A possible explanation for the peculiar correlations in the angular distribution of -ray bursts

It has been recently discovered that the angular autocorrelation function ofgamma-ray bursts exhibits sharp peaks at angular separations of $\lesssim4\deg$ (Quashnock and Lamb 1993), and at $\gtrsim 176\deg$ (Narayan and Piran1993). While an excess of very close pairs of bursts can naturally arise fromburst repetition or from a spatial correlation of the burst sources, a physicalexplanation for an angular correlation on a scale of $\simeq 180\deg$ seemsinconceivable. We show that both sharp peaks in the correlation function can be explained bya possible bias in the determination of the burst positions. A generic way isdescribed in which the suggested bias can be introduced into the burstlocalization procedure, either through instrumental imperfection or through thesoftware analyses. We apply Monte Carlo simulations to show that the observedcorrelation function can be reproduced by the suggested effect. We demonstratethat the results can nicely agree with the observations even if only a fractionof the bursts are subject to the bias. We emphasis that the only motivation for suggesting the existence of thisbias are the features found in the angular autocorrelation function. It doesnot rule out the possibility that bursts repeat. The natural way in which suchbias, if it exists, explains both sharp peaks, and the various conceivablecauses for the origin of this bias, make the bias hypothesis worth considering.Comment: 5 pages + 5 figures, Latex, Preprints availabl