3. Data analysis
The lightcurves in B, V and R bands are shown in Fig. 1a,b,c. Light fluctuations are seen in each lightcurve, with amplitudes up to 0.2 mag. These light variations are erratic in the early decline, and later they appear as small light oscillations with typical timescales of about 10 days (see also Fig. 5a).
Fig. 2 reports the X-ray and the B lightcurves in a common flux scale. The comparison between the X-ray peak flux (Kitamoto et al. 1992) and the optical flux computed from the spectra at maximum light (see Fig. 3) leads to a ratio of 103, which is typical for LMXBs and outbursting SXTs (Tanaka & Lewin 1995).
A linear fit on the decay lightcurves has been performed to estimate the decline rates in the B, V and R bands. Table 3 reports the decay rates for each photometric band in different parts of the lightcurve.
Table 3. Decay rates (in mag d-1) of the B, V and R outburst lightcurves of GU Mus computed at different phases of the decline
Two bumps at 50 and 70 days after the maximum are clearly visible (see also Fig. 5a). The second bump is consistent with the secondary maximum observed in the X-ray band (see Fig. 2) and reported by Kitamoto et al. (1992), while the first one might be associated to a small feature in the X-ray lightcurve (see Fig. 1, upper panel, of Kitamoto et al. 1992; see also Fig. 4 of Greiner et al. 1994). This behaviour is typical of SXTs during the decline (Chen et al. 1993, van Paradijs & McClintock 1995).
The decay starts again in the April-May period, and strengthens in the second part of 1991. It can be also noticed from Table 3 that, after March 1991, the decay in the B is slightly faster than the ones in V and R.
However, if we integrate our photometric data with the magnitude determinations reported by King et al. (1996) and with the nightly averages of the Bailyn's (1992) observations, we notice another local maximum in the V lightcurve around Julian Day (hereafter JD) 2448460, i.e. 200 days after the X-ray peak: thus, the increased luminosity can be correlated with the tertiary X-ray maximum (see Fig. 2) observed by Ebisawa et al. (1994).
The presence of minioutbursts during the late decline of MM Vel (=X-Ray Nova Velorum 1993; Bailyn & Orosz 1995) and V518 Per (=GRO J0422+32; Chevalier & Ilovaisky 1995, Callanan et al. 1995), would indicate that they are a not unusual feature in these objects, though for these two systems this phenomenon has been purely optical and not associated with a restart of the X-ray activity.
The evolution of () and () colors during the decay is shown in Table 2. It is noteworthy that, while the () color is roughly constant around 0.25 from the beginning of the outburst until mid-May 1991, the () seems to decay (0.00118 mag d-1) as the decline proceeds. The mean () color fairly agrees with the value given by Bailyn (1992) for the month of April 1991. We also note that the object appears to become bluer around JD 2448320 and 20-25 days later, that is during the two bumps in the early B lightcurve (see Fig. 5a).
At the beginning of January 1992, the colors return to their pre-outburst values (see Della Valle et al. 1991).
The most prominent emission lines are H , H , the N III +He II blend at 4640-4686 Å and, from March 1991 (Fig. 4; first spectrum from top), N II 7217. The fluxes of these emission lines are reported in Table 4. The only evident absorption lines are the interstellar NaD doublet at 5890 Å and the telluric lines at 6850 Å and 7600 Å (the latter has not been included in the figures).
The mean value of the EW of the NaD absorption is Å, in good agreement with the estimate of Della Valle et al. (1991). According to the relation between the EW of NaD and the () color excess given by Barbon et al. (1990), we find () +1.3 at quiescence, consistent with the spectral classification by Remillard et al. (1992) and by Orosz et al. (1996).
The spectral evolution of the nova during the decline is mainly characterized by the weakening and the broadening of the H emission, as shown in Fig. 5b,c. An increase of the FWHM's is actually observed also in the He II 4686 emission, while the behaviour of the 4640 blend is more uncertain due to the broadened profile and to the bad signal-to-noise ratio (see Fig. 5c). On February 21 and on March 21, H seems to be embedded inside a wide and shallow absorption. This is not unusual for SXTs (see e.g. Callanan et al. 1995, Masetti et al. 1997, Bianchini et al. 1997).
The profile of H , close to maximum light, shows a shallow absorption filled in by an emission core. During the secondary X-ray maximum (Fig. 4; third spectrum from top) only the emission component is shown.
The He II and 4640 blend fluxes decay monotonically with time but rise up during the X-ray `reflare'. At that time (see Fig. 5b,c) also the EW and FWHM of He II show a "jump".
The continuum becomes less steep in the blue part as the outburst proceeds; actually, the last spectrum of the run appears flatter than the other ones. At this stage, the disk emission is still predominant, indeed we do not see any absorption feature of the secondary, since the magnitude of the system is still around and the secondary is a K3-K5 dwarf (Orosz et al. 1996).
© European Southern Observatory (ESO) 1998
Online publication: December 8, 1997