THE 1ST DETAILED ABUNDANCE ANALYSIS OF GALACTIC BULGE-K GIANTS IN BAADE WINDOW

We report analysis of echelle spectra (R = 17,000, S/N almost-equal-to 50) of 12 Galactic bulge K giants in Baade's window from the sample o f Rich ( 1988). We perform an abundance analysis for 11 stars ranging from [Fe/H] = -1 to [Fe/H] = 0.45. The accuracy of our abundance scale is confirmed relative to the disk super-metal-rich stars via the meta l-rich giant mu Leonis, and to the Galactic globular clusters using a star in NGC 5927. We investigate the effects of CN blends on abundance analysis for stars more metal-rich than [Fe/H] = -0.3, through spectr um synthesis of the region from 6140 to 7600 angstrom, including in ex cess of 8000 (CN)-C-12 and (CN)-C-13 lines. We confirm Peterson's asse rtion that blends of atomic lines with CN vibration-rotation lines can affect abundance results at R = 17,000; failure to reject affected Fe lines would have resulted in a spurious 0.3 dex increase in the deriv ed Fe abundances of our most metal-rich stars. We find that one of the most metal-rich Baade's window stars (BW IV-167) has approximately th e same metallicity as the super-metal-rich disk giant mu Leo. The abun dance scale of Rich (1988), based on measurement of strong Fe and Mg l ines, is almost-equal-to 0.3 dex high but may agree with our results i f Rich's [M/H] actually measures [(Fe + Mg)/H]. Using our abundances t o transform Rich's [M/H] distribution, we find that the mean [Fe/H] is -0.25 for the bulge, slightly less than, but not appreciably differen t from, the mean [Fe/H] for solar neighborhood G and K giants. Surpris ingly, this is not in conflict with the Galactic metallicity gradient found for the solar neighborhood. The extant data on [Fe/H] with Galac tocentric radius are completely consistent with a zero gradient within the solar circle. Infrared flux temperatures for a sample of 34 Baade 's window K giants observed by Frogel & Whitford ( 1987) show that the bulge (J - K) versus T(eff) relation is identical to that of the sola r neighborhood. We assert that the observed infrared colors reflect th e actual effective temperatures, and that infrared colors affected by molecular blanketing cannot explain why the bulge giant branch is as b lue as that of 47 Tuc. Verification of the (J - K) temperature scale, and use of our average [Fe/H] results, suggest that the Frogel & Whitf ord ( 1987) infrared H-R diagram requires that the mean mass of Baade' s window giants can be no less than about 1.1 M.. Abundances for the a lpha-elements are based on spectrum synthesis of lines due to Mg, Ti, Ca, and Si. We find a puzzling pattern in the trends of element abunda nce ratios with [ Fe/ H 1. Mg and Ti are enhanced by almost-equal-to 0 .3 dex relative to solar over almost the full [Fe/H] range. In contras t, Ca and Si closely follow the normal trends for disk giants, with so lar ratios occurring above [Fe/H] almost-equal-to -0.2. The enrichment pattern of bulge giants sets them clearly apart from disk giants of t he same [Fe/H]. For a subset of the sample we estimate [O/Fe] almost-e qual-to O using the [O I] line at 6300 angstrom; the absence of a carb on abundance indicator requires us to assume a solar C/Fe ratio. We sp eculate that the high Mg abundances seen in ellipticals (Worthey, Fabe r, & Gonzalez 1992) may reflect a common enrichment process for bulges and ellipticals. We further propose that the high Ti abundances may l argely account for the tendency for Baade's window M giants to have la ter spectral types than solar neighborhood M giants of the same infrar ed color (Frogel & Whitford 1987). Other noteworthy results include en hanced Al abundances similar to those seen in globular cluster giants, and (tentatively) enhancement of the r-process element Eu in the meta l-rich bulge stars. The iron-peak and s-process elements appear normal relative to Fe. We find one super-Li-rich giant, BW I-194. We also id entify BW I-155 as a spectroscopic double. This work also presents the following results of general use to investigators in the field: (1) I n order to obtain a complete set of g f values for our Fe i lines, we find transformations between numerous log g f sources. In total we pro vide a list of 208 atomic g f values, for Fe I and numerous other spec ies. (2) We present a list of uncontaminated continuum regions appropr iate for use in metal-rich stars, as well as atomic lines clear from s evere CN blending at a variety of metallicities.