Various calibrations of the absolute magnitudes of RR Lyrae stars as a function of [Fe/H] are reviewed and compared. The steep luminosity dependence on metallicity of dMV(RR)/d[Fe/H] = 0.37 previously derived from the periods, temperatures, and masses of RR Lyrae stars from the pulsation equation is the same as has recently been obtained by Buonanno, Corsi, and Fusi Pecci from main-sequence fitting of their faint-cluster photometry. In this paper we also find a similar slope for the RR luminosity-metallicity dependence using a new calibration of the position of the zero-age main sequence for stars of different metallicity. However, the very high sensitivity (and therefore relative uncertainty) of the slope determination to the positions of the zero-age main sequences for different [Fe/H] values is shown. Calibration of RR Lyrae luminosities determined from the period-shift data studied earlier, when combined with adopted precepts for RR Lyrae masses as a function of [Fe/H], gives MV(ZARR) = 0.39[Fe/H] + 1.27 for the zero-age horizontal-branch RR Lyrae stars, determined from the pulsation method. This slope dependence on metallicity is nearly the same as that determined from the main-sequence fits made by Buonanno, Corsi, and Fusi Pecci and from our own version of the main-sequence fitting method, but is more than a factor of 2 larger than dMV(RR)/d[Fe/H] = 0.19 found by combining several recent calibrations using the Baade-Wesselink method. At the moment it is not possible to decide between the various calibrations. Ages of the halo clusters studied by Buonanno and coworkers have been determined from the absolute magnitudes of the main-sequence turnoff points found in various ways that use the several RR Lyrae MV(RR) calibrations, and also directly from main-sequence fits independent of RR Lyrae star luminosities. With no [O/Fe] enhancement as [Fe/H] decreases, adopting MV(ZARR) = 0.39[Fe/H] + 1.27 from the period-shift pulsation analysis gives a mean age for this cluster sample of 17 Gyr with virtually no dependence on metallicity. However, if there is an oxygen enhancement of [O/Fe] = +0.6 for [Fe/H] < -1, the mean age decreases by 23% to 14 Gyr, again nearly independent of [Fe/H]. On the other hand, if the slope dependence of MV(RR) on [Fe/H] is as small as 0.19, as given by the current data from the Baade-Wesselink method, then an age variation of the globular cluster system of ∼20% would exist over the metallicity range for [Fe/H] of - 0.8 to -2.2, the metal-rich clusters being younger. Determination of the main-sequence turnoff luminosity directly from main-sequence fits gives an age for the globular cluster system of 19 Gyr for no oxygen enhancement, and 15.5 Gyr for a constant oxygen abundance enhancement of [O/Fe] = +0.6, again with no measurable dependence on metallicity. Each of these absolute age values is uncertain by 20% if the estimates of absolute luminosities, either of the RR Lyrae stars or of the main-sequence turnoff points, are uncertain by 0.2 mag.
