Historically, the debate on the star formation history of ETGs was framed in terms of two competing scenarios: hierarchical clustering (e.g., White & Rees 1978, Searle & Zinn 1978), according to which these galaxies were assembled through the merging of less massive structures formed at high redshift; and monolithic dissipative collapse (e.g., Eggen et al. 1962, Larson 1974), whereby massive ETGs were formed at very high redshift by means of a rapid gravitational collapse. Deciding between these two scenarios was one of the main motivations behind attempts to measure stellar ages and abundances in ETGs. Today there is little question that galaxies formed hierarchically in a -CDM universe, and while that historical debate has been settled, studies of unresolved stellar populations have acquired renewed importance, as they provide much needed constraints to increasingly sophisticated galaxy formation models.
Early attempts at dating stellar populations from applications of stellar population synthesis models to observations of integrated light were based on photometric or low-resolution spectrophotometric observations (e.g., O'Connell 1980, Gunn et al. 1981, Renzini & Buzzoni 1986) and high-resolution photographic spectroscopy (e.g., Rose 1985). They led to promising, yet not entirely conclusive results, due to limitations of the early models and/or uncertainties associated with the age-metallicity degeneracy (e.g., Renzini 1986, Worthey 1994). The latter is a manifestation of the similar dependence of the temperatures of main sequence and giant stars of a given stellar population on age and metallicity, which causes the integrated colors of stellar populations, particularly in the optical and near-UV, to respond in similar ways to variations of these two parameters.
It was only after the systematic modeling of Balmer lines as (relatively) clean age indicators, initiated by Worthey (1994), that reliable quantitative estimates of mean luminosity-weighted 1 stellar ages became available. The method relies on the dependence of Balmer lines, such as H 486 nm, on the temperatures and luminosities of turnoff stars - which are higher in younger stellar populations. For A-type stars and cooler, Balmer line strength is positively correlated with temperature, so that the lines in integrated spectra of stellar populations older than a few 100 Myr are stronger for younger ages. Spectroscopic ages based on H for large numbers of ETGs, both in the field and in clusters, suggest that many of them have undergone recent star formation events (e.g., Trager et al. 2000, Kuntschner 2000, Caldwell et al. 2003, Denicoló et al. 2005, Thomas et al. 2005). Trager et al. proposed a scenario where the bulk of the stars in their sample galaxies were old, and only a very small fraction of their stellar populations had young ages (~ 1 Gyr). Because the latter are brighter in the optical, they weight the mean ages towards lower values. Although plausible, this schematic scenario could not be verified by the Trager et al. data because of the degeneracy between the age of the young component and its relative contribution to a galaxy's total stellar mass budget, which cannot be broken on the basis of H and metal-line indices alone.
With the inclusion of additional age indicators, such as higher-order Balmer lines in the blue (Leonardi & Rose 1996, Worthey & Ottaviani 1997, Schiavon 2007), stronger constraints were placed on the age distributions of stars. These studies confirmed early suggestions that a small fraction of the stellar mass budget (few %) is required to be young in order to match the data. They also ruled out claims that Balmer line strengths in the spectra of ETGs required the presence of metal-poor stars with blue horizontal branches (e.g., Freitas Pacheco & Barbuy 1995, Maraston & Thomas 2000, Lee et al. 2000). Models including metal-poor stars cannot simultaneously match all Balmer and metal lines in the 400-530 nm range (see also Trager et al. 2005).
Perhaps the cleanest evidence for the presence of young/intermediate-age stellar populations was provided by Galex observations of ETGs by Yi et al. (2005). Analyzing UV-optical colors of ETGs from the Sloan Digital Sky Survey, Yi et al. found that approximately 2/3 of their sample present strong UV fluxes which cannot be explained by hot horizontal branch stars. They estimate that approximately 1-2% of the stellar mass is in the form of young stars. Because they constitute a tiny fraction of the stellar mass budget, young stars are vastly outshone by older populations in the optical, where they can only be detected on the basis of accurate spectrum synthesis of very high S/N spectra (e.g., Schiavon et al. 2004, Schiavon 2007, Graves et al. 2007) - while being relatively easy to detect in the UV.
More recent estimates based on a combination of Galex photometry with larger SDSS samples indicate that as much as 10% of all stellar mass in ETGs today were formed since z ~ 1 (S. Yi, this volume). While a fascinating debate is ongoing regarding what triggered and quenched this relatively recent star formation (e.g., Kuntschner et al. 2010, Zhu et al. 2010, Sánchez-Blázquez et al. 2009, Serra et al. 2008, Schawinski et al. 2007, Kaviraj et al. 2007, Graves et al. 2007), information on the history of formation of the remaining ~ 90% of the stellar mass is relatively scanty, possibly because it requires observations of large samples at higher redshifts than so far possible and/or a far more thorough assessment of the fossil record in the abundance patterns of stars in the nearby ETGs. The latter is the topic of the next section.
1 See Trager & Somerville (2009) for a discussion of the relation between luminosity- and mass-weighted ages/abundances with those obtained from comparison of line indices with single stellar population synthesis models. Back.