The metallicity distribution of unbiased samples of stars is one of the basic observational elements against which theories of the chemical evolution of the Galaxy are tested. Lower mass (K-type) stars are a critical group, in this context, because, in the disk's lifetime, they have not significantly evolved away from the main sequence. Most comparisons between observations and theory have been done using the metallicity distribution of volume limited samples of G dwarfs, which have been available, using photometric abundance determinations, since the '70s. In particular, the sample of Pagel & Patchett 1975 , with subsequent corrections, has for a long time been the standard local abundance distribution against which models of galactic chemical evolution have been tested. It has been evident since the early days that simple models of chemical evolution of the Galaxy, which assumed zero initial metallicity and a chemically homogeneous disk, predicted far more low metallicity G dwarfs than what is actually observed. This apparent deficit of low metallicity dwarfs, which has become known as the "G dwarf problem", has been recently mitigated by the inclusion, in the theoretical interpretation, of important effects such as disk heating, which decreases the number of older stars in the solar neighborhood, as they diffuse away toward larger and larger scale heights, and of an initial disk metallicity. However, the problem has not completely disappeared. The reader is referred, for a more detailed discussion, to the review of Rana 1991 where the classical G dwarf problem is extensively discussed, together with the available observational material. As Rana 1991 notices, however, "it is necessary to check that the metallicity distribution for dwarfs of other spectral types do indeed conform with that of the G-dwarfs, given the same age-metallicity relationship. This cannot be accomplished without complete volume-limited data".
Homogeneous metallicity data on volume limited samples of cool dwarfs are not available in the literature. To study the metallicity distribution of spectral types other than G Rana & Basu 1990 used a collection of heterogeneous data, which included 60 K dwarfs, almost all of which came from the catalog of [Fe/H] determinations of Cayrel de Strobel et al. 1985. The uncertain selection criteria of the Cayrel de Strobel et al. 1985 (and subsequent versions) catalog, whose aim is simply to collect all the [Fe/H] determination in the literature, are likely to result in biases in the sample. In particular, as already noted by Rana & Basu 1990 , the catalog is likely to be biased toward lower metallicity stars, which tend to be "interesting" for several reasons, and thus studied more often. A volume limited sample is in principle free from selection biases, although comparison with theoretical predictions has to include a correction for the already mentioned disk heating effect. This has to be done on the basis of an age-metallicity relationship and of a model of galactic dynamics. By being model dependent such corrections obviously introduce additional uncertainties.
In the present work we have analyzed the spectra of a volume limited sample of 91 G and K dwarfs, determining their [Fe/H] abundance using the available Fe I lines. These spectra were already used to study the lithium abundance of the sample stars (Favata et al. 1996 ). The sample is free from selection biases, and has been analyzed in a coherent way. It thus forms, with respect to collections of heterogeneous data, an improved starting point for comparison with theoretical models of galactic chemical evolution. In the present paper we compare the resulting metallicity distributions with the work of Rana & Basu 1990 , in particular determining the slope of the age-metallicity relationship and the spread in metallicity at a given age, and determining whether the metallicity distributions for G and K are compatible with the same set of galactic evolution parameters.
© European Southern Observatory (ESO) 1997
Online publication: May 26, 1998