* title:
  a bit too vague (what kind of diagnostics ?)
  
--I've added "spectral" to the title.

* abstract:
  "It is difficult to distinguish between pure Seyfert and pure LINER sources."
  Is this caused by a predominance of accretion activity in the LINERs
  of the sample ? Some well-studied LINERs have been found to actually be
  starbursts; would these occupy a location closer to that of HII galaxies ?
  For the LINERs included in the sample, do we have independent evidence
  to help us distinguish between stellar and non-stellar activity ?
  
--To my knowledge we do not have such independent evidence.

  "Compared to starbursts, extranuclear regions separate even further [...]"
  How do we interpret that ? Should we say something about the fact that
  extranuclear regions are cleaner representatives of HII regions than
  starburst nuclei, because their stellar populations and ISM structure
  are less complex ? This was also the case with near-IR line diagnostics
  (in your PIFS paper) indicating younger ages for extranuclear regions,
  because they trace a single burst, as opposed to the average of multiple
  star formation episodes for nuclei.
  
--I've added words to this effect to the Summary.

* 2.2:
  - A factor of 1000 is missing in the given ionizing star temperatures.
  - Rather than saying "100-fold range in radiation field intensity", can
  you give the actual range limits, since you quantify the other listed
  parameters ?
  - Where does the H2/HI mass ratio come from ? This quantity is hard
  to obtain and deserves a small description.
  - Similarly, we should give references for the derivation of all the
  quantities.

--I'm using the numbers from Kennicutt et al. (2003), Section 3.2.  I have added
  the actual ISRF numbers and the factor of 1000 in T_eff.  I have added that the
  H2/HI numbers are CO-based, but we can ask Rob for more details if necessary.
    
* 3.1:
  "High-resolution spectroscopy was obtained in SH, LH and SL".
  SL was of course not used for high-resolution spectroscopy but to measure
  PAH EWs.
  
--This has been corrected.

  The nuclei with enlarged SH maps should be marked clearly in the table.

--I prefer not to mark these since are are not utilizing the enlarged maps in
  this work -- I'm using a consistent 23"x15" aperture for all sources.
  
* 3.2:
  A data table should be given for the archival data as well, with references.
  
--A table has been added including source name, source type, and reference.  The
  table will be published as an electronic table, meaning it will appear in its
  abbreviated form in the print journal (and fully in the electronic version).

* 3.3:
  "Spitzer high-resolution observations do not always detect the underlying hot
  dust continuum."
  This is incorrect. The continuum is detected. The problem is that its true
  brightness cannot be recovered because, owing to the way in which the "dark
  current" is removed, the background/foreground brightness is not recoverable
  without dedicated observations in the same module, which are not obtained
  by SINGS.
  
--Your point is taken.  However, there are still some faint star-forming sources
  for which the continuum is not detected but lines persist (e.g. NGC 3031 Munch 1).
  But I have rephrased the wording, and emphasized that we do not have sky
  observations for subtracting the sky continuum.

* 4.1 and 4.2:
  These paragraphs do not actually describe results, and would be better placed
  within a new section entitled "Measured quantities" for instance.

  The measurements are listed in Tables 1 and 2, not only 2.

--Corrected.

* 4.3:
  There is confusion between the Seyfert, LINER and AGN classifications.
  In my mind, AGN is generally synonymous of Seyfert. The text seems to adopt
  a larger definition, or to take for granted that all LINERs have hidden
  Seyfert activity, which may not be the case.
  
--The text has been cleared up, saying "LINER or Seyfert" instead of "AGN".

  Since the infrared line ratios are measured in 23" x 15", it would maybe be
  instructive to do the infrared-optical comparison in a fairer way, using for
  the optical the 20" x 20" extractions. How does the diagram in Figure 2 evolve
  going to this larger aperture ? I expect to see more transition objects which
  you say are lacking using the 2.5" x 2.5" aperture.
  
--I have put in the figure using 20"x20" extractions.  There are indeed more
  transition objects.

  "We turn to the mid-IR where we will perhaps more convincingly determine whether
  optical classifications are appropriate."
  This does not sound like a sensible expectation, for the reasons you give
  explicitely: 1) the optical ratios are measured in smaller apertures, hence
  should be able to produce cleaner separations between the different nuclear
  types ; 2) the extinction in the optical is modest, whereas you said extinction
  effects are the main motivation to go from optical to IR diagnostics ;
  3) the separation between nuclei in Figure 2 is already as convincing as
  one could wish.
  
--Now that we are using the 20"x20" extractions, the additional transitional 
  objects make this discussion more compelling.

* 4.4.1:
  "PAH features are quite prominent throughout much of the ISM save for regions
  characterized by exceptionally hard radiation fields."
  To be more exact, it should be said here that if observations are obtained at
  spatial resolution high enough to separate HII regions from PDRs, it is seen
  that PAHs are absent from the HII regions themselves but prominent in the
  surrounding PDRs just outside. Yet, HII regions are not characterized by
  exceptionally hard radiation fields.
  
--This is a sensible suggestion.  The text now indicates that PAHs are low for the
  cores of HII regions.

  It should maybe be pointed out that only the top-left and bottom-right extremes
  of Figure 3 allow a clean separation between Seyfert-classified and HII-classified
  regions, i.e. that the zone where types are mixed together is quite large.
  - How does this zone compare with what one would obtain in the optical with
    an equivalent aperture ?
  - The shift of Seyfert nuclei toward the location of HII nuclei is easy to
    understand in terms of larger aperture. But the shift of HII nuclei
    toward the location of Seyfert nuclei cannot be caused by any such effect.
    Could part of the mixing be due to:
    (1) a fraction of the objects having [OIV] powered by WR stars ? For instance,
    there is one HII nucleus (with [OIV] detection) having a very high [OIV]/[NeII]
    ratio. What is this object ?
    (2) varying volume filling factors of HII regions and PDRs (i.e. extreme
    starbursts with higher HII/PDR ratios than usual would shift towards the left,
    as would low-metallicity regions) ?

--Additional discussion along these lines has been added.  The HII source with a high
  line ratio is NGC 1705 (a known starburst or post-starburst galaxy).

  Was the photospheric continuum removed before deriving the PAH EWs ?
  Nothing is said about this, but this is important.
  
--The stellar continuum was not removed since this is not possible for the
  archival data that are also included in the Figure.

  I find the points with upper limits confusing and hard to mentally excise from
  the diagrams. Is it possible to show the same diagrams without the non-detections,
  for clarity ?

--I've substantially revised the portrayal of the data in the Figures.  To make 
  things easier to read, I've removed the error bars from the literature points, and
  I've only shown the statistical uncertainties for SINGS data.

  Looking closely at the [SiII]/[NeII] graph of Figure 3, I am not really
  convinced of the existence of a real trend in this ratio with activity type.
  There are many HII nuclei with similar [SiII]/[NeII] ratios as Seyferts.
  The only clean separation I see is between the Galactic HII regions and the
  rest, and this can easily be explained by a lower PDR contribution in these
  resolved objects.
  The origin of [SiII] may change going from HII nuclei to Seyfert nuclei,
  but I do not agree that this can be exploited practically, because the [SiII]/[NeII]
  ratio does not vary in a systematic way, as long as the beam covers a large
  physical area.
  Have you tried different [SiII]/X ratios, where X is another line ?
  Which one would provide the most useful behavior ?
  
--I have tried a variety of ratios and SiII/NeII provides one of the cleanest
  separations.  The typical SINGS HII ratio is 0.8 while it is 1.2 for Seyferts, a
  50% difference.
  
  "An argument based on interstellar density provides another possibility for
  the low [SiII]/[NeII] ratio in AGN."
  also in following sentences in the same paragraph
  low -> high
  
--This has been corrected.

  What is the physical motivation for adopting the limiting values that you
  give for the mixing curves ? Can they be compared with what is observed
  in some prototypical objects with "pure" Seyfert emission and "pure" HII
  emission ? To what kind of conditions do they correspond in models ?
  
--The motivations for the mixing curve anchors were purely empirical.  I originally
  used SINGS objects as the anchors, but the SINGS sample simply doesn't fill 
  enough parameter space to provide useful anchors.  Leitherer and Kewley et al. are
  working on a paper that compares models to SINGS line data.  They would prefer to 
  keep the model analysis in their paper.

* 4.4.3:
  I don't think that the sequence Seyferts - HII nuclei - HII regions in the
  [SIII]/[SiII] ratio could be explained by metallicity effects, since the
  Galactic HII regions are as much separated from the HII nuclei as the LMC
  HII regions are, but the SINGS extranuclear HII regions cover the same
  location as the HII nuclei. A more fundamental difference may be in the fraction
  of PDR included within the beam, i.e. the closest regions are the best resolved
  and have all chances of having higher HII region/PDR ratios.
  The differences in size of the probed physical regions should be discussed
  somewhere. The first explanation proposed for Seyferts (X-ray dominated
  regions) is also more plausible than a systematic metallicity difference
  between them and HII nuclei.
  
--These suggestions have been incorporated into the text.

  The excitation potential is given only for [OIV], [NeII] and [SiII].
  It should be given also for [SIII] (for Figure 5).
  
--This is found near the top of Tables 1 and 2.

* 4.4.4:
  As written, this paragraph is a bit confusing. The main point is that HII nuclei
  have stronger [SIII] line to continuum ratios than Seyfert and LINER nuclei,
  both for the 18.7 micron transition and the 33.5 micron transition.
  Yet, your formulation gives the impression that the [SIII]_33.5 / [SIII]_18.7
  line ratio is higher in HII nuclei, which is not true.
  The linear fits do not demonstrate anything and can be omitted. The information
  provided by the lines of constant n_e is enough.
  
--The two fits separate along the diagonal, showing that HII regions/nuclei have
  higher line-to-continuum ratios.  I have reworded the text to make this more
  clear.

  Can you compare the derived electronic densities with literature values for
  luminous starbursts/AGNs, and for nearby HII regions ?
  
--A comparison with literature values/ranges has been added.
---------------------------------------------------------------------------------
1) Measurements for N1377:
You derive a 6.2 micron PAH EW of (0.02 +- 0.01) micron ; yet, I do not
detect any PAH at 6.2 microns (but possibly one at 6.3 microns).
Likewise, H_alpha and [OIII] are not detected in the spectra that John
Moustakas sent me, and H_beta is seen only in absorption. I do not
understand how N1377 can appear as a detection in all four lines of the
[OIII]/H_beta vs. [NII]/H_alpha diagram.
N1377 is a prime example of an unclassifiable galaxy (at least using the kind
of diagnostics studied here), and it doesn't make sense to say that it is
a LINER or Seyfert. Veilleux et al. (1995) were wise enough not to attach
a label to it.

--Good point.  Yes, it is at 6.3 microns.  Moreover, I shouldn't trust this 
  measurement as a PAH measurement, since it's difficult to see much evidence
  for other PAH features.  I will remove this from the table and list it as not 
  detected.

--Moustakas shows S/N greater than or equal to 3 for these four lines. 
  I will try to track down this discrepancy between the data John sent you
  and the data he sent me.  I tentatively have added a footnote indicating
  that this is likely a deeply buried source.

2) "No sources appear to be heavily buried in the optical so presumably none
of the classifications are skewed by large amounts of dust".

Since N1377 is part of the sample, I infer that it is not a deeply obscured
source. It actually has tau_V ~ 70 or more, and absolutely nothing of the zone
of activity escapes in the optical. 

--I have placed a footnote cautioning the reader about NGC 1377, as described
  above.

Two other galaxies of the sample are found by JD to have high optical depth
at 10 microns (tau > 2), namely N1266 and N3198.
Using a plausible extinction law, these galaxies have A(V) ~ 17 and 34.
Yet you quote a maximum extinction A(V) ~ 4.1 for N1266. 
The H_alpha/H_beta decrement is typically biased in favor of low extinction
values, because it does not probe the heavily obscured regions but, if such
regions exist, only foreground emission. This may be the case here.
It would be particularly interesting to locate these two galaxies in the
diagrams.

--I compute A(V) using a screen model, whereas JD uses a "mix" model,
  which yields larger values.  An A(V)_mix of 17 and 34 respectively
  correspond to A(V)_screen of ~2.8 and 3.5   I have suggested to JD that
  he emphasize this distinction.

  I understand your concern over NGC 1377.  Indeed it must have a lot of
  extinction in the nucleus.  But how do we reconcile this with Moustaka's
  data?  I could qualify the statement to emphasize that these extinctions
  correspond to a 20"x20", or simply say that ancillary data for NGC 1377
  indicate that it is likely very buried and thus the optical line
  detections are only probing the foreground emission.