Top panel: distribution of the estimated errors in the [Fe/H] as quoted in Crestani et al. (2022, their Table 2). The latter work is based on high dispersion spectra (R=35 000) for 143 RRL. Bottom: distribution of the estimated errors in the photometric [Fe/H] values inferred by Dékány & Grebel (2022). Photometric [Fe/H] values are inferred from the shape of the light curve, by means of a machine learning algorithm trained (indirectly) on the spectroscopic measurements by Crestani et al. (2022).