* I moved back to France, could you add to my affiliation my current institute ? : Laboratoire d'Astrophysique de Marseille Traverse du Syphon - Les trois lucs BP8 13376 Marseille cedex 12, France --Done * Introduction : "with normal galaxies systematically exhibiting redder slopes." I am not sure about the "systematically" used here and elsewhere in the description of the IRX-beta relationship. May be you should you say "redder slope for the same infrared-to-ultraviolet ratio": you could also say that they have the same colors (except for red early-type), but lower infrared-to-ultraviolet ratio. I'll come back to the subject of the IRX-beta thing below. --I think you mean to quote 'significantly' instead of 'systematically' I've reworded the text a bit here. * Introduction : "... of a large, diverse sample of nearby galaxies,..." I have the feeling that you could say in the introduction how the SINGS sample was designed to reproduce the variety of morphology, luminosity, etc... This information is in the section 2, but section 2 is so short that I would move the crucial points on the sample in the introduction, and whatever is left in section 2 could be easily merged in the current section 3. --My inclination is to keep Section 2. * Section 3 : about the extended source aperture corrections. I must say that I found this part a bit unclear : how were determined A,B,C for equation 1 ? What is f_{measured} exactly ? My understanding from later in the text is that fluxes were measured within an aperture which is given in table 1. This is actually said in section 3.2 concerning optical data : "global optical fluxes are extracted using the same aperture used for the IRAC and MIPS global flux extraction". If this is true, then this aperture should be mentioned in the beginning of section 3 speaking of IRAC fluxes. But then, why an "aperture" correction is needed ? I may be asking something stupid, but I was a bit confused. --I've moved the optical data, subsection to appear before the IR data subsection, so the reader gets the aperture information earlier. Your questions are legitimate, and I hope the interested reader will go to Tom Jarrett's website that I footnote in this subsection. That website says: "These corrections not only account for the "extended" emission from the PSF itself, but also from the scattering of the diffuse emisssion across the IRAC focal plane." I have added this to the text. * Section 4.1: about the fitting. You start by saying that the solid curve is the sum of a dust and stellar model. Then you say that the dust is fitted by MIPS, and the stellar curve is fitted to the 2MASS data. It is a bit unclear which is fitted by what, because I see in the plots that at least for some galaxies, the sum is fitting very well the 2mass points, with a contribution from both dust and star light. Thus, my guess would be that you 1) Fitted the dust with MIPS 2) Added the stellar light so that the star + dust model fits well the 2MASS data. --I fit dust curve to MIPS only, and the stellar curve to 2MASS only. I agree that sometimes the stellar+dust matches well the 2MASS data, but that only means the dust contribution was negligible in such cases. I don't see any examples of where the stellar curve falls below the 2MASS data, a situation that would imply that I fit stellar+dust. About the stellar curve, it seems that you fixed the age, metalliciy, IMF, thus the only parameter you are fitting is only the amount of mass you put in it ? Would it be terribly hard to fit seds for let's say various star formation time-scale and the current age of the universe, instead of a 900 Myr continuous star formation that does not seem to me representative of the old stellar population of galaxies ? --It's a very reasonable suggestion. But this paper was negotiated by the SINGS and GALEX teams to be a data paper, with a subsequent paper to focus on fitting stellar population synthesis models, etc. * Section 4.2 : "note also the broad spread in the ultraviolet data compared to that in the infrared". Well, since in this plot, everything is normalized in K, the impression we get could be misleading: if normalizing in u, we would not see as much variation in the uv. What can be concluded from this plot is that there is a larger spread in UV light per unit stellar mass versus infrared light per unit stellar mass. It might also be true in an absolute sense, but we cannot see it in the figure: we would need some sort of absolute scale to say that. --Good point. I have altered the text. * Section 4.3: e1 is indicative of a galaxy with a low infrared to uv, and e2 represents a high infrared to ultraviolet... Could we give something a bit more interpreted : may be comparing them to real spectra to give an idea of how much they are similar to real galaxies, are e1 and e2 "extreme" cases (very dust free, very dusty), or more averages of a class ? Thinks like that. I am not very satisfied with just eigenvectors which have not much physical sense. --I have rewritten this sectin in an attempt to have more physical interpretation. * Section 5.1: Inclination. I was a bit surprised by the result as it sounds very strange that edge on galaxies would not have more extinction. May be you should stress more that there is very few very inclined (edge-on-ish) galaxies. Within the range of inclination where most of the SINGS galaxies are, the model from Tuffs 04 actually predict a mild increase. I agree that there is a lot of dispersion so that other effects are more important in deriving extinction for this mix of galaxies. However, if you look for some types (e.g. if you look only from Sa to Sbc, you may have something after all... --Veronique and Tom Jarrett were happy to see the result, but you and Chad Engelbracht were surprised. I agree with you that there may be a trend, but it should only become obvious for nearly egde-on systems. * Section 5.2 About the IR/UV vs morphological type : Armando's atlas fig 5-f shows a similar plot (with small difference in the wavelength used) that is may-be worth looking at and mentioning. --Done * Section 5.3 This section presents an interesting case, I was just wondering if you checked that resolution effects do not affect your conclusion. I mean: if the dwarfs/irregulars are systematically closers than the other ones, you may see more details, and then they may appear clumpy. If on the other hand, big spirals are more distant on average, you may see only bigger structures, where things could be less clumpy. I guess just checking the distances of dwarfs/irregulars vs spirals would be enough to be sure... --I've added distance to the current Fig 16 - it shows no distance dependence. * Section 5.4 "... and the later the spiral Hubble type the larger the specific star formation rate and the smaller the infrared-to-ultraviolet ratio." I have the feeling that this contains two facts: - The specific star formation rate increases with Hubble type", what has been thought for a long time: for instance, it is a consequence of the sketches of the star formation histories given in Sandage 1986 A&A,161,89, I guess you could give this reference. You can also see Boselli et al., 2000 for some plot of the current to past average star formation rate b (similar to specific star formation rate) - You have shown in previous plots that the uv/infrared ratio depends on the Hubble type. Thus I think you could plot in fig 17 another panel showing the SSFR vs type. It will be clear that the three quantities: type, uv/infrared, and SSFR are linked together. Now, I don't know if you can say that one of the quantity depends on another one : my feeling is that they all derive from the evolution of the galaxies, may be decided by a combination of their mass/ environment. --Thanks. I have added a reference to that nice Sandage paper. I think I prefer to not add another figure displaying T vs SSFR, since that information is built into the current SSFR plot. * Section 5.5. As I said before, I have the feeling that saying that the relation is more scattered and shifted to the red for normal galaxies than starbursts is a bit of an oversimplification. I would suggest to give more details on the various irx-beta that we observe in this paper and others, what may help the reader to understand where the various type of IRX-beta relations happens, and why they may be different. I give below a few indications of what I have in mind that you can use if you want for your paper: - uv slope vs uv/infrared in small regions of individual galaxies usually present more scatter (e.g. Calzetti in M51). That may be because there is diffusion effects: the infrared getting out of a region might be dust heated by the UV coming from a distinct nearby region. - Along profiles in spirals (boissier 06): the scatter is small, as you say because we avoid problems with bulges. With respect to study like Calzetti, we have the advantage of azimuthally averaging over large areas, i.e. removing small scales "diffusion" problems - In integrated star-forming galaxies : they usually describe a relatively tight relation with lower infrared/uv than starburst for the same color. This can be seen in your fig 18. Actually if you remove E and S0 galaxies, you will find that the scatter is not so large. My feeling is that it is even smaller than for the starbursts that you show, but you may want to quantify this ! - In Early type galaxies (E, S0), there is much more scatter and the UV color is indeed redder: this is actually what you would expect for quenched star formation rate histories, what is probably correct for this sort of galaxies. --I agree: my discussion was a bit uniformed and cryptic. I have added some more discussion along the lines you suggest above. Thanks a lot. By the way, the Kong et al. data are definitely more scattered than the Calzetti et al. data, which in trun appears quite tighter than SINGS spirals. Do you have any idea why the Kong data is more scattered than Calzetti data? * References: Boissier et al., 2006 is now accepted. You may find it there: http://serweb.oamp.fr/perso/boissier/preprintprofiles/ms.pdf --Thanks