Abstract: Maybe change to "estimated fraction of dust" or "regions is estimated to be", just to be clear that these are model predictions. *** Done Intro: I have been dinged by referees for not defining "cold" as soon as possible in the papers, mainly because of the argument that there are two components (T~20K and T~7K) that fit the word "cold". So perhaps you could add "particularly cold dust (T=xK)" in that first paragraph. *** Done Section 3.1 - "built in the scan" --> "built in to the scan" *** Thanks Section 3.2 - In the description of the SPIRE data reduction, you mention that the galaxies are masked out for the additional background subtraction. Is the plan to leave the size of these masked apertures for the data release documentation rather than putting them here? Did you really mean "naive" mapper? It might be useful to know what the average shift of the map coordinates is when matched to MIPS 24um (again, unless that is to go in the data release docs). *** Let's remind Chad to include this information when we write up our data release documentation. Chad gave me the wording on the HIPE reductions. He confirmed that he meant "naive". But I simply removed it to avoid confusion. Section 4.1 - "The suite...allow" --> "The suite...allows" *** Thanks Section 4.2 - "The data processing...remove" --> "The data processing...removes". No "and" after the semi-colon. *** Thanks. I checked my grammar book, and it says "and" is ok as a concluding conjunction when listing three or more items. Section 4.3 - You mention that some of the most difficult galaxies to fit are the dwarfs. What about NGC6946, where the cirrus is really high and it's not clear (at least at SPIRE wavelengths) where the galaxy ends and cirrus begins? Does that make the fit difficult or does it just introduce slightly larger error bars on a galaxy that is bright? *** Good point. Relatively speaking, NGC 6946 is much easier than the dwarfs b/c any errors errors in the sky estimation are negligible compared to the overall galaxy flux. Section 4.4 - I though the Galametz argument was that the most metal poor galaxies have their peak beyond 160um, not the metal rich? It is worth qualifying your trends with the fact that there are just far fewer galaxies in the sample with low metallicities. *** The Galametz argument is subtle, but my description of her work is accurate: metal rich galaxies exhibit *smaller* dust masses when submm data are included, and that's b/c their IR SEDs peak *beyond* 160um and thus submm data are needed to accurately model their emission. Interestingly, Bruce gave me slightly updated models, and now the trend comparing metal-rich to metal-poor is slightly stronger. --> I removed terms like "marginal trend" in my description. Section 4.6 - I agree that the dwarfs you listed have excess at 500um compared to models. But I also thought I saw a few that weren't dwarfs, like NGC5474, NGC4236, NGC3049. NGC1316. So I guess I'm not sure that I agree that "all such instances" are dwarfs. These examples aren't dramatic like the dwarfs, but then HoII isn't dramatic either. Could the argument perhaps be that some galaxies of all types show excess, but a higher percentage of the dwarfs do? *** Fair enough. I changed the wording to not exaggerate things w.r.t. only dwarfs showing the excess. Section 5 - "differences...reflects" --> "differences...reflect" *** Done Section 5 - Again, I'd try to make sure that 'cold' and 'very cold' are defined by temperature where possible. Here and in the intro... It seems to me as if the motivation is mainly written as finding the dust mass and properties of galaxies where that has been hard before, which is true. What's missing to me is the bigger picture of why finding these masses/properties is important, i.e., the physical motivation. Their properties are tied to their internal chemical evolution and their external evolutionary histories. An object that has a huge amount of metals tied up in large quantities of dust, especially on their outskirts where it is presumably undisturbed, says a lot about its merger history (likely no history of mergers). And it also says that the object underwent a burst of star formation in the past just to produce all those metals. Then you can tie its current star formation rate to the star formation rate it had to have had in the past in order to produce all that dust. So I'd be in favor of seeing a bit more discussion of the physical motivation. *** Nice. I worked some of this into the Introduction. Section 5 - "attributabe" --> "attributable" *** Thanks Section 5 - At the very end... Maybe that much dust in dwarfs is because the dust at ~20K is mainly heated by UV photons escaping from HII regions. Some of the dwarfs only have very clumpy small SF regions, so the distances between regions where the UV photons could be leaking is large. *** This argument reminds me of some work I did with SINGS data, clumpiness, escape fraction, etc. I'll keep this in mind when I put the finishing touches on the summary.