2006ApJ...646..407R

We measure the rates of type I X-ray bursts from a likely flux-limited sample of 37 nonpulsing Galactic transients observed with RXTE during 1996-2004. These sources are well categorized in the literature as either neutron star systems or black hole candidates. Our goals are to test the burst model for neutron stars and to investigate whether black holes have event horizons. Target selection is one of several differences between the present study and the investigation of the event horizon question by Tournear and coworkers. Burst rates are measured as a function of bolometric luminosity, and the results are compared with augmented versions of the burst model developed by Narayan & Heyl. For a given mass, we consider a range in both the radius and the temperature at the boundary below the accretion layer. We find 135 spectrally confirmed type I bursts in 3.7 Ms of PCA exposures for the neutron star group (13 sources), and the burst rate function is generally consistent with the model predictions. However, for the black hole groups (18 sources), there are no confirmed type I bursts in 6.5 Ms of exposure, and the upper limits in the burst function are inconsistent with the model predictions for heavy compact objects with a solid surface. The consistency probability is ∼2x10^–7 for dynamical black hole binaries, falling to 3x10^–13 with the added exposures of black hole candidates. Furthermore, there are systematic spectral differences between the neutron star and black hole groups, supporting the presumption that physical differences underly the classifications in our sample. These results provide indirect evidence that black holes do have event horizons.