Following the discovery by Quashnock and Lamb (1993) of an apparent excess of $��$-ray burst pairs with small angular separations, we reanalyze the angular distribution of the bursts in the BATSE catalogue. We find that in addition to an excess of close pairs, there is also a comparable excess of antipodal bursts, i.e pairs of bursts separated by about 180 degrees in the sky. Both excesses have only modest statistical significance. We reject the hypothesis put forward by Quashnock and Lamb that burst sources are repeaters, since it is obvious that this hypothesis does not predict an excess of antipodal coincidences. Lacking any physical model of bursts that can explain the antipodal pairs, we suggest that the two excesses seen in the data are either due to an unusual statistical fluctuation or caused by some unknown selection effect.

6 pages, standard tex, 2 figures can be obtained from the authors by e-mail on request