Abstract. There are several points I will make below which might go into abstract, if they pan out. I think the fraction of resolved and unresolved 24um flux from spirals is interesting, especially if you do not count the nucleus. This measures how much of the radiation from recent star formation in GMCs is absorbed by the natal cloud, versus how much escapes to the diffuse ISM. We might also compare that with how much UV escapes the galaxy. Also, I think we can get a measure of the interstellar radiation field (G_0) in the FUV from the GALEX measurements. I show how below. Can we say anything about how much dust is heated by the UV versus how much from optical or near IR photons from stars? --I sent you the UV fluxes and effective radii from which 50% of the UV light derives. We also have the optical and NIR data available. Can you use this information to constrain it? p.2. Intro. line 2. It seems like the sentence about diffuse interstellar bands is irrelevant to this paper, and could be omitted. --But it's such a fun fact! second paragraph, 1st sentence. This interests me very much, and I think we should give a reference here (to the statement that much of the radiation intercepted by interstellar grains originates in the ultraviolet. What I do note from our SEDs is that the nu F_nu in the FIR is always similar to the nu F_nu in the ultraviolet. Is this a coincidence or is this telling us that the dust is heated often by the UV? --But that isn't true for low IR/UV galaxies, as evidenced by Fig 9. But otherwise, yes, it seems that often the bulk of the dust is heated by UV. 3 lines from bottom. "exhibiting redder slopes THAN STARBURST GALAXIES" --Thanks p.3, section 2, line6 "infrared to optical color" Maybe use 'ratio" instead of "color"? --Thanks p.4. In the IR data section, you might mention the beam size for the MIPS 24um, and what this corresponds to at 10 Mpc in linear diameter. This helps interpret the "resolved" vs. "unresolved" discussions later. --Good point. I've taken your lead and added the linear resolution at 24um in the Section and paragraph that introduces why 24um might be the best wavelength at which to probe this resolved-unresolved idea. section 3.1. line 3. you might mention the bandwidth of the GALEX filters which are centered at 1528 and 2271. --Done Also, I was interested in how big were the UV emitting regions typically in our Galaxies. That is, what angular diameter, or linear diameter would encompass 50% of the UV flux we report. With this knowledge, and the UV fluxes we report, you can get a handle on the interstellar UV radiation field within this diameter in the Galaxy. I have calculated analytically that G_0 ~ 3 (1'/Theta)^2 [nu_FUV f_FUV/(10^{-13} W m^{-2}] where G_0 is how I defined it for PDRs (G_0=1 means a flux between 912-2000 of 1.6 x 10^{-3} erg cm^{-2} s^{-1}), Theta is the angular diameter of a circular beam that encompasses the observed FUV flux, and nu_FUV f_FUV is as you give in the paper. --I sent you the f_FUV and Theta data. p.6, section 4.1. line 6. I think you should describe what alpha _SED is in a bit more detail here. --I've added a bit of explanation end of 1st paragraph. I find the stellar curve very confusing and misleading. In the early type galaxies especially, the nu F_nu of the "missing UV" (that is, the difference between the model and the observed UV) is much greater than the nu F_nu of the dust in the IR. This shows the model is no good, so it only serves to confuse the reader in these cases. If the UV is missing, and thus absorbed, it should come out in IR. --You raise a good point, but I'm not sure how to remedy it. You're right: a 1 Gyr continuous SF model is not appropriate for ellipticals. On the other hand, this paper was born out of a careful compromise between the SINGS and GALEX teams, one that mandated this to be a non-sophisticated data paper. A more sophisticated analysis that would include fitting stellar population sythesis models, etc to the data is expected to come in Paper II of this SINGS+GALEX series. I like the idea of having the same model from galaxy to galaxy as a reference, so I've added to the text a caveat that this model is likely not applicable to ellipticals. p.7. section 4.2 third paragraph, 1st sentence. Define what the "bins" are. --Done p.8. section 4.3, 1st paragraph. You say the analysis covers 3/4ths of the galaxies, but in Table 1 the superscript + which indicates which systems are involved is added to much less than 3/4ths. Maybe missing from page 2 of the Table. --Correct. Thanks for catching that. 2nd paragraph. 89% and 7%. In figure caption of figure 10 it says 84% abd 10%, In the Conclusion (section 6) you say 88% and 7%. I think a little more discussion of what these percentages physically mean is in order, here and in conclusion. --I've rewritten this section in an attempt to be more physical. I've also fixed that percentage discrepancy. Thanks. p.9, section 5.2, line 3. "changing significance of RECENT star formation" perhaps (although you may object not for ellipticals!). I did not like this sentence in general, to me change in observed UV reflects how the dust is distributed and how much there is. THis is especially true of the uv to IR ratio. I feel like it less reflects the star formation and the ultraviolet luminosity to the overall energy budget in galaxies. In fact, I was not sure what you meant. --I've tried to reword this to include the spirit of your comments. line 7 Typo "is due excess to an excess" line 8 "due to a relative paucity of ultraviolet photons" I do not see why it is ultraviolet. The grains could get their IR from optical or near IR photons, especially in ellipticals. --Thanks - I've made these changes. same paragraph. It is about here that it struck me that the TIR/UV ranged from 0.3 to 8, much less range than the optical to TIR, and it suggested to me that UV may heat grains often. --The lowest bin has a ceiling of 0.3 (now 0.4), and the highest bin has a floor of 8. But the full range spans <0.1 to >100, similar to the full range spanned by the optical/infrared ratio. Also, didn't Sangeeta Malhotra do a paper on ellipticals based on ISO data as part of our ISO Key Project team that discussed the origin of the UV in ellipticals? I may have been on that paper! --You were second author! In that paper we said: "We estimate the UV radiation expected from the old stellar populations.... In three out of four galaxies the predicted UV falls short by a factor 2-3..." This would require modeling for this paper, something again that has been mandated to appear in Paper II in this series. p.10. section 5.3. line 3. I did not see the relevance of the statement about less line blanketing leading to harder radiation fields. --I'm trying to make a smooth segue from the end of Section 5.2, which talks about IR/UV in irregulars, to the beginning of Section 5.3, which introduces the ratio as a function of FIR color. So my point is that low metallicity irregulars show hotter dust since they have harder radiation fields. I add diminished line blanketing as a reason. Next paragraph, line 1, "the wedge". On the phone you answered my question about whether these galaxies with low 70/160 were low luminosity. Apparently you said there are a number with high luminosity. You might make this point, as I was worried that if they were low luminosity, then the upper left part of the curve has high TIR/UV but low TIR, implying very low UV. Then, GALEX might not detect the UV and we would not have points in this region. However, if as you say, it is not observational bias, then perhaps we should speculate that the reason for this vacant part of parameter space is that to have high TIR/UV, you need lots of dust opacity. But lots of dust tends to lead to star formation in the dusty region, which heats the dust and so it is hard to have low 70/160. --I've added hundreds of nearby galaxies from the just-accepted GALEX Atlas of Nearby Galaxies. Those data show the same distribution, further supporting that this distribution is not an observational selection effect. I've also rewritten this portion of this section to incorporate your above scenario. line 5. "this distribution". Other than the wedge, it looks pretty random. I do not see strong evidence that high 70/160 leads to hotter dust which your next sentence seems to imply. But then you hedge afterwards--I would write this more clearly. Also, perhaps you want to change a later sentence to "Such clumpy galaxies would hence show comparatively low IR to UV ratios, BUT A RANGE OF 70/160 DEPENDING ON WHETHER LOCAL CLUMPS OR MORE DIFFUSE DUST ABSORBS THE BULK OF THE STELLAR PHOTONS. In the next sentence, you might add "high infrared-to-ultraviolet ratios AND HIGH DUST T. --Done p.11. Perhaps I am misguided, but I have argued on a number of our SINGS papers that the 24um may be from single photon heating of very small grains (to explain the good linear correlation with star formation, etc). Therefore, even though it may correlate with HII regions, it need not come from within the HII region. paragraph just before section 5.4. On the resolved and unresolved at 24 um by MIPS. I would give beam angular diameter and also the linear diameter this corresponds to at 10 Mpc. Also, it is of great interest for those of us who model the ISM of other galaxies to know what the interstellar radiation field is, and how high a flux of radiation a typical GMC receives. To know this, we need to know what fraction of the luminosity of the massive stars born in GMCs is absorbed by that natal GMC (the unresolved IR), versus how much is absorbed by the diffuse ISM (the resolved IR). We would also like to know what fraction of the UV starlight escapes the galaxy (which GALEX should tell us). One thing that might reduce scatter in your plots is to ignore the nucleus, and only plot the resolved versus unresolved emission for regions outside the nucleus. This helps answer the questions above. --I added the beam size at 10 Mpc. As for the nuclear vs disk analysis: I might sound like a broken record, but believe it or not such an analysis has been reserved for Papers III and IV in this series... p.12 2nd to last paragraph of section 5.4, starting "Figure 17.." In the 2nd sentence, "known to be unresolved at 24um" are you referring to sources with the bulk from the nucleus? --Yes. I've reworded it to be more obvious. last paragraph of section 5.4. I do not think you have explained why the larger the specific star formation rate, the smaller the IR to UV ratio in later types, and why it is the reverse in early types. If you increase the ratio of IR to UV, you have higher extinction. Thus, if ellipticals show increased ratios of IR/UV with the star formation rate, this is because the amount of dust must be increasing with the star formation rate. It is not because there is more UV. More UV would not change the ratio if the dust extinction stayed the same. Similarly, in later types, if increasing the star formation rate decreases the IR/UV ratio, then what we are saying is that the increased star formation rate leads to more holes for UV to escape. This is not said clearly. --Thanks for the input. Yeah, my initial idea doesn't seem to pan out. I've changed this paragraph along your suggested lines. p.13, section 5.5, last paragraph. Our points seem to deviate from the Cortese solid line in figure 18. They seem to be more on the right and below the line. Is this true--if so, should we comment on it? --Good point. They only deal with types later than S0a, and as you point out, those are exactly the galaxies causing all the havoc. I've added a note about this. p.14. section 6., 1st paragraph. Note that if dust is heated mainly by UV or by recent star formation, there must be som unresolved IR for every galaxy, since the natal cloud will absorb some fraction of the luminosity of the young stars. End of that paragraph. Since highly variable in H alpha to infrared, I do not see why they should show "low infared to ultraviolet ratios". Seems like the ratios will be variable too. --I've added 'global' to 'low IR/UV' to make clear my intention. Figures 1-8. Not sure why this is not just Figure 1 (continued). --I just find this approach easier to format in LaTeX. On figure 10, y axis nu fnu Is this in log? --No on figure 14. A scale in angle and linear size would be helpful. --Done In figure 17. SO galaxies are all over the place. i wonder why? --Maybe because they are so diverse: LINERs, starbursts, quiescent, ...