The Discovery of Gamma-Ray Bursts

Summary

Satellites built to detect nuclear explosions in space recorded brief flashes of gamma rays that came from neither the Earth nor the Sun. It took thirty years to learn that the flashes come from other galaxies, and fifty to see one arrive with the gravitational waves of two neutron stars merging.

The explanation is secure in outline: collapsing massive stars make the long bursts, merging neutron stars at least some of the short ones. The exceptions are not settled. In July 2025 a burst repeated over the course of a day, and no known scenario fully explains it.

What is documented

The satellites. In 1959 the Los Alamos laboratory was charged with developing a satellite system to detect nuclear devices exploded in space. The US Air Force launched the first Vela satellites in October 1963, in a series prompted by that year's nuclear test ban treaty, and further pairs followed until 1970. Each pair shared a circular orbit about 120,000 kilometres from the centre of the Earth.

The first burst. On 2 July 1967 the satellites Vela 4a and 4b recorded a burst of gamma rays that also triggered the older Vela 3 satellites. It was found two years later, in 1969, when Ray Klebesadel and Roy Olson of Los Alamos looked back over the Vela 4 data shortly before Vela 5 was launched.

Why the announcement waited. According to J. T. Bonnell's history of the discovery, drafted in 1995 and published with Klebesadel in 1996, the satellites of 1967 could not time a burst's arrival precisely enough to fix its direction. The later Vela satellites could. In 1972 Ian Strong, also at Los Alamos, was asked to look at Klebesadel's files, and Klebesadel, Strong and Olson used the better timing to work out directions. Their paper appeared in The Astrophysical Journal in June 1973. It reported sixteen bursts recorded between July 1969 and July 1972, with photon energies of 0.2 to 1.5 MeV and durations from under a tenth of a second to about 30 seconds. The directions ruled out the Earth and the Sun as sources. The 1967 burst was not among the sixteen.

No counterpart, no distance. No burst could be matched to anything seen at other wavelengths, and without a counterpart there was no way to measure a distance. NASA's Compton Gamma Ray Observatory, launched on 5 April 1991, carried the Burst and Transient Source Experiment (BATSE), which detected about one burst a day. In January 1992 Charles Meegan and colleagues reported on 153 of them: spread evenly across the sky, but with fewer faint bursts than a uniform population would give. They found the combination inconsistent with any known population of objects in the Milky Way and possibly consistent with cosmological distances. By the end of the observatory's nine-year mission BATSE had recorded 2,704 bursts.

The debate. In April 1995, in the Smithsonian auditorium where Harlow Shapley and Heber Curtis had debated in 1920, Bohdan Paczyński argued that the bursts were at cosmological distances. Donald Lamb argued that they came from fast-moving neutron stars in a distant corona around the Milky Way, and that the evidence added up to "a strong case for the Galactic hypothesis." Martin Rees moderated.

The afterglow. After a burst on 28 February 1997, GRB 970228, the Italian-Dutch X-ray satellite BeppoSAX detected a fading X-ray "afterglow", the first such detection for any burst. Less than 21 hours after the burst, van Paradijs and colleagues found a fading optical source in its error box, apparently in a faint galaxy. On 8 May 1997 came GRB 970508. Metzger and colleagues found absorption lines in the spectrum of its optical counterpart, from material along the line of sight at a redshift of 0.835, so the source was at least that distant: billions of light years away. They called the result the first direct limits on the distance to a gamma-ray burst.

Two kinds of burst. Bursts fall into two classes: long ones, lasting more than about 2 seconds, and short ones. An unusual and very luminous supernova, SN 1998bw, appeared in the error box of a burst of 25 April 1998, GRB 980425. A very energetic supernova appeared at the position of GRB 030329, a burst of 29 March 2003, and Jens Hjorth and colleagues found that it had exploded within a few days of the burst, which they took as strong evidence that the collapse of a massive star's core can give rise to one, "thereby favouring the 'collapsar' model." Short bursts were harder to locate. The first to be placed quickly and precisely, GRB 050509B in May 2005, lay near an elliptical galaxy with no star formation, where a merger of neutron stars or black holes would be expected.

The merger. On 17 August 2017 the LIGO and Virgo detectors recorded gravitational waves from two neutron stars merging, GW170817. About 1.7 seconds after the waves arrived, NASA's Fermi Gamma-ray Space Telescope recorded a short burst, GRB 170817A, which the INTEGRAL satellite also detected. The source was in the galaxy NGC 4993, about 130 million light years away. LIGO's announcement said that astronomers could now say with certainty "that at least one source of short gamma ray bursts in the universe is merging neutron stars."

The exceptions. GRB 211211A, in December 2021, lasted about a minute, which makes it a long burst by the usual rule, yet it was followed by what its discoverers reported as a kilonova, the glow that marks a merger of compact objects. On 2 July 2025, fifty-eight years to the day after the first Vela burst, Fermi recorded three bursts from a single source over several hours, and the Einstein Probe was found to have seen the source active almost a day earlier. Observations with ESO's Very Large Telescope and the Hubble Space Telescope placed it in a galaxy perhaps a few billion light years away. One study measured about 25,000 seconds of gamma-ray emission, the longest burst yet seen.

The same satellite programme later recorded the optical double flash of 22 September 1979 over the southern Indian Ocean, a different event with its own record (AR-0083).

Leading explanations

Sources in distant galaxies. Established. By 1998 Galama and colleagues could write that afterglows and redshifts had "established that γ-ray bursts lie at extreme distances."

Long bursts: the collapse of massive stars (the collapsar model). Established for most long bursts. The supernovae found with GRB 980425 and GRB 030329 are the direct evidence. A 2022 paper in Nature opens by stating the standard division: long bursts from the core collapse of massive stars, short bursts from the merger of two compact objects.

Short bursts: mergers of neutron stars. Established for at least some. A host galaxy without star formation pointed to a merger in 2005. GW170817 showed it directly in 2017.

Neutron stars in a corona around the Milky Way (Lamb, 1995). Superseded. It fitted the BATSE sky map, which could not distinguish between the two hypotheses. The redshifts of 1997 settled the question for the bursts measured.

The outliers (open). For GRB 250702B, Andrew Levan, Antonio Martin-Carrillo and colleagues favour a white dwarf torn apart by an intermediate-mass black hole. Eliza Neights and colleagues propose a black hole falling into a helium star and consuming it from within. ESO said that "no known scenario can completely explain this new GRB." None of these proposals is established.

Current status

Later explained. The anomaly of 1973 was a flash of unknown origin at an unknown distance. Both questions now have answers, each from direct observation: the bursts come from other galaxies, and the two main engines are the collapse of massive stars and the merger of neutron stars. What remains open is at the edges: bursts that break the duration rule, and GRB 250702B, whose nature was unresolved at the time of this review. A new kind of source would add a class of burst; it would not reopen the question of where the bursts are.

Last reviewed: September 2026. Records are provisional. Where the evidence changes, the entry changes. Found an error? Tell us.