The main reason to measure the ages of GGCs is their intimate connection to the knowledge of timescales and processes of Galaxy formation and early evolution. Leaving aside the difficult issue of determining the absolute ages (the most important item concerning the cosmological impact, see Gratton et al. (1997) and VandenBerg, Stetson and Bolte (1996) for a complete review), there are several reasons to improve the reliability of the derived relative ages, at least.
First of all, if we start from the evidence (Zinn 1993, and references therein) that the halo and disk populations display distinctly different rotational properties (slowly rotating the former, rapidly rotating the latter) and we consider that GGCs at different galactocentric distances appear to show similar, though less significative, kinematical differences, it is fundamental to know the actual age differences as they set the basic timescale in the whole Galaxy formation process.
Second, in recent years, a growing group of "young" GGCs has been detected which (based on the Main Sequence TO properties) seem to be Gyr younger than the bulk of GGCs having similar metallicities (see for references Buonanno et al. 1994, Fusi Pecci et al. 1995). These clusters (Pal 12, Ruprecht 106, Arp 2, Terzan 7, IC 4499) 1 are actually quite small and appear located on great circles in the sky suggesting possible connections to satellites of the Milky Way, like the Magellanic Clouds and the Sagittarius dwarf spheroidal (Sgr dSph), and it is thus under debate their possible peculiar origin somehow related to interactions between these satellites and the Galaxy (see for references and discussions Lin and Richer 1992, Buonanno et al. 1994, Ibata et al. 1994, Mateo et al. 1995, Sarajedini and Layden 1995, Fusi Pecci et al. 1995).
Thirdly, even by restricting the sample to more "classic" GGCs, it is very important to study how the differential ages vary within the Milky Way with varying metallicity, galactocentric distances, etc. as this is crucial to a complete description of the chemical and dynamical history of the Galactic material.
As recently reviewed for instance in The Formation of the Galactic Halo ...Inside and Out (Morrison and Sarajedini, eds. 1996; see also VandenBerg, Stetson and Bolte 1996), a massive number of papers have dealt with the problem of GGC (relative) age determination using different methods. However, the uncertainties affecting the results are still too large to yield a truly satisfactory answer to most questions. In particular, for various reasons discussed below (a) it is almost impossible to make precise differential comparisons of the ages of a wide GGC sample, because even well settled procedures cannot be applied to clusters having different color-magnitude diagrams (CMD) morphologies, and (b) restricting the comparison to cluster-pairs, very discrepant results are frequently obtained. For instance, the two clusters NGC 288 and NGC 362, generally considered to be a pair of clusters having different ages by Gyr (Bolte 1989, Green & Norris 1990), have recently been re-analysed by Catelan and de Freitas Pacheco (1994) pointing out the difficulties to interpret the observed features just in terms of age-differences. Similar conclusions are obtained comparing this couple of cluster to NGC 2808 (Rood et al. 1993) and to NGC 1851 (Stetson et al. 1996), two clusters with clearly bimodal HB's.
The first traditional technique to estimate relative ages for cluster-pairs having the same metallicity, the method (Iben & Faulkner 1968), is in principle very robust essentially because the clock, i.e. the TO luminosity calibrated in terms of age by the theoretical models, is based on relatively "well known and properly checked" input physics (see Renzini & Fusi Pecci 1988, and references therein).
Nevertheless, when applied to the bulk of GGCs, the use of the observable -the magnitude difference between the TO-point and the horizontal branch (HB) at the corresponding color- suffers of two major disadvantages: (1) the apparent magnitude of the TO can hardly be measured to better than mag even from high quality CMDs, mainly because the TO-region is almost vertical at the TO-point; (2) there are clusters (like M92, M13, NGC 6752, NGC 6397, 47 Tuc, NGC 6553 etc.) which present CMDs totally depopulated in the horizontal portion of the HB, making the estimate of the HB luminosity hardly better than an educated guess. Moreover, even if the HB is populated, one should properly take into account the (generally small) effects due to evolution off the Zero Age Horizontal Branch (ZAHB).
To reduce the impact of these two effects, we propose here to use a new observable measured on the CMD and to follow a new semi-empirical approach to derive the relative ages of GGCs. In particular, this new approach passes through an empirical calibration in metallicity of the the second traditional method for estimating cluster relative ages, i.e. the so-called -method (VBS, SD) which can yield accurate age comparisons only for clusters having strictly similar metal content.
© European Southern Observatory (ESO) 1998
Online publication: April 20, 1998