Severe new limits on the host galaxies of gamma-ray bursts

The nature of gamma-ray bursts (GRBs) remains a complete mystery, despite the recent breakthrough discovery of low-energy counterparts, although it is now generally believed that at least most GRBs are at cosmological distances. Virtually all proposed cosmological models require bursters to reside in ordinary galaxies. This can be tested by looking inside the smallest GRB error boxes to see if ordinary galaxies appear at the expected brightness levels. This Letter reports on an analysis of the contents of 26 of the smallest regions, many of which are from the brightest bursts. These events will have z < 0.4 and small uncertainties around the luminosity functions, K-corrections, and galaxy evolutions, whereas the recent events with optical transients are much fainter and hence have high redshifts and grave difficulties in interpretation. This analysis strongly rejects the many models with peak luminosities of 10(57) photons s(-1) as deduced from the log N-log P curve with no evolution. Indeed, the lower limit on acceptable luminosities is 6 x 10(58) photons s(-1). The only possible solution is either to place GRBs at unexpectedly large distances (with z > 5.9 for the faint BATSE bursts) or to require bursters to be far outside any normal host galaxy.