Stars in globular clusters are essentially coeval and - with very few exceptions - have all the same chemical composition, with only few elements breaking the rule. As such, globular clusters are the best approximation to simple stellar populations (SSP), and therefore offer a virtually unique opportunity to relate the integrated spectrum of stellar populations to age and chemical composition, and do it in a fully empirical fashion. Indeed, the chemical composition can be determined via high-resolution spectroscopy of cluster stars, the age via the cluster turnoff luminosity, while integrated spectroscopy of the cluster can also be obtained without major difficulties. In this way, empirical relations can be established between integrated-light line indices (e.g. Lick indices as defined by Faber et al. 1985) of the clusters, on one hand, and their age and chemical composition on the other hand (i.e., [Fe/H], [ $\alpha$ /Fe], etc.).

These empirical relations are useful in two major applications: 1) to directly estimate the age and chemical composition of unresolved stellar populations for which integrated spectroscopy is available (e.g. for elliptical galaxies and spiral bulges), and 2) to provide a basic check of population synthesis models.

Today we know of about 150 globular clusters in the Milky Way (Harris 1996), and more clusters might be hidden behind the high-absorption regions of the Galactic disk. Like in the case of many elliptical galaxies (e.g. Harris 2001), the Galactic globular cluster system shows a bimodal metallicity distribution (Freeman & Norris 1981; Zinn 1985; Ashman & Zepf 1998; Harris 2001) and consists of two major sub-populations, the metal-rich bulge and the metal-poor halo sub-populations.

The metal-rich ( ${\rm [Fe/H]}>-0.8$ dex) component was initially referred to as a "disk'' globular cluster system (Zinn 1985), but it is now clear that the metal-rich globular clusters physically reside inside the bulge and share its chemical and kinematical properties (Minniti 1995; Barbuy et al. 1998; Côté 1999). Moreover, the best studied metal-rich clusters (NGC 6528 and NGC 6553) appear to have virtually the same old age as both the halo clusters and the general bulge population (Ortolani et al. 1995a; Feltzing & Gilmore 2000; Ortolani et al. 2001; Zoccali et al. 2001,2002; Feltzing et al. 2002), hence providing important clues on the formation of the Galactic bulge and of the whole Milky Way galaxy.

Given their relatively high metallicity (up to ${\sim} Z_\odot$ ), the bulge globular clusters are especially interesting in the context of stellar population studies, as they allow comparisons of their spectral indices with those of other spheroids, such as elliptical galaxies and spiral bulges. However, while Lick indices have been measured for a representative sample of metal-poor globular clusters (Burstein et al. 1984; Covino et al. 1995; Cohen et al. 1998; Trager et al. 1998), no such indices had been measured for the more metal-rich clusters of the Galactic bulge. It is the primary aim of this paper to present and discuss the results of spectroscopic observations of a set of metal-rich globular clusters that complement and extend the dataset so far available only for metal-poor globulars.

Substantial progress has been made in recent years to gather the complementary data to this empirical approach: i.e. ages and chemical composition of the metal-rich clusters. Concerning ages, HST/WFPC2 observations of the clusters NGC 6528 and NGC 6553 have been critical to reduce to a minimum and eventually to eliminate the contamination of foreground disk stars (see references above), while HST/NICMOS observations have started to extend these studies to other, more heavily obscured clusters of the bulge (Ortolani et al. 2001).

High spectral-resolution studies of individual stars in these clusters is still scanty, but one can expect rapid progress as high multiplex spectrographs become available at 8-10 m class telescopes. A few stars in NGC 6528 and NGC 6553 have been observed at high spectral resolution, but with somewhat discrepant results. For NGC 6528, Carretta et al. (2001) and Coelho et al. (2001) report respectively $\rm [Fe/H]=+0.07$ and -0.5 dex (the latter value coming from low-resolution spectra). For [M/H] the same authors derive +0.17 and -0.25 dex, respectively. For NGC 6553 Barbuy et al. (1999) give $\rm [Fe/H] =-0.55$ dex and $\rm [M/H] =-0.08$ dex, while Cohen et al. (1999) report $\rm [Fe/H] =-0.16$ dex, and Origlia et al. (2002) give $\rm [Fe/H] =-0.3$ dex, with [ $\alpha$ /Fe] =+0.3 dex. Some $\alpha$ -element enhancement has also been found among bulge field stars, yet with apparently different element-to-element ratios (McWilliam & Rich 1994).

Hopefully these discrepancies may soon disappear, as more and better quality high-resolution data are gathered at 8-10 m class telescopes. In summary, the overall metallicity of these two clusters (whose color magnitude diagrams are virtually identical, Ortolani et al. 1995a) appears to be close to solar, with an $\alpha$ -element enhancement [ $\alpha$ /Fe] $\simeq +0.3$ dex.

The $\alpha$ -element enhancement plays an especially important role in the present study. It is generally interpreted as the result of most stars having formed rapidly (within less than, say $\sim$ 1 Gyr), thus having had the time to incorporate the $\alpha$ -elements produced predominantly by type II supernovae, but failing to incorporate most of the iron produced by the longer-living progenitors of type Ia supernovae. Since quite a long time, an $\alpha$ -element enhancement has been suspected for giant elliptical galaxies, inferred from the a comparison of Mg and Fe indices with theoretical models (Peletier 1989; Worthey et al. 1992; Davies et al. 1993; Greggio 1997). This interpretation has far-reaching implications for the star formation timescale of these galaxies, with a fast star formation being at variance with the slow process, typical of the current hierarchical merging scenario (Thomas & Kauffmann 1999). However, in principle the apparent $\alpha$ -element enhancement may also be an artifact of some flaws in the models of synthetic stellar populations, especially at high metallicity (Maraston et al. 2001). The observations presented in this paper are also meant to provide a dataset against which to conduct a direct test of population synthesis models, hence either excluding or straightening the case for an $\alpha$ -element enhancement in elliptical galaxies. This aspect is extensively addressed in an accompanying paper (Maraston et al. 2002).

The main goal of this work is the measurement of the Lick indices for the metal-rich globular clusters of the bulge and of the bulge field itself. Among others, we measure line indices of Fe, Mg, Ca, CN, and the Balmer series which are defined in the Lick standard system (Worthey & Ottaviani 1997; Trager et al. 1998). In Sect. 2 we describe in detail the observations and our data reduction which leads to the analysis and measurement of line indices in Sect. 3. Index ratios in globular clusters and the bulge are presented in Sect. 4. Index-metallicity relations are calibrated with the new data in Sect. 5 and Sect. 6 discusses the index variations as a function of galactocentric radius. Section 7 closes this work with the conclusions followed by a summary in Sect. 8.

**Table 1:** General properties of sample Globular Clusters. If not else mentioned, all data were taken from the 1999 update of the McMaster catalog of Milky Way Globular Clusters (Harris 1996). $R_{\rm gc}$ is the globular cluster distance from the Galactic Center. $r_{\rm h}$ is the half-light radius. E_(B-V) and (m-M)_V are the reddening and the distance modulus. $v_{\rm rad}$ the heliocentric radial velocity. Note, that our radial-velocity errors are simple *internal* errors which result from the fitting of the cross-correlation peak. The real *external* errors are a factor $\sim$ 2-4 larger. HBR is the horizontal-branch morphology parameter (e.g. Lee et al. 1994).
GC	$R_{\rm gc}$ [kpc]	[Fe/H]	$r_{\rm h}$ [arcmin]	E_(B-V)^a	(m-M)_V	$v_{\rm rad}^{b}$ [km s^-1]	$v_{\rm rad}$ [km s^-1]	HBR^c

NGC 5927	4.5	-0.37	1.15	0.45	15.81	$-130\pm12$	$-107.5\pm 1.0$	-1.00^d
NGC 6218 (M 12)	4.5	-1.48	2.16	0.40	14.02	$-46\pm23$	$-42.2\pm 0.5$	0.97^d
NGC 6284	6.9	-1.32	0.78	0.28	16.70	$8\pm16$	$27.6\pm 1.7$	1.00^e
NGC 6356	7.6	-0.50	0.74	0.28	16.77	$35\pm12$	$27.0\pm 4.3$	-1.00^d
NGC 6388	4.4	-0.60	0.67	0.40	16.54	$58\pm10$	$81.2\pm 1.2$	-0.70^e
NGC 6441	3.5	-0.53	0.64	0.44	16.62	$-13\pm10$	$16.4\pm 1.2$	-0.70^f
NGC 6528	1.3	-0.17	0.43	0.56	16.53	$180\pm10$	$184.9\pm 3.8$	-1.00^d
NGC 6553	2.5	-0.34	1.55	0.75	16.05	$-25\pm16$	$-6.5\pm 2.7$	-1.00^d
NGC 6624	1.2	-0.42	0.82	0.28	15.37	$27\pm12$	$53.9\pm 0.6$	-1.00^d
NGC 6626 (M 28)	2.6	-1.45	1.56	0.43	15.12	$-15\pm15$	$17.0\pm 1.0$	0.90^d
NGC 6637 (M 69)	1.6	-0.71	0.83	0.16	15.16	$6\pm12$	$39.9\pm 2.8$	-1.00^d
NGC 6981 (M 72)	12.9	-1.40	0.88	0.05	16.31	$-360\pm18$	$-345.1\pm 3.7$	0.14^d

^a: Taken from Harris (1996).
^b: This work.
^c: Horizontal branch parameter, (B-R)/(B+V+R), for details see e.g. Lee et al. (1994).
^d: Taken from Harris (1996).
^e: Taken from Zoccali et al. (2000).
^f: Due to very similar HB morphologies in CMDs of NGC 6388 and NGC 6441 (see Rich et al. 1997), we assume that the HBR parameter is similar for both globular clusters and adopt $\rm HBR=-0.70$ for NGC 6441.

1 Introduction