We have refined the period for the variability in HD 50896, P=3.7650 days, which produces a phase-dependent modulation of three different spectral quantities and the photometric light curves which have coherent properties over a timespan of 15 years. This shows that the underlying process responsible for the periodic variations has a high degree of repeatability, and thus cannot be associated with wind instabilities or any other such mechanism which would not yield coherent variations over such long timescales. This does not mean that wind instabilities are absent, nor does it mean that other potential phenomena such as magnetic structures and non-spherically symmetric outflows cannot be present. In fact, some of these mechanisms must be present, as is evident from the secular variations observed, but the coherence over such a long timescale can only be compatible with a binary scenario or one of a rotating star with a fixed, non-spherically symmetric wind distribution. The rotation scenario requires a spatially stable structure displaced from the rotational axis in order to produce the variations (cf. the jets proposed by Mattews et al. 1992). To be stable, these structures would be expected to be symmetric with respect to the rotational axis, which would produce a double wave in the phase-dependent variations. There is, indeed, evidence for two maxima and minima, at least in the N V 4620 WAF and the He II 4686 kurtosis phase curves, but they are not equally spaced. Two maxima are at times present in the photometric light curves as well. A discussion of mechanisms which could give rise to non-spherically symmetric outflows from a star such as HD 50896 is given in Morel et al. (1998), although it is yet to be shown that such mechanisms can remain stable over such long timescales as the ones we now have shown to persist.
The coherent variations allow us to revisit the binary hypothesis for HD 50896. According to this scenario, the periodic changes in the N V 4604,4620 lines can be explained as the result of the perturbation of a portion of the WR wind by X-ray emission from the companion and/or a shock cone associated with the companion. This phenomenon is known as the Hatchett - McCray effect (Hatchett and McCray, 1977) who first predicted its existence. Because these N V lines arise deep in the WR star's wind, the companion's perturbing effect has to be very far reaching, unless its orbit is very close. Since the X-ray flux levels are not very high (cf. Stevens & Willis 1988), the former seems very unlikely. Rather, if HD 50896 contains a companion, its orbit must lie close to the base of the stellar wind, near the region where the N V 4604,4620 lines are formed. If the companion is a collapsed object, a mechanism to inhibit accretion and thus limit the X-ray flux could be invoked (Lipunov 1982). If the companion is not collapsed, it remains to be explained how its interaction with the WR primary produces such significant variability.
© European Southern Observatory (ESO) 1999
Online publication: June 30, 1999