Example (subhalo 577372) illustrating the recovery of ∇[Fe/H](τ) from $\mathrm{Range}\stackrel{\sim }{[\mathrm{Fe}/\mathrm{H}]}(\mathrm{age})$. Left: Density distribution of the [Fe/H]–R_birth plane for stars born 2 − 3 Gyr ago. The correlation between the [Fe/H] running median and R_birth is given in the top right corner. The fitted true birth gradient is shown as the black line, and the maximum and minimum [Fe/H] used to determine the range are shown as the dashed red lines. Right: ∇[Fe/H](τ) versus $\mathrm{Range}\stackrel{\sim }{[\mathrm{Fe}/\mathrm{H}]}(\mathrm{age})$ for the same galaxy, with each point representing a different age bin (width 1 Gyr). $\mathrm{Range}\stackrel{\sim }{[\mathrm{Fe}/\mathrm{H}]}(\mathrm{age})$ is determined by looking at the range in [Fe/H] across all age bins, and then normalized so the minimum range is given a value of 0, and the maximum range is given a value of 1. The correlation between $\mathrm{Range}\stackrel{\sim }{[\mathrm{Fe}/\mathrm{H}]}(\mathrm{age})$ and ∇[Fe/H](τ) for this galaxy is −0.87, indicating the evolution of the gradient can be recovered well from $\mathrm{Range}\stackrel{\sim }{[\mathrm{Fe}/\mathrm{H}]}(\mathrm{age})$.