We scanned 2068 days of BATSE daily records and found 1243 non-triggered events which can be classified as classic GRBs (Kommers et al. 1998 found 837 non-triggered GRBs per 2200 days). We found 1374 bursts which were triggered by BATSE (Kommers et al. 1998 detected 1393 BATSE triggered events), and missed near 350 BATSE triggers: some of them are in data gaps, some are too short to be detected with 1 s time resolution. We also found 3780 test bursts out of about 6800 added to the data. All test bursts were identified and excluded from the final sample. The peak flux distribution of real events found in the scan and classified as GRBs is presented in Fig. 1.
Data gaps and periods of a high ionospheric background are taken into account so the efficiency is normalized to whole elapsed time of CGRO operation.
|Figure 1: Differential peak flux distribution of detected GRBs. Thick histogram - the distribution of BATSE triggers detected in the scan (1374). Thin histogram - all bursts detected in the scan (2617). Both distributions are given before correction for the efficiency is applied|
|Figure 2: Hardness - peak flux plot for all detected GRBs. Solid line - median value of the hardness ratio for triggered GRBs; dotted line - the same for non-triggered GRBs|
The hardness - peak flux scattering plot shown in Fig. 2 demonstrates that new weak bursts give a direct continuation of the distribution of stronger GRBs. The well-known brightness-hardness correlation (Mallozi et al. 1995; Nemiroff et al. 1994), is clearly visible. Nemiroff et al. (1994) found the hardness ratio for 3/2 channels to vary by factor 1.54 from weak to strong bursts (64 ms resolution peak flux). In our case of 1024 ms resolution many short hard bursts fall to the weak end of the distribution reducing the correlation. With the same reason weak triggered bursts are slightly harder than non-triggered bursts. Nevertheless we see a reasonable global correlation where median hardness varies by factor 1.58 in approximate agreement with Nemiroff et al. (1994).
The resulting distribution in absolute units is presented in Fig. 3 in comparison with BATSE from Meegan et al. (1998) (in arbitrary normalisation) and that from Kommers et al. (1998) (in absolute units).
We fit the distributions with the simplest hypothesis of the standard candle non-evolving GRB population: the best fit gives = 50.2 at 22 degrees of freedom. For the first time the simplest cosmological model cannot fit data. The fit will be even worse if we use the star formation rate curve as the GRB evolution scenario. The rejection of the standard candle hypothesis is not surprising, however this is still an achievement because the cosmological fit of the becomes conclusive.
We thank Aino Skassyrskaia, Andrei Skorbun, Eugeni Stern, Vladimir Kurt, Kirill Semenkov, Stas Masolkin, Alex Sergeev, Max Voronkov and Felix Ryde for valuable assistance. We acknowledge Andrei Beloborodov and Juri Poutanen for useful discussions.
This work was supported by the Swedish Natural Science Research Council, the Royal Swedish Academy of Science, the Wennergren Foundation for Scientific Research, and a NORDITA Nordic Collaboration Project grant. RFBR grant 97-02-16975 supported one of the authors (D.K.).
|Figure 3: The differential distribution in absolute units for all events (2617) (1243 non-triggered) detected in the scan. Data of Kommers et al. (1998) are shown by circles, BATSE-3 distribution from Meegan et al. (1998) is shown by diamonds (in arbitrary normalisation). Our data are corrected for the "test bursts'' efficiency which is for bright bursts (due to data gaps and bad background intervals) and smoothly declines to the lowest peak fluxes|
Copyright The European Southern Observatory (ESO)