The combination of Bologna (, ; this work) and La Palma (, ; van Paradijs et al. 1997) data taken on February 28 could provide very important informations to understand the nature of this transient, so far unique in the optical bands. These observations are not simultaneous, therefore they can give some insights into the problem of the light variation and into the related one of the energy distribution. We consider here three different hypotheses: a) the brightness was constant during the four hours between the two sets of observations, one obtains , , and ; such color indices are not consistent with the spectral energy distribution of any known astrophysical object and imply strong variations with the wavelength. b) a fading during the time span between Bologna and La Palma observations; in this case the above result would be strengthened. c) the possibility of an increasing emission. In the latter case, we can interpolate the V spectral flux density from B and R values at the time of the Bologna observations. We assume tentatively a linear flux density / distribution. The conversion from magnitudes to fluxes was done using Table 9 by Fukugita et al. (1995). We obtain erg cm-2 s-1 Å-1, corresponding to . In the same way, from La Palma V and I magnitudes we derive a R spectral flux density of erg cm-2 s-1 Å-1, which gives . Thus, the total R flux at maximum was not less than 15 times that of the extended object. These figures imply a flux increase of 1.9 times in V and R between the two sets of observations, corresponding to a variation of -0.7 mag in both bands, with a mean rate of -0.2 mag hr-1. The conclusion seems to be inescapable: either the OT has a very bizarre spectrum, or if it has a more normal spectral distribution it must display an increase of brightness between Bologna and La Palma observations. In the framework of the foregoing hypothesis, this result is significant at a 2 confidence level. Being cautious for the observational error, we suggest that the optical luminosity increased at least until March 1.0 UT. Indeed a quick computation shows that, in order to have constancy or a decrease in brightness with a 3 confidence level between Bologna and La Palma data, we should have observed the OT at least at on February 28.8; this value is outside the 3 interval centered on the observed magnitude . We then conclude that the hypothesis of a non-increase in brightness can be rejected with a confidence level of almost 4 .
We can now refine our first-approximation figures. On February 28, the Bologna B value was acquired about 50 minutes later than the R one, therefore we might infer that the B magnitude of the OT at the time of the R frame was actually brighter by 0.2 mag. This implies a simultaneous color index of 1.0; correspondingly, one has and (interpolated at the same time). Bearing this in mind, it results that in a time span of about 4 hours the V magnitude of the OT decreased of 0.8 mag; this corresponds to a flux increase of a factor 2.1.
HST data (Sahu et al. 1997) are particularly relevant for understanding the mid-term behaviour of the OT. Therefore, by using the method described above, we computed the R flux densities and magnitudes; we obtained for March 26 and for April 7. From these data it is evident a substantial reddening of the OT, which changed from and on February 28.99 to and on March 26.4. The fluxes corresponding to these interpolations, together with those of other relevant B, V, R and I measurements and with the fluxes of the X-ray transient source SAX J0501.7+1146 given by Costa et al. (1997b) are reported in Table 2.
Table 2. X-ray, B, V (Johnson), R and I (Cousins) fluxes of the event. B and R values were computed by subtracting from the data of Table 1 in the days from February 28 to March 3 the fluxes of the extended object and of the nearby star. Effective wavelengths are reported. We adopted the band widths published by Fukugita et al. (1995). All fluxes are in units of 10-15 erg cm-2 s-1. Data between parentheses are inferred from our interpolations. See text for further details
In spite of the poor quality of our observation of March 1.8 UT we can state that, on the basis of a comparison with other objects in the frame, at that time the OT had faded below the level of our first detection. Therefore we can fix the time delay between the -ray event and optical maximum in the range . Correspondingly, the duration of the first fading phase is . During this time the fading rate is 1 mag day-1. This fixes tight constraints on both the rising and fading optical rates. The presence of an early phase of rapid fading is implicitly confirmed by the R band observations of Metzger et al. (1997a,b) who found that the total R magnitude has slowly faded by mag in one month (March 6-April 6).
If we assume for the R band luminosity a power law decay , a limit can be determined by using the R flux (interpolated as before) from the HST data of March 26.4 (Sahu et al. 1997); the optical data of the first days require . An index is also found for the X-ray decay behaviour. An exponential decay law, similar to that of X-ray bursts, results in a decay time , to be compared with deduced by Palmer et al. (1997) for the X-ray emission. The ratio is 4 , if the luminosity values are taken at their respective observed maxima (at least 13 hours apart). By scaling the X-ray flux at the time of optical maximum (with a decay law), we obtain . On March 3 X-ray and optical observations were nearly simultaneous: we can fix an upper limit of 1.2 to the ratio.
It is interesting to note that Castro-Tirado et al. (1997) did not see anything noteworthy in the error box of another Gamma-Ray Burst, GRB 970111, just 19 hours after the -ray event. This indicates a significant difference in the optical behaviour of GRB 970111 and GRB 970228.
© European Southern Observatory (ESO) 1997
Online publication: March 26, 1998