3 Classification of triggers

For each of the 4485 triggers, we derived sky positions on the assumption that they corresponded to the emission from a point source. We used the response function for the eight detectors (cf. Pendleton et al. 1995), and assumed a Band et al. (1993) type spectrum with $\alpha = -1$, $\beta = -2$, and break energy E₀ = 150 keV. The positions were derived by testing around 10,000 positions on the sky and minimizing $\chi^2$ of the counts for all eight detectors. We ignored the effects of Compton scattering on the Earth's atmosphere or the spacecraft. Figure 3 shows the celestial coordinates of all 4485 triggers.

It is clear from an inspection of Fig. 3 that at least four types of triggers are present. Both CygX-1 and Nova Persei 1992 are prominent. The sun is clearly shown by solar flares along the ecliptic. There is a fourth component which appears more or less isotropic.

  

Table 1: Classification of 4485 triggers

Plots of the number of triggers versus time close to the sky positions of CygX-1 and of Nova Persei 1992 show that both were detected during well-defined periods of activity. We used our positions of Nova Persei 1992, which totally dominated activity in its part of the sky for around 60 days, to evaluate the distribution of position errors. Based on these results, we eliminated from consideration all triggers within 23 deg of the sun, of CygX-1, and of Nova Persei 1992 during the periods that these sources were active, see Table 1.

We accept as GRBs all triggers whose onset was within 230.4 s of those listed for GRBs in the BATSE catalog (cf. Meegan et al. 1997 and the World Wide Web version maintained by the BATSE Burst team). For the remaining triggers, we inspected each of their time profiles from -300 to +400 s relative to the trigger time. In our judgment of the nature of these events we were guided by descriptions by the BATSE team of magnetospheric events (cf. Fishman et al. 1992; Horack et al. 1992), by the value of $\chi^2$ of the solution, by the dispersion of the positions obtained during each second that the burst was brighter than the limiting flux, and by a derivation of the angular motion of the source (which for a GRB should be consistent with zero). In the process, we noticed that 44 of the remaining triggers had spectra softer than any listed in the BATSE catalog, and that they were all within 35 deg of the sun. On the basis of this evidence, we rejected these as GRBs. The results of this excercise are shown in Table 1. We accepted 404 GRBs that are not listed in the BATSE catalog.

There are 130 GRBs listed in the BATSE catalog that we have not detected, while as stated we find 404 GRBs that are not in the catalog. The difference in content between the BATSE catalog and our sample is partly a consequence of differences in the de-activation of the trigger following a burst or around a defect in the data, and partly caused by statistical differences in the different backgrounds used.

Based on the entire search procedure, including the time windows of exclusion set up around bad data, the exclusion of particular geographic areas and of some sky areas in the classification procedure, and the part of the sky occulted by the Earth, we estimate that the sample of 1422 GRBs represents effectively 2.0 years of isotropic exposure by BATSE, for an annual detection rate of 710 GRBs per year.