1. p2 top par, line 3: the text "in different portions of phase space" probably is not necessary (the implied meaning is sufficient). --Done 2. p2 top, same par: It might be useful to mention here the basic approach taken by Genzel et al (namely correlating an ionization- sensitive index vs PAH strength). And that our higher sensitivity allows us to not only extend the exploration of this type of diagnostic but also to investigate combinations of fine structure lines alone. --I added the Genzel blurb to the introduction. I did not add that we are the first to investigate combinations of fine structure lines since this was done with ISO data. 3. p3, Section 3.1: I really think the paper would be served well by one or two figures that show example spectra, certainly at least one covering the whole spectral range, or alternatively 2 or 3 examples of the different types we see (HII region, Seyfert, LINER, etc). This will lend reader confidence to the quality of the data we are working with, and illustrate the richness of the spectra. --Done. I've also made a linear-linear version of the plot, but I personally like the log-log version better. 4. pp3-4, Sections 3.1-3.2 and beyond: I found the order of sections here a bit confusing, in that we begin to discuss the character of the spectra before we describe the observations, reductions, etc. Then again I can see how you want to refer to PAHs, forbidden lines, etc, as part of the data sections. One possible solution would be to introduce the two types of features in the introduction (see comment 2 above), then follow with the data sections (now 3.3-3.6), and use the current 3.1-3.2 to open Section 4. Just a thought.... --They've been rearranged. 5. p4, Sec 3.3, middle par: I think the reference you want is Kennicutt et al (2003), not 2005, right? --Yes 6. p4, end of Sec 3.3: One thing that is not described much here is the extraction procedure and what regions the 1D spectra really refer to. This is important, especially given that we work with different spectral resolutions across the wavelength range. --This has been updated - please let me know if it will suffice. 7. same place: Also, shouldn't we also say something about the uncertainties in the relative line fluxes (we now only cover absolute fluxes)? --Done 8. Sec 3.4: A general question that will come up later-- are there any published or astro-ph level IRS team spectra that we should be including in the comparison? Or is this a touchy area? I am only thinking about using published fluxes, not remeasuring spectra. --I have included all relevant Spitzer data published to date (e.g. Weedman et al, Armus et al, Haas et al) 9. p5, Sec 3.6, line 2: The correct reference is Moustakas and Kennicutt (2005, ApJ, submitted). May be in press by the time we finish this draft. --Thanks 10. p6, end of Sec 4.1: In answer to your questions, yes, I think we should mention the extinctions. --Done 11. Fig 1 and end of Sec 4.1: One thing from this figure that impressed me is the clean separation between the HII regions and the AGNs in the optical BPT diagram. I wonder if this is worth mentioning here? It's not as if we have a bunch of transition objects that are difficult to interpret. --I have replaced this BPT diagram with a more appropriate version that uses 20"x20" apertures, similar to our Spitzer extractions. The discussion has changed a bit somewhat, as there are more transition objects. 12. Fig 2 and p6, bottom par: The behavior of [OIV] and [SiII] in these two diagrams seems to be somewhat similar. What happens when you plot the two lines (ratioed to [NeII]) against each other? -- log ( OIV/NeII ) = 2.6[+/-0.4] * log ( SiII/NeII ) - 1.2[+/-0.1] See http://physics.uwyo.edu/~ddale/research/sirtf/lines/DR3/lines.html 13. p7 top: The behavior of SiII reminds me a lot of [OI] in optical spectra of AGNs. I wonder if the explanation is similar-- basically the effect of a power-law ionizing spectrum providing a flatter distribution of nebular ionization in the surrounding clouds? In any case it would be good to discuss our results with someone who has worked on AGN spectra in the optical, because I have a sense that we may be re-inventing some wheels here in our interpretation. -- Armus says that it's long been known that AGN can show copious emission from low-ionization lines, too. 14. p8, middle par, and Fig 5: Danny I don't completely understand this diagram. Isn't the ratio of the two SIII lines a density diagnostic? In which case you expect a roughly linear correlation, with any deviations correlating with electron density? Or am I confusing these with another line? I guess I don't get the reason for this diagram. --I have revised this, directly addressing this as a density diagnostic. 15. Fig 1: We do have optical spectra for some of the HII regions in this sample. Should I look them up so we can add those points to this diagram? I am afraid we probably don't have all (e.g, many of the objects in NGC 6946). -- That would be great! 16. Fig 2: I am curious about why you chose the Sombrero as your prototype AGN for the mixing line. It does have the lowest PAH strength, but its ionization is below most of the other AGNs. Why not use one of the AGNs with higher OIV and SiII instead. Then I suspect that your mixing line goes through more of the points. Likewise you chose for the HII region an object with one of the lowest excitations. Why not one that is closer to the center of the distribution on the right? --I now simply use values typical of the ends of distribution, not individual galaxies as anchors. 17. Fig 3: I guess this repeats a comment, but I am curious whether OIV/NeII shows the same behavior as NeIII/NeII-- or when you replace SIV/SIII with the same index? In other words is there a continuous shift as you to up the ionization energy ladder? --There is a strong trend, and a real nice separation between AGN and star-forming. See http://physics.uwyo.edu/~ddale/research/sirtf/lines/DR3/lines.html 18. Fig 4: Maybe a big point here. There clearly is a separation of the AGNs from the HII regions and even the HII nuclei. But the diagram also shows that the HII nuclei are intermediate between the AGNs and the disk HII regions. I wonder why? Could it be because of higher metallicity or is there some buried AGN component even in the HII nuclei? Worth thinking about and discussing in any case. -- Good comment. In fact, this AGN->HII nuc->extranuc trend is a main point of at least 3 Figures in this paper. I have added more discussion centered on metallicity differences and the possibility of weak AGN contribution to HII nuclei. 19. Same fig: I guess one thing I wonder about is whether just for the HII regions do any of these indices correlate with abundance? A quick way to check is to plot them against the optical R23 ratio, for the ones that have R23 values. Have you compiled these? If not I'll try to look them up and make a plot. This probably is well beyond the scope of this paper, but if possible I would like to have a better intuitive feel for whether there is a sequence just among the HII regions, following abundance or whatever. -- I have not compiled metallicities for the HII regions. So far I just have what Moustakas and you have provided for the SINGS nuclei. 20. Fig 5: I will try plotting something like NeIII/24 vs NeII/24 or SiII/24, to try to reproduce an analog to OIII/Hbeta vs NII/Halpha. 21. Figure 2: Is it possible to overplot the Genzel et al relation (I think they fitted a curve) to our data? I realize it involves a conversion of PAH indices but since you remeasured a bunch of their galaxies this might be possible. --Done ------------------------------------------------------------------- 1. Author list: Please list me with dual affiliations, Cambridge first and Steward second. The relevant IoA information: Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge CB3 0HA, United Kingdom --Done 2. Abstract: This really is a matter of taste, but I would have been a bit more specific in the latter part of the abstract, mentioning a couple of points specifically: - that our results confirm the efficacy of the combination of a high ionization line index vs PAH strength as a useful AGN vs star formation discriminant, following Genzel and others - that we find the ratio of [SIII]33.5/[SiII]34.8 is another excellent AGN vs star formation discriminant I guess I feel especially strong about the first point, because as it stands the abstract gives the impression we are investigating this area for the first time. And if you add the first point you probably want to add the second for balance. --Done 3. Abstract again: As I mention later, I actually find our result of modest densities in the nuclei to be quite interesting, and as such it might be worth mentioning in the abstract as well. But check below first to see whether you agree. --Done 4. p10 very bottom and top of p11: This sentence really does not explain why SIII/SiII works, instead it almost restates the same point twice. But then again I don't know if you want to write the few sentences on why hard ionizing sources give rise to extended PDR regions. At a minimum I would replace "X-ray dominated" with "X-ray ionized", and remove "dense" later in the sentence (after all our own measurements show that these regions are not very dense). You could also add a phrase to the effect "in analogy to the increased strength of [OI]6300 emission in AGNs (reference BPT or Veilleux and Osterbrock). --Done 5. pp12-13, end of Sec 5, or last science paragraph of Sec 6: Maybe this is not a surprise to anyone, but one interesting implication of Figure 6 is that the characteristic densities we infer from the infrared lines are comparable to those one would derive from optical spectra (e.g., [OII] or [SII] doublets). This rules out any model in which there are dense clumps of highly obscured gas and dust residing in the emitting regions, especially in the circumnuclear environments of the galaxies. As you say in the last par in Sec 6 perhaps this is not too surprising, but I wonder if we want to make this point more directly?? Might be worth running by Lee or someone more familiar with nuclei. --Text along these lines added to the abstract.