version 3.1:

introduction:
End of page 2 and of 2nd par: the sentence ends in a strange way: ...
from recent (versus current) star formation, and (references). Is that
correct?

--Thanks for catching that!

3rd par: you say that you want to explore how TIR/UV varies with
morphology, color (i imagine "dust color" since you never consider
optical or near-IR colors) and geometry: why not luminosity, which seems
to be the most important parameter characterizing late-type galaxies?
The now "a la mode" downsizing-effect, which is nothing else than a
strong relation between galaxies properties and their total mass (or H
band luminosity) could be easily tested in your sample. Just to remind
you, we have shown in a series of papers that most of the properties of
late-type (but also several of early type) galaxies strongly correlates
with luminosity (see for instance Gavazzi et al 1996 A&A, 312, 397;
Boselli et al 2001, AJ, 121, 753; if you want i can send you a complete
list of papers): star formation history, molecular and atomic gas
properties, structural properties, metallicity etc; a check between the
TIR/UV flux ratio vs luminosity has been presented in and extinction
Cortese et al 2006. It would be easy for you to test also whether your
results depends on a more quantitative parameter than morphological type
such as mass/luminosity.

--I have added a plot of TIR/FUV vs L_H, along with an entire new
  section that discusses this.  Thanks for your help on that.

The data:
I do not understand why you want to present the radio data (20 cm) while
you never use them in your analysis. If it is just for completeness, you
might also find data at 6 and 50 cm; in my older work on UV to radio SED
of Virgo cluster galaxies (Boselli et al 2003, A&A, 402, 37, that you
might want to read for comparison or for some ideas), i presented such
data. They allowed me to estimate the radio spectral slope for instance.
After reading your manuscript i did not understand why you want to
present them if you do not use them.   
On the other hand i see that several times you use IRAS and ISO data,
but you never give reference to these data. This might be important
since sometimes i noticed an important discrepancy between the ISO and
Spitzer data (see for instance NGC 2915, or NGC 6822). Why these data
are so different? i imagine you discuss it in your previous SINGS paper,
you might just send the reader there...
By the way, i also noticed that you use only ISOCAM data: for the 3
galaxies that we have in common, NGC 4450 (VCC 1110), NGC 4569 (VCC
1690) and NGC 4579 (VCC 1727) you might also find ISOPHOT data in our
paper Boselli et al 2003, A&A, 402, 37. It might be even interesting to
compare their SED.

--I was wondering if anyone would point out that I compile radio data
  but do not utilize them in a figure, for instance.  You're right: I
  include them merely for completeness.  I have a reference to Dale et
  al. (2005) for the ISO and IRAS data.

Sect 4:

4.1; 
I do not understand why you plot as reference 1 gyr continuum star
formation event, that never fits the data (it is in fact a non realistic
scenario for most of these normal galaxies). Since you can correct the
UV to mid-IR stellar SED, why not fitting them with more realistic SFH
such as a delayed exponentially declining star formation rate (Sandage
law)? That is what we have done in Boselli et al 2003, (or Gavazzi et al
2002, ApJ, 576, 135), and the results that we have found are strongly
encouraging. I thus would suggest you to do the same. If not, just for
reference it might be more interesting to put a black body at 3000K
fitting the cold stellar component (e.g. Boselli et al 1998, A&A, 335,
53)

--The group of us SINGS+GALEX people that formed the "Working Group"
  for this paper definitely considered these possibilities.  But from 
  the start, this paper was mandated to be the inital 'data paper'
  (Paper I) for joint SINGS+GALEX data.  Paper II is to incorporate
  more sophisticated stellar population synthesis modeling, radiative
  transfer, etc.  Moreover, I agree that a 1 Gyr continuous SF model
  isn't ideal for several SINGS systems, but it is reasonable for many,
  and when it is reasonable the discrepancy between the model and the
  UV data gives a visual feel for the amount of extinction.

2 of the galaxy that you mention with deviating mid data from the fit
are NGC 6822 and 2915, where the ISO data are also deviating from  
the Spitzer data. Is this a problem of the model or of the data?

--Good point.  The ISO and Spitzer data do seem the most discrepant for
  these two galaxies.  I looked at these data before, and I've looked at
  them again.  The 2915 data come from the ISO archives.  Roussel and I
  have both measured their fluxes and get consistent answers.  The 6822
  data come from a careful analysis Deidre Hunter carried out for the
  ISO Key Project.  I don't know why there are the discrepancies.  But
  rather than simply remove the data, I think we shoud leave them in
  ('let the data speak for themselves').

4.2;
first paragraph: we have reconstructed templates of UV to FIR SED
according to different criteria in Boselli et al 1998, A&A, 335, 53
(UV/K) and Boselli et al 2003 (morphological type, luminosity).  

Why do you use a "non-standard" TIR/UV definition? generally people use
for UV only the 1500 A (see for instance the recent GALEX papers of V.
Buat, Boissier et al, Gil de Paz et al or Cortese et al). I would say
that using your definition is quite dangerous since it makes your
results not easily comparable to those obtained by others, in particular
those obtained by the GALEX team. To avoid any problem and
misunderstanding i would suggest you to use the standard definition (the
one given in Cortese et al 2006). 

--There are two figures in the current version of the manuscript that
  can be directly compared to literature results: TIR/FUV vs UV slope
  and TIR/FUV vs H luminosity.  So I use the canonical TIR/UV definition
  for these two figures.  I chose to use both FUV and NUV in the other
  figures since I like the idea of representing more of the total UV
  emission, and because it should to improve the S/N in the ratio.
  
3rd par: .. the spread at UV wavelengths is presumably significantly
affected by variations in dust content: content or dust extinction? i
would say extinction. 
The following sentence: low metallicity systems are known to be
deficient in PAHs emission: we have shown this results earlier on with
ISO data: Boselli et al, 2004, 428,409, you might want to cite this
paper.

--Good point.  Fixed.

sect 5: 

1st par: i would rather say the TIR/UV is mostly related to the
extinction and, to a second order, to the SFH, metal content, geometry
etc

--OK, rearranged.

5.1: using the TIR/UV flux ratio we never found any relation with galaxy
inclination in any galaxy sample: i checked in our Boselli et al 2003
and Cortese et al 2006 paper but i did not find any reference to that
point; i imagine that you can find this result in the old Buat and Xu
papers of the nighties... (check with Veronique Buat).

--Thanks.  I couldn't find a reference to IR/UV vs inclination in any
  Buat paper, but I have asked her in case I missed something.

5.2: 
line 7: ...galaxies is due excess to an excess...: is that correct in
english?

--Thanks for catching that.

Here you should mention all the results of works on TIR/UV vs morphology
of Buat et al....

--I have added Buat & Xu 1996, which includes IR/UV vs T type.

I would also expect here to see something in relation to luminosity....

--A new section dealing with H band luminosity has been added.

Last paragraph: you say that irregulars are less extinguished than
spirals, but what about the extreme point at the top right of fig. 12?
is this M82, which is an extremely extinguished I0?

--Yes, that's M82.  I have added this caveat to the text.

5.3; 
first sent.: once again M82 might be an exception

--See note above.

1st par: is your result (dust temp vs type) consistent with our (Boselli
et al 2003)?

--It is difficulty to discern the curve representing Irrs in your Fig 3,
  but the data in your Table 9 show Irrs having larger 60/100 ratios.
  Reference added.

The definition of SSFR is unclear (and uncorrected) for me: first of all
this relation (5) is not valid for E and S0-S0a, where both the TIR and
the UV contribution is not due to recently formed stars. You will thus
overestimate the SFR in these objects.  In early spirals TIR might be
contaminated by the dust emission heated by older stellar population: 
such is the case in NGC 4569 on which i am working right now.  I think
that you might find a more appropriate TIR+UV SFR calibration adapted
for SINGS like galaxies in the works of V. Buat, see for instance
Hirashita et al 2003, but you could ask Veronique whether they have a
more updated calibration.  I would also specify that L(1500) is 
uncorrected for extinction, although this might be obvious.

--Using the Hirashita "best" SFR equation, I get a plot that is quality
  very similar (the data shift a bit to the left).  So I have added text
  that explains this, and references the Hirashita paper/equation along
  with their parameters for nearby star-forming galaxies.
  I have tweaked the text to indicate that the uv luminosity is not
  corrected for extinction.  I have also noted that this approach is an
  overestimate for early types.

You also mention that the near-IR Ks band luminosity is an indicator of
the stellar mass: we have written a paper on the dynamical mass vs. H
band luminosity, Gavazzi et al 1996, A&A, 312, 397, which is to me the
appropriate reference.

--Reference added.  Thanks.

In the description of Fig 17 i would suggest you to remove the E S0 and
S0a, for which, as explained before, you are not measuring a SSFR.
Although mentioned in sect 5.2, it would be better to well state that
the UV flux of these early type galaxies is not due to star formation,
and as a consequence their TIR/UV ratio is not necessarily a dust
extinction indicator. If your early type galaxies are, as i can imagine,
"normal" early types, their UV emission, in particular that at 1500 A,
is due to evolved stars, and not young stars (UV upturn). This is the
case in most of the nearby ellipticals (see for instance Donas et al
2006 on the GALEX webpage, or Boselli et al 2005, ApJ, 629, L29). The Yi
et al and Rich et al star forming ellipticals, which are in any case a
minority, have been detected at 2300 A, and are present in their samples
just because of the adopted UV selection and of the automatic
morphological classification adopted in the definition of the samples.
We do not have them (or are really unfrequent) in optically selected
samples such as yours. 
I insist on this point since i have the impression that you always treat
E and S0-S0a along the paper as spirals, where the UV and dust emission
is due to young stellar populations; here it is not the case.

--I have added a couple of caveats related to the fact that the SSFR is
  likely overestimated for SINGS early-type galaxies.  NED lists several
  of our early types as having HII or starburst nuclei (N1377, N1705,
  N5866), though.

table 1: i do not see NGC 3034, which is although present in Table 2.
Why?

--I have added it to be complete.  I originally excluded it from this
  table since the Spitzer photometry has been rendered unusable due to
  saturation.

table 2 and 3: flux densities are all given in Jy, except NUV and FUV:
why? i would put everything in the same units (mJy?).

--I used mJy since, e.g., N1266 FUV would be 0.00005+/-0.000007 Jy.

table 4: the 2nd legend note for table 4 is not the same at page 25 and
26; to avoid confusion i would put "note" only once at the end of the
table

--I have restructured the tables.

fig: whenever you discuss TIR/UV as a dust extinction indicator, i would
remove E and S0, or specify it in the legend.

fig 17: legend: where is the 30% error in the Ks vs mass relation taken
from? 

--From the Bell et al. (2003) article, Fig 18.

fig 18: if you are interested in, i might ask Luca Cortese to send you
the data we have for our star forming sample to include in this figure,
although i understand it already is quite dense

--Thanks, but yeah, it's already too dense.