5 Discussion

The following discussion is based on the parameters derived with the first method (using both "low'' and "high'' excitation potential transitions). Opting for the parameters derived with one of the two other methods (at the exception of V-I temperature diagnostic) would not significantly alter the conclusions.

IS Vir is a giant star of near-solar metallicity. Its position in the H-R diagram (using $T_{\rm eff}$ determined in this work and M_v derived from Hipparcos parallaxes) is plotted in Fig. 2 (circle), together with four isochrones of respective ages 0.5, 1.0, 5.0 and 10 Gyr, derived using solar metallicity (Bertelli et al. 1994). The mass and bolometric magnitude tabulated by Bertelli et al. (1994) for a 1 Gyr old star with $T_{\rm eff} = 4730$ K, $M = 2.06~ M_\odot$ and $M_{\rm bol} = 0.42$, correspond to a surface gravity of $\log g = 2.67$, in perfect agreement with the spectroscopic gravity derived here. This area of the H-R diagram is too dense to determine the evolutionary stage of the star, i.e. whether it is at the beginning of the giant branch or a giant of the clump, or its age (error bars on  $T_{\rm eff}$ are still compatible at the 1.5 $\sigma$level with a 5 or 10 Gyr clump star). Its chemical mix is somewhat different from the the solar one, with Na, Mg, Al and Ca exhibiting overabundances of 0.1 to 0.2 dex with respect to Fe. The Na and Mg abundances should be considered with caution, since they both rely on a single line of the curve of growth "plateau''. The ratio [Ni/Fe] is somewhat lower than in the Sun.

V851 Cen is a slightly metal-poor evolved star. Its position in the H-R diagram (Fig. 2, square) shows that it is a several Gyr old star, starting the ascension of the giant branch. The 10 Gyr isochrone leads to a gravity of $\log g = 3.25$ for a temperature of 4700 K, in reasonable agreement with the spectroscopic determination. Its chemical composition displays some similarities with the one of IS Vir, with (stronger) overabundances of Na, Mg, Al and Ca ranging from +0.2 to +0.5 dex. In addition Si, Sc and Ti are also over-abundant by +0.15 to +0.25 dex.

V851 Cen is one of the stars studied by Randich et al. (1993). With a value of B-V = 1.07 they derived an effective temperature of $T_{\rm eff} = 4650$ K, very close to our (B - V) photometric determination ( $T_{\rm eff} = 4670$ K) and 50 K colder than our first method estimate. They found a gravity larger by $\Delta \log g = 0.60$. The largest discrepancy between our analysis and theirs concern the metallicity, as they derived an iron abundance of ${\rm [Fe/H]} = -0.6$, $\simeq$-0.5 dex lower than our determination.

As a caveat, it should be reminded that abundances have been derived, here, under the assumption of an atmosphere which can be represented through a single plane parallel model and that the lines were formed under LTE conditions. Clearly, the behaviour of the colour indices described above shows that this assumption is likely to be a simplification of reality in these stars. On the other hand, it is reassuring to note that three different methods lead to converging results. However, as shown in Tables 6 and 7, while the absolute abundance values depend on the effective temperature, most abundance ratios (i.e. Na, Mg, Al, Ca, Ti, Co and Ni over Fe) appears relatively insensitive to the assumed temperature. Therefore, even in an atmosphere which is a superposition of models with different effective temperatures, the abundance pattern should remain largely unchanged. In general, all the other analyses of abundances in active stars have also been carried out under the same assumptions, so that the resulting bias, if present, should be common to all works.

Figure 2: The position of IS Vir (circle) and V851 Cen (square) in the H-R diagram using the temperatures derived here, plotted with four solar metallicity isochrones of respective ages 0.5, 1.0, 5.0 and 10.0 Gyr (from top to bottom) from Bertelli et al. (1994).