1997A&AS..121...95C

We have obtained high resolution, high signal-to-noise ratio CCD echelle spectra of 10 bright red giants in 3 globular clusters (47 Tuc, NGC 6752 and NGC 6397) roughly spanning the whole range of metallicities of the galactic globular cluster system. The analysis of this newly acquired material reveals no significant evidence of star-to-star variation of the [Fe/H] ratio in these three clusters. Moreover, a large set of high quality literature data (equivalent widths from high dispersion CCD spectra) was re-analyzed in an homogeneous and self-consistent way to integrate our observations and derive new metal abundances for more than 160 bright red giants in 24 globular clusters (i.e. about 16% of the known population of galactic globulars). This set was then used to define a new metallicity scale for globular clusters which is the result of high quality, direct spectroscopic data, of new and updated model atmospheres from the grid of Kurucz (1992, priv. comm.), and of a careful fine abundance analysis; this last, in turn, is based on a common set of both atomic and atmospheric parameters for all the stars examined. Given the very high degree of internal homogeneity, our new scale supersedes the offsets and discrepancies existing in previous attempts to obtain a metallicity scale. The internal uncertainty in [Fe/H] is very small: 0.06dex (24 clusters) on average, and can be interpreted also as the mean precision of the cluster ranking. Compared to our system, metallicities on the widely used Zinn and West's scale are about 0.10dex higher for [Fe/H]>-1, 0.23dex lower for -1<[Fe/H]←1.9 and 0.11dex too high for [Fe/H]←1.9. The non-linearity of the Zinn and West's scale is significant even at 3σ level. A quadratic transformation is given to correct older values to the new scale in the range of our calibrating clusters (-2.24≤[Fe/H]_ZW≤-0.51). A minor disagreement is found at low metallicities between the metallicity scale based on field and cluster RR Lyrae variables (via a new calibration of the {DELTA}S index) and our new cluster metallicities. It could be tentatively ascribed to non-linearity in the [Fe/H]-{DELTA}S relationship. The impact of new metallicities on major astrophysical problems is exemplified through a simple exercise on the Oosterhoff effect in the classical pair M 3 and M 15.