You mention that your aperture measurements will not take into account the (likely) real differences between the stellar/NIR and FIR profiles within the context of the aperture corrections you formulate based on 3.6 micron imaging. But I think it would be good to raise this point elsewhere in the text as well (since this is always a concern when measuring fluxes in apertures, as opposed to with `curve of growth' techniques). Relatedly, I'm curious about how different the 500 micron scalelengths are (if at all) from those in the shorter FIR bands in the galaxies with excess 500 micron emission. If the disk scalelength of the 'coldest dust' is much smaller than in the other wavebands, this could superficially elevate the longer wavelength fluxes relative to the those in bluer bands (if these all have more comparable scalelengths, say). Do the 500 micron-excess galaxies have much shorter disks at 500 microns? I've looked quickly at NGC 5408, and there's hardly anything at 500 mu m beyond the central concentration, nothing extended in the 350/500 ratio map. I'd be interested to know what you think. *** These systems are fainty and patchy, and there doesn't seem to be any difference in the 500um disk scale lengths in these systems. But to do an accurate estimate you'd need to smooth everything to 500um resolution, and analyze the elliptical isophotes as a function of radius (which is what Leslie is doing). Also, do these galaxies show any change when alpha or Umax (the only parameters in the Draine & Li models you've kept fixed) are allowed to vary? It might not be unreasonable that the power law of starlight intensities is different if (for whatever reason) the SF regions in the shallower potentials/or at the lowest metallicities are different. I don't have a sense for how little these affect the fits here (or in Draine et al. 2007), or whether or not they could even have an influence on how the SED looks at the one longest wavelength, so again, I'd be curious to hear your thoughts. *** Allowing Umax to vary makes very little difference (e.g., < 10% difference in the dust mass). Allowing alpha to vary is not something I have set up to do with my code, but experience with my own SED models makes me think it would be similar to allowing Umin to vary. Also, even though they'll be in Aniano et al. 2011 it would be really nice to add a table of the values from the Draine model fits (although I can see this might mean discussion outside the current scope of the paper). *** We've negotiated to have such a table in Aniano et al.