Abstract. Despite a rich diversity in observational properties, gamma-ray bursts (GRBs) can be divided into two broad categories based on their duration and spectral hardness -the long-soft and the short-hard GRBs. The discovery of afterglows from long GRBs in 1997, and their localization to arcsecond accuracy, was a watershed event. The ensuing decade of intense study led to the realization that long-soft GRBs are located in star forming galaxies, produce about 10 51 erg in collimated relativistic ejecta, are accompanied by supernovae, and result from the death of massive stars. While theoretical arguments suggest that short GRBs have a different physical origin, the lack of detectable afterglows prevented definitive conclusions. The situation changed dramatically starting in May 2005 with the discovery of the first afterglows from short GRBs localized by Swift and HETE-2. Here I summarize the discovery of these afterglows and the underlying host galaxies, and draw initial conclusions about the nature of the progenitors and the properties of the bursts.
HISTORY AND MODELSThe detection of short-duration gamma-ray bursts (GRBs) dates back to the Vela satellites [1]. However, only in 1993 the short bursts (with T 90 ∼ < 2 s) were recognized as a separate sub-class from the long GRBs, and were furthermore shown to have on average a harder γ-ray spectrum [2]; hereafter I will refer to short-hard bursts as SHBs. The short durations of SHBs suggest that they are unlikely to result from the death of massive stars, for which the natural timescale (the free-fall time) is significantly longer, t f f ≈ 30 s (M/10 M ⊙ ) −1/2 (R/10 10 cm) 3/2 .Instead the main theoretical thrust has been focused on coalescing compact objects -neutron stars and/or black holes (DNS or NS-BH) -as the progenitors of SHBs [3,4,5,6,7]. In this context the duration, which is set by the viscous timescale of the gas accreting onto the newly-formed black hole, is short due to the small scale of the system. Other progenitors have been proposed in addition to the DNS and NS-BH systems, namely magnetars, thought to be the power source behind soft γ-ray repeaters [8], and accretion-induced collapse (AIC) of neutron stars [9,10]. The magnetar model is unlikely to account for SHBs at cosmological distances, since even the 2004 Dec. 27 giant flare from SGR 1806-20 would only be detected by BATSE or Swift at ∼ < 50 Mpc; magnetars may contribute to a local population of SHBs. The AIC model has not been investigated in detail, but in the case of a white dwarf was shown to be too baryon rich to produce GRBs [11].Naturally, without a distance and energy scale, or an understanding of the microand macro-environments of SHBs, it is nearly impossible to make any quantitative statements about their progenitors or the detailed underlying physics. As in the case of