5. Comparison with previous abundance determination
We have searched both the Cayrel de Strobel et al. 1992 (an updated version of Cayrel de Strobel et al. 1985) and the Taylor 1994b (and Taylor 1995 and Taylor 1996 extensions) catalogs of abundance determinations for objects in our sample and have compared in detail our results with previously available abundance determinations. The two catalogs used are of a somewhat different nature. While the Cayrel de Strobel et al. 1992 catalog is essentially an annotated bibliography of literature data, the series of papers from Taylor is part of an organic attempt at re-assessing the literature data on [Fe/H] in cool stars. To this end, the [Fe/H] literature values have been converted to a coherent temperature scale (the same one adopted in the present work), thus forming a much more homogeneous data set, against which more meaningful comparisons are possible. For the purpose of assessing the consistence of our [Fe/H] values with previous literature data we have enhanced the data set discussed here with another set of stars (from Favata et al. 1997 ), which have been taken with the same instrument, and have been reduced and analyzed together with the data discussed here. While they do not form a volume limited sample, and are not discussed here, they can still be used for the purpose of assessing the whole analysis procedure.
We have searched the Taylor series of catalogs for stars in common with our extended sample, and found 35. One object (GJ 27) has been excluded from the comparison as the literature photometric and spectroscopic abundance determination appear strongly discrepant. We have also added a star (GJ 309) for which the abundance has recently been determined by Flynn & Morell 1997 . For the 35 objects in common with Taylor and with Flynn & Morell 1997 , we have plotted (Fig. 2) our [Fe/H] values against the ones from the Taylor catalogs. Also plotted are the "ideal" 45 deg relationship (dotted line) as well as a least square fit to the data points. Our [Fe/H] values are on the same scale as Taylor's catalog, as the slope of the mean relationship is 1.028 and the zero-point shift is only 0.02 dex. The RMS scatter of the difference between our [Fe/H] values and Taylor's is 0.09 dex. The data points in common with literature values cover the complete range of metallicities represented in the present work, from the most metal-poor object up to the most metal-rich ones.
Some works in the literature also report the actual equivalent widths for some lines in common with our analysis. When possible, we have compared them in detail. Abia et al. performed an abundance analysis on a number of cool stars using data from the same spectrograph as the one used here, although with an older Reticon detector, and in the same spectral region. Our sample has 3 stars (GJ 19, GJ 559A, GJ 559B) in common with the Abia et al. sample, and for these 3 stars the equivalent widths for the Fe lines in common agree with the equivalent widths determined in the course of the present work to %.
Pasquini et al. 1994 have observed a sample of early G dwarfs in the same spectral region around the Li I line, also with the same instrument and detector combination, with 8 objects in common with our sample. They have also performed a [Fe/H] analysis, however using only two Fe lines, the ones at 6703.57 and 6705.10 Å. While we have decided not to use the 6703.57 Å line in our abundance analysis because of the considerable non-LTE effects in the cooler stars (which however appear to be negligible in the hotter stars studied by Pasquini et al. 1994 ), we have compared the equivalent widths reported by Pasquini et al. 1994 for these two lines with ours, finding that they also agree to better than 7% (peak to peak relative dispersion). Two of the stars in common between our work and Pasquini et al. 1994 have differenced in the derived metallicities as large as 0.2 dex. Given the small difference in the measured equivalent widths, these difference are to be attributed partly to the differences in the model atmosphere parameters (specially the ), and partly to the different set of model atmospheres used. The effective temperatures of Pasquini et al. 1994 differ in some cases from the ones adopted here by as much as 100 K, while the values are mostly close to the main sequence ones adopted in the present work, except in one case (GJ 77) for which Pasquini et al. 1994 assume , i.e. significantly above the main sequence.
© European Southern Observatory (ESO) 1997
Online publication: May 26, 1998