Comparison of fluorescence efficiencies (with geometric factors) and observed fluxes of Ni I (left) and Fe I (right). Model error bars are taken from the Monte Carlo iterations of the fluorescence model within the stated Einstein A coefficient uncertainty. An absolute error of ±40% was assumed for the observed fluxes (R. Meier et al. 1998). The linear fit used to compute the column densities is shown in red for each atom.