6. Variations on a time scale of a few weeks
Mennickent et al. (1994) called attention to the fact that the star exhibits cyclic brightness variations with a pseudoperiod of about 25 d in the epoch when it brightens. My analysis of the semiaplitude and mean RV of the local 1 372 RV curves also indicated that these variations occur on a time scale of weeks. I attempted to find a consistent periodicity in the range from 30 to 50 d which would give a meaningful phase curve. I used only local determinations from data sets spanning no more than about 10 d. A period of 34 675 0 037 was detected not only in the mean RVs and semi-amplitudes but also in the O-C from the local RV maxima with respect to the 1 372 which were analyzed earlier and which were suggested to vary with a cycle of about 5600 d! A formal orbital solution for a fixed eccentricity of 0.8 (a free solution is unstable because of a small number of data points) leads to the periastron passage at HJD 2442788.088. (Note that the position of the periastron passage is not substantially affected by my rather arbitrary choice of the value of e since it is the shape of the RV curve which basically defines it.) Fig. 10 is the phase plot for all three considered quantities, with the calculated periastron passage adopted as phase zero.
Using the program HEC13, based on Vondrák's (1969, 1977) smoothing technique I prewhitened all blue and yellow observations for the long-term light variations. Then, I combined the more numerous blue-magnitude O-C residuals with the yellow residuals from the Hipparcos observations to obtain the largest possible data set. I subjected these O-C residuals to a period analysis, separately for the data from epochs without large changes and from active epochs. It is useful to realize that all the residuals from quiet epochs lie within 0 03 which is the range identical to the range of the most stable comparison stars recorded by Hipparcos (see above). One can note, however, that the O-C residuals vary systematically over a period of several weeks. The period search was, therefore, carried out over the period range from 15 to 50 d. It indicated a number of possible, but weakly detected periods between 18 and 50 d. On the other hand, the best period detected in the data from active epochs is 34 666 0 011, identical to that found from spectral data. Several other possible periods between about 35 and 42 d were also detected. There are no comparably good periods in the range from 0 4 to 15 0.
© European Southern Observatory (ESO) 1998
Online publication: May 15, 1998