4. Radial velocities and a preliminary orbit
Fig. 4 plots all available radial velocity data for HD 51066 and reveals long-term variations consistent with a wide but rather eccentric spectroscopic binary. Included in the figure are the data from Table 1 and additional 15 velocities from Fekel (1998), partly already mentioned in Henry et al. (1995b), that were also taken with the KPNO Coudé feed telescope plus the TI CCD. The probable errors are between 1.5-4.0 km s-1 for the TI CCD data and 0.8-1.5 km s-1 for the F3KB-CCD data (Table 1). The cause for the different uncertainties is the three times larger wavelength range of the F3KB CCD and the resulting narrower cross-correlation function. Note that significantly better radial velocities are hard to achieve because of the relatively large line broadening and the variable line profiles due to spots.
Altogether, 66 radial velocities are used to compute a preliminary orbit with the help of a modified version of the differential-correction program of Barker et al. (1967). The orbital period is estimated by eye to be 10 yrs based on a systemic velocity of around -19 km s-1 (the dotted line in Fig. 4) and a time of periastron passage in late 1994, and remains fixed throughout the iterative solution for the orbital elements. The formally best solution gives a standard error of an observation of unit weight of 1.7 km s-1 , thus comparable to our observational uncertainties, but the O-C residuals for the eight velocities from 1991-1993 are systematically too high (by 1.2-3.2 km s-1 ). Thus, we emphasize that our orbit is just preliminary and based on a rough estimate of the period. The elements are given in Table 3 and the computed velocity curve is plotted in Fig. 4 along with the observations.
© European Southern Observatory (ESO) 1998
Online publication: July 20, 1998