DO GAMMA-RAY BURST SOURCES REPEAT?

ABSTRACTThe demonstration of repeated gamma-ray bursts from an individual source would severely constrain burstsource models. Recent reports (Quashnock & Lamb 1993; Wang & Lingenfelter 1993) of evidence for repeti-tion in the first BATSE burst catalog have generated renewed interest in this issue. Here, we analyze theangular distribution of 585 bursts of the second BATSE catalog (Meegan et al. 1994). We search for evidenceof burst recurrence using the nearest and farthest neighbor statistic and the two-point angular correlationfunction. We find the data to be consistent with the hypothesis that burst sources do not repeat; however, arepeater fraction of up to about 20% of the observed bursts cannot be excluded.Subject headings: gamma rays: bursts 1. INTRODUCTION The isotropic and inhomogeneous spatial distributionderived from BATSE burst data (Meegan et al. 1992; Fishmanet al. 1994) severely constrains possible Galactic distributionsand argues in favor of sources at cosmological distances.Although neutron stars in an extended Galactic halo were con-sidered as an alternative to cosmological models, the obser-vational constraints are now so severe that a halo origin ofbursts appears unlikely (Hakkila et al. 1994; Hartmann et al.1994; Briggs et al. 1995). In contrast, cosmological modelsnaturally explain the observed isotropy and inhomogeneity,but some of these are incapable of producing multiple burstsfrom one source.The absence of an excess of overlapping error circles in pre-BATSE burst localizations provides a model-dependent upperlimit of ~ 10 years on the burst repetition timescale (Schaefer& Cline 1985; Atteia et al. 1987). Several recent reports,however, have cited evidence for repetition on much shortertimes. Quashnock & Lamb find an excess of close neighbors inthe IB burst catalog, and conclude that a large fraction ofclassical bursts repeat on timescales of order months(Quashnock & Lamb 1993). Wang & Lingenfelter (1993) claimevidence for repetition with recurrence times perhaps as shortas days from one particular location (0855 — 00). Both Quash-nock & Lamb (1993) and Wang & Lingenfelter (1993) interprettheir results as evidence for multiple repetitions, of about fiveobserved bursts per repeating source. Also, the coincidence ofthe locations of GRB 940301 and GRB 930704 determinedwith COMPTEL has a 3% chance probability, suggesting apossible repeater (Kippen et al. 1995).To investigate burst repetition, we test the null hypothesisthat " the angular distribution of gamma-ray bursts (GRBs) isconsistent with the isotropic distribution," i.e., equal probabil-ity per unit solid angle. Tests of isotropy have varying sensiti-vities to clustering, which can indicate the presence of