3. The orbital period
Our radial velocity data are not suited for a good period determination. They are, however, compatible with the 1380 day periodicity found by Mayall (1940). The precision of Mayall's period was put into question by the work of Iijima (1985). Although he found the same periodicity, his interpretation of spectroscopic changes as periodic attenuation by the cool giant was inconsistent with Mayall's ephemeris in the sense that his minimum phase was shifted by about half a period with respect to that of Mayall. Because the period is of prime importance in our study and accurate periods can only be obtained from data with a long time base we re-determine the periodicity of the light variations from the available data.
3.1. The light curve of BX Mon
For our analysis of the light variations of BX Mon we employed two data sets.
The first set consists of the original data from Mayall's laboratory journal kindly provided by The Harvard College. These brightness estimates are based on photographic plates and cover the period from 1890 to 1940 (see Mayall 1940). The dataset consists of 731 detections. The error of a measurement is magnitudes and the amplitude of the variations amounts to magnitudes.
The second data set has been kindly provided by the Royal Astronomical Society of New Zealand (RASNZ). It consists of the visual brightness estimates reported on a regular basis by the variable star section of the RASNZ. These data cover the years 1989 to 1995.
Although separated by more than half a century both light curves are similar (Fig. 1c,d). They show a relatively narrow maximum and a wide flat minimum with approximately the same periodicity.
A period analysis applied to the combined data sets yields two possible periodicities, and . Our analysis excludes Mayall's period of 1380 days because the maxima of the old (Mayall) and the new light curves (RASNZ) are out of phase by about half a period. As mentioned above, this problem was already encountered by Iijima (1985). A reanalysis of Mayall's data alone confirms the 1380 day period, but the uncertainty is rather large and includes the two periods determined by us with the combined data set.
The ambiguity in the orbital period can be resolved if we include the IUE data into the analysis and examine the UV eclipse behaviour.
3.2. Eclipse effects in the UV
The complete set of IUE archive data is shown in Fig. 2. In two IUE measurements we see a strong flux attenuation (Table3) which can be interpreted as eclipses of the hot component by the cool giant. The two observations are separated by 4347 days or a little more than 3 periods. We compared phase plots of the integrated IUE fluxes from and for and . Using puts an IUE spectrum with high flux level between the two IUE spectra with strongly reduced integrated fluxes. This behavior is not consistent with an eclipse interpretation of the IUE flux reduction. Only an orbital period of is consistent with an eclipsing system as shown in Fig. 1a,b). This interpretation implies, that the eclipse phase lasts in the far UV for about 150 days. Such long UV-eclipses, about 10% of the orbital period, are characteristic for symbiotic systems. Our eclipse interpretation is also strongly supported by the measured radial velocities (see Sect. 4.1). Throughout this paper we will use the period .
© European Southern Observatory (ESO) 1998
Online publication: July 20, 1998