Washington photometry of globular cluster giants: The most metal-poor clusters

We present photometry on the Washington system for 125 stars in 10 of the most metal-poor Galactic globular clusters. Luminosity class, temperature, and metal abundance are derived for each star. A new temperature calibration is presented, based on recent model atmosphere calculations by Bell & Gustafsson [MNRAS, 236, 653 (1989)]. Metal abundances are determined utilizing a recently improved calibration which corrects for the decreased metallicity sensitivity of cooler giants and introduces a new temperature index, which significantly reduces the effects of photometric errors and reddening uncertainties in determining abundances for very metal-poor stars. Nevertheless, the derived abundances are rather uncertain due to the high reddening and photometric error sensitivities of the system for cool, metal-poor stars and because reddenings are large and/or poorly known for several of the clusters. For NGC 6397, M15, M30, and M55, the abundances are well determined (uncertainties of 0.2-0.3 dex, including internal and reddening errors) and in very good agreement with the abundance scale of Zinn [ApJ, 293, 424 (1985)]. For five other clusters, including M68 and NGC 2298, 4833, 5897, and 6101, we derive metallicities that are substantially less than those given by Zinn, although the uncertainties are larger, typically 0.3-0.5 dex. The discrepancies are particularly significant for NGC 5897 and 6101, which have small and relatively well-determined reddenings and derived metallicities as much as 0.8 dex lower than nominal values, with [Fe/H]∼ -2.5. Since the Washington and Zinn metallacity scales are shown to be in good overall agreement, we discuss in detail possible reasons for the individual discrepancies. Reddening and photometric errors, and expected errors in the respective calibrations can account for most of these, although we cannot rule out the possibility that at least several of these clusters are indeed substantially more metal deficient than previously indicated. The likelihood of intermediate metallicities for some of the discrepant clusters is discussed. Confirmation of the metal abundances of these clusters must await the results of detailed spectroscopic studies now underway. We also derive new (B - V)_0,g values for a large sample of metal-poor clusters and show that this parameter loses abundance sensitivity below [Fe/H]∼ -1.85, regardless of the metallicity scale used or the (B-A)_0,g source.