It was immediately clear that the time profiles of the bursts were very different, with some events lasting a few ms and others that lasted for more than 1 h [2]. In one case, variability in a timescale of 0.2 ms has been measured [3], implying an emission region of linear dimensions less than 60 km, in the absence of relativistic motions. The distribution of GRB durations has been studied in the past [4, 5] and a bimodal distribution was noticed. This has been confirmed recently [6] and the events have been classified in two classes: short events (< 2 s) and long events (> 2 s). The short GRBs are apparently harder than the longer ones. In general, there is no evidence of periodicity in the time histories of the GRBs, and only three of them have shown clearly to repeat. These three sources have been named Soft Gamma-Ray Repeaters (SGRs) and they are set apart as class different from the classical GRBs: they show multiple repetitions and soft spectra (blackbody spectra with temperatures of few keV). Two of the SGRs are associated with supernova remnants.
The typical GRBs fluences (the integral of the flux over time) observed by the different experiments lie in the range 10E(-4)-10E(-8) erg/cm2. Exceptionally, GRBs with fluences up to 10E(-3) erg/cm2 have been observed. The spectral capabilities of some instruments revealed that there were a large diversity of spectra. The continuum of the spectra is very broad, usually very hard and non thermal. In a few cases, soft bursts (kT = 1.2-1.6 keV) have been detected [8] whereas in GRB 940217 there is emission up to 18 GeV 1.5 h after the onset of the event [9]. Some events show spectral breaks at E = 2.5 MeV [10] and other indicate spectral curvature at higher energies [11]. For some GRBs, the KONUS experiment on Venera 11 and 12 (among others) reported absorption lines at 30 and 60 keV and emission lines at 400 keV [12], interpreted as the redshifted electron-positron annihilation line (511 keV) in the presence of a strong magnetic field. But perhaps the most striking spectrum is the one for the GRB that ocurred on February 5, 1988 (GRB 880205). Observations by GINGA revealed spectral features [13] consistent with cyclotron harmonics, that were taken into account as an indication of a strongly magnetized neutron star origin. However, the BATSE experiment failed to detected any absorption features in 142 GRBs, so the reality of the lines is still open [14].
Several methods have been used for determining the position of the bursts. Positions with error of several degrees have been determined when combining the relative intensities of the burst when seen by different detectors in the same spacecraft. Each detector will detect N x cos(theta), where N is the number of counts that would be seen if the burst arises in right angle towards the detector, and theta is the angle at which the burst is seen by the detector. In other cases, imaging instruments or rotation collimators like WATCH on GRANAT, can reach an accuracy of the order of 1 degree. But the best positions (square arcminutes error boxes) are achieved when the differences in arrival times by widely separated spacecrafts (the three Interplanetary Networks) are considered. The KONUS experiment determined the error boxes for more than 150 GRB by means of the first method. It gave the first indication that GRB sources were isotropically distributed in the sky and therefore they did not follow the distribution of the bright X-ray sources towards the Galactic Plane [12, 16]. This result has been confirmed for the weakest events by the BATSE experiment on the Compton Gamma-Ray Observatory [7, 17]. Thus it was possible to quantify the degree of isotropy of GRBs sources in the sky by means of the dipole and quadrupole moments with respect to the Galactic Centre and the Galactic Plane. The last BATSE measurement of the dipole moment of the angular distribution of 1121 GRBs gives <cos(theta)> = -0.013 +/- 0.019. The V/Vmax test is less affected by the uncertainty in the detection threshold. For each particular burst seen by a detector, a number is given. This is the rate between the maximum hypothetical volume within which the burst could be detected by the instrument (at minimum intensity) and the inner volume V to which the source can be restricted on the basis of its observed intensity. This number is just V/Vmax. For instance, if a distribution of sources is spatially uniform, isotropic and static in an euclidean geometry, the V/Vmax values will be uniformly distributed between 0 and 1, and the mean value of V/V_max will be 0.5. If the sources are distributed over a volume which is significantly smaller than the maximal volume from which sources of this luminosity could be detected, then <V/Vmax> will be smaller than 0.5. This could be the case for a galactic population of sources. For 520 BATSE bursts, <V/Vmax>=0.321 +/- 0.013 [19], which is less than 0.5, the expected value for a uniform distribution of sources. Therefore, it seems that we are at the centre of a distribution of sources, and we do see some confinement, but what we can not estimate is the size of such a distribution. Recently, it has been shown that no single population of sources situated in the Galactic disk, or halo can explain the BATSE observations. One model that is still in agreement with the data is an extended spherical Galactic corona [22]. Other possibilities are cosmological models and multiple-population models.