Reply to Two Comments on "Dark matter searches going bananas the contribution of Potassium (and Chlorine) to the 3.5 keV line"

We respond to two comments on our recent paper, Jeltema & Profumo (2014). The first comment by Boyarsky et al. confirms the absence of a line from M31 in the 3-4 keV energy range, but criticizes the energy range for spectral fitting on the basis that (i) the background model adopted between 3-4 keV is invalid outside that range and that (ii) extending the energy range multiple features appear, including a 3.5 keV line. Point (i) is manifestly irrelevant (the 3-4 keV background model was not meant to extend outside that range), while closer inspection of point (ii) shows that the detected features are inconsistent and likely unphysical. We demonstrate that the existence of an excess near 3.5 keV in the M31 data requires fitting a broad enough energy range such that the background modeling near 3.5 keV is poor to a level that multiple spurious residual features become significant. Bulbul et al. criticize our use of WebGuide instead of the full AtomDB package. While a technically correct remark, this is only a red herring: our predictions are based on line ratios, and not on absolute emissivities; line ratios, for atomic transitions with similar peak temperatures, are largely temperature-independent, thus the line ratios we employed to draw our conclusions are substantially correct. Bulbul et al. also present a new analysis of their data at lower energy, which excludes a significant Cl contamination to the 3.5 keV line. Cl emission was however predicted to be subdominant in our original study. Both of the Bulbul et al.'s criticisms are thus inconsequential to the conclusions of our original study. Finally, we demonstrate that the multi-temperature models employed in Bulbul et al. are, in fact, inconsistent, based on the Ca XX to Ca XIX line ratio: we show that the overestimated cluster plasma temperatures they employ lead to gross underestimates of the K XVIII line emissivity.