Absorption Line Indices

We measured the equivalent widths (EWs) of a number of absorption lines, following the definitions given by Worthey et al. (1994) and Worthey & Ottaviani (1997). The somewhat limited spectral coverage of the Jones spectral library prevented us from measuring several interesting line indices, such as Ca4455, Fe4531, C$_2$4668, and all indices redder than $\sim$ 5400 ${\rm\AA}$ (but see Section 2.2.1). Nevertheless, the remaining Lick/IDS indices that can be modeldeficiencyed on the basis of the Jones spectral library still provide us with a rich set of spectral indicators which are sensitive to the ages of old stellar populations, as well as to the abundances of key elements for the understanding of galaxy chemical evolution, such as iron, magnesium, calcium, carbon and nitrogen.

Another limitation of the Jones library refers to its coverage of stellar parameters, whereby some important loci of stellar parameter space are not represented with sufficient density. In order to address this deficiency, and enhance the robustness of our fitting functions in those stellar parameter regions, we decided to supplement our data with index measurements from Worthey et al. (1994) for stars hotter than 7000 K, M giants and K-M dwarfs. For that purpose, we need to determine the conversion between our EWs and the Lick/IDS system. A detailed recipe to perform this determination has been given by Worthey & Ottaviani (1997), and is followed here. The most important part of the conversion involves degrading the resolution of the Jones spectra (1.8 ${\rm\AA}$) to match the lower, variable resolution of Lick/IDS spectra (8.5-11 ${\rm\AA}$). This was achieved by gaussian-convolving the Jones spectra in order to match the Lick/IDS resolution. The resolution of the original Lick/IDS spectra at the central wavelength of each index was obtained from graphical interpolation in Figure 7 of Worthey & Ottaviani (1997). In Table 1 we provide the resolution FWHM assumed for each index.

Equivalent widths are somewhat dependent on the software used to perform the actual measurements. In the initial stages of this project, all index measurements were performed using a script based on the IRAF bplot routine (i.e., splot in batch mode). Unfortunately, however, we later realized that bplot did not consider fractionary pixels. That means that the wavelengths of the pseudo-continuum and passband definitions actually employed in the measurements were not the input numbers, but were instead the wavelengths of the pixels that were nearest to those of the original definitions. That error introduced in our EWs systematic effects that were a function of the actual grid of wavelengths defined by the dispersion solution for each spectrum and which, of course, were more severe for lower resolution spectra. As a result, the index measurements had to be retaken, this time using the LECTORprogram, by A. Vazdekis, and all the numbers in this paper (in particular, the index fitting functions, see Section 3) had to be re-derived.