ABSTRACT:

I would probably be more explicit regarding what we mean by "24um
morphology". Perhaps you can just write "... shows that the 24\,$\mu$m
morphology (smooth, clumpy, or point-like) can be a useful ..."

--Done

1 INTRODUCTION:

2nd paragraph: It looks great. I think it would be good to emphasize
in the second paragraph that the larger scatter in the IRX-beta
obtained when using normal star-forming galaxies is due in part to the
fact that integrated properties are used. Indeed, a recent paper
submitted by Boissier et al. (2006) shows that this scatter is reduced
when azymuthally-averaged radial profiles data (where the contribution
from the bulge or an AGN can be identified an excluded) are used.

--Done

3rd paragraph: I would remove the sentence "Below we present global
..." and replace it by a new paragraph describing the content of the
following sections: "Section X presents the sample. The analysis of
the images is described in Section Y, etc..." or something like that.

--Done

2 THE SAMPLE:

Instead of "and star-forming nuclei" I would write "and starburst
(SBN) nuclei". There are two "span a wide range" in the same sentence,
you can perhaps replace the latter by "and adequately sample other
..."

--I incorporated your second suggestion.

  I favor the term star-forming instead of starbursting for the SINGS
  sample since it is more generic; to me, 'burst' implies very bright,
  whereas several SINGS nuclei are actually pretty wimpy.

3 THE DATA:

From what I remember from a previous version only data to shorter
wavelengths than 2.2um had been corrected for Galactic
extinction. That has now been changed, right? All IRAC data have been
also corrected according to the caption of Figure 1.

--Correct.  I decided to do things uniformly.

3.1 ULTRAVIOLET DATA:

Answering one of Luciana's concerns, the UV magnitudes were obtained
from the galaxies' surface brightness profiles which were measured
after properly masking all field stars and background galaxies. You
can refer here to the GALEX Atlas paper. BTW, the GALEX Atlas paper
has now been accepted for publication. I hope to post the paper next
week in astro-ph. I will send you the number as soon as I have it.

--Done

2nd paragraph: I do not know if the new objects added could have
changed the percentages given in this paragraph regarding the
differences between asymptotic and D25-aperture magnitudes but it
would be good to check it.

--Thanks for the reminder.  Still 14%.

The SINGS objects that have not been observed by GALEX yet but that
are planned are: NGC1377, NGC3184, NGC5033, IC4710. Regarding the
objects that will never be observed, these are NGC5408 and NGC6946. Of
these, NGC6946 could perhaps be done using a petal pattern observation
but it would be quite time consuming.

--Thanks!

We should perhaps mention here that there are few galaxies that were
not part of the GALEX Atlas sample. In particular 5 of the galaxies I
sent you yesterday (M81DwA, NGC3773, NGC4254, NGC4725, & NGC6882) and
some others that at the time the Atlas was put together had been
observed only in NUV (NGC1705 & NGC2976). All these objects, however,
were reduced and analyzed following the same exact procedures as those
for the galaxies in the Atlas.

--Done

3.2 OPTICAL DATA:

3rd paragraph: Regarding other of Luciana's concern. I think it would
be good to provide in either in Table 1 or 2 the parameters of the
elliptical apertures used for the Spitzer and optical-NIR
magnitudes. As you commented in your e-mail this might require that we
somehow estimate how reliable is comparing these aperture magnitudes
with the UV asymptotic magnitudes. I would propose that you send me
the parameters of these ellipses and I estimate from my groth curve
what would be the UV aperture magnitudes within those ellipses. I am
quite sure that the ellipticity and PA of the ellipse for these
objects are close enough to those used for the UV growth curve
computation so this test is meaningful.

--The elliptical apertures are now listed in Table 1.  I sent you the 
  apertures; we can compare the two methods when you extract the UV
  fluxes using these apertures.

Below, you say that the field stars were identified based on the
3.6/8.0 and 8.0/24 flux ratios. I suppose these stars were further
checked by eye. What about the stars that were visible at optical
wavelengths and not in the IRAC bands? Were they removed
interactively? If so, it would be good to mention it here.

--The IRAC data are quite deep, and we didn't get carried away with
  removing very faint objects, so we rarely removed significant optical
  sources that weren't also removed in the IRAC data.

3.3 RADIO DATA:

I see no reference SCUBA 450um and 850um data given in Table 3. I
might have missed.

--I reference Dale et al. (2005) in the opening paragraph of Section 3
  for the SCUBA data.

"for 53 of of" -> "for 53 of"

--Thanks

4.1 GLOBAL BROADBAND SPECTRA ENERGY DISTRIBUTIONS:

I guess the stellar curve does not include any dust extinction, right?
We should perhaps say it explicitly.

--Done

4.2 SED BINNED BY THE INFRARED-TO-UV RATIO:

1st paragraph: Although I suppose the UV flux in the TIR/UV is
computed as nu*f_nu I say it explicitly.

--Done

3rd paragraph: When talking about the PAH emission and the
low-metallicity systems I think that in order for the sentence to make
sense we should add that these low metallicity systems, because their
lower dust content, are included in the bin with TIR/UV<0.3 (I suppose
that is the case), which is the one showing the comparatively faintest
MIR emission.

--Good point.  Done.

4.3 PCA:

Since finally only half of the sample made it to the PCA, I add a flag
either in Table 1 or 2 to let the reader know what particular galaxies
are used for the PCA.

I have been thinking about the interpretation of the two first
eigenvectors. The way I see it (this is just an opinion) the shape of
first eigenvector indicates that the specific star formation rate
(sSFR, or alternatively, the birthrate parameter) or some other kind
of current_SFR/past_SFR estimator is the main responsible for the
variation of the SEDs from galaxy to galaxy and the dust content is
the second one. If this is the case it seems that the variation in the
SED of galaxies is mostly due to sSFR (84%) and to a lesser extent to
dust (%10). Would you agree on that? If so, it would be interesting to
say that these two numbers might change and even reverse (in terms of
their contribution to the variance of the SEDs) if the selection
criteria are different or if we are dealing with high-redshift
objects, where perhaps variation in the sSFR are not that
critical. Just a thought.

--I added something along these lines to the discussion.

At the end of this section I miss a couple of sentences connecting to
the next section. In order words, is there anything in the results of
the PCA that makes us now to study the parameters governing the
variation in the Infrared-to-UV ratio or we are just analyzing another
important issue affecting to the interpretation of the SEDs of
galaxies? If the latter, adding a paragraph describing the focus of
each of the sections at the end of the introduction would certainly
help to guide the reader.

5. THE INFRARED-TO-UV RATIO:

5.1 INCLINATION:

I would add a concluding sentence to this section, saying, for
example, that Figure 11 indicates that for moderate inclination angles
the inclination does not to dominate the optical thickness of galaxies
(there is certainly a better and more proper way to say this).

--Done

5.3 FAR-INFRARED COLOR:

disribution -> distribution

--Thanks

This is not a comment to the paper but I thought it would be good to
bring it up. In the attached plot (part of the GALEX Atlas paper) I
show the distribution of FUV-NUV color vs. the absolute magnitude in
the FUV. As you can see for star-forming galaxies, spirals (green) and
irregulars (blue), a cone-shaped diagram is also found at bright M_FUV
values. Note that FUV-NUV is also a measure of the dust extinction. I
am not sure if these two things are related in the sense that brighter
galaxies in (observed) M_FUV would also have hotter 70/160um FIR
color but it would be nice to check it.

5.4 ULTRAVIOLET SPECTRAL SLOPE:

Since both in the Atlas paper and in Boissier et al. (2006, submitted)
we both discuss on the IRX-beta plot applied to normal star-forming
galaxies you might want to add these references to the list give in
the 1st paragraph.

--Done

2nd: It might be interesting to mention that by making use of
azymuthally-averaged radial profiles (as in Bossier et al. 2006,
submitted), where the regions dominated by emission from the bulge or
an AGN can be easily identified and excluded, the relation between
TIR/UV and UV-slope tightens up compared with the one obtained using
the integrated properties.

--Done

ACKNOWLEDGMENTS:

Could you please add "AGdP is financed by the MAGPOP EU Marie Curie
Research Training Network and the Spanish Programa Nacional de
Astronom\'{\i}a y Astrof\'{\i}sica under grant AYA2003-01676." ?
Thanks!

--Done

FIGURES:

Figure 1: I would start the caption with "Globally-integrated
0.15-850\,$\mu$m spectral energy distributions for the SINGS galaxy
sample." and then in the captions of Figure 2-8 I would just say
"Globally-integrated 0.15-850\,$\mu$m spectral energy distributions
for the SINGS galaxy sample (cont.)."

--Done

Figures 2-8: I would put the left edge of each plot starting at a
little bit shorter wavelengths, sometimes the FUV point is hard to
see.

--Done

Figure 9: There are points both at 24um and 25um (IRAS, I suppose),
right? If so, I think we should mention the fact that we use the IRAS
25um data somewhere in the text.

--Done

Could you please make the symbols in this plot (especially in the
legend) somewhat larger?

--Done

Figure 13: Axes labels are missing (you knew that already!)

--Fixed

Figure 14: It would be good to mention when discussing on this figure
what part of the 24um morphology dependence of the TIR/UV and 70/160
ratio might be due to nuclear activity, especially in the case of the
point-like sources.

--Done

Figure 15: How do you fell about providing the best linear fit to the
f_24_resolved/f_24_unresolved vs. f_70/f_160 data points in this
figure?