Introduction: It is stated that FIR/submillimeter data is crucial for an estimate of dust mass, but doesn't the bottom panel of figures 4 and 5 (and the small offsets and scatter quoted) suggest that wavelengths longer than 160 micron actually don't change things very much? Except perhaps for dwarf galaxies, where S+H dust masses then get close to violating dust/gas ratio constraints. *** Your interpretation of those figures is correct. On the other hand, I want these data to be relevant; the spin here is that submm data are crucial for *accurate* dust masses. I've tried to emphasize this a bit with "long wavelength data are crucial for probing relatively cool 15--20~K dust with any accuracy, and dust emitting at this temperature range makes up the bulk of the dust mass in a typical star-forming galaxy". 3.3, Sky subtraction: 1) Why did you use many small apertures instead of an annulus? Is it because of MW cirrus emission (which is not discussed)? *** There was no real particular reason for the choice, besides the fact that I'm accustomed to using multiple small apertures. I do mention cirrus a couple times, though. 2) In later sections (S4.1 paragraph 3, S4.3 Paragraph 5, possibly S4.6) you note that results for some faint galaxies depend crucially on the sky subtraction, which may not be accurate, above and beyond the stated errors. Is it possible to flag in some way the most outstanding of these cases? The contribution of systematics in the sky background estimation may also deserve some discussion in 3.4. *** I invoke "sigma_sky" in Section 3.4, but I do quote the sigma_sky values for PACS and SPIRE in Sections 3.1 and 3.2. I hope that is sufficient! 3.4 Aperture Photometry: 1) As Chad mentioned, I wonder about the removal of background galaxies from the source aperture while they are included in the sky apertures. Is this only bright galaxies that would be avoided in the selection of sky apertures? *** Thanks for the useful comment. I've reworded the sky subtraction discussion to clarify that I do indeed remove obvious/prominent background galaxies from the sky apertures. My original point was: when there is literally a sea of background galaxy detections at 250um, I can't possibly properly ID and remove every one, from either the galaxy ap or the sky aps. So I left them in everywhere. But the text was unclear on that. So I now just focus on explaining that I remove obvious background galaxy contamination everywhere. 2) Pardon my density, but by standard deviation of the surface brightness fluctuations do you mean just the standard deviation of the values in all the sky pixels, or the standard deviation of the mean, i.e. the former divided by N_sky^(1/2)? *** It's from all the sky fluctuations. 4.4, Fit Parameters 1) The wording of the 20% increase in the fraction of dust heated in PDRs is a little ambiguous. Since that fraction is sometimes expressed as a percentage, an incautious reader might take that to mean that the typical gamma goes from ~5% to around 23%, instead of from 5% to 6%. But perhaps it is not worth clarifying. 2) the increased average gamma is interesting, but the scatter is actually smaller than for both M_d and U_min. You could make a case that even though the average parameter ratios are near 1 for these two, the Herschel data is actually making a bigger difference on a galaxy by galaxy basis for U_min and M_dust. I suppose to make that case you'd have to worry about what the uncertainties are, which is difficult to quantify, but the trends of both these parameter ratios with 70/160 micron flux suggests that there is a real effect here. It just happens to average to 1 for the KINGFISH sample. *** That's interesting. I've made a note of this in the text. 3) The metallicity dependence of the dust mass determination seems to come almost entirely from the dwarfs, which as discussed in the introduction and 4.6 may have a submillimeter excess that is not due to cold dust, as that would pose problems for the dust to gas ratios. It appears that the dust masses actually don't change much in high metallicity galaxies when including herschel data, at least on average, which would not seem to be consistent with Galametz 2011. Can you tell whether the emission peak is occurring longward of 160 micron preferentially for the high metallicity galaxies? *** If you compute the statistics for the higher Z galaxies, the dust mass is 10% lower when Herschel data are included, which works in the same sense as the result in Galametz et al. 2011. Maud and I have looked into this for KINGFISH, specifically investigating the SED shapes as a function of Z. There really isn't an obvious trend in the peak wavelength with Z, and that's partly conveyed by my plot of 250/500 vs 70/160, color coded according to Z. 4.5, Comparison to BB: 1) Since gamma is typically less than 10% of the dust mass, adding another 70% of dust mass from the inclusion of hot dust alone seems difficult. It would be almost enough though to reconcile the BB fits including 70 micron with beta=2. Is there any constraint in the Draine and Li models on a parameter related to beta? *** My understanding is that that parameter is essentially fixed, according to observations of the Milky Way. 2) It seems strange to me that, as stated, inclusion of the 70 micron data point alone does not affect the BB masses much, or that fixing beta but for lambda > 100 micron doesn't affect the results, but doing both gives you substantially larger dust masses. Does the fit quality substantially improve or decrease when doing these things? Are the BB fits better or worse than the Draine+Li fits in the relevant bands? What happens if you fit the Draine and Li models to just the lambda> 100 micron data? While the ratio as given is interesting because many people in the literature use BB fits, I think drawing conclusions about why the fits are different first requires a more fair comparison of models. *** I think Karl Gordon pinged the entire KINGFISH team about his forthcoming paper that compares dust masses derived using different techniques. I understand what you are saying -- my view was to not go into a whole lot of detail like Karl will do, but to simply compare two now-popular ways to estimate dust mass. Yes, it may be an unfair comparison to incorporate more photometry for the DL07 fits, but that's the way people are doing it in the literature. I've added some discussion that I think helps to elucidate the discrepancy in the dust masses: "Figure~\ref{fig:Draine_v_BB} shows a primary reason for the discrepancy: even when limited to $\lambda \geq 100\micron$ photometry, single-temperature blackbody fits overestimate the dust temperature, thus underestimating the dust mass. The single-temperature model does not account for the contribution of warm dust emitting at shorter wavelengths and the temperatures are driven towards higher values in the attempt to fit both the short and long wavelength far-infrared emission."