We found five 2-8 s bursts similar to the burst-like events reported by Kuulkers et al. (1995). The times and the properties are given in Table 3. In Figs. 1c and e and Fig. 3a it is indicated where in the Z track the bursts occured. The occurence of the burst seems to be uncorrelated with the overall intensity level, the orbital phase, or the location in the Z track (although no burst were detected on the FB).
For the ratio, , of the average persistent flux to the time-averaged flux emitted in the bursts (which reflects the ratio between the gravitational and nuclear burning energy per gram of accreted matter for thermonuclear bursts), we could only obtain lower limits (Table 3), since there are many interruptions by SAA and/or Earth occultations. They were calculated using the average count rate from the start of the last data gap to just before the burst. The other typical burst parameters , the ratio of the mean persistent pre-burst flux and net peak burst flux, and , the ratio of the total integrated net burst flux and the net burst peak flux (a representation of the burst duration) were also calculated (see Table 3). The count rates used in order to calculate the burst parameters were corrected for deadtime, background and aspect. Burst IV was observed during time when the satellite was slewing to the source. Therefore, during that time the collimator transmission was low ( 50 %) and the uncertainty on the count rate large. The count rate for burst IV was not corrected for the overestimation of the count rates in the low photon energy bands, due to the reflection of low energy photons (below 6 keV) agains the collimator walls (see Sect. 2.1).
Due to the low time resolution of the data obtained during the occurence of bursts I, II and V, we did not examine the burst profiles and spectral properties of these events. The burst profiles, for different energy bands, of bursts III and IV are shown in Figs. 10 and 11, respectively. These bursts do not show evidence for spectral cooling as would expected for bona fide type I bursts. Instead, it is clearly visible that burst III shows evidence for spectral hardening: the higher the energies, the broader the event. Also, the post-burst count rate is higher than the pre-burst count rate. This effect is most prominent at the higher energies. Although these phenomena are not seen during event IV, no evidence for spectral cooling is seen either. In the nine bursts in the EXOSAT data of Cygnus X-2 hints for spectral hardening were already found (Kuulkers et al. 1995), although the poor statistics made a definite conclusion impossible. Taking all bursts into account we conclude that these events are probably not bona fide type I. We searched for QPOs and periodic oscillations during burst III and IV and found none. Statistics were insufficient to set meaning full upper limits.
© European Southern Observatory (ESO) 1997
Online publication: June 5, 1998