Pereira et al. (1995) determined an orbital period of SY Mus based on about 2500 visual estimates by the observers of the Variable Section of the Royal Astronomical Society of New Zealand (RASNZ). Their data set covers orbital cycles, and they derive a period of .
In Fig. 1 we show IUE UV light curves corresponding to periods , , and days. For all four curves, is set to the value derived in Sect. 3.2. The IUE SWP10188L spectrum is separated by 9 orbits from the spectra of similar phase. It is taken at ingress and shows already a heavily attenuated stellar continuum. In the light curve corresponding to , this spectrum is flanked by spectra showing almost no continuum attenuation (see Fig. 1). We find, that an orbital period of is required to obtain a continuous UV light curve. This is only larger than the value derived by Pereira et al. (1995). The upper limit for the period inferred by the UV light curve is .
The visual estimates of SY Mus do not show any erratic variability. The mean visual magnitude of the light curve also remained stable during 40 years (Pereira et al. 1995). Different atmospheric extensions due to stellar pulsations are therefore unlikely to be the cause for the attenuation in the SWP10188L spectrum.
We also analyzed the updated time series of the visual estimates kindly provided by the RASNZ. Our sample has been collected in the years 1954-1999 and covers 26 orbits, which is 4 orbits more than has been analyzed by Pereira et al. (1995). The number of measurements increased from to . This is a significant improvement, as the amplitude of the variability is only 1.6 times the error of a single measurement. Based on the visual estimates, we derive a new photometric period of . The combination of and , leads us to adopt .
In order to derive an accurate time of mid-eclipse for SY Mus, we perform a least squares fit including our new 21 RV-measurements and the 9 RV-measurements of Schmutz et al. (1994). This yields the orbital parameters listed in Table 4. The derived orbital periods for the eccentric and the circular solutions are in good agreement with a period of , as derived in Sect. 3.1. The orbital solutions are consistent with the previous analysis of Schmutz et al. (1994).
Table 4. Orbital parameters of the M star in SY Mus. gives the Julian date when the M-giant is in front of the white dwarf. For the eccentric solution gives the time of periastron passage.
To test whether the eccentric solution is significantly better than the circular solution we compare the sum with the expected value . The number of free parameters for the eccentric solution is , for the circular solution it is . The number of measurements is . The mean deviations listed in Table 4 indicate an observational error of the radial velocities of . The eccentric solution is thus not significantly better than the circular solution. The circular solution and the radial velocity data are shown in Fig. 2. As the theory of tidal forces in binaries (Zahn 1977) also predicts the orbit to be circular, we will calculate the phase of a given date JD according to the circular fit with the period from Sect. 3.1:
To derive an error for the time of mid-eclipse, , we have done a series of least squares fits, where for each fit the period was set to a value in the range . We find that fits where differs by more than from JD lead to M-values outside the expected range . The phases of all attenuated IUE spectra in Table 2 (except SWP10188L) have an uncertainty .
© European Southern Observatory (ESO) 1999
Online publication: August 25, 1999