Abstract - I think that the Abstract needs 11HUGS and ANGST in it. I realize that it is a bit awkward to define acronyms in an abstract, but the proposal was based on those two surveys and needs to be connected to them by the various automatic and non-automatic (astro-ph readers) readers. The first few sentences in the Sample section say things pretty succinctly, so maybe a modified form of that? --I've reworked the abstract. Global - the references to in prep works often have these large spaces in them E.g., Lee at al. 2009b). using a \ after the al. will let latex know that this is an abbreviation, not an end of a sentence. --Thanks for the neat tip. last paragraph in the intro in most cases you can replace the word Section with \S --I usually do that, so thanks for the reminder. Section 2, first para The bounds represent the ranges beyond which -> Beyond these bounds --Thanks Section 2, fourth para dramatically differ -> differ dramatically (I think) --I've actually read up a bit on this mini controversy. It seems there is no consensus among academics. But I'll give you this one. :) I find the two "branches" in the L(TIR) vs. L(TIR)/L(B) diagram pretty amazing. Would a larger sample find galaxies that lie between the two branches, so that galaxies with large L(TIR) at a particular L(TIR)/L(B) are just outliers on a broad distribution and the selection bias of the SINGS sample just populated that upper branch out of proportion? Regardless, those large L(TIR) are special. I didn't find a similar figure in Dale et al. (2007), so I guess you did not discuss it there? --I was surprised, too, when I made that figure. For the SINGS proposal we made similar plots (Fig 5 Kennicutt et al. 2003), and the strange bi-model SINGS distribution can also be seen there. It must be an artifact of SINGS sampling many bright (famous) galaxies, plus a bunch of really faint targets to sample more of parameter space. To test whether SINGS is indeed just sampling the upper and lower envelopes, I made a similar IR vs IR/opt plot of the IRAS 1.2 Jy survey. The database is vague about what exactly is "opt", but the distribution fills in between the upper and lower envelopes, as expected. So I think this diagram confirms my joking and unfair description of SINGS: a bunch of really bright and famous spirals, plus some ellipticals and dwarfs. ..... Now that I look again at the LVL and SINGS and IRAS 1.2 Jy surveys in IR vs IR/opt, here's an alternative take -- the volume-limited, nearby sample is picking up many faint dwarfs, dwarfs that IRAS essentially had no chance of detecting except for the closest examples. --> With Spitzer sensitivity and the proximity of the LVL sample, LVL preferentially samples the lower envelope (and SINGS still samples the upper envelope). I've added some thoughts along these lines to the text. Section 3.1 second para BCD -> basic calibrated data --Thanks section 3.3 utilize -> use --OK section 4.2.1 1st para Why would foreground stars or background galaxies be H-alpha rich? The background galaxies are not likely to show emission in a rest wavelength H-alpha filter. --Thanks for pointing that out. What I meant was that if something shows up in H-alpha, then we know to keep it, to *not* remove it like a background or foreground object. I've fixed the wording. section 4.2.1 2nd para insert space between 160 \mum and (Engelbracht --Thanks section 5.3, 1st para alpha_SED is introduced in a single sentence with a reference to Dale and Helou. Perhaps a bit longer intro with typical range of expected values would help the reader to better understand the meaning of alpha_SED? --You're right, and I've fixed that. [ I always feel a bit awkward describing details of my own work, as if I'm blowing my own horn too much. ] But you're right. section 5.4 3rd para To the references starting with Engelbracht et al. 2005 discussing the PAH/vsg - metallicity correlation, I would like to add Jackson et al. 2006, not only because it is part of the debate, but also because the spatially resolved nature of the nearby galaxies will allow more progress on this issue (as indicated in the next sentence). \bibitem[Jackson et al.(2006)]{2006ApJ...646..192J} Jackson, D.~C., Cannon, J.~M., Skillman, E.~D., Lee, H., Gehrz, R.~D., Woodward, C.~E., \& Polomski, E.\ 2006, \apj, 646, 192 --Thanks section 5.5 2nd para the term birthrate parameter is introduced in quotation marks. I think it would be more useful to introduce it with a definition as the ratio of the current star formation rate to the past average star formation rate with a reference to something Kennicuttian (probably Kennicutt, Tamblyn & Congdon 1994). --Done (you weren't the only one to suggest this!) section 5.5, 3rd para according to the age inferred from the theoretical ... I think that something more like "mean age" or better "the age of a continuously star forming system" more accurately describes the number on the RHS of fig 13 --Thanks Figure 6 In the lower right hand panel, in the -11 to -11.5 bin, the detection rate is about 8%. The number of galaxies in that bin is 7. I think that this is equal to a half of a detection. Is that correct? What does that mean? --Very perceptive. It's because I'm averaging the detection rates at 5.8 and 8.0um for that particular histogram, and the detection rate for that bin is 0/7 at 5.8um and 1/7 at 8.0um. I've improved the figure caption to explain this. Figures 8 and 9 The images labeled "H-alpha" are H-alpha + continuum and not continuum subtracted. One could argue for including either (or both), but the better correspondence to the 3.6 micron image as opposed to the 24 micron image means that the R-band light is dominating the image and not the H-alpha emission. Maybe a 3 x 3 with an R-band, a continuum subtracted H-alpha, and a 2MASS J or H image (it would nice to have a 2MASS image in there for comparison since we compile 2MASS data)? --I replaced the Ha+continuum with just Ha (something Liese suggested). Figure 12 Adding labels of "starbursts" and "normal galaxies" to the two curves would make this figure much more immediately accessible --Good point Figure 13 Wouldn't it be possible to do a similar RHS labeling for the bottom half of the figure? --Yes, I put some time into that and ultimately was unsatisfied. The "problem" is that the two types of birthrate parameters give different timescales for the same model. I could make the timescales look similar if I chose different models for the two panels, but that is also unsatifsying. Since my goals are more conceptual than quantitative, I'd rather leave off the conflicting interpretations. Clearly there is a problem confronting the models with the data, but that's a battle best left for another study.