Fig. 2:
a) It is not clear to me from that figure that shows metallicity color coding to deduce that Sc spirals have mainly coolor colors.  Would it be worth to add two more panels with a color coding for morphology type?

*** I prefer to stick with two panels and using symbols to denote morphology.

b) I am wondering of what is the implication of the right hand panel to the recent development in the higher-z community to advocate a single IR SED, e.g. Elbaz et al. 2010. Would the KINGFISH galaxies agree with that proposed global SED within its uncertainties? Are only a sub-class of galaxies (morphological or metallicity selected) that do not agree with such a SED?  Addressing some of these questions might be interesting in the light of the high-z community

*** Excellent point.  Perhaps one interpretation is that the KINGFISH sample doesn't follow the "IR Main Sequence" in its entirety.  Elbaz and I have argued about this generally.  For example, he quotes a tight correlation between 8um and TIR, whereas my SINGS and LVL publications have shown a rather large dispersion in L(8)/L(TIR), around 0.4 dex (e.g., Figure 11 and Table 4 of Dale et al. 2009; we do find similar dispersions in L(24)/L(TIR), however ...).  One reason for our differences, I believe, is that he analyzies higher-redshift surveys, where one is likely to find similar types of galaxies due to their relatively high IR luminosities.  In contrast, SINGS/KINGFISH samples lower L (and lower Z) galaxies.  I'd rather not rehash this argument in the context of this paper, in part because the focus here is on the Herschel data, and not comparisons of L(TIR) using extrapolations from monochromatic Spitzer indicators at 24 and 8um.

Fig. 3:
c) I personally find the presentation of the SED fits too complicated/colorful. Maybe a simple B&W figure using black for all symbols and the final fit and grey-scale for the model components would do. I was trying to relate the symbol colors to the individual model components and got really confused in the first place.