We analyse the BATSE 3B catalog using the pair-matching statistic. This statistic counts only the burst pairs which may have originated from the same source, so it is less likely to yield false detections of ``repeating bursts than the nearest neighbor and correlation function statistics. Even in the ideal case when repeating is the only possible source of burst correlations, the pair matching statistic is more sensitive to repeating bursts than these other statistics particularly for models which predict faint or multiple burst repetitions. We find that the BATSE 3B data set contains no excess of matched burst pairs over the expectation from a sample with random positions. We also apply the pair-matching statistic to the bursts that previously appeared in the BATSE 1B catalog which now have improved positions and position errors in the BATSE 3B data set. Previously, these bursts had exhibited some peculiar position correlations that were interpreted by some as evidence for burst repetition, but we find that these correlations have disappeared with the improved BATSE 3B positions.
Download