paper2_comments.txt

Dear Coleman,

I enjoyed reading your paper. I learned a lot, and there is great
stuff in there. Appended are my comments. I have a few concerns, but
I think it's in good shape overall. FYI - I will be moving to
Connecticut the last week of August, but otherwise available through
the fall.

Cheers,
Sarah

%start of comments

General:

(1) You have included a lot of detail about the specific methodology
you used for determining the E(B-V) and a_lambda distributions for the
quasars. I am assuming that this is something you developed, and
given the amount of work that appears to be involved and the likely
utility of this methodology for other people, you should promote it
(mention it explicitly and in more detail) in the abstract so it won't
be just quasar people that find it. If this is a well-known
methodology that you just adopted for this project, then there is way
too much detail in section 3, and it should be cut substantially.

(2) The combination of stronger [OII] emission (likely from star
formation in the host galaxy) and weaker Balmer line emission in the
lower-z (because the optical spectra are available) intrinsically red
quasars makes me suspicious that the host galaxy continuum has been
under-subtracted from this set of objects. This would both make the
continuum redder and decrease the EW of the Balmer lines. 
----
Coleman -- one thing that you could do is to do it again *without*
any host subtraction just to see how much effect is has. 
----
Specifics
about the host galaxy subtraction need to be outlined, e.g., what
template was used? how was the template normalized? Using a z=0
elliptical template (if this is what was done) is problematic for high
z in particular (even in a quiescent galaxy the stellar populations
will be at least a few Gyr younger). I think a referee is likely to
have similar questions about this part of the conclusions.
----
Sarah -- can you suggest better templates as a function of redshift? 
----
(3) I am not sure if this is within the scope of this paper, but I'd
be interested in discussing further what the E(B-V) distribution can
tell us about dust in the host galaxy vs. dust in the vicinity of the
quasar. 
----
You'll be interested in the "multiple scatterings" dust model results that
Karen asked us to run too. 
----
The link between BAL quasars that are very blue and very dust
reddened and their CIV BAL features tells us that that dust must be
associated with the quasar. For the ones with absolutely no dust,
that means that there exist dust-free lines-of-sight through the
hosts, which is really remarkable. The fraction of dust-free quasars
should be quantified. Are those most likely to be "post-feedback"
types of systems? How else would you get no dust through the kpcs of
the galaxy ISM? Quiescent stellar populations have AGB stars, and
local ellipticals have some mid-IR emission from dust. Are there any
other regimes in the alpha_lambda vs. E(B-V) parameter space that can
give some insight into the quasar dust vs. host dust balance?
----
I think that we can answer this to some extent just by looking at
the E(B-V) distribution from the MW for the full SDSS sample.  I
did this really quickly and found that
E(B-V)<0.01 is 4.2% of SDSS quasars
<0.02 is 32%
<0.05 is 84.5%

So, how rare they are depends on what we'd call "no" dust.   But I'd
also say that, even with our attempts to break the degeneracy,
"no dust" just means that we can't distinguish between a little dust
and a change in the power-law slope. 
----
Specifics:

================
Sec. 2
--------------
Even with the reference to K13, would suggest including a short para.
that specifies the S/N of the match detections to GALEX, UKIDSS and
[particularly] 2MASS to make clear that photometric errors and
Eddington bias [for 2MASS] are not affecting the colours of the
individual quasars significantly. [cmk: only needed in paper, not thesis (all one section in the thesis)]

Similarly, I don't see any discussion of the dispersion in the individual
SEDs due to intrinsic variability, which is not insignificant given some
of the epoch differences between SDSS and the other surveys. The inter-survey
colours are definitely affected. That sais I don't expect there is a significant
impact on the conclusions, although the dispersion in the spread of
the intrinsic power-laws is definitely measurable when I have used [for
example] SDSS and UKIDSS. So, suggest short para. making clear that results
are not affected by variability. [cmk: not sure where to put this...]

Further on

A comment, about which you can't really do anything, although it does sort
of relate. Ages ago I determined the shape of the extinction curve for
1200-3000A using the SDSS quasar spectra and the photometric data
longward of 3000A. I really should publish the result. While the shape
is certainly SMC-like the slope in the UV below 2000A is definitely not
as steep as SMC. Suspect that one can see that in some of your c1 - c2
plots.

Similarly, when I generate a model SED to reproduce my version of your
modal colours one definitely needs two power-laws. one relatively blue
(alpha_nu~=-0.5 below ~2800A) and one with alpha_nu~=-0.1 out to a
micron. Guess the manifestation in the plots is a degree of curvature
in the distribution of points in the c1-c2 plots? 

================

p. 12 

How does presence of BALs affect "curvature"? Could the flux missing
from the CIV band induce "curvature"? [cmk: not sure]


Fig. 14

lowest panel, explicitly indicate the SBB on the panel (overplot
power-law continuum). [cmk: not sure this makes sense for this plot]

p. 15 

A few words are warranted here on how limb-darkening of the accretion
disk would be expected to affect the colors. If the dust is intrinsic
to the quasar, then more reddening should go with more limb-darkening
(assuming the dust originates in the plane of the accretion disk). [cmk: I have no idea what this is]

Do any of the intrinsically bluest quasars have no dust? Does the
fraction of quasars with E(B-V) consistent with 0 change with
intrinsic color? Examining this could reveal something about host
galaxy dust. [cmk: this is answered in the marginaliztion plot...]
---------

4. The discussion is slightly weak, in my opinion, due to the lack of
interpretation comments.   Did you intend that, or are you hoping that
some of us will kick in some stuff?  I am wondering, e.g., why some
quasars have bluer intrinsic continua than red quasars - do I remember
a link with inclination, maybe an old Netzer paper?  That might
explain the stronger CIV, given that the continuum contribution would
decrease as cos theta from face on to edge on.  Although, this would
not explain why Hbeta is weaker when the spectrum is redder....

%end of comments