Astron. Astrophys. 360, 1077-1085 (2000)

3. Radial velocity measurements

In the majority of our spectra the lines of the two components are well separated, and we were able to obtain measurements of the individual radial velocities of the primary (A) and the secondary (B) stars. A sample spectrum is shown in Fig. 1. To obtain an independent check on the radial velocity measurements, two additional spectra were obtained with the new high-resolution spectrograph FEROS (see Kaufer et al. 1999) on the ESO 1.5m telescope at La Silla. These spectra have a lower ratio and have not been used for elemental abundance measurements.

Fig. 1. Example of spectrum of CS 22873-139 observed with the CASPEC spectrograph in the region of the Mg I triplet at a phase showing a large shift.

As a result of the very low metallicity of CS 22873-139 rather few lines are measurable on the spectra, thus we estimate that the accuracy of the radial velocities lies between 1 and 2  for the primary component (A), and about 3  for the secondary (B), the lines of which are extremely faint. The results are given in Table 2. We note that several telluric lines are visible in the EMMI and FEROS spectra, and thus the radial velocities could be corrected for small systematic errors which might be present. In the CASPEC spectra no telluric lines could be used, thus a small systematic error in the radial velocities cannot be excluded.

Table 2. Radial velocity observations of CS 22873-139

The better resolution and time coverage provided by the ESO observations allow us to improve the determination of the period and check the other parameters of the binary orbit. Preliminary orbital solutions showed that Preston's velocities for the primary star (A) were, on average, 3 kms^-1 below ours (and Preston later explained this shift). As the latter were checked with the telluric absorption lines (see above), we added 3 kms^-1 to Preston's velocities and made a new least-squares solution for the orbital elements of star A, for which Preston's velocities are presumably least affected by blending with lines of the other star. Fixing e and , we then solved again for the orbital elements, giving Preston's observations half weight of the ESO data for star A and retaining only the ESO observations for star B, for which Preston's data show large scatter and only marginal resolution of primary and secondary lines.

The result is given in Table 3 and shown in Fig. 2. As will be seen, we find a slightly lower eccentricity and larger and than Preston. However, the results depend critically on an apparent systematic shift of 3-4 kms^-1 in the new CASPEC data (we find a mass ratio ; including the CASPEC data leads to a considerably smaller value). Until more definitive data become available, we shall assume in the following.

Fig. 2. Spectroscopic orbits of CS 22873-139. Small dots and plusses denote Preston's velocities (corrected by - see text), large dots and circles the new ESO velocities for star A and B, respectively. The two curves have been computed with the orbital elements of Table 3.

Table 3. New orbital elements for CS 22873-139 (see text)

© European Southern Observatory (ESO) 2000

Online publication: August 23, 2000
helpdesk.link@springer.de