All Tables

Table 1: Basic stellar parameters of the reference stars. The iron abundance is derived from Fe II lines with the microturbulence fulfilling the usual line strength constraint (see text). [ $\alpha$ /Fe] = 0.4 was adopted in the computation of the metal-poor atmospheres, for Procyon ( $\equiv$ HD 61421) a solar mixture was assumed. Except for Procyon (data from Allende Prieto et al. 2002), surface gravities are from H IPPARCOS parallaxes with masses derived from the tracks of VandenBerg et al. (2000). [Fe/H] = $\log~($ Fe/H $)_\star - \log~($ Fe/H $)_\odot$ .
Table 2: Comparison of effective temperature determinations for the five reference stars. TI99 = Thévenin & Idiart (1999), F00 = Fulbright (2000), AAM-R96 = Alonso et al. (1996), GCC96 = Gratton et al. (1996), KSG03 = this work. The mean offset is computed for the four metal-poor stars according to $\Delta$ $T_{\rm eff}$ (study - KSG03). Significant offsets with respect to our temperature scale are encountered in the studies of Thévenin & Idiart (1999) and Fulbright (2000).
Table 3: Fe I and Fe II abundances for the reference stars and the Sun for different assumptions concerning the efficiency of hydrogen collisions. Differential abundances [Fe/H] $_{\rm diff}$ are given, except in the case of the Sun where log $\varepsilon$ (Fe) is specified. $\sigma$ refers to the line-to-line scatter which is a factor $\sqrt n$ larger than the error of the mean. Note that the scatter in [Fe II/H] $_{\rm diff}$ is significantly reduced. Like in the case of HD 140283 discussed above, a strong imbalance between Fe I and Fe II results for all metal-poor stars if $S_{\rm H}$ = 0 is assumed. Instead, the ionization equilibrium points towards $S_{\rm H}$ = 3. Procyon cannot be fitted by any model.
Table 4: Atomic data of Fe I lines used in the analysis of the reference stars. Column 1 gives the multiplet number, Col. 2 the wavelength in Å, Col. 3 the lower excitation energy, Col. 4 the differential log gf value with respect to log $\varepsilon$ (Fe) $_\odot$ = 7.51, Col. 5 the employed log C₆ value. The products log $gf\varepsilon _\odot$ that the log gf values presented here imply differ slightly from those presented in Paper II. This is because a global minimization was performed in Paper II to determine the log C₆ values, whereas lines were fitted individually here on the basis of the minimization procedure. A radial-tangential approximation to macroturbulence (Gray 1977) was used for the analysis of the Sun, a Gaussian otherwise. The equivalent widths (in mÅ) in Cols. 6-10 are of low quality, but only enter the analysis in the derivation of the microturbulence. We discourage from using these values for abundance analyses.
Table 5: Atomic data of Fe II lines used in the analysis of the reference stars. Column 1 gives the multiplet number, Col. 2 the wavelength in Å, Col. 3 the lower excitation energy, Col. 4 the differential gf value with respect to log $\varepsilon$ (Fe) $_\odot$ = 7.51, Col. 5 the employed log C₆ value. A radial-tangential approximation to macroturbulence (Gray 1977) was used for the analysis of the Sun, a Gaussian otherwise. The equivalent widths (in mÅ) in Cols. 6-10 are of low quality, but only enter the analysis in the derivation of the microturbulence. We discourage from using these values for abundance analyses.
Table 6: Different sets of stellar parameters for Procyon. The radius is determined from $T_{\rm eff}$ and $M_{\rm bol}$ and has to be compared with the fundamental value of R = (2.07 $\pm$ 0.02) $R_\odot$ . $\Delta_{\rm ~HIP}~=\unboldmath~100~(d_{\rm~spec}~-~d_{\rm ~HIP})/d_{\rm ~HIP}$ .
Table 7: Stellar parameters derived using Fe I/II in non-LTE as a gravity indicator with $S_{\rm H}$ = 3. The spectroscopic distance $d_{\rm spec}$ is calculated from $\log~\pi_{\rm spec}=0.5~([g]-[M])-2~[T_{\rm eff}]-0.2~(V+{\rm BC}+A_{V}+0.25)$ , where [X] = $\log ~(X/X_\odot)$ . $\Delta_{\rm ~HIP}~=\unboldmath~100~(d_{\rm~spec}~-~d_{\rm ~HIP})/d_{\rm ~HIP}$ . The oxygen abundance is derived by means of profile analysis of the IR triplet in non-LTE (Reetz 1999), the magnesium abundance from profiles of weak optical lines in non-LTE ( $\lambda \lambda$ 4571, 4702, 4730, 5528, 5711 Å, Zhao et al. 1998).