In this paper we compare the predictions of a detailed multi-zone chemical evolution model for elliptical galaxies with the very recent observations of the galaxy NGC 4697. As a consequence of the earlier development of the wind in the outer regions with respect to the inner ones, we predict an increase of the mean stellar [] ratio with radius, in very good agreement with the data for NGC4697. This finding strongly supports the proposed outside-in formation scenario for ellipticals. We show that, in spite of the good agreement found for the [] ratio, the predicted slope of the mass-weighted metallicity gradient does not reproduce the one derived from observations, once a calibration to convert indices into abundances is applied. This is explained as the consequence of the different behaviour with metallicity of the line-strength indices as predicted by a Single Stellar Population (SSP) and those derived by averaging over a Composite Stellar Population (CSP). In order to better address this issue, we calculate the theoretical ``G-dwarf'' distributions of stars as functions of both metallicity ([Z/H]) and [Fe/H], showing that they are broad and asymmetric that a SSP cannot correctly mimick the mixture of stellar populations at any given radius. We find that these distributions differ from the ``G-dwarf'' distributions especially at large radii,except for the one as a function of [Mg/Fe]. Therefore, we conclude that in ellipticals the [Mg/Fe] ratio is the most reliable quantity to be compared with observations and is the best estimator of the star formation timescale at each radius.(abridged)

15 pages, 10 figures, ApJ accepted