Astron. Astrophys. 335, 995-1002 (1998)

4. Discussion

The variability of early-type stellar winds is usually attributed to inhomogeneities or large-scale structures evolving on time scales of several hours to a few days. Large-scale structures in the wind may be connected to photospheric variability through a modulation of the stellar wind by the combined effect of stellar rotation and a moderate magnetic field. An alternative explanation for a photospheric connection could be the instability inherent to radiatively driven stellar winds in the case of a photosphere undergoing non radial pulsations.

Kaper et al. (1997) suggest that the variability in their sample of bright O stars is related to so-called Corotating Interacting Regions. Such corotating stream structures are expected as a result of azimuthal variations in the properties of the outflow as would arise near bright or dark spots on the stellar surface (Cranmer & Owocki 1996). A large-scale wind structure could also result from a strong magnetic field that would confine the wind towards the magnetic equator (Babel & Montmerle 1997). If the magnetic axis is tilted with respect to the rotational axis, the line emitting region near the magnetic equator will be seen under varying inclinations, resulting in periodic line profile variability. Such a configuration is suggested by Stahl et al. (1996) to explain the strictly periodic variability found in the case of HD 37022 (O7 V).

Our observations reveal strong line profile and intensity variations in the spectrum of the O8 I(f) star HD 192639. The changes seen in the absorption lines appear to be correlated to the variability of the He ii 4686 line, suggesting that the phenomenon probably arises in the stellar wind.

The double-peaked shape of the He ii 4686 emission component as well as its variability indicate that the wind of HD 192639 is not spherically symmetric. The variability in the wind of HD 192639 affects the entire He ii 4686 profile and is therefore most probably related to large-scale structures in the low velocity part of the wind rather than to small-scale inhomogeneities.

The characteristic timescales are a key issue for the study of variability in early-type stars. The present data are clearly not sufficient to perform a detailed time series analysis and to find out if the phenomenon observed in the spectrum of HD 192639 is periodic or not. Nevertheless, some of the He ii 4686 profiles we have observed were already seen during earlier snapshot observations by other observers (Underhill 1995a, Herrero et al. 1992), suggesting that the phenomenon is stable over several years. Furthermore, the properties of the He ii 4686 line during the 1997 observing campaign (Fig. 2) suggest that the recurrence time scale could be of the order of 4 - 6 days, although we cannot exclude a somewhat longer time scale. A crude Fourier analysis of the characteristics of the He ii 4686 line over our entire dataset (i.e. including the 1995 and 1996 snapshot observations) yields a recurrence time of days. However, since this time scale is comparable with the total time spanned by our 1997 dataset, the meaning of the Fourier analysis is rather difficult to assess. We find no trace of the 2 day recurrence marginally present in the 1986 observations of the C iv 5801, 5812 doublet by Fullerton (1990). On the other hand, during the same limited observing run (6 nights), Fullerton (1990) derived a time scale of days for the variations of the He i 5876 line, a value which is more compatible with our results.

Herrero et al. (1992) derive a radius of 19.5 for HD 192639. Estimates of the projected rotational velocity range from 103 km s^-1 (Conti & Ebbets 1977) over 110 km s^-1 (Penny 1996) to 125 km s^-1 (Herrero et al. 1992). From the two most extreme estimates of , we derive an upper limit on the actual rotational period of 7.9 to 9.6 days respectively. Considering the critical break-up velocity and using the stellar parameters derived by Herrero et al. (1992), we can set a lower limit on the rotational period of 3.10 days. The variability of HD 192639 might therefore well be connected to the stellar rotation.

Additional observations over a longer time span are needed to further investigate the properties of the variability seen in the spectrum of HD 192639 and to find out whether or not the phenomenon can be explained by one of the above mentioned scenarios.

© European Southern Observatory (ESO) 1998

Online publication: June 26, 1998
helpdesk.link@springer.de