We submitted the paper in Sept 1999, the following is the exchange of referee reports and emails from the editor. The paper required over 9 months to referee, in the end only minor changes were made to the text. As you can see from the exchange, there were several problems in the process: 1) the editor (Huchra) clearly never read the paper and simply felt his job was to forward the referee reports, 2) the referee (anonymous) believes, for some strange reason, that the entire sample is due to confusion with other sources (implying that you can point Arecibo at any place in the sky and get a detection).
Here is the first referee report, 2 months after submission.
November 9, 1999
Dr. James M. Schombert
Department of Physics
461 Willamette Hall
University of Oregon
Eugene, OR 97403-1274
Dear Jim:
Attached please find the referee's report on your manuscript entitled
" Gas Rich Dwarfs from the PSS-II III. H\,I Profiles and Dynamical Masses."
The referee recommends eventual publication in the Supplements after a number
of confusing issues are clarified. Most of the referee's suggestions
relate to presentation and should be easy to fix. Any place where the
referee is confused is guaranteed to cause problems for other readers.
Please upload the revised version of your manuscript to the ApJ site
and send me a note detailing your response to the referees comments.
Continue to refer to ApJ MS#50463. Please do *not* e-mail me the manuscript!
Take care,
Sincerely,
John Huchra
Scientific Editor
===============================================================================
Title: Gas Rich Dwarfs from the PSS-II III. HI Profiles and Dynamical Masses
Authors: Eder & Schombert
Summary: This paper presents valuable Arecibo HI measurements of fluxes and
linewidths for a collection of dwarf and LSB galaxies. It also presents
angular sizes and eccentricities for a subset of the galaxies based on
optical CCD images, and some derived HI-to-optical comparisons. Various
general conclusions are drawn about the nature of the galaxies in these
samples that are sometimes well supported and other times not.
As this paper is part of a series and is not the final analysis paper
(which is expected to follow this one), one of its main purposes should be
the thorough and accurate description of the samples and observations it
presents. The current version of the paper does not succeed in this regard.
The paper does include quite a bit of analysis, some of which is a natural
extension and exploration of the observations presented, but other parts of
which appear to need considerably more examination before the arguments
would be persuasive.
My recommendation is that the authors elaborate on the observations and
shift some of the more-poorly supported parts of the analysis in this paper
to their planned analysis paper where there will be more space for a
complete explanation of their ideas. Most of the data presented here are
primarily of archival value and therefore most appropriate for the
Astrophysical Journal Supplements.
The following are more specific problems with the paper:
"Dwarfs": This sample clearly contains a wide variety of galaxy sizes,
masses, and luminosities like previous HI flux-limited samples of LSB
galaxies studied at Arecibo. Calling them all "dwarfs" is not supported by
the information presented, and just confuses the discussion of galaxy
properties. It seems reasonable to say that the small subset of the objects
with HI masses below a few times 10^8 solar masses are probably dwarfs, but
more convincing evidence needs to be presented for the rest. Calling this
"the most extensive, uniform sample of 'normal' dwarfs available" seems
unwarranted.
Opening paragraph: On p.3 a statement is made about "the technological
advancement is the existence of the Arecibo telescope..." The authors
appear to be hunting for a dramatic introduction, but this statement comes
out sounding awkward and silly.
Samples 1-3: Sample 3 appears to be intended as a test of the completeness
or reliability of Sample 1, but the galaxies that overlapped or were
different from Sample 1 are not identified in Table 1 (for instance by
writing 1 and 3 in the last column), and the meaning of the samples is lost
because random galaxies from the earlier samples that had not been observed
are apparently thrown together with sample 3 because they were observed at
the later time. It would make more sense to identify the samples according
to how they were chosen as opposed to when they were observed.
Figure xx: Possible differences between the samples could be explored by
comparing the galaxies' properties. On page 7 such a comparison is
suggested, but the "Figure xx" that is supposed to demonstrate the
similarity was not provided. To support their claim that Sample 1 had a
higher detection percentage because they observed only the larger galaxies,
they should report the detection percentage of galaxies larger than 27" in
Sample 2. This would present an opportunity to present the detection rate
as a function of angular size, which would be interesting.
Velocity ranges: It is highly puzzling to read that the search range for
undetected galaxies was 1000 to 10,000 km/s on p. 8 (or occasionally 0 to
12,000 km/s according to the notes), when earlier it is stated that the
observations were centered at 4000 km/s, covered a total velocity range of
8000 km/s, and extended to 8120 km/s. The math does not add up or make
sense. Was a 9000 km/s range actually observed sometimes? Do most of the
undetected galaxies have a possibility of being at <1000 km/s? Why
wouldn't repeat observations cover more like 0 to 16,000 km/s since the
bandpass allows 8000 km/s at a time? Were standard observations centered at
4120 km/s? These are important details if the significance of the
non-detections is ever to be accurately interpreted.
Line Widths: The line width is first said to be measured at "50% of the
mean flux" (p. 8), which normally signifies 50% of the mean flux density
within the width of the profile. But the next sentence says a polynomial
was fit between 15% and 85% of the respective horn on each side, which
seems odd since the 50% of MEAN flux might occasionally fall below the 15%
of PEAK flux density, depending on the shape of the profile. Later on p. 9
the heliocentric velocity is said to be measured at 50% of peak level. It
is possible that the authors mean all three of these things, but it would
be a strange and unlikely mixture of measurement methods. I'm guessing
that the measurements were made at 50% of the mean of the peak flux density
in the horns of the profile.
Confusion in Detections: The authors say they have examined the plates for
possible sources of confusion, but this does not appear to have been done
carefully. On just the first page of spectra 409-7, 409-19, 475-3 and 476-3
all have typical profile shapes for confused sources, and inspection of the
DSS images confirms that there are multiple sources close enough to cause
confusion. None of these has a note indicating that confusion has been
examined or ruled out, and worse, the line widths in at least three of the
cases are almost certainly spuriously large.
Double vs. Single-Peaked Profiles: The authors quote a statistic on p. 17
saying that 34 of the 246 profiles are single-peaked. This puzzles me
because I quickly count that many in just the first several pages of the
profiles. In addition, there are a number where I am suspicious of
confusion that may be single-peaked as well (like the ones mentioned
above). It is important to keep in mind that these are noisy spectra, so
even a truly single-peaked profile is likely to have some additional shape
to it, but should not necessarily be considered double-horned.
Size of HI disk: While it is probably true that the HI extends well outside
of the optical disks of these galaxies, it is at best only statistically
correct and needs to be demonstrated for this sample. While some authors
have demonstrated this for other samples, these were selected in quite
different ways, and it is dangerous to simply adopt their correction
schemes. Uniform application of a single factor between optical and HI
sizes to galaxies spanning several orders of magnitude in HI mass and
optical luminosity is not reasonable.
Dynamical Masses: The dynamical masses quoted in this paper are suspect for
several reasons: applications of very crude size estimates of the HI
radius; use of profile widths that are sometimes confused; and application
of a formula for disk systems which is inappropriate for systems with large
turbulent motions. These masses are, at best, only accurate averaged over
large subsets of the sample. I would encourage the authors to take the most
extreme M/L objects and observe them at the VLA to confirm their mass
estimates. For now, the discussion here about non-baryonic matter (p. 22)
is wildly speculative.
HI Surface Densities: Since the HI sizes are questionable, the HI surface
densities are too. Thus it is also not worth spending much effort with
arguments about critical star-formation densities. What would be a more
valuable analysis is a comparison of, for example, the ratio of HI mass to
optical luminosity relative to other samples, like the dwarf samples of
Hoffmann et al. or Schneider et al., or to HSB galaxy samples, like de
Jong's.
Optical Data: Since the paper makes a number of comparisons involving the
optical luminosities of these galaxies, the optical magnitudes should be
included in Table 2 along with the optical radii and eccentricities. It
should also be made clear whether "Ecc" in the table is actually the
geometric definition of eccentricity, the ellipticity, or the axis ratio
(which is the parameter that is normally listed).
There are numerous misspellings, verb/subject disagreements, stray
characters, and TeX errors--too many to list here. For the resubmitted
version of this paper please be sure to note that % symbols in TeX cause
the end of a line to be eliminated from the printed text.
While many of the comments are useful to clarify the presentation, the comments about confusion and double-horned profiles left us at a loss on how to respond. Clearly confusion is not a problem for an LSB sample. The objects are selected off of PSS-II plates, its easy to see when there are neighbors. Companions are noted in the Notes to the Tables. Our method of classifying double or single horned is outlined in the text, others with different opinions on how to classify profiles can look at the data themselves (which was the purpose of the paper, to get the profiles in the literature and out to the public).
So we opted to simply make as many of the minor changes as we could and hope the referee would be satisified. We were wrong. Here is our response to the referee report.
Dear John:
Here's the resubmission of MS#50463, Eder and Schombert LSB Dwarfs. The
referee report was very helpful, I believe I made all the changes requested.
Jim
---------
We thank the referee for a very useful report to strengthen the
presentation of our paper. We have corrected all of the problems noted by
the referee. While there is a final analysis paper in the works, the M/L
analysis is not a part of that study and we would like to preserve that
section in this paper since it may be the only time the issue is addressed
with this dataset.
The following is a response to the individual issues raised by the
referee:
"Dwarfs": We disagree with the referee that low HI mass is the definition
of a dwarf. We have discussed this problem at length in Schombert et al.
1997 and Pildis et al. 1997. We find that total magnitude and HI mass are
insufficient for the definition of a dwarf galaxy. Although the two
combined values provide you with a measure of the baryon mass of a galaxy,
the dark matter component is unknown and a define of dwarf by total mass
would require this information. Instead, we prefer the size of a galaxy,
either by scale length or isophotal radius, as the criteria for dwarf. To
this end, as discussed by Schombert et al., the irregular morphology and
small metric radii are to set this limit and we have expanded the text
to discuss this point.
Opening paragraph: The offending sentence is dropped.
Samples 1-3: Sample 3 was not for completion, it was to finish up an area
that contains objects which were missed in the first runs (due to
limitations in the pointing of Arecibo) or objects we did not have optical
data for and were given lower priority during the first two runs. All the
samples were selected the same way off of PSS-II plates, the Arecibo
observing is constrained by the operational parameters of its pointing.
We have divided the samples by observing run in the tables so that we can
keep track of calibration changes, telescope sensitivity and whatnot. We
have clarified this is the text.
Figure xx: Oops, we apologize. In the passing of the paper from Arecibo
to UOregon, this section was cut and pasted. As mentioned above, this
section was to tabulate our observing program. There is no attempt at
completeness in this project, only at maximizing the sample of LSB
dwarfs. We have rewritten this section.
Velocity ranges: This was another mistake in the editing by 2 authors
(looks like 3 authors is optimal for catching inconsistences). The
correct velocity range is 0 to 8,000, and the text has been corrected to
make this clear. We did, on occasion, search out to 12,000 because we had
made statements in previous papers that distant, late-type disks have a
similar appearance to nearby dwarfs, but that we could differentiate them
based on their symmetry. We selected a few of these objects for higher
velocity searches and have included them here only for completeness of the
dataset. This was discussed in the text and noted for each object in the
Notes. These are not used in later papers concerning the properties of
dwarfs.
Line widths: The referee is correct that the line width was 50% of peak
flux, this has been corrected in the text.
Confusion in Detections: We had examined the region centered on 10
arcmins for each galaxy on the PSS-II plates while constructing the
catalogs and searched NED for cataloged galaxies. Possible companions are
placed in the Notes to Table 1. Neighbors of unknown velocity are
mentioned in the notes. Neighbors of much higher or lower velocity are
ignored. We have also noted at least two cases of LSB galaxies with LSB
neighbors. As such, we can not guarantee that every galaxy is isolated.
Skewed profiles may also be due to poor pointing (actually the pointing is
fine, our coordinates are hand measured and only good to 15 arcsecs). We
have amended the text to include these possibilities.
Double vs Single: This raises an interesting issue concerning the nature
of HI profile shapes. Our method was to assume a double horn profile in
all cases where it was not clearly a single peak (i.e. a plateau shape).
In addition, we assume all very narrow profiles were single. Both these
assumptions may be in error, and we were certainly negligent for not
outlining our assumptions in the text. Of course, if we classify the
noisy spectra as single peaked, then another referee would complain that
we overestimated the number of single peaked spectra.
As a compromise, we have changed the classification for the noisy
spectra with possible plateau shapes to `D:' and noisy spectra without
evidence of plateaus to `S:'. The profiles are divided into two types,
with peaks less than or greater than 20 mJy. The following is a breakdown
of the types
S S: D D:
peak > 20mJy 33 (37%) 1 ( 1%) 50 (57%) 6 ( 7%) 90
peak < 20mJy 28 (18%) 8 ( 5%) 90 (58%) 30 (19%) 156
-------------------------------------------------------
total 61 (25%) 9 ( 4%) 140 (57%) 36 (15%) 246
The number of double horned profiles still dominates the sample at 70%.
The ratio of clear classification also gives 61 S to 140 D, also 70%. We
have added this analysis to the text. We also noted in the text that when
we followed up on a handful of noisy plateau-like spectra, they all turned
out to be double-horned. It is our estimate that the error goes in favor
of double horned profiles.
Size of HI Disk: We apply a statistical correction. However, a second
referee would criticize our method if we didn't make this correction. Our
VLA images (another paper in preparation) for four objects shows that they
follow the same relationship for optical to HI size as outlined by Hoffman
et al., but we are not ready to publish that data at this time. While it
is undemonstrated at this time that the correction is the same as Hoffman
et al., there are numerous studies that demonstrate that HI is more
extended than the optical light in all late-type galaxies. Therefore, we
believe it is more honest to make the correction and state in the text
what we have done.
Dynamical Masses: The calculation of dynamical masses would only vary a
small amount if turbulence, rather than rotation, equations are used.
While our discussion is speculative, there is no deception here. We have
stated all the corrections and methods we have applied. A reader can take
the raw data and refit using their own prescriptions. As this sections
stands now, we believe these values to best represent the M/L of these
galaxies.
HI Surface Densities: We have added the comparison of HI mass to
luminosity as suggested by the referee.
Optical Data: We have used axial ratio (b/a) and have modified to the
text to state this. The inclusion of the optical luminosities is an
excellent idea and we have modify Table 2 to include them.
Our hopes that these changes were sufficient were for not, as seen by the second referee report.
January 13, 2000
Dr. James M. Schombert
Department of Physics
461 Willamette Hall
University of Oregon
Eugene, OR 97403-1274
Dear Jim:
Attached please find the referee's report on the revised version of
your manuscript entitled " Gas Rich Dwarfs from the PSS-II III.
HI Profiles and Dynamical Masses," ApJ MS50463.
The referee recommends eventual publication in the Supplements but
doesn't think you did a very good job of answering his initial
objections. The list of problems presented has to be dealt with
before publication.
Please upload the revised version of your manuscript to the ApJ site
and send me a note detailing your response to the referees comments.
Continue to refer to ApJ MS#50463. Please do *not* e-mail me the manuscript!
Take care,
Sincerely,
John Huchra
Scientific Editor
===============================================================================
Title: Gas Rich Dwarfs from the PSS-II III. HI Profiles and Dynamical Masses
Authors: Eder & Schombert
The authors have made some good efforts to clarify the text of this paper
in their revisions, but unfortunately they have not gotten to the heart
of the problems with their data. I outline below several major problems
based on an examination of a small fraction of their data. I am discouraged
that they did not even follow up on the problem cases I pointed out in my
first referee's report (which I elaborate on below), and so I have little
confidence in the remaining data. Therefore I cannot recommend this paper
for publication in its current form.
I still think this paper contains some valuable data, but it also contains
some plainly erroneous data. There are also several fundamental flaws in
the data reduction that render the claims about dynamical masses meaningless.
Unless the authors are prepared to correct these problems, publication
of the paper would be a mistake.
Dwarfs:
Many of the objects selected in this paper are not dwarfs by ANY measure.
The authors' response that low HI mass does not define a dwarf is obvious,
but they seem to miss the point that many of their sources have fairly
large diameters, HI masses, luminosities, and dynamical masses. Calling
them dwarfs, calling this "the most extensive, uniform sample of normal
dwarfs," and implying that this sample differs from previous ones that
have "Unfortunately...been contaminated by large numbers of luminous low
surface brightness spirals," are all misleading. The presence of these
sources does not ruin this paper, and some of the large LSB galaxies appear
to be interesting in their own right. Mislabeling them is a problem, and
the authors need to do a lot more work to show that this is a "uniform"
sample and define what they mean by that.
WHY THIS IS IMPORTANT: The authors use approximations for estimating
dynamical masses and HI surface densities that are only appropriate for
dwarfs. Applying these approximations (virial mass formula on p.14; HI
extent estimate on p.15) to other types of galaxies will give erroneous
results.
Confusion:
The authors have restated a claim in their response that they had checked
for confusion, but they have done nothing to correct even the obvious cases
I pointed out in my first report. Four of the HI profiles on just the first
page of spectra had indications of confusion, such as shoulders on the
profiles, large asymmetries, and three "horns." The authors respond that
asymmetries may be due to coordinate uncertainties of 15". With Arecibo's
beam size, a 15" or even 30" offset could not produce any of these features.
What's more, all of the sources I identified had obvious nearby companions
on the Digitized Sky Survey. These were companions of similar angular extent,
and to my eye probably at the same distance. Since the authors apparently
did not think it worth checking again, I looked on NED for galaxies nearby
the listed LSB source in the four cases I pointed out to them. Three of them
had nearby galaxies with either similar redshifts or unknown redshifts, and
the fourth is clearly a pair of LSB objects:
409-7: NED search: SRGb 062.12 1.5' away 7333 km/s (vs 7060)
409-19: NED search: 2MASX1 J0013057+295238 1.7' away (unknown v)
475-3: NED search: 2MASX1 J0104336+222928 1.5' away (unknown v)
476-3: pair of LSB objects < 1' apart
According to the authors' representations, all of these sources should
have had footnotes. Since they did not, the care taken with the entire
sample is open to question.
WHY THIS IS IMPORTANT: HI measurements that are confused will give
inaccurate or spurious line widths and HI fluxes. Sources with these
problems must be removed from any subsequent analyses of dynamical
masses or HI properties.
High M/L sources:
Since the authors insist on including their M/L analysis in this paper, I
took a closer look at two of the sources with extreme values, and found
further mistakes:
582-1: LSBC582-2 1.9' away (which IS listed in the footnotes) is a much
brighter and larger galaxy that has (to within the errors) the same
HI velocity and linewidth. Treating these parameters as though they
come from a much fainter source gives a meaningless result.
572-2 Low S/N spectrum. None of the HI parameters is certain, but even
assuming they were, the dynamical mass does not make sense:
From table 1: V0 = 3670 km/s, so dist = 48.93 Mpc
W50 = 144 km/s
From table 2: R25 = 9.9", so radius = 48.93Mpc x 9.9/206265 = 2.35 kpc
b/a = 0.23, so sin(i) = 0.97 (assuming q0=0)
putting this together using formulas from page 14:
Vrot = (W50/2 - 1.18 sigma)/sini = 61.86 km/s (for sigma = 10)
Mdyn = 2.3e5 (Vrot^2 + 3sigma^2) (2.3 R25) = 5.13e9
So how did the authors get log(Mdyn)=10.96?
The discrepant results for the dynamical mass in the second source had me
puzzled until I ran across a statement on p.14 that "sin i is the inclination
equivalent to the axial ratio of the outer isophotes of the dwarf galaxy."
This appears to mean they assume sin i = b/a, which is obviously incorrect,
but which does give the mass they quote. It is well known (e.g. Mihalas &
Binney) that cos i = sqrt{[(b/a)^2-q0^2]/[1-q0^2]}. To make sure that this
was what they had done, I checked the other sources and found they all make
this incorrect assumption; I also checked several of the DSS images to be
sure that Table 2 indeed lists b/a as the authors now state.
WHY THIS IS IMPORTANT: Obviously, confusion of sources and mistakes in
fundamental equations mean the dynamical mass estimates are invalid.
Note also that the second galaxy is more flattened than the authors'
assumption for q0 permits. In fact over half of the sources in Table 2
have b/a values that are not permitted by the assumed value of q0 used
in the dynamical mass estimates. This means the virial mass formula
they use is inappropriate.
The number of errors I have identified while examining just a small fraction
of the data from this paper indicates that the authors need to reevaluate
their work. I do not think it worthwhile reviewing the paper any further
at this time until the bad data are better identified, and the erroneous
analysis and indiscriminate use of data and formulas are corrected.
I always enjoy referees' who wish to tell the rest of us their personal agendas. The search in NED for 2MASS sources is great. The 2MASS extended source catalog is not quite a galaxy catalog, its more of a not-star catalog, i.e. anything not-a-point source is in it. All the rest were clearly background galaxies. So we were at a loss to reply to this report. So we asked for a second referee. Normally this is done in any case when a stand-off is declared. For some reason Huchra chose to ignore us and send the paper back to the original referee.
Here's our reponse to the 2nd report.
John - Here is our 2nd response to the referee. While some of the comments were useful, the criticism on completeness is completely wrong. The objects the referee uses as examples of confusion in our sample are absurd. What is most annoying is that, as science editor, you should have seen this in the report. One of the objects complained about is from a CfA poor cluster survey and have no relevance to an HI survey. The others are 2MASS objects (of which you are on the science team) and are clearly background objects or not even galaxies. This paper is a straight forward presentation of our HI data and a simple analysis of the virial masses of the small subset of objects with optical data. We have presented the data with precise notes. We have outlined our prescriptions, formulas and assumptions. Many may disagree for astrophysical reasons. And they may write their own papers reanalyzing our data to their hearts content. But this referee is overly consumed with his/her own agenda and is simply wrong in his/her interpretation of our data. While I appreciate the work the referee believes he/her is doing, we do not understand many of his/her comments and we have made very little substantial changes to our paper (other than to correct a stupid program error in our Tables, we are thankful to the referee for pointing that out). We do not believe the referee can contribute any further in this process. And we request that you read our response and make on editorial call on whether this, primarily data paper, is ready for publication in the Supplements. Jim Schombert ---------------------------------------------------------------------- Our response to referee report #2, Gas Rich Dwarfs from the PSS-II We thank the referee,again, for a very useful report to strengthen the presentation of our paper. We apologize if the referee feels we have not responded to his/her criticisms. We have made a serious effort in this draft to either correct or address their concerns. Most of the referee's concerns are on the quality of the data and its reliability. We hope in the following response that we can convince the referee that we have made every possible effort to present the data in a clear and unambiguous manner. We also ask the referee to remember that this is the presentation of all the data from our HI work. There are several objects in the tables which are pairs or confused or not dwarf-like in their properties. However, these were outlined in the Notes to the Tables and they were culled from the sample before the mass analysis section. Dwarfs: Perhaps we didn't state our conclusion clearly enough. While there are some large objects in the sample, only 3 have scale lengths greater than de Jong's average for disk galaxies of 3.5 kpc. A vast majority (over 58%) of the sample are a factor of two smaller than the mean scale length given by de Jong. We present all the objects we have detected in HI in this paper for completeness. There is no attempt to present the large LSB disks as dwarfs. Our point, in the discussion section, is not directly concerned with the internal structure that makes a galaxy a dwarf. Rather, we are interested solely in an observational technique for selecting a sample by surface brightness and irregular morphology which produces a sample of object which are primarily dwarfs as defined by size and luminosity. None of the objects sample selected for the dynamical mass estimates have magnitudes greater than -20, none have sizes greater than 20 kpc. Over 80% have sizes less than 15 kpcs. We argue, and have done so in the first two papers of this series, that this is a concrete method to maximize a dwarf survey. In order to help clarify this problem, we have eliminated all objects greater than 15 kpc from the mass analysis. This makes no difference to our conclusions or final plots. Hopefully the referee will now allow us to continue to treat this as a dwarf galaxy sample, in the purest sense, by size, magnitude, mass and M/L. Our fourth paper in the series addresses the differences between dwarfs and disks in more depth, and in that paper we will be arguing that dwarfs and disk are clearly visible as two separate populations in the size-luminosity portion of the fundamental plane. However, for the sake of getting the HI profiles into the literature in advance of that paper, we will trim the sample down to only those objects who are without any reasonable doubt, dwarf in metric size, dwarf-like in morphology and dwarf in luminosity. We have removed any comment, note or discussion of the sample in the context of other dwarf surveys, and we have eliminated any comparison of our work to other work on dwarfs. Confusion: The fear here seems to be that the sample is highly contaminated by nearby sources. It is extremely unlikely that we could have selected fuzzy things off of the PSS-II, gone to the telescope, pointed and found a nearby, HI-rich galaxy to be detected instead of the selected source. It is well known for many years that Arecibo is not confusion limited. We appreciate the referees concern for the accuracy and quality of the data, but many of the irregularities noted by the referee in the HI profiles are due to low S/N, poor coordinate knowledge or multi-sources as noted in the table notes. The examples the referee notes (D409-7, D409-19, D475-3 and D476-3) are interesting and we apologize for dismissing them in the previous report. Although we dispute his/her claim that the profiles are irregular in light of their S/N. D409-7 and D475-3, for example, are fine examples of a classic double-horn profile. With respect to confusion, the referee notes that D409-7 has an object (SRGb 062.12 from a poor cluster survey) and D409-19, D475-3 have two objects from the 2MASS extended object catalog, according to NED. Quick inspection of the NED DPOSS images demonstrates why they are not mentioned in the notes to the tables. SRGb 062.12 is, according to NED, clearly a late-type galaxy, S0-like, and would not be expected to be detected at Arecibo. In fact, its redshift is known from an optical spectrum (a poor cluster is rich in S0's and ellipticals). In addition, D409-7 was not a member of the optical sample, there was no CCD imaging, therefore any comments on confusion would be based on a judgment made will inspecting the plate material. That judgment was that there were no HI-rich objects in the field. Future researchers are welcome to search around our objects, however, the primary goal of this paper is to present the data. The two 2MASS objects are from the recently released extended source catalog. Although they are marked as `galaxies' in the NED catalog, they can be classed as extended for a number of reasons. They may be unresolved double stars. They may be a very faint background galaxies near a star. In fact (since one of the authors was a 2MASS team member many years ago), the 2MASS objects should be classed as `non-star' rather than galaxy. Again, inspection of the NED sky images reveals these are clearly very small, elliptical-like (if galaxies at all) sources and not something to be concerned about in a low redshift, HI survey. D476-3 looks like a double galaxy on the NED images, but it is one object on high contrast plates with an irregular morphology and two knots about 45 arcsec apart. This was noted in a previous paper and appears to look like two objects in the NED image; but it is one and fits in the Arecibo beam. It is also not a member of the mass subsample, there is no CCD image to explore its morphology further. While these other objects would have been interesting to add to our table notes, they do not have a bearing on detection or confusion at Arecibo. We also note that these catalogs are very recent and were published as our paper was submitted to the ApJ. We apologize for not including them for interest, but we feel this is the role of NED and other astronomical software. For the sake of expediting this dataset for publication, we have performed an extensive search of the NED dataset (at the cost of dozens of manhours). We have expanded the table notes to include any nearby galaxies, within 10 arcmins of our LSB objects, with the exception of 2MASS, radio and QSO sources. The 2MASS data gives no indication of size or distance, and an near-IR survey is going to highly biased to finding ellipticals and S0's (i.e. HI-poor objects). We hope the referee agrees to our request to exclude them. We also note that any competent observer can check NED during follow-up work. We wish to point out that, in particular, none of the objects used in the dynamical section of this paper has any confusion associated with it, nor are they Malin objects or background spirals. The mass subsample was culled by size and morphology to be dwarf-like, and CCD images were obtained for luminosity and size measurements. Those same CCD frames had sufficient field size to resolve any confusion problems. High M/L sources With respect to D582-1, we do not understand the referee's comments. D582-1 is at 7065 km/s, D582-2 is at 4575 km/s. This is very clear in the HI profile figure. Although they are 1.9 arcmin apart, they are not near each other in 3D space. We do have an error in the inclination corrections for the mass calculations. We store the value of ellipticity (1-b/a) which, of course, relates to cos i. We have corrected the tables and apologize for the confusion. In addition, we accidentally carried over the SDK comment about the intrinsic shape of dwarfs (q=0.58). We derive no conclusion about the intrinsic shape of LSB dwarfs and simply follow the SDK prescription for mass estimates which explicitly uses the axial ratios. We have added text to note how these prescriptions and values might change for different assumptions of intrinsic shape and internal kinematics (rotation or not). Most importantly, we simple present the relevant measurables in our tables and apply a straight forward calculation. We outline our assumptions and future researchers are free to use our data, and different assumptions, to draw their own conclusions.
The next report contained the same complaints (single minded referee). However, there was hope. If we simply add to the tables a note that every source might have a background galaxy we were home free! This seemed dumb, so we tried another approach.
March 21, 2000
Dr. James M. Schombert
Department of Physics
461 Willamette Hall
University of Oregon
Eugene, OR 97403-1274
Dear Jim:
Attached please find the referee's report on the revised version of
your manuscript entitled " Gas Rich Dwarfs from the PSS-II III.
HI Profiles and Dynamical Masses," ApJ MS50463. I know you're mad at me
for sending this back to the referee, but that person has made some
extremely useful comments, especially about confusion, that you should
consider in a revised version. The referee has been enormously careful
in going through your data and attending to his concerns *will* make
this a better paper, even if it is destined for the Supplements.
Please upload the revised version of your manuscript to the ApJ site
and send me a note detailing your response to the referees comments.
Continue to refer to ApJ MS#50463. Please do *not* e-mail me the manuscript!
Take care,
Sincerely,
John Huchra
Scientific Editor
===============================================================================
Title: Gas Rich Dwarfs from the PSS-II III. HI Profiles and Dynamical Masses
Authors: Eder & Schombert
The authors of this paper have made some definite progress from the
previous version.
The addition of more detailed notes on the individual sources is a big
improvement. The paper still has some problems, but they can be remedied
with a few more changes. The comments below are lengthy, but the changes
I recommend are not all that extensive, and I am replying as rapidly as
possible to help expedite the publication of this paper.
The authors' frustration in trying to get this "data" paper published is
obvious from their comments. Believe me that I too am frustrated at
having had to spend many hours refereeing this paper--much more than is
usually necessary in my experience. However, since they insist on
carrying along a weak analysis of mass-to-light ratios, I feel obligated
to point out the problems with that analysis.
Confusion:
The authors dismiss most of the confusion questions raised, saying "It
is extremely unlikely that we could have selected fuzzy things off of
the PSS-II, gone to the telescope, pointed and found a nearby, HI-rich
galaxy to be detected instead of the selected source. It is well known
for many years that Arecibo is not confusion limited." What they say may
be true for a random pointing, but actually it is well known that
galaxies tend to cluster, and dwarfs are frequently found in the
vicinity of larger galaxies. Confusion at Arecibo is very common for
these sorts of systems.
The authors state: "D409-7 and D475-3, for example, are fine examples of
a classic double-horn profile." They should look again.
The shape of 409-7 looks like two separate peaks superimposed, not a
classic double-horned profile. Since there is a second galaxy in the
beam with a similar velocity, there is a very good probability that the
dwarf is one of the horns and the nearby disk galaxy is contributing the
other. I assume the authors are aware that there are many examples of SO
galaxies with lots of HI, and even some ellipticals, so exclusion of
objects based on morphology is a weak argument. In the other galaxy,
475-3, the horns are highly asymmetric with a weak high-velocity
shoulder; in my experience this indicates a disk system observed far
(2-3 arcmin) off-center (probably comprising the strong low-velocity
horn and the high-velocity shoulder) and a narrow line source at an
overlapping velocity, which is probably the dwarf galaxy which the
authors intend to observe.
The authors also dismiss the 2MASS associations in NED, since 2MASS
extended sources are not always galaxies. I only pointed out the 2MASS
names since they were in NED, but as I said in my original report, these
were galaxies I saw on the DSS. They have angular sizes and separations
that made me strongly suspicious that they could cause confusion. The
more significant point is that I identified the HI profiles that looked
confused first, and then examined the DSS and found that the optical
images confirmed my suspicions.
What concerns me is that if these kinds of possibly-erroneous values
enter the literature without clear warning flags, they will create
continuing confusion for others who are even less familiar with
subtleties in HI profile interpretation. It is incumbent upon authors
publishing HI detections to provide that sort of expertise. Moreover,
the paper would be much more credible if it stuck to a clean sample for
claims about fractions of sources with double horns, dynamical mass
estimates, etc.
My specific recommendation is that the authors add a column to Table 1
with an asterisk (or something similar) to indicate sources where
confusion is possible. Even though there are written notes on many of
these sources, they are buried in a mass of other information, so the
typical reader is unlikely to realize that the HI profile parameters may
not be associated with the named source.
High M/L sources:
The authors replied "With respect to D582-1, we do not understand the
referee's comments. D582-1 is at 7065 km/s, D582-2 is at 4575 km/s.
This is very clear in the HI profile figure. Although they are 1.9
arcmin apart, they are not near each other in 3D space."
Apparently the authors are confused by their own nomenclature. The notes
in their previous draft gave a velocity for 582-2 that was close to
582-1, but they changed it in the current draft. The "582-2" I was
referring to is the one listed in NED, which is apparently "F"582-2. The
"582-2" in this paper is a much more distant galaxy, not 1.9' away as
the note now erroneously states. Since this nomenclature is likely to
cause confusion for other users as well, the authors should probably put
a "D" (or whatever letter they use to distinguish these various surveys)
in front of all of the galaxy numbers in the tables.
Note that D582-1 now has the highest M/L value in figure, certainly
belying the claim that "none of the objects used in the dynamical
section of this paper has any confusion associated with it."
Now that the inclination corrections are being applied correctly, I have
looked at the other high M/L sources to see if they support the claims
being made. D570-6 has the next-highest M/L, but it has an absurdly weak
HI profile for inclusion in the analysis here. This signal is weaker
than dozens of features ignored in the nominal baselines of other of the
HI spectra. D564-11 has a high surface brightness companion (a 2MASS
source in NED again), which maybe should be ignored or maybe not. Again
the analysis would be cleaner if this type of suspect source was
omitted. (The preferable alternative would be to observe at the
positions of both sources to determine which the HI is associated with,
but it is too late for that.) The next-highest M/L source, D646-11,is a
very low-luminosity source which has a Gaussian profile, and for which
it is clearly inappropriate to be using the same dynamical mass formula
as the rotating sources.
I am trying to make a few points here. (1) There ARE problems with
confusion in the analysis. In some cases the confusion is only
suspected, but then these cases should be excluded until further
observations clarify the situation. (2) The sources with the most
extreme M/L values in the population are too questionable to adequately
support the abstract's claim that the M/L is as high as 200. (3) If the
formula being used is not necessarily appropriate, then the results
should be used in only a statistical sense, and even then only
cautiously. All of these problems raise serious questions about the
claim that "these dwarfs [are] the most dark matter dominated objects on
galactic scales."
My suggestion to the authors is that they refrain from such extreme
claims when their data are sometimes dubious. A much more believable
argument would be to take the average M/L for the double-horned systems
after dismissing any of the suspect objects. I think the results will
still point to this sample being an exceptional set of galaxies. As it
stands, the analysis presented has too many flaws to support the
authors' claims that these objects "reverse the trend for higher M/L
values" for larger structures, and that they are "the highest Mdyn/L
galaxies as a class of objects." Besides the problems I've pointed out
with some of their highest claimed M/L values, it's not clear whether
their higher mean M/L is just a reflection of using a larger
extrapolated HI radius than other authors have used.
Double-Horned Sources:
The authors quote statistics for double- vs. single-horned profiles that
are skewed in favor of classification as being double-horned. At the
moderate or low S/N levels found in most of these spectra, a
single-horned profile will usually end up looking asymmetric or
double-horned, both of which earn a double-horned label according to the
paper. By contrast, it is very unlikely that a double-horned spectrum
will end up looking single-horned in the presence of noise
This is a simple thing to simulate, which I have done by taking
single-peaked Gaussian profiles of similar width and resolution as in
this paper, and adding Gaussian noise. A surprisingly large percentage
of the resulting profiles appear to have two peaks or to be asymmetric.
I urge the authors to do the same kind of simulation for themselves
before drawing conclusions about this sample. It is clear to me after
doing these simulations that only a few of the profiles in the first
panel of spectra have a high probability of being double-horned (and two
of those I would blame on confusion).
Whether or not the authors accept this advice, they really must list
their classifications (double, single, or uncertain) in Table 1 so that
others can make comparisons with their results.
Other Problems:
The spectra in Figure 1 are not presented in numerical order.
Figure 8 is a little difficult to interpret since so many of the objects
fall into the lowest bin. It might be clearer if presented in
logarithmic intervals as are most of the other histograms.
Table 2 is missing a column of data.
Many misspellings, typos, duplicated words, and TeX errors remain:
"viral theorem"; "varys"; "beatdown of the soda machine"; "display
display"; "its" instead of "it's"; omitted subscripts; etc.
We respond with a plea for the editor to do something before the paper is watered down to nothing. Hopefully by adding another column to the main data table we can please the referee's pet agenda.
John - Here is my response to the 3rd referee report. There's a reason I asked for a different referee. It's because this ones comments on confusion are blindly stupid. I have shown this to four other HI researchers and they are completed befuddled on what the referee is attempted to demonstrate. And they are, quite frankly, very confused on why you, as editor, can't sort through this. I must insist that this be our final draft. The comments the referee makes are trivial and do nothing to improve the science of the paper. Please make a decision on whether to accept or reject the paper as it stands. Jim ------ We are pleased the referee believes we have made progress, and we hope that this will be the final round of review. Confusion: We are sorry that we appear to dismiss the referee's comments on confusion out of hand. We have a reason to dismiss them because we believe they are in error. Let me try and summarize the argument here. This is a survey of LSB things, picked by morphology and surface brightness. Almost all of the HI detections are between 2000 and 8000 km/sec (27 to 106 Mpc). The objects range from 30 arcsecs to 1.5 arcmins in angular size. The referee's worry/fear is that we are not detecting the LSB thing, but rather something else. The something else can *only* be a nearby, off beam something. Consider the two logical possibilities here: 1) There is some nearby HI-rich galaxy and the LSB thing is a dwarf associated with it (but the HI signal is from the primary galaxy). If this is the case, where the dwarfs are 0.5 to 1.5 arcmins, than we have missed some nearby normal galaxy which would be around 3 to 10 arcmins in angular size (the typical dwarf is 1/5 the size of a normal spiral, Sandage 1980) within the 10 to 15 arcmin side-lobe beam area. We may not have the sharpest eyes, but somebody would have seen these galaxies. If confusion is a serious problem for this sample, then we have a very large population of 3 to 10 arcmin gals which are not in any known catalog of galaxies. This is highly improbable. 2) The HI detection is from some other galaxy in the Arecibo beam, and the LSB thing is not associated with it. The referee produces a list of objects from NED. Inspection of these objects shows them to be things between 10 to 40 arcsecs in size (obviously, or they would be in the UGC). This means we have discovered a sample of spiral or elliptical galaxies, 10 to 40 arcsecs in size, between 2000 and 8000 km/sec. That would make these galaxies between 1 and 15 kpc in diameter! The smallest disk galaxies on record. We strongly object to this as an argument for confusion in our data. The referee continues to make a series of errors in his/her "redshift by eye" method where he/she compares the size of our LSB dwarfs with other galaxies (from NED) on the DSS. The DSS does not display the full size of the dwarfs due to its limited surface brightness levels. Thus, the referee is selecting background objects and confusing the apparent size off DSS images of our LSB dwarfs, which vastly underestimates their true angular size. The referee also focuses on the asymmetry of many of the profiles and how this also indicates confusion. There is an alternative explanation to the asymmetric profiles ... the galaxies are asymmetric. Most dwarfs have irregular optical and HI images and, thus, a majority of LSB and dwarf galaxies have asymmetric HI profiles simply due to the fact that the gas is distributed in an irregular fashion. While the referee states that it is their `experience' that this is due to confusion, it is our `experience' that this is a characteristic of LSB galaxies. We must also object to introducing subjective comments into our paper. For example, the referee wishes us to state that D409-7 is not a double-horned profile. It is. There is a high probability that this is a pair of LSB galaxies (as noted in the Notes to Table 1), but that is irrelevant to the classification of the HI profile. It is double-horned by the criteria that we have outlined in the paper. Much like a baseball umpire, we call them as we see them. If the referee has alternative classifications or explanations, we welcome them to collect VLA imaging and publish their own paper. However, we have stated our methods, and listed our results. This is what is mandated under the standards of the objective scientific method. We are at a loss on what to do to satisfy the referee's needs concerning confusion. Situations, such as D409-7, are noted in the Notes to the Table. While we could add a column to our Table indicating possible confusion or asymmetry, we have no idea what criteria to apply so that the referee won't complain we had missed this or that galaxy. In addition, what possible merit is there to adding a column to the Table when we are publishing figures (make that, attempting to publish) of the actual profiles for the reader to look at. The reader can decide for themselves what level of asymmetry or profile shape for their own research needs. The referee believes it is incumbent upon us to provide guidance to the reader. We believe the opposite, that we should present the data in a straight forward manner *without* imposing our own subjective interpretations in form of guessing why the data is as it is. The referee's comments on various individual objects are the seeds to future work with the VLA or optical spectroscopic but, quite frankly, have no place in this paper. On this philosophical difference, we must request the editor make a decision. High M/L sources: Concerning our `weak' analysis, we have stated the method that we use. This method has been used by numerous other authors. We'd love to hear about the referee's method of determining mass but, again, we have stated our method and published our analysis. If the referee has a superior method, then we strongly urge him/her to publish their own paper on the subject. Otherwise, the referee has not demonstrated to us where we are in error. His/her comments are laced with adjectives such as "suspect" or "feel" which may provide him/her with motivation to do follow-up work, but we do not understand how to incorporate these comments into our paper. Despite our above objection, we have culled the M/L sample of all objects which appear "suspicion" in a completely subjective fashion. We have no idea if this will satisfy the referee. We have eliminated any statements in the text that attempt to place these results in context to other work. We have rewritten the conclusion to use only short, simple, plain descriptions of the analysis plots. Double Horned Sources: We have added a column to Table 1 indicating our classification of single, double or unclassified profile shape. The simulations the referee suggests are interesting, but we must object (again) to overlaying the data with interpretation based on computer models. We strongly feel that the data, and classification, should be presented first. While it is true that classification may be skewed towards double-horned profiles, that is not the purpose of this paper and will be examined in the later analysis papers. We have stated our assumptions in the mass calculations concerning rotation and double-horned profiles. Anyone can take the data and reduce the value using a gaussian calculation. It is our data and our choice on how to calculate the mass (as long as we make this assumptions clear to the reader). Again, we invite the referee to write their own paper re-analyzing the data with their own assumptions. Other Problems: We no longer care about typos, TeX errors or figure problems. We believe the paper is sufficiently plain and uninteresting as it stands.
We basically have no edited out all the interesting stuff, which makes no difference because the analysis papers were already in press. Apparently we wore the referee down but not before he has a chance to place one more agenda item in his report.
June 5, 2000
Dr. James M. Schombert
Department of Physics
461 Willamette Hall
University of Oregon
Eugene, OR 97403-1274
Dear Jim:
Attached please find the referee's report on the second revised
version of your manuscript entitled " Gas Rich Dwarfs from the PSS-II III.
HI Profiles and Dynamical Masses," ApJ MS50463. Sorry this took so long.
Progress. The referee has only a few small corrections to suggest.
If you make these, you're home free. Patience is a virtue.
Please upload the revised version of your manuscript to the ApJ site
and send me a note detailing your response to the referees comments.
Continue to refer to ApJ MS#50463. Please do *not* e-mail me the manuscript!
Take care,
Sincerely,
John Huchra
Scientific Editor
===============================================================================
Title: Gas Rich Dwarfs from the PSS-II III. HI Profiles and Dynamical Masses
Authors: Eder & Schombert
With a couple of minor fixes, this paper is now in appropriate shape for
publication.
The comments by the authors in their reply are not particularly relevant
or accurate, and they remain confused about several points. However, it is
not worth responding in detail since they have removed the most
objectionable material from the paper.
I found two significant errors that need to be corrected:
(1) In the footnote for D582-1, the velocity of F582-2 is incorrectly given
as 4575 km/s instead of 7043 km/s. This is important because F582-2 then has
the same velocity as the D582-1, strongly implying confusion.
(2) The statement on p.18 that "sin i is the inclination equivalent to the
axial ratio of the outer isophotes of the dwarf galaxy" is a holdover from
the mistake in calculating inclinations that I pointed out a couple of drafts
ago. It would be sufficient to replace "sin i" with "cos i" (in the statement,
not the formula), although the authors may wish to state this in some
clearer way.
Once these changes are made, the paper does not need another round of review.
I add one further comment, since the authors ask, about dynamical mass
calculations. Obviously, dynamical masses are difficult to define because the
mass depends on the radius at which the measurement is made. I objected
to the claim that their sample showed the largest M/L of any sample of
galaxies because other samples usually measure dynamical masses in different
ways. The authors of this paper multiply their optical radii and therefore
their masses by 2.3 as a sort of statistical correction. Even among high
surface brightness galaxies many studies find HI extending to 4 or 5 times
the optical radius, so if a similar statistical correction was made to bright
galaxies' radii and masses (which is easily justified by a variety of other
mass indicators at large radii for nearby bright galaxies), it could just as
easily be shown that bright galaxies have much higher M/L values than is
usually quoted. Since the comparative statement was removed, my objection
is also removed.
Morale of the story, double check typos and request an editor who reads the paper.