sec 1 par 2: change obtain secure useful -> obtain reliable few challenging -> few partially-transparent *** Done sec 1 par 2: change that conveniently bridge -> that bridge *** Done sec 1 par 4: change Milky Way and nearby galaxies. -> Milky Way. Weingartner & Draing 2001 did use the LMC and SMC extinction law to produce dust models for those two galaxies, but the DL07 models that we are using here were based entirely on observations of the nearby Milky Way (including submm observations by COBE-FIRAS; more on this below). *** Done sec 3.1 par 1. Do you want to define \epsilon_{\rm cal}? Perhaps it is clear enough from the usage, but you might want to change \epsilon_{\rm cal} to \epsilon_{{\rm cal},\nu}. *** Done sec 3.1 par 3: change fits -> FITS is one fourth -> is $\sim$ one-fourth *** Done sec 3.2 par 1: change differ from -> are less than *** Done sec 3.2 par 2: This paragraph states that 6 galaxies were observed by the Herschel Reference Survey. Should state here whether the observing procedure was the same as, or different from, that just described for the KINGFISH observations. *** I checked into this, and HRS used three scans for these six galaxies (versus either two or four for KINGFISH observations). I summarize this now in the text. sec 3.2 par 3: The text states that The raw KINGFISH SPIRE data... Is this meant to imply that the raw SPIRE data from the HRS were processed differently? Should make it clear whether the processing was the same or different. *** Good catch. No, that was not my intent. I removed "KINGFISH" to avoid confusion. sec 3.2 par 3: change by multiplying -> by dividing *** Thanks sec 3.3 par 1: change In addition, the sum of the sky -> The sum of the sky *** Done sec 3.4 Table 1: You have integers from -2 to 90 in the "Optical Morphology" column of Table 1. What do these numbers signify? You list "2a" and "2b" in Table 1, but never define what a and b are (I presume 2a and 2b are the extent of the ellipse along the major and minor axes). I think it would be good to add a column with D_25. I suppose that it is considered to be obvious, but I think the table note should be changed from The position angle is... -> The position angle of the major axis is... Given that the position angle measured E of N should vary from 0 to 180 deg, why are some of the position angles larger than 180? (e.g., NGC1316, with PA=230. Shouldn't this just be 230-180=50?). *** All these have been fixed sec 3.4 par 1: You use elliptical apertures, and then speak of the "aperture diameter" in the text. Do you mean \sqrt(4ab) ? If so, to make this clear perhaps change to read The average ratio of aperture diameter to the de Vaucouleurs D_{25} optical diameter is \sqrt(4ab)/D_25 = 1.45 What is the meaning of \sigma in "1.45 (sigma=0.45)" ? *** This has been fixed; I meant "major axis". And I've explained in the text that sigma corresponds to the 1 sigma dispersion in the ratio. sec 3.4 par 3: The method used to compute the aperture correction is only correct if the galaxy profile in the FIR is the same as at 3.6um. Since 3.6um is starlight, and the FIR is emission from dust, it is likely that there will be appreciable differences in the emission profiles. Perhaps this should be mentioned here. *** Done sec 4.1 par 3 and Fig. 1: I think it would be better if Fig. 1 showed ratios of the **actual** values of F_\nu reported by MIPS and PACS, with no correction. Then, you could put a dotted line at 1.06 for the 70/70 plot, and 1.015 for the 160/160 plot, as the estimate of what might be expected for some representative dust emission spectrum. *** Done. Note that due to Marc Sauvage's presentation in Paris, I have now adopted PACS extended source photometry correction factors of order 10%. sec 4.1 Fig. 1: It might be interesting to label a few of the more extreme outlier points. *** I'm having a student carefully look at differences in what Gonzalo and I are doing in terms of global photometry. I suspect based on this analysis we will both change our photometry a bit here or there. I'll wait until things are firmed up before I consider labeling outliers. sec 4.1 par 3: You asked where what the 1.06 and 1.015 mean. The way I estimated these ratios was to take the MIPS and PACS response functions, and the way they claim to be calibrated, then calculate for a dust model what each of these instruments should report, and take the ratio. I just looked again at our model predictions. Here are two examples PACS70/MIPS70 PACS160/MIPS160 ------------- --------------- q_PAH=0.060, U=1 (delta func): 1.086 0.983 q_PAH=0.060, U=0.7 (delta func): 1.079 0.986 q_PAH=0.060, U= 1 - 1e6 (alpha=2): 1.093 0.980 q_PAH=0.060, U=0.7 - 1e6 (alpha=2): 1.093 0.981 The values 1.06 and 1.015 in the "color effects" report that I prepared in Jun 2010 (attached here) were intended to be representative values, but that report showed that the ratios varied quite a bit as a function of starlight intensity U. *** Thanks sec 4.1 par 4: I would add text: ...This latter issue will be revisited in \S 4.6. -> ...This latter issue will be revisited in \S 4.6. The right-hand panel of Fig. 2 shows clearly that the galaxy SEDs do not form a simple one-parameter sequence of "cooler" to "hotter" dust. Galaxies are more complicated, with mixes of dust temperatures that vary from one galaxy to another. *** Done sec 4.1 Fig. 2: I suggest using larger points. *** Done sec 4.3 par 1: I agree with Maud that it would be helpful to remind the reader of how the DL07 model got its submm opacity. Here is some possible text to follow at the end of par 1: The DL07 dust models use the far-infrared and submm opacities for graphite and amorphous silicate from Li & Draine (2001). Li & Draine (2001) used the graphite opacity from Draine & Lee (1984), but made small modifications to the amorphous silicate opacity. The imaginary part of the amorphous silicate dielectric function $\epsilon_2(\lambda)$ was adjusted in order for the model to better match the average high Galactic latitude dust emission spectrum measured by COBE-FIRAS (Wright et al 1991; Reach et al. 1995; Finkbeiner, Davis, & Schlegel 1999). The adjustments were relatively modest: $\epsilon_2(\lambda)$ was unchanged for $\lambda < 250\micron$, and modified by less than 12\% for $250 < \lambda < 1100\micron$. With this dielectric function for the amorphous silicate component, the DL07 model gives generally good agreement with the observed submm emission from the Milky Way diffuse ISM. Thus the DL07 model has in effect been "tuned" to reproduce the diffuse emission from the local Milky Way. The dust model used here will be referred to as DL07, but in fact has two small changes: (1) there have been some small changes in some of the PAH band parameters, and (2) the graphite dielectric function has been modified to broaden out an opacity peak near 30um. These changes are described in Aniano et al (2011). *** Done. Thanks. sec 4.3 par 3: The restriction to U_min = 0.7 was reasonable when there was no data longward of 160um, but when SPIRE data is available to constrain the dust model, it is appropriate to allow fits with smaller values of U_min. In fact, when we include SPIRE500 data, Gonzalo and I allow U_min to go as low as 0.01, although the best fits never go this low. I haven't surveyed all the pixels of all the galaxies, but, for example, at MIPS160 resolution: galaxy lowest U_min NGC0628 0.4 NGC6946 0.15 (just a few pixels out of 1353 pixels) So I would suggest allowing smaller values of U_min . I think that 0.01 would be reasonable. I expect that in some cases the best-fit U_min value will fall below your current limit of 0.7, and the estimated dust mass will correspondingly increase. Incidentally, I'm not sure what version of the DL07 models you are using. There have been a few small tweaks to the model since the data that I posted on my website. There were a few small glitches because every once in a while my code for finding the temperature distribution function for small grains had a numerical failure that I had not previously noticed (Caroline Bot asked about some anomalies in the SEDs, which I tracked down to this problem). Anyhow, I've fixed every problem that I know about, recalculated everything, and extended to lower values of Umin for completeness, and extended to higher values of q_PAH to allow a higher "ceiling", and have calculated SEDs for a uniform grid of q_PAH values, running from 0.000 to 0.120 in steps of 0.010 Also, I now use a finer grid in U values for calculating the dust temperature distributions and emission spectra. If you want the new "library", it can be downloaded from ftp:ftp.astro.princeton.edu/draine/dale/spec20.tgz (spec20.tgz is a gzipped tarfile; it is fairly large, 923 MB). If you would like me to repeat the DL07 model-fitting using your photometry, I can easily do this if you would like; you would just need to put the photometry into the format that my fitting code wants to read. *** Thanks! I used your latest models, and allowed U_min to reach 0.01. I also interpolated such that qPAH varies by 0.1%. sec 4.3 par 4: The notation in eq. (5) is not good: f_nu on the LHS is a flux density, but the f_nu on the RHS are emissivity per unit mass. I suggest the following change in notation: f_\nu^{\rm diffuse}(j_M,U_{\rm min}) -> p_\nu^{(0)}(q_{\rm PAH},U_{\rm min}) f_\nu^{\rm PDR}(j_M,U_{\rm min},\alpha) -> p_\nu(q_{\rm PAH},U_{\rm min},U_{\rm max},\alpha) with the following explanatory text $p_\nu^{(0)}(q_{\rm PAH},U)$ and $p_\nu(q_{\rm PAH},U_{\rm min},U_{\rm max},\alpha)$ are, respectively, the emitted power per unit frequency per unit dust mass for dust heated by a single starlight intensity $U$, and dust heated by a power-law distribution of starlight intensities $dM/dU \prop U^{-\alpha}$ extending from $U_{\rm min}$ to $U_{\rm max}$. *** Done sec 4.3 par 4: change subtended by the starlight. -> subtended by stellar photospheres. *** Done sec 4.3 par 4: gamma and (1-gamma) are not dust *heating* fractions, they are dust *mass* fractions. Change $\gamma$ and $(1-\gamma)$ are the fractions of the dust heating that arise from photo-dissociation regions where $U>100$ and... -> $\gamma$ and $(1-\gamma)$ are the fractions of the dust mass heated by the "power-law" and "delta-function" starlight distributions, respectively. The $U=U_{\rm min}$ component is thought of as the dust in the general diffuse interstellar medium. The "power-law" starlight distribution allows for dust heated by more intense starlight, such as in the intense photodissociation regions (PDRs) in star-forming regions. For simplicity, emission from dust heated by $U>U_{\rm min}$ will be referred to as the "PDR" component, and the emission from dust heated by $U=U_{\rm min}$ will be referred to as the "diffuse ISM" component. *** Done sec 4.3 par 4: I think I would delete the sentence In this context the... $f_\nu^{\rm diffuse} *** Done sec 4.3 par 4: change to account for -> to allow for *** Done sec 4.4 Fig. 4: The ordinate labels do not correspond to the plotted quantity. I would change the labels to something like gamma(SST+HSO)/gamma(SST only) and similarly for q_{PAH}, U_{min}, and M_{dust}. *** Done (as close as I could given the space limitations) sec 4.4 par 2: ratios are shown as a function of oxygen abundance from Moustakas et al, but they of course give 2 abundances. I suggest using the PT05 abundances (which I consider to be closer to reality). In any case, the text should say which abundances are used. Maybe the KINGFISH paper will use only one abundance (that's what I voted for) in which case that would be the abundance scale to use (it would probably then be appropriate to cite both Kennicutt et al 2011 and Moustakas et al 2010). *** There actually was a statement that the metallicities were on the PT05 abundance scale. However, I have reinforced this by adding this to the figure caption as well. sec 4.4 Fig. 5: The caption should state what oxygen abundance scale is used. *** Done sec 4.5 eq. (8): This equation is not correct. The RHS should read \frac{f_\nu D^2}{\kappa(\nu_0)} \left[x B_\nu(T_1)*(\nu/\nu_0)^{\beta_1}+ (1-x)B_\nu(T_2)*(\nu/\nu_0)^{\beta_2} \right]^{-1} Are you in fact planning to make fits with eq. (8)? If not, then it could perhaps be omitted. Note that the adopted notation is potentially misleading. M_{dust_2} is the total dust mass, but T_{d_2} is the temperature of *one* of the two dust components. If this equation is to be retained, I suggest simply omitting the subscript on M_{dust} in eq. (7) and eq. (8). *** I fixed the equation and the notation. sec 4.5 par 2: You give 110um as a lower cutoff for the photometry used to constrain the blackbody. I think we are generally using 100um as the nominal wavelength for the middle PACS camera. *** Fixed sec 4.5 Fig. 6: I would prefer to see "Draine & Li 2007" changed to just DL07 in the ordinate label. The caption also gives the middle PACS band as 110um; I think it should be 100um. *** Done sec 4.5 Fig. 7: I suggest changing "Draine Li (2007)" in the legend to "DL07 model". *** Done sec 4.5 Fig. 7: The caption refers to 110-500um photometry, which should be 100-500um. *** Done sec 5 par 2: change including the fraction $\gamma$ that derives from photo-dissociation regions and the complementary fraction $1-\gamma$ that comes from the diffuse interstellar medium. -> including the fraction $\gamma$ of the dust mass that is located in regions with $U>U_{\rm min}$, and the complementary fraction $1-\gamma$ that is located in the general diffuse interstellar medium. *** Done sec 5 par 2: gamma is a mass fraction, not the fraction of the dust heating: change representing the fraction of the dust heating in intense fields where $U>100$, is $8 \pm 1$\% smaller when Herschel data are included in the fits. -> representing the fraction of the dust mass located in regions with $U>U_{\rm min}$, is $8 \pm 1$\% smaller when Herschel data are included in the fits. For $\alpha=2$, the fraction $f_{\rm PDR}$, which we define to be the fraction of the total dust luminosity contributed by regions with $U>100$ is given by eq. (29) of DL07: \begin{equation} f(L_{\rm dust},U>10^2)= \frac{\gamma \ln(U_{\rm max}/10^2)} {(1-\gamma)(1-U_{\rm min}/U_{\rm max})+ \gamma\ln(U_{\rm max}/U_{\rm min})} \end{equation} *** Done, plus I added in a quantification of how f_PDR changes with and without Herschel data (which is what I assumed you meant to suggest). sec 5 par 4: change more sophisticated Draine \& Li (2007) curves to fully represent the dust emission spectrum from the near-infrared through the sub-millimeter. -> Draine \& Li (2007) dust model to represent the dust emission spectrum from the near-infrared through the sub-millimeter, with (for a given value of $q_{\rm PAH}$) a single dust opacity function $\kappa_\nu$, but allowing for a distribution of starlight heating intensities and resulting dust temperature distributions. *** Done