I'm assuming that there is a very deliberate attempt here not to step on Gonzalo's toes. *** Correct Substance 1) The introduction tackles cold dust as they key reason to get SPIRE/Herschel data but this seems a bit fringe to me. It's worth addressing but the main reason we want the data is that dust masses cannot be modeled to better than about a factor of two accuracy without long wavelength data, as you all showed in Draine+ 07. The cold dust thing is a provocative interpretation of some interesting IR color trends, but as you note already in the intro it breaks basic things like the metallicity. I'd suggest a minor shift in focus to call out that we need the sub-mm to see 15-20 K dust, which makes up the bulk of the dust by mass, with any accuracy. *** Agreed. I made a small modification along these lines. 2) It might be nice to give an idea of what the relative contributions of statistical / sky noise and calibration uncertainty are. Put another way, I didn't see you quote the calibration uncertainty anywhere. Might it be worth adding? *** The calibration uncertainties are listed (5-7%). 3) I'd object that the definition of U_min as quiescent is interpretive. This isn't a huge problem, but formally U_min is the radiation field experienced by all dust to make their model work. In a galaxy with light dominated overwhelmingly by star forming regions U_min will not be 'quiescent' or 'diffuse' in any meaningful sense. It's just the radiation field floor for most of the visible dust emission. My only concrete suggestion here is to separate interpretation and definition a bit on page 11 near Equation 4. You don't need to lose the interpretation, just maybe clarify that "U_min maybe interpreted as ..." or something. (The discussion at the top of page 12 is clearer on this). *** I now say "The $U=U_{\rm min}$ component may be interpreted as the dust in the general diffuse interstellar medium." 4) Something is confused in pages 12-13. You indicate that you don't fit U_min and then that you do fit U_min. I suggest to double check the language to make clear that you do fit it. *** Good catch. I forgot to include U_min in my listing of free parameters on p. 12. 5) I'm missing something on page 13: if you're going to bring up the peak wavelength claim in the Galametz paper, can't you test this? What are the peak wavelengths found for the KINGFISH SEDS? What band, then what wavelength, using the fits... seems worth stating if you're going to discuss this claim. *** Well, we only have a few low-Z systems, and they are faint. I point to Maud's paper because she can address this issue better with her sample of more (and in many cases brighter) low-Z galaxies. 6) Are you sure that it's the sub-mm data changing the low-metallicity points around and not the better constraint on U offered by having the 100 micron point (often missing from IRAS, right?). *** Excluding the 100um point in the fits doesn't change my conclusions. 7) You claim a better fit using the Draine models vs. the blackbody, particularly in the conclusions. Can you quote the statistics to back this up? I think the f-test is the right approach here, given the additional parameters in DL07. *** I've re-written the text that compares the two types of fits (see my response to your Question 9). 8) You've presumably pinged Karl regarding the SPIRE LMC/SMC results? It might be worth a cross-reference here as you discuss the dwarfs. Aguirre et al. TOPHAT results (2001) also find high SMC fluxes at long wavelengths. *** Karl and I have talked. I've included several LMC/SMC citations in the Introduction, and I've now added a reference to Aguirre et al. (2003). Thanks. 9) I don't think I really buy the last paragraph or so. Your dwarf fits all return pretty high U_min ~ 1 (up to 5 in DDO 053 or 6 in HoII, right?) ... where in the Draine model is this cold dust hiding then? I may be missing something but I don't see how the claim that the DL07 models fit better because they include cold dust matches the actual calculation, which doesn't have cold dust in it. *** Hmmm...as usual you make me think. In the plot M_dust(BB)/M_dust(DL07) vs 70/160, there's a slight trend to worse agreement for lower values of 70/160. But KINGFISGH dwarf galaxies do not preferentially lie at low 70/160, which is what I think you are presuming (see the figure plotting 250/500 vs 70/160). Here's my new text for explaining the difference earlier in the paper -- let me know if this helps to clarify: 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. Nitty Gritty 1) Sample Section ... "All are sub-LIRG" but by TIR standards NGC 2146 is in fact a LIRG, just barely. *** Modified. 2) Should Spitzer, Herschel, etc. by italicized? *** Fixed 3) Out of curiosity is this "the" KINGFISH Scanamorphous blurb? It would be nice to concentrate into one place to keep data sections compact. *** I guess not, because there's a slightly different version in Kennicutt et al. 2011. 4) Is it worth giving a date for "most recent calibrations" in S3.5? If this were to change, would the date help reproducibility? *** Done 5) Figure 2 totally needs bigger labels. The others look great. *** Fixed