The present results, taken at face value, indicate that the cooler K dwarfs may have had a different chemical evolution than G dwarfs, in particular showing, in their present day metallicity distribution, a deficit of lower metallicity stars. Before any conclusion is drawn from this observational fact, it is important to stress that the samples discussed here are relatively small, so that the two metallicity distributions (for cooler and hotter dwarfs) are only different at the 98% confidence level.
Assuming nevertheless that the difference between the two metallicity distributions is real, and not simply a statistical artifact, or a consequence of biases present in the parent sample, the implications can be far-reaching. If we assume that the simple model that we have used is a good representation of the reality, the different model parameters for the two sub-samples imply that the initial metallicity of the disk was significantly different for cooler and hotter stars. Such a conclusion is, for obvious physical reasons, unlikely, and it would assume the existence of a "cosmic conspiracy" that brings the metallicity of cooler and hotter dwarfs together at the current epoch. It is however interesting to compare our result with Fig. 8 of Eggen 1996 , which shows the relationship between the chromospheric age and the photometrically derived metallicity for a sample (not randomly chosen) of Gliese stars in the range of spectral types K5V-M4V (i.e. cooler than considered here). The change in mean metallicity along the disk's lifetime for the sample shown there is of only dex, i.e. a value very similar to our result. The spread in Eggen's data is however very large. If we discard the possibility of different initial abundances in the disk for stars of different masses, we have to question the assumptions of the chemical evolution model.
One of the assumptions, namely the one of identical present-day metallicity, is strictly speaking untested, as all the published accurate spectroscopic determinations of metallicity in Hyades stars (as well as in other young and intermediate age open clusters) are limited to stars hotter than K (see Taylor 1994a ). It is thus in principle possible that the assumption of identical present day abundance for G and K stars is not correct. However, while certainly it would be desirable to have good metallicity determinations for cool Hyades dwarfs, to verify this assumption, the near coincidence of the upper range in metallicity for the cooler and hotter dwarfs (see Fig. 3) can be taken to indicate that the present-day metallicity is not strongly mass-dependent for G and K dwarfs.
The other key assumption of the simple model used here is the constancy of the stellar birth rate, across the Galaxy's lifetime, and for the whole range of masses. This is clearly a simplistic assumption, tantamount to assuming that the IMF is a truly universal function, independent of the physical condition of the star formation environment. In fact, observational evidence is accumulating showing that the IMF from different star-forming regions and embedded young clusters is not universal. Meyer 1996 has shown, for example, that the IMF's of two embedded clusters (NGC 2024 and the Ophiuchus cloud core) are actually different, and that more in general the ratio of the number of intermediate (1.0-10 ) to low (0.1-1.0 ) mass stars in various embedded clusters is not constant, and ranges between and 0.45, while in the Miller-Scalo IMF (the one observed for the field) has a value of 0.17. On the basis of the available observational evidence Meyer 1996 suggests that higher stellar density embedded clusters would tend to form more intermediate mass stars, and lower density ones would form more low mass stars.
It is not unreasonable to expect that one of the important parameters, driving the development of the star forming process, may be the total metallicity of the parent cloud. In fact, the cooling of a molecular aggregate (the main star forming environments in our Galaxy) is dominated by the radiation of molecules containing elements heavier than He, so that it does not seem far fetched that a different initial abundance may lead to different conditions in the collapsing cloud and thus to a different mass spectrum in the resulting stellar population. Our observed metallicity distribution would seem compatible with a scenario in which low mass stars preferentially form in higher metallicity clouds, while intermediate mass stars form more efficiently in lower metallicity clouds, producing an apparent deficit of low metallicity low-mass stars. Such a scenario would imply a mass-dependent age-metallicity relationship and an age-dependent IMF. In this context, it is also suggestive that the present day mass function (which should be identical to the initial mass function for stars cooler than mid-G, given that their main sequence lifetime is longer than the disk's age) is not smooth in the mass range discussed here, but rather it shows a bump at masses corresponding to and , implying "two mass scales in the process of star formation" (Rana 1991 ). The position of the first bump corresponds with the mass range of the cooler sub-sample discussed here.
Again, the sample on which this paper is based is relatively small. Given that the implications of the present results are potentially far-reaching, an effort should be made to enlarge the sample, determining the metallicity of a larger volume limited sample of cool dwarfs. Also, the metallicity of the cooler dwarfs in open clusters (specially the Hyades) should be accurately determined, testing the key assumption that the present day metallicity is not a function of stellar mass.
© European Southern Observatory (ESO) 1997
Online publication: May 26, 1998