Abstract

We determine iron abundances for an average of two to three giant stars in each of eight metal-poor Galactic globular clusters. We present equivalent width measurements derived from high-resolution (lambda/DELTAlambda approximately 25,000), high S/N ratio (45 less-than-or-equal-to S/N less-than-or-equal-to 100 per pixel) echelle CCD spectra. The abundances are determined using line analysis via latest generation model atmospheres and synthetic spectrum techniques. We derive the following [Fe/H] values (relative to a solar Fe abundance of 7.5): - 2.17 +/- 0.05 (standard error of the mean, based on two stars) for M68, -1.71 for NGC 4833 (one star), -1.59 for NGC 6144 (one star), -1.99 +/- 0.01 for NGC 6397 (six stars), -1.58 +/- 0.04 for NGC 6752 (three stars), -1.96 +/- 0.04 for M55 (two stars), -2.23 +/- 0.04 for M15 (two stars), and -2.10 +/- 0.08 for M30 (two stars), with no evidence for intracluster metallicity variations. We estimate the total errors to be of order 0.15 dex, including internal and external sources. These values are in excellent agreement with previous results of similar high quality. The low end of the globular cluster metallicity scale is now well established by such spectroscopic data and support the Zinn (1985) scale, with an error of 0.15 dex. The Washington scale of Geisler et al. (1991), is also generally supported, but the errors are larger (approximately 0.23 dex), as expected from the high reddening and reduced metallicity sensitivity of this photometric system at these low abundances. However, several of the clusters suggested by the Washington photometry of Geisler et al. (1992) as being more metal-poor than their Zinn (1985) values are instead found to have intermediate metallicities. We find that the mass-metallicity limit imposed by assuming the self-enrichment of metal-poor globulars by supernovae provides a boundary that is in good agreement with the observed distribution.