The accuracy of stellar atmospheric parameter determinations: a case study with HD 32115 and HD 37594

We present detailed parameter determinations of two chemically normal late A-type stars, HD 32115 and HD 37594, to uncover the reasons behind large discrepancies between two previous analyses of these stars performed with a semi-automatic procedure and a "classical" analysis. Our study is based on high resolution, high signal-to-noise spectra obtained at the McDonald Observatory. Our method is based on the simultaneous use of all available observables: multicolor photometry, pressure-sensitive magnesium lines, metallic lines and Balmer line profiles. Our final set of fundamental parameters fits, within the error bars, all available observables. It differs from the published results obtained with a semi-automatic procedure. A direct comparison between our new observational material and the spectra previously used by other authors shows that the quality of the data is not the origin of the discrepancies. As the two stars require a substantial macroturbulence velocity to fit the line profiles, we concluded that neglecting this additional broadening in the semi-automatic analysis is one origin of discrepancy. The use of FeI excitation equilibrium and of the Fe ionisation equilibrium, to derive effective temperature and surface gravity, respectively, neglecting all other indicators leads to a systematically erroneously high effective temperature. We deduce that the results obtained using only one parameter indicator might be biased and that those results need to be cautiously taken when performing further detailed analyses, such as modelling of the asteroseismic frequencies or characterising transiting exoplanets.