We study the stellar populations of a sample of 14 elliptical galaxies in the Virgo cluster .
Using spectra with high signal-to-noise ratio ( S/N \lower 2.15 pt \hbox { $ \buildrel > \over { \sim } $ } 100 Å ^ { -1 } ) we propose an alternative approach to the standard side-band method to measure equivalent widths ( EWs ) .
Our Boosted Median Continuum is shown to map the EWs more robustly than the side-band method , minimising the effect from neighbouring absorption lines and reducing the uncertainty at a given signal to noise ratio .
Our newly defined line strengths are more sucessful at disentangling the age-metallicity degeneracy .
We concentrate on Balmer lines ( H \beta , H \gamma , H \delta ) , the G band and the combination [ MgFe ] as the main age and metallicity indicators .
We go beyond the standard comparison of the observations with simple stellar populations ( SSP ) and consider four different models to describe the star formation histories , either with a continuous star formation rate or with a mixture of two different SSPs .
These models improve the estimates of the more physically meaningful mass-weighted ages .
Composite models are found to give more consistent fits among individual line strengths and agree with an independent estimate using the spectral energy distribution .
A combination of age and metallicity-sensitive spectral features allows us to constrain the average age and metallicity .
For a Virgo sample of elliptical galaxies our age and metallicity estimates correlate well with stellar mass or velocity dispersion , with a significant threshold around 5 \times 10 ^ { 10 } M _ { \odot } above which galaxies are uniformly old and metal rich .
This threshold is reminiscent of the one found by Kauffmann et al .
in the general population of SDSS galaxies at a stellar mass 3 \times 10 ^ { 10 } M _ { \odot } .
In a more speculative way , our models suggest that it is formation epoch and not formation timescale what drives the Mass-Age relationship of elliptical galaxies .