Astron. Astrophys. 344, 607-613 (1999)

3. The eclipse

In PVN95, it was shown that the shape of the continuum exhibits large differences when spectra are taken few months apart. In their Fig. 1, the continuum shortwards of 1500 Å shows strong extinction due to Rayleigh scattering (see Isliker et al., 1989). If Rayleigh scattering were the only source responsible for the attenuation of the continuum flux shortwards 1500 Å, the observed spectrum could be fitted by the expression,

where represents the wavelength dependent Rayleigh cross section, and is the unattenuated black-body continuum flux of the source with the parameters R=0.12 and T=105 000 K.

However, as it was demonstrated by PVN95, there is an additional wavelength independent attenuation, which can be expressed in terms of the relation

where Q^c() is the additional phase dependent attenuation. In Table 5 we list for the eclipse observations the values of the attenuation factor for the stellar and nebular components and the column densities of Rayleigh scattering, .

Table 5. Column densities of neutral hydrogen, , and continuum attenuation factors 1-Q^c, for various phases . The attenuation factors 1-Q^c are given for the stellar and nebular components.

The new ultra-violet observations of SY Mus analyzed in this paper are the first that well cover the phases close the total eclipse of the hot star. Inspecting Table 1, it is apparent that the observations were performed in order to also probe the extinction of the shortwave continuum between 1200 Å and 1500 Å where the effects of Rayleigh scattering should be present. This is indeed the case as demonstrated by the values of the column densities given in Table 5. However, there is an asymmetry of the values of the column densities derived before and after the total eclipse. This is better seen in a plot of versus , where is the reduced phase as defined in V91, (Fig. 3).

Fig. 3a and b. Observed column densities of SY Mus plotted against the reduced phase as defined in Vogel (1991) a . In b the reduced phases were corrected by 0.035. The open circles represent the points at phases 0.944, 0.930 and 0.917 and the filled circles the points at phases 0.091, 0.107, 0.119, 0.130, 0.161 and 0.176.

The asymmetry of the observed column densities may reflect a situation in which the distribution of the neutral hydrogen is not symmetric around the cool star. The erratic behavior of the emission lines close to the eclipse (see Fig. 2) might be a further indication of a region with complex structure. An alternative explanation for the observed asymmetry of the column density is that the ephemeris should have corrections. Notwithstanding our observations do not allow us to precise the instant of minimum, the observed asymmetry of the column densities suggests a negative phase shift of 0.035 to the observed phases. This change is smaller than what was previously determined in PVN95, when the phase was corrected at about 0.08 positive from Kenyon & Bateson (1984) ephemeris. This small phase shift is enough to allow the study of the giant's cool wind on the basis of the observed hydrogen column densities.

An argument for phase shifting would come if we analyzed only the egress data. Inspecting Fig. 3a, we notice that at the reduced phase =0.09 the total eclipse would be still in progress. From Eq. (6) below, this would result in the very unlikely situation of total eclipse occurrence at an impact parameter of b=2.33 giant's stellar radii. Aside from that, the analysis of only the egress data, using formula (7) below, would indicate a wind velocity of only 1% of the terminal velocity at the distance of 6.0 cool giant's stellar radii. Such an analysis would lead further to an extremely low mass-loss rate of the order of 10^-13 /yr assuming a wind terminal velocity of 30 km s^-1.

© European Southern Observatory (ESO) 1999

Online publication: March 18, 1999
helpdesk@link.springer.de