a black hole. Such a system loses orbital energy as a result of the emission of gravitational radiation, and so the two objects spiral toward each other and merge into one. Just as in the collapsar scenario, the result is the formation of a single black hole surrounded by a disk.

Many celestial phenomena involve a hole-disk combination. What distinguishes this particular type of system is the sheer mass of the disk (which allows for a gargantuan release of energy) and the lack of a companion star to resupply the disk (which means that the energy release is, more or less, a one-shot event). The black hole and disk have two large reservoirs of energy: the gravitational energy of the disk and the rotational energy of the hole. Exactly how these are converted into a fireball is not fully understood. It is possible that a magnetic field, 1015 times more intense than Earth's magnetic field, builds up during the formation of the disk. In so doing, it heats the disk to such high temperatures that it unleashes a fireball of photons, neutrinos and plasma. The fireball is funneled into a pair of narrow jets that flow out along the rotational axis.

In addition to collapsar and compact-object merger models, it should be noted that there are other models for the central engine of a GRB. One involves the extraction of energy from an electrically charged black hole. In this scenario, both the prompt and afterglow GRB emissions are a result of the fireball sweeping up the external medium.

There is quite a bit of evidence to support the hypothesis that collapsars account for the long GRBs. In particular, the association of long GRBs with supernovae is a point in favor of collapsars, which, after all, are essentially large supernovae. Furthermore, long GRBs are usually found just where collapsars would be expected to occur--namely, in areas of recent star formation within galaxies. A massive star blows up fairly soon (a few million years) after it is born, so its deathbed is close to its birthplace.

The evidence is growing that compact-star coalescence accounts for the short-duration GRBs. This mechanism is not expected to produce a supernova, and, indeed, an association between short GRBs and supernovae has not been found. Also, the decay of the orbit of a pair of compact stars is a process that occurs on a range of timescales, from tens of millions to billions of years. In the former case, the merger will occur close to where the stars in the compact pair were born. In the latter case, the pair will drift around its host galaxy, and so the final coalescence is unlikely to have an association with any star-forming region. Such a mixed association of short GRBs with star formation is exactly what was found after Swift and HETE-2 discovered and localized the x-ray afterglows of several short bursts in 2005.

We still do not completely understand the differences between long and short GRBs. For example, the recent burst GRB060614 was a bright, well-observed, nearby event that does not fit cleanly into either category.

All these findings have shown that the field has the potential for answering some of the most fundamental questions in astronomy: How do stars end their lives? How and where are black holes formed? What is the nature of jet outflows from collapsed objects?

Blasts from the Past

ONE OUTSTANDING question concerns the dark, or ghost, GRBs. Of the roughly 200 GRBs that have been localized and studied at wavelengths other than gamma rays, about 90 percent have been seen in x-rays. In contrast, only about 50 percent have been seen in visible light. Why do some bursts fail to shine in visible light?

There are several effects that can make a burst dark. One explanation is that these GRBs lie in regions of star formation, which tend to be filled with dust. Dust blocks visible light but not x-rays. Another intriguing possibility is that some of the ghosts are GRBs that happen to be very far away. The relevant wavelengths of light produced by these bursts would be absorbed by intergalactic gas. To test this hypothesis, measurement of the distance via x-ray or infrared spectra will be crucial. A third possibility is that ghosts are optically faint by nature.

High-sensitivity optical and radio investigations have identified the probable host galaxies of some dark GRBs. Most of them are at moderate distances, favoring--for these events--the dust explanation. But one of them, pinpointed by Swift, is at high redshift in the dark universe region.

Another mystery concerns a class of events known as the x-ray-rich GRBs, or simply the x-ray flashes. Discovered by BeppoSAX, later confirmed by HETE-2 observation and reanalysis of BATSE data, and currently observed by Swift, these bursts represent 20 to 30 percent of GRBs. They give off more x-radiation than gamma radiation; indeed, extreme cases exhibit no detectable gamma radiation at all.

There are three possible explanations for the x-ray flashes. One is that the fireball is loaded with a relatively large amount of baryonic matter such as protons, making for a dirty fireball. These particles increase the inertia of the fireball, so that it moves more slowly and is less able to boost photons into the gamma-ray range. Alternatively, the x-ray flashes could be typical GRBs with jets that are pointing just out of our view, so that only the less collimated and less energetic x-rays reach us. A third possibility is that at least one of the x-ray flashes appears to be associated with a less extreme supernova explosion than is usual for normal long GRBs. There is speculation that, in this case, a neutron star, not a black hole, was formed in the supernova.

The next step for GRB astronomy is to accumulate observations of hundreds of bursts of all varieties to flesh out the data on burst, afterglow and host-galaxy characteristics. This effort is being spearheaded by the Swift satellite, thanks to its multiwavelength capabilities and its ability to quickly and autonomously reorient itself to better observe a burst with its high-resolution instruments. Swift's sensitivity to short-duration bursts has been a major factor in understanding this poorly studied class.

Another goal is to probe extreme gamma-ray energies. GRB940217, for example, emitted high-energy gamma rays for more than an hour after the burst, as observed by the Energetic Gamma Ray Experiment Telescope on the Compton Gamma Ray Observatory. Astronomers do not understand how such extensive and energetic afterglows can be produced. The Italian Space Agency's AGILE satellite, expected to launch in 2007, will observe GRBs at these high energies. The supersensitive Gamma-Ray Large Area Space Telescope mission, also scheduled for launch in 2007, will be key for studying this puzzling phenomenon.

Other missions, though not designed solely for GRB discovery, will also contribute. The International Gamma-Ray Astrophysics Laboratory, launched on October 17, 2002, is detecting more than 10 GRBs a year. The proposed Energetic X-ray Imaging Survey Telescope will have a sensitive gamma-ray instrument capable of detecting thousands of GRBs. The Explorer of Diffuse Emission and GRB Explosions (EDGE) is proposed to observe GRBs as cosmic beacons--to study the early stages of the universe and its evolution over time.

The field has experienced a series of breakthrough years, with the discovery that GRBs are immense explosions occurring throughout the universe. Bursts provide us with an exciting opportunity to study new regimes of physics and to learn what the universe was like at the earliest epochs of star formation. Space- and ground-based observations over the coming years should allow us to uncover the detailed nature of these most remarkable beasts. Astronomers can no longer talk of bursts as utter mysteries, but that does not mean the puzzle is completely solved.


NEIL GEHRELS, LUIGI PIRO and PETER J. T. LEONARD bring both observation and theory to the study of gamma-ray bursts. Gehrels and Piro are primarily observers--they were the lead scientists, respectively, of the Compton Gamma Ray Observatory and the BeppoSAX satellite. Leonard is a theorist, and like most theorists, he used to think it unlikely that the bursts were bright enough to be seen across the vastness of intergalactic space. I have to admit that the GRBs really had me fooled, he says. Gehrels is chief of the Astroparticle Physics Laboratory at the NASA Goddard Space Flight Center and lead scientist of the Swift satellite. Piro is director of research at the Institute of Space Astrophysics and Cosmic Physics of the INAF (National Institute of Astrophysics) in Rome. Leonard works for ADNET Systems, in support of missions at Goddard.