4 Variability

4.1 Photometry

BI Lyn has been found to be variable in light with an amplitude of $\sim$0.1 mag and a period of 0.33818 d (Lipunova & Shugarov 1990, 1991). Reports of other periodicities (280 s: Lipunova & Shugarov 1991, 1117 s: Kuczawska et al. 1993) have not been confirmed. These find possible explanations in a CV model as the orbital period and oscillations of the white dwarf respectively. An orbital period of 0.34 d was not detected spectroscopically (Orosz et al. 1997). The 0.34 d period cannot be associated with the cool star rotation; a star with $R=9~\mbox{$R_{\odot}$ }$, $\mbox{~\em T$_{\rm eff}$ }=5840~\mbox{K}$ and $\mbox{~log $g$ }=3.2$ has a minimum rotation period of 2.6 d.

A period of $\sim$0.3 d can be explained by pulsations (radial or non-radial) in a luminous hot helium star. A corollary would be V2076 Oph (Lynas-Gray et al. 1987), a luminous helium star with $\mbox{~\em T$_{\rm eff}$ }\sim32~000~\mbox{K}$, $\mbox{~log $g$ }\sim3$ and non-radial pulsation periods $\sim$0.7 and 1.1 days. The observed periods of these strange-mode pulsations (Saio 1995) depend critically on the stellar mass, luminosity and which modes are most excited. V2076 Oph is thought to have a mass $\sim$ $0.7~\mbox{$M_{\odot}$ }$ (Lynas-Gray et al. 1987).

4.2 Radial velocity

Given the resolution and inhomogeneity of the spectra, the short time base over which observations were obtained and the rotational broadening of the cool star spectrum, no significant changes in the radial velocity of either component could be identified (Table 2). The mean difference in velocity of the two components is negligible. High-resolution studies of a specific spectral region over an extended interval will be necessary to establish the orbital period and velocity amplitudes in this system. These will be extremely important observations and should be undertaken as a matter of urgency since they will give the mass ratio directly and independently from the spectroscopic analysis.

4.3 H$\alpha$

Variations in H$\alpha$ are more difficult to interpret reliably, particularly since the promised report of 1995 March observations (Wade & Potter 1995; Orosz et al. 1997) has still to appear. The available data describe a broad emission with a variable and narrower absorption component. We observed a change in the absorption strength of H$\alpha$ between 1998 and 1999, but not during the short interval of the 1999 observations (Fig. 7). A rapid increase in absorption strength has been reported elsewhere (Wade & Potter 1995). Such phenomena have previously been seen in hydrogen-deficient supergiant binaries (e.g. $\upsilon$ Sgr) on a timescale comparable with their orbital periods.

$\upsilon$ Sgr is a single-lined spectroscopic binary with an orbital period of 138 days (Wilson 1914). The primary is an early-type supergiant with $\mbox{~\em T$_{\rm eff}$ }\sim12~000~\mbox{K}$ (Dudley & Jeffery 1993) and an extremely low hydrogen surface abundance (Schönberner & Drilling 1983). Being of the 6th magnitude, it has been scrutinized for over a century (Campbell 1899). It is interesting to review some of the reports concerning H$\alpha$:

• "Intensity variations were found in the emission and absorption of hydrogen lines'' (Seydel 1929);
• "At H$\alpha$ a very broad, weak emission band, roughly 8 Å in width, can be seen'' (Greenstein 1943);
• "One strong P Cygni outburst was visible in the hydrogen lines in the envelope of the system. A double strong, deep, absorption at H$\alpha$ appeared...'' (Greenstein 1950);
• "Stationary emission lines of [...] H$\alpha$ occur in the envelope of the system'' (Greenstein 1950);
• "Many investigators have reported that the displaced H$\alpha$ absorption line appears during an interval of about 40 days centered about the phase... at which the primary is farthest from us (e.g. Bidelman 1949; Hack 1960)'' (Nariai 1967);
• "... the gas flow was not observed during cycles 43 to 98... It is suggested that the jet appears periodically...'' (Nariai 1967);

More recent data (Frame et al. 1995) confirm the strongly variable nature of H$\alpha$ absorption and emission. Absorption episodes are roughly correlated with orbital phase although they may switch off for several cycles at a time. The preferred model is that of a supersonic jet (Nariai 1967) generated by material passing through the inner Lagrangian point on to the secondary and which eclipses an extended envelope around the primary responsible for the broad emission. Similar phenomena have been seen in the three other known hydrogen-deficient binaries KS Per (Nariai 1972), V426 Car = CPD $-58^{\circ}2721$(Frame et al. 1995) and HDE320156 = LSS4300 (Frame et al. 1995).

Although the data are sparse, the qualitative behaviour of H$\alpha$ in BI Lyn and $\upsilon$ Sgr may be similar. Consequently a similar interpretation involving a circumstellar envelope and a gas-stream flowing through the inner Lagrangian point may be appropriate. The rapid increase in absorption reported in H$\alpha$ (Wade & Potter 1995) would correspond to eclipse of the circumstellar material by the gas stream. Clearly, more detailed work including determination of the orbital period and the phase-dependency of the H$\alpha$ profile will be required to confirm this hypothesis.