Although a majority of astronomers / cosmologists seem to favour a Big Bang model, there is a significant minority that does not accept that redshift is mainly related to cosmological distance and due to expansion of the Universe. This minority includes Arp, Narlikar, the Burbidges and others. And me. ;-)

So it would be sensible to consider a definition of redshift in terms of various components that include both sides and then argue about which components actually exist. We actually observe wavelengths or frequencies of light rather than redshifts, so that the terms generally have the form (1+z) which is the relative wavelength and the various components are multiplicative. By this I mean that if there is a gravitational redshift and a velocity redshift and a cosmological redshift then the three (1+z) components must be multiplied together to get the final (1+z) that is observed. So here is a list of various proposed redshifts and a notation for them:

z_g Gravitational redshift. This is well understood and not disputed at smaller scales although there might be factors that can be debated at large scales in connection with say missing mass and galaxy rotation curves. However this is not in dispute in the current proposal.

z_v Velocity redshift. By this I will mean only local peculiar velocity of motion, not expansion of the Universe. The Doppler equations are known and not disputed by either side.

z_c Cosmological redshift. By this is meant a redshift that is proportional to distance and results from expansion of the universe, or in the alternative model from changes of particle mass over time (Narlikar Variable Mass Hypothesis).

z_i Internal redshift. This has been proposed by Arp as an additional component which is not accepted by big bang cosmology. If z_i can be shown to exist then it undermines standard cosmology because redshift then has an origin that is not due to expansion but some other causes. To be sure that z_i is not zero requires proving the real association in 3D space of objects with very different redshifts that cannot be explained by gravitation or velocity. I suggest that z_i > .01 gets suspicious and z_i > .1 is indisputable proof.

z_o Observed redshift. It is then expressed as:

(1+z_o) = (1+z_g) * (1+z_v) * (1+z_c) * (1+z_i)

To summarize, the above formula would be accepted by both big bang and alternative cosmologies, except that z_c has a different interpretation as to cause and z_i is always 0 in big bang cosmology.

This is a rare case where the big bang has less parameters than alternatives. It is worth mentioning that having more parameters confers an advantage on a theory which has nothing to do with its merits (it can be called "curve fitting"). Therefore it is reasonable that a high standard of proof be required to accept the additional parameter. In particular it would be useful also if the alternative theory had other measured parameters that correlated with z_i so that a reasonable and coherent explanation for its existence is offered.

According to Arp, most galaxies only have small values for z_i and this is certainly required so that scatter diagrams of redshift versus various measures such brightness are found to be decently correlated. For galaxies we might agree that in general, z_g, z_v and z_i are all very small, say <.001 typically and certainly <.01 in the vast majority of cases.

However for quasars the alternative model proposes that z_i may be very large, often of the order of 1 or even more. Quite clearly such differences cannot possibly be accommodated within z_g or z_v and so demonstration of quasars with such deviant internal redshifts would disprove so-called standard cosmology. It is worth mentioning that there are classes of active galaxies that fall between these extremes and might have z_i of >.01 and so be difficult to reconcile with big bang, but this proposal will deal only with the more extreme class referred to as quasars or QSO.

I do not intend to get into what the definition of a quasar is, leaving that entirely to the astronomers. It is not relevant to the arguments offered here. Now to those arguments, and a new proposal that I shall put forward.

On my web site I have a page which mentions this and I quote from there to give some references: (I will edit out some material)


This proposal is that there exists demonstrable real associations between galaxies and quasars that are at very different redshifts. That alone disproves the cosmological nature of quasar redshifts and totally undermines all the interpretations of the big bang.

In addition I propose a new test that will make this even clearer. The new test is to look at samples of quasars and galaxies that have very different redshifts (say > .01) and are very nearby in the sky (specifically that they are unlikely to be chance associations by statistical arguments) and to test the two models by the following procedure.

Make a scatter diagram of galaxy redshift versus quasar brightness.

Just to make it perfectly clear, the things being plotted are taken one from the quasar and one from the galaxy.

Consider the expected outcomes if each of the rival theories is correct.

Big Bang: If the galaxy and quasar are really at very different distances and not really associated with each other in space, then there is absolutely no real relationship between the galaxy redshift and the quasar brightness. The result should be that objects will be scattered over a rectangle with zero correlation coefficient.

Alternative: If the quasars really are associated with much closer by galaxies then the galaxy redshift is a better measure of the quasar true distance than the quasar redshift is. This means that the scatter diagram should be tighter than the scatter diagram of quasar brightness versus quasar redshift.

If either of these results happens then it is a very clear proof of that theory as regards the redshifts of quasars. There are other possible outcomes (intermediate correlations) which would indicate that both theories are wrong.

Would astronomers agree that this proposal is a valid test and a very clear result should be obtainable?

Would they agree that if the alternative outcome above is found then it does disprove the big bang?

__________________
Never attribute to malice what can be adequately explained by ignorance or stupidity.
Isaac Asimov

I'm not sure that it would be that simple. The deeper surveys have dramatically expanded the number of quasars at fainter redshifts. Even in some of Arp's more recent papers you find he has to apply magnitude cuts to identify the quasars he argues are associated with local galaxies. You also have to deal with the fact that in Arp's model, quasar luminosity increases as redshift decreases, so quasars at the same distance of a local galaxy would still have a range of magnitudes - perhaps too much scatter for deciding between the models as you've proposed.

But it wouldn't hurt to investigate it and see what you find. You also should look at a paper published in ApJ Letter last year by Morley Bell which contrasts the quasars - which don't seem to show much of a magnitude redshift relation, with radio galaxies - which do show a fairly tight magnitude redshift relation.

Quote:

Would they agree that if the alternative outcome above is found then it does disprove the big bang?

Quote:

You should also keep in mind that if quasars were shown to be local, that could be interpreted such that the universe expands and there is an underlying Hubble law upon which the quasar intrinsic redshifts would be superposed. So, I'd argue that the answer is "no" - even if there was compelling proof that quasars are local, the Big Bang theory could still be valid.

__________________
"The scientist who asks the right question reconnoiters a new patch of the unknown, and may, with luck, bring it within the constricted but expanding boundaries of the known."

~Timothy Ferris (The Red Limit) 1982

Hi Jim. What is new is just looking at one theme, galaxy-quasar associations, and a new proposal to make a test that should have an extremely clear result either in favour of the big bang or the alternative proposal of Arp. I don't think that anyone has ever thought of this proposal before. Specifically, to test galaxy redshift against a nearby quasar's brightness to see if it correlates. Regards, Ray

Yes, it is true that even in the Arp model there will be genuine line of sight apparent associations just as in the big bang model. There need to be measures taken in advance to address this sensibly. In particular, it seems that true associated quasars are at a spacing of the order of 50 kPc from galaxies in the alternative case. This is a reasonable basis for selection to see if it makes the scatter less of redshift versus brightness.

If a suitable survey is used, then the probable number of line of sight pairs can be computed from statistics. If the number found is much greater than that then it already favours Arp. Furthermore, in the alternative analysis, the scatter diagram should show two logical groups, the true associations and the line of sight ones. The true associations should be on a relatively tight line compared to the random line of sight ones.

By reducing the angular separation that is considered a pair, the result should move gradually from a mixture with more random associations to more real associations, thus making the scatter diagram tighter. These are all additional components of the test that make the result clearer.
If there can be found a way to make quasars have a tighter relationship then it must be that it is getting to something more fundamental. Logically, various other attributes can then be developed from quasar measurements to work out accurate indicators of true internal brightness.
Well, if you can show that huge redshifts can result without any velocity of expansion involved for one class of objects, then it makes sense to abandon it as an idea altogether. Many other problems are then solved.

http://arxiv.org/abs/0704.1631v1
http://www.journals.uchicago.edu/doi/abs/10.1086/522337
I assume that the above paper is the one that you refer to.

Just one more thought Jim. The papers I quote are not by Arp. One is by a team including Burbidge and Narlikar. Don't know if that is relevant. It is about making a sensible new test that should have particular outcomes for different cosmologies and seeing what results. Ray

Moderator Jim

Thank you for the link to the background discussions about the quasar/galaxy/red shift issue.

I looked at the links you suggested and found that the graphical information as presented by rtomes to be a valuable addition to the discussion.
Redundant? Perhaps. Clearly stated? Indubitably.

I also believe that the intent of rtomes is to provide background information for his theoretical model.

Again, thank you for the links. The background information in the lengthy “Arp” discussion will also provide some of the necessary caveats to any alternative model and is a valuable resource.

Thank you,

Snowflake

Sounds fair enough. Let's (everyone) try to stay on this topic, quasar-galaxy correlations and this specific proposed test.

It's the concept, not the originator or author, so it's not directly relevant.

__________________
Never attribute to malice what can be adequately explained by ignorance or stupidity.
Isaac Asimov

*sigh* Here we go again. I can't put together a full reply as I need to walk the dog and eat dinner, but I'll just make a few comments (of course, that usually means I'll spend much way too much time on this... oh-well).

As I've mentioned to you several times before, this is your first problem. "What a quasar is observationally" is absolutely relevant to the discussion: different selection techniques can overlap by less than 20%.

And here's something that should make you think a little bit: the difference between a "quasar" and an "active galaxy" in many catalogs is based on an absolute magnitude cut. Do you know what that means?

I really should try to finish that Quasar thread... If it'll be read...

As to the plot that you showed from "Associations between Quasi-stellar Objects and Galaxies", do you know how much more we understand about AGN since 1990, and how many more objects there are? Selection and identification of quasars prior to ~2000 was generally very haphazard, so I would put very little trust in such plots from before that time.

How will you select quasasrs that are "unlikely to be chance associations by statistical arguments?" The (optically selected, broad-line) quasar density on the sky is ~10 per square degree to i=19.1. That's quite a lot. It means one would expect >100 (optically selected, broad-line) quasars in the vicinity of, say, Andromeda (though M31 isn't part of the main SDSS survey, I hope you get my point).

Also notice the very important: optically selected, broad-line. Radio, X-ray and IR selection find different objects, and give somewhat different densities.

So, how you gonna pick 'em?

__________________
"What do you care what other people think?" -- Richard Feynman
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Feynman, at the conclusion of his Challenger report

Also, to just make a couple of quick comments about Bell 2007:

Dear FSM, how lazy was the reviewer? Seriously, how the heck (when mentioning this paper to a colleague, I used rather stronger language) did this make it into ApJ? It's terrible.

I've said elsewhere that Bell doesn't understand selection, and this absolutely proves it. He doesn't even understand what samples went into the catalog he used!

Hmm... If I go to SIMBAD and look at the names of the objects included in Véron-Cetty & Véron, 2006, I see several at the top of the list with "SDSS" in the name. Seems pretty clear to me. And I didn't even have to read their paper to determine that!

Strike one! Though, in truth, that's enough to completely chuck it, I'd say.

Huh... Really? This is news to me. Guess there were a whole slew of posters at the latest AAS meeting that were completely irrelevant...

That statement is true for certain classes of AGN, given certain types of observations. And it is certainly true for luminous (optically selected, broad line) quasars. But for AGN in general? Not for any commonly used definitions of the term. I'm not familiar enough with VC-V to know whether this is a safe assumption for their catalog, but considering Bell didn't even know whether the catalog included SDSS quasars (and it took me literally 2 minutes to check, most of which was waiting for SIMBAD to load), I'm definitely not trusting his word on it.

Strike two!

Really, those two should be enough to toss this one out on the curb. But, since I started counting strikes, here's a good choice for number 3: his use of Stickel et al. 1994 to define the "radio galaxy" sample. Certainly FIRST is a better choice for looking at radio sources: it is uniform (as opposed to a literature search) it goes down to 1 mJy (compared to 1 Jy), has 5" resolution and includes about a million sources (compared to ~600). And it overlaps completely with SDSS, so there are optical identifications with spectra for many of the sources (a very trivial and stupid check gives ~40,000 SDSS spectra of FIRST sources). I have no idea why someone would use a much older survey to define a sample cut.

Unless Bell isn't aware of FIRST.

Which is odd, because it's one of the best large area radio surveys in currently existence.

I sense a trend here...

That makes three. Can we burn it now?

__________________
"What do you care what other people think?" -- Richard Feynman
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Feynman, at the conclusion of his Challenger report

While this preamble contains many assertions that I would question, if only because they are such poor summaries as to be quite misleading, it is essentially irrelevant to the ATM idea(s) at the heart of this OP*, which is:As has already been noted, it is all too easy to show this assertion is non-science.

For now, I shall simply add that, as it stands, this proposal scrambles the general and the particular ("galaxies and quasars" - all galaxies? and all quasars? all galaxies and some quasars? ...), misuses the word 'prove' (astronomy, as a branch of science, is not mathematics), and that its conclusion ("totally undermines all the interpretations of the big bang", emphasis added) does not follow, even if its premises were to be established.
As a first, qualitative, draft, this seems ... sloppy, but perhaps workable.

Clearly, it needs a lot more work before it can be actionable.

(to be continued)

* Except for the 'definition of a quasar' ... this is, of course, central to the ATM idea, and will (no doubt) be visited many, many times in the life of this thread.

Would you please state whether you are prepared to answer questions on this, and address challenges to it?

For avoidance of doubt, if the answer is in the affirmative, I have many questions that I think are direct and pertinent.
Assuming this is not a rhetorical question, then you already have answers from two BAUT members who have papers published in relevant, peer-reviewed journals ... and both answers are in the negative (it is not a valid test (though it may be interesting), clear results are unlikely to obtainable (at least as the proposal is currently formulated)).

Personally, I think a random sample of astronomers would give answers that were similar ... though I suspect many would be couched in rather colourful (shall we say) language.

Let me ask a question though: if even questions on the consistency (etc) of definition of key terms ("quasar", "galaxy", for example) are off the table - much less consensus on those definitions - what grounds do you have for thinking that the proposal could be "a valid test and a very clear result should be obtainable"?
If no questions, or discussion, on core terms in the proposal are to be permitted, I should think they'd agree any outcomes could have no more than the most tentative of a status ...

Again, if no questions on, or discussions of, the definitions of the key terms are to be permitted, what grounds do you have for thinking that "if the alternative outcome above is found then it does disprove the big bang"?

I'm not sure if you used the link in rtomes' post to access the Bell paper, or read the actual published one.

ArXiv has two versions; rtomes' link is to v1.

Your three strikes apply to v2 (apparently), though I think strike 1 is a little more subtle than as you have written it .... in any case, the three strikes provide a nice, concrete example of one class of difficulties a concrete attempt to examine rtomes' proposal would encounter, and also shows just how difficult (if not impossible) it would be to reach "a very clear result" (rtomes' explicitly stated goal) if no discussion of definitions were permitted.

I read the published version. I would expect that the first version is worse (though I'm not really certain how that is possible).

I'm curious what you mean in that regard. It seems quite clear to me that Bell did not understand the catalog he used in writing the paper. That makes the entire thing unreliable, at best.

Garbage in -> Garbage out.

__________________
"What do you care what other people think?" -- Richard Feynman
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Feynman, at the conclusion of his Challenger report

Depending on your mood, I think you'll find it (v1) good for laughter or weeping (HINT: it is, IMHO, considerably worse than v2).The take-away from strike 1 remains the same; the subtlety comes from the fact that Bell seems to use 'AGN galaxies' inconsistently (or insufficiently clearly) in the paper - mostly it is clearly intended to include quasars (or does so explicitly), but sometimes he seems to regard them as a separate class of object from quasars (despite writing, on p3 of the arXiv v2 preprint, "the sources listed as quasars and AGN are clearly the same"). This seems to come from his use of VCVcat, which uses both terms. The part you quoted, in full, is as follows [my annotation is in square brackets]:

Ahhh... I see your point. I didn't pay enough attention to that paragraph. My bad.

So, at least he was aware that the catalog contained some SDSS sources. But not knowing whether the catalog contains "AGN galaxies from SDSS" which he never defines, or being aware that many of the "quasars" in SDSS have obvious galaxies around them in the SDSS photometry is nearly as bad.

So I'll need to replace the exact wording of that strike with something else, but I think you've made it quite clear what it is. Thanks Nereid.

__________________
"What do you care what other people think?" -- Richard Feynman
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Feynman, at the conclusion of his Challenger report

Either you did not read Bell’s paper carefully when you made the above claims or … well its better to just assume you didn’t read his paper carefully. Starting with your first Bell quote it is important to go back and add in the full statement that you snipped so that there is proper context:

Bell does know that the VCV contains SDSS quasars. You claim he doesn’t even though he says “It contains the quasars found in the SDSS …” in the very sentence you selected to argue he doesn’t. In addition, Bell was specifically saying that while the VCV contains SDSS quasars, Seyferts, and BLLacs, it is uncertain how many AGN galaxies from the SDSS survey are in the VCV - and he gives a reason. The color selection technique is biased against detecting AGN galaxies with low luminosity nuclei.

In addition, Bell starts section 3.1 with:

It is hard to understand how you could claim that “Bell didn’t even know whether the catalog included SDSS quasars.” And from this carelessness you then claim that the referee was lazy?! Actually two referee’s – not one:

Since Bell did know that the VCV includes SDSS quasars I don’t know what to make of this argument. Did you understand the context of Bell’s statement?

Again – having the full context of the statement you selected helps:

His argument was that we can be pretty sure that the AGN listed in the VCV are AGN as opposed to non-AGN. Was his argument correct? If not, or if not that simple, then in order to be a relevant argument against the content of his paper you would need to adopt the position that a significant portion of the VCV catalog objects listed as QSO’s and AGN are objects that may not actually be AGN – and therefore could potentially impact Figure 2 of his paper. Is that the position you’re taking? It’s your supposed strike, so it’s a very relevant question.

It would also help to know what AAS presentations you’re referring to. Do said papers present anything that makes it likely that the sample of QSO’s and AGN in the VCV are contaminated by a large portion of non-AGN?

Does any of this warrant a rejection of the Bell paper from publication? You see – you’ve twice now illustrated a questionable attitude toward refereeing this paper. First, you incorrectly argued that Bell didn’t know SDSS QSO’s were in the VCV and then suggested the referee’s were lazy and that the paper should’ve been rejected – based upon a snippet from a sentence in which Bell mentioned QSO’s from the SDSS. It is really not very good to have your critique contradicted by the very same sentence upon which you base your critique.

Then you make the error of suggesting a strike (and presumably basis for rejecting the paper) that you fail to justify with specifics. Implicit in the way you have made your arguments appears to be a notion that any debatable statement is grounds for immediate rejection of the paper. How about this – when the referee report is written, the referee asks the author to clarify or put some qualifiers on that sentence about AGN being easy to distinguish from other galaxies? Or … if the reviewer thinks that - contrary to what Bell was actually arguing - the VCVcat could be contaminated by a large sample of non-AGN, the reviewer ask Bell to go through the VCV sample and verify the objects are AGN – culminating in a re-plot of Fig. 2 after eliminating any contaminating non-AGN galaxies.

But rather than suggest improvements to the paper, you just throw out the whole paper based upon a flaw that you’ve not demonstrated is a flaw and may just be your own careless reading of the paper. What you’re illustrating is the kind of refereeing that gives refereeing a bad name.

Has anybody compiled a radio galaxy sample from FIRST? The original 1997 is cited 316 times in ADS. I didn’t really see anything among that list that is a sample of radio galaxies fitting the purpose of Bell – but I may have missed something so feel free to check for yourself. The Stickel et al paper is cited 109 times as recently as 2007.

However – let’s say he could’ve improved the radio galaxy sample with FIRST as opposed to the Stickel data. Do we reject the paper over that? How about asking the author to do the work of pulling a sample of radio galaxies from the FIRST data rather than just outright rejection? Would the FIRST data make a significant difference in the overall result? Shouldn’t we know that before we advocate rejection?

The refereeing process can involve constructive suggestions. When it doesn’t, if referee’s carelessly look to reject a paper rather than suggest improvements, it does a disservice to everyone. Yes, sometimes papers need to be rejected, but not with the types of arguments you’ve made. Your strikes were either incorrect or ideas that could be presented to the author as avenue’s to explore for improvement.

__________________
"The scientist who asks the right question reconnoiters a new patch of the unknown, and may, with luck, bring it within the constricted but expanding boundaries of the known."

~Timothy Ferris (The Red Limit) 1982

OK, there is a lot of interesting discussion and I accept that there may be some issues that are too tricky for me as a non-astronomer. I think that others are better to try and resolve those questions.

However I think that this is an interesting test because it is a new way of looking at it and directly tests big bang versus Arp and each has a specific outcome that is very clearly different.

Agreed that quasar definition must be a factor, but people are making catalogs based on some definitions. I think that such discussions should happen before making the catalogs. OK, you might agree with that but say it hasn't been done. In that case there needs to be agreement on what is definitely a quasar rather than a maybe. It should be a similar definition to what was used for quasars used by Arp when he did his earlier studies, because that is what we are testing. So any advice on how to achieve that would be most welcome.

Of course there are issues about how the universe develops and the changing nature of quasars over time. Both sides will agree that quasars are a stage of development but the details will be different.

For this proposal, I think it best to use relatively low redshift quasars that are relatively bright, but great care must be taken not to bias the results. If Arp is right that they are ejected from galaxies at much lower redshifts, then the galaxy sample can also be rather low redshift ones but with enough range to have the possibility of a sensible scatter diagram.

I think that the statistical problem of such a test are best solved by making a model for each of the alternative theories using Monte Carlo techniques and reasonable assumptions that are agreed by that side. Such data should be able to reproduce existing scatter diagrams and statistics about number of objects by redshift. Then the models are each used in the Monte Carlo test based on the two alternatives - line of sight coincidence for big bang and real association for Arp model to make scatter diagrams in advance of the test being done. The Arp test would need to allow for some true line of sight coincidences, but these should vary with the closeness of the limit set for separation between galaxy and quasar. In the Arp case the real separation does seem to be ~50 KPc so setting an angular limit that varies through 500 KPc, 200KPc, 100 KPc, 50 Kpc should show a steady progression from many to few line of sight objects.

I had a further thought on the test that perhaps overcomes some of the statistical problems and means that even if brightness limits or redshift limits are used in selecting the sample, a sensible test can still be made.

If, however selected, a sample of galaxy-quasar observed to be close on a line of sight but with significantly different redshifts is used to make a scatter diagram of galaxy-redshift versus quasar-brightness, there can be a control test where the galaxies and quasars are shuffled to get different partners still in galaxy-quasar pairs but not together in the sky.

In the case of the big bang being correct, the shuffling should have no effect on the scatter diagram because they are all random associations to start with. In the Arp alternative case, the shuffling should produce a significantly different scatter diagram. So the difference between the these two scatter diagrams tells us which model is better. Very similar implies big bang is correct, very different implies Arp is correct.

I think that by this means many problems of how to select the sample are removed, because the argument is true however the sample is selected as long as it isn't hand picked by an Arp believer. It is still OK to use Arp logic to determine the sample selecting.

One more thought related to this, but an issue with modern catalogs which should be addressed even apart from this proposal. In large surveys that use optical fibers to do many simultaneous measurements of objects, there are limits on how close the fibers can be in the field. This means that quasars very close to galaxies are less likely to be sampled. This is a real bias against alternative cosmologies that needs to be addressed. It seems that as areas of sky are sampled multiple times to get repeat measurements there is a need to make sure that all objects are sampled at least sometimes. Is this fully appreciated and is it in fact practiced?

See my note to Nereid above: I missed that part of the paragraph when skimming the paper, and already admitted that fact. But Nereid makes the point that my comment is still valid, though in a different way. And Bell has a history of not understanding the data he uses: see my comment in this BAUT thread about his complete misunderstanding of the SDSS selection function.

I did not do a thorough reading of the paper. But it did not take a detailed reading to determine that it was a rather typical intrinsic redshift paper, in that the author did not understand the catalog that he used to draw his conclusions from. For something like this, definitions are vitally important.

It still means that Bell doesn't understand the catalog. I'd love to see Bell define the distinction between "Seyferts", "SDSS quasars" and "AGN galaxies from the SDSS."

Heck, I'd love to hear him define "AGN galaxy."

As I said, I'm not familiar enough with VCV (and cannot directly access A&A, though I'm trying to get a copy of the paper) to know the details of their catalog. Knowing what I do about the authors, I trust that their contamination fraction (from non-active sources) is low. But Bell gives no reasons to believe that is the case, besides his comment that it is easy to distinguish AGN from non-AGN. As I said, that is true for some classes of AGN, and some types of observations. Bell provides no reasons to trust that VCV contains only those sources that are easy to distinguish.

Keep in mind, that it is also very important to understand which AGN are missing from VCV, as that can have a very important impact on Bell's claims. He says nothing about this. I took his comment to read both ways: since they are easy to distinguish, VCV should be both low in contamination (it probably is) and complete (it probably isn't). I am aware of the context of his statement (only with regards to contamination), but he says little of value about completeness.

Not necessarily, but there are always posters about new methods of finding AGN. Bell's comment suggested that he believed AGN and "other types" (whatever that means) of galaxies are easy to distinguish. Sometimes yes, sometimes no. The onus is on him to convince us that VCV contains the easy ones, or was careful enough in excluding others. Even just a quote from their paper about their selection methods would suffice.

Again, I already admitted I'd missed that. It's rather embarrassing, but it happens. Need I grovel upon the floor?

I was not a reviewer for the paper, did not read it particularly carefully (I was too busy exploding over various bits of nonsense) nor do I particularly care about the results of it: I am already on record (see the link above, if not elsewhere) as stating that the intrinsic redshift crowd is essentially irrelevant in modern astronomy. This paper does nothing to assuage me of that notion.

If you have evidence to the contrary, I'd love to see it.

All of those are great suggestions. But as I said, I wasn't the referee, and I'm not refereeing it now, just pointing out some stupid things in it that make me think the whole thing is suspect. If you want to go and check on those points, feel free, but I certainly won't.

My central point is this:

Bell doesn't understand the data that went into his paper!!!

There are plenty of subtleties in AGN selection that could completely mess up his results. He discusses only a couple of the simplest, and completely ignores the rest. If he doesn't understand the creation of the catalog that he used to derive his results, why should I believe any of the results?

Define "radio galaxy" for me, and I'll tell you. I do know that more than 20,000 FIRST sources have unique matches to SDSS galaxies that have spectroscopy (SDSS DR4 - there are more in DR6, but I haven't bothered to count yet).

I have no problems with the Stickel paper itself: it looks quite useful for certain purposes. But it was also a very biased sample of sources, by the nature of how it was created. If one wanted to compare two, supposedly different classes of objects, wouldn't one want to use uniform catalogs?

And, Bell says nothing about whether any of the Stickel sources are AGN of any kind (many, if not most, of them are). If the Stickel sources look just like some of his "AGN galaxies" spectroscopically, why should we believe that they should be treated differently from the VC-V sources? Bell does so because it is convenient for his "model."

It was not my goal to referee the paper. But all of the things that I brought up should have come up in the review. And they make the results highly suspect, as I've noted above.

Do you believe otherwise?

__________________
"What do you care what other people think?" -- Richard Feynman
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Feynman, at the conclusion of his Challenger report

This is a very important aspect of large surveys. To give an example of the work that goes into this, check this page from the two degree field galaxy redshift survey. To properly test if your sample obtains significant results, you'd generate mock catalogs of the expected quasar distribution, taking into account the survey masks, make 'observations' of the mock catalog and compare to your observed results.

How important is it, to get a statistically robust conclusion, that the full, fine details of the test(s) be specified before any sampling, selection, etc is begun?

For example, if the researcher ('semotr', shall we say) haphazardly (or not) tries out various catalogues, selection criteria, statistical tests, and so on until he finds something that he feels - for whatever stated, unstated, conscious, or unconscious (or combination) reasons - is 'best' (or just 'good enough'), to what extent are any conclusions compromised?

Consider cases where semotr's path towards the choice of exact test(s) to be done avoids (scrupulously or not) any use of real data, the use of some real data; again whether such use is focussed, random, or anything else.

How important is it that semotr's process of deciding what test(s) to use (in all its full details) itself be fully public and documented?

What role should double blind protocols play, if any, in the conduct of the test(s), once semotr has decided what they will be?

Note that the things which go into any tests are fixed and immutable, in the sense that semotr cannot 'run the experiment again' (there's only one 3C273, for example).

Such discussions occur all the time when generating catalogs of objects. Astronomers are very much aware of the problems that can crop up when producing catalogs of observations. I've given you links in the past. Try looking up the SDSS quasar target selection and quasar catalog papers. Here's something to get you started.

There is, in the astronomical community. A quasar is a supermassive black hole that is accreting at a high rate (typically at or near the Eddington limit, though not always). The expected observational properties essentially come out of that definition (and its corollaries and caveats). One of the features that sometimes goes into the observational definition is an absolute magnitude cut (do you know what that means?), which is somewhat arbitrary, but also traditional, and generally for optical selection.

Why don't you ask Arp what his definition was?

Fair warning, though (and I've said this plenty of times before): there's more than one way to find quasars. Here's a nifty one.

matt.o gave an excellent reply to this (SDSS also has mock catalogs for quasars and galaxies). I'd like to point out a different direction, and it is the AAS talk I just linked above about the millionth quasar:

Of course, understanding how such a photometric catalog is created requires understanding what a quasar looks like spectroscopically.

But, addressing your more general question: you (or someone else proposing a similar idea) need to give "the mainstream" a reason to care about whether "quasars very close to galaxies are less likely to be sampled." If Bell's paper is any indication of the general quality of such arguments, astronomers will continue to not care.

In fact, we actually do have reasons to care: it's one way to find multiply lensed systems, measure the quasar-galaxy cross-correlation function and look for absorption systems (among other reasons). But, with a density of ~10 per square degree in SDSS, there is generally no problem selecting "enough" quasars: only about 10% of the fiber plugs were needed for quasar targets, and fiber collision for quasars was much less frequent than it was for galaxies.

__________________
"What do you care what other people think?" -- Richard Feynman
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Feynman, at the conclusion of his Challenger report

parejkoj, first let me say thanks for the great response!

I missed that response before I posted. Probably was in the process of wading through the 316 citations of the 1997 FIRST paper.

Not just Bell - is there a standard consensus set of definitions for the various types of AGN (QSO's, BLLac's, Sy1, Sy2, Sy3, Liners ...) that AGN researchers have adopted? What criteria? Obviously spectra are critical, but in the past researchers have arbitrarily define qso's as having visual absolute magnitudes <-23.0. I've seen objects in SIMBAD classified as QSO/Sy1.

Bell (or anyone else) can't be confused about it if there is a standard set of definitions put out by the AGN specialists. Certainly if you read enough papers, you get a general sense of what the AGN specialists work with - but perhaps given the huge volumes of data now available with SDSS, 2dF, 2MASS, FIRST ... somebody should put together a paper devoted to outlining for the non-AGN specialists just how to define AGN samples - and a few cautions about the selection biases in the various catalogs. If the AGN specialists struggle with it, why would we expect Bell to come up with his own definition. He's trusting the VCV classifications.


I didn't read it to mean that it is probably complete. In fact, his later statement that "but whether the current VCVcat contains many AGN galaxies found in the SDSS galaxy survey is unclear" suggests that he does not believe the VCV is complete. What he does say is that:

Now a potential contributing factor in the brevity of some of his explanations could simply stem from the fact that he submitted the paper to ApJ Letters - which has a 4 page limit. Perhaps that was the wrong choice, but if you're trying to keep the paper of the proper length for that journal, and you still want to get to the hypothesis you're testing, you have to be brief.

But he could've gone with ApJ and taken the time to provide more explanation too - sometimes an author wants the paper in Letters rather than the main journal.


This goes back to my earlier comment - he's essentially talking about Seyferts, BLLac's, QSO's, and radio galaxies. How are these things defined? If the major catalog devoted to AGN calls an object a Seyfert 1 galaxy, why shouldn't Bell trust that? The VCV has been around a long time and is repeatedly being updated.

Which AGN are easy and which ones are hard to identify? And under which types of observations (spectroscopically, in different wavelengths, at different redshifts???) How would the excluded AGN potentially change his figure 2?

And again going back to the question of when to reject a paper - what is wrong with presenting what one finds from the VCV? You've just asked some great questions - questions that could be investigated and presented in future papers. But nobody is going to ask those questions without some starting point. His paper would in his opinion I'm sure be some groundwork for further investigation. You have to start somewhere.

I'm not convinced his paper proves what he's suggesting, but I do find it interesting that the radio galaxy sample follows a fairly tight Hubble relation whereas the QSO's and other AGN seem not to. Even if the reason is not intrinsic redshifts, certainly understanding why this is the case is something that deserves an explanation. And it was noted in the 1972 Sandage paper Bell cites - the much larger VCV catalog doesn't change that - just adds more data points. Maybe the explanation is as mundane as selection effects or intrinsic luminosity spread rather than more exotic explanations such as intrinsic redshifts - but bell's paper points out that there is something to explain on some level.

Certainly not, I hadn't seen your admission yet when I posted that. The problem I had was that Peer review is too important to be taken lightly or carelessly - as anyone who has submitted or reviewed papers should understand. The peer reviewer may be asked to review a paper that he/she immediately takes a dislike to, but the reviewers job is to identify the flaws and make recommendations. Since reviewers can recommend revisions rather than outright rejection - the reviewer should make every attempt to do that first.

Granted BAUT is not peer review, but when you question the efforts of two reviewers and suggest the paper should not have been published (and apparently had worse to say to a colleage), you should not be careless in your reasons - and you also should be willing to assess whether or not the reasons you're proposing warrant outright rejection or not. That's all I'm saying.

Frankly, I don't understand the careless, quick to reject reviewer -and they're out there. Anyone that reviews a paper has gone through the process of submitting a paper. I'm sure most researchers must have a few rejections in their history. It takes guts to submit your work for critique. And when you get that e-mail from the journal editor saying "referee report" you don't want to open that and find that the reviewer didn't even carefully read your paper.

Intrinsic redshifts shouldn't be relevant in modern mainstream astronomy. There's not enough convincing evidence that intrinsic redshifts exist. But there is evidence enough for some to explore the topic. I fail to see the crime in that. As I pointed out to Ray Tomes, some - such as Bell - advocate that the universe expands and intrinsic redshifts are superposed on top of that expansion. If Bell and a few others want to see if there is additional evidence for that possibility - where's the harm?

But it is not clear that the "stupid things" you've pointed out are reasons for rejection. How many referee's does it take to hit every possible suggestion that is out there. Bell thanks two referee's for suggestions that helped improve the paper. V1 and V2 are very different.

One referee can say X and the next will say X is a bad idea because researchers in the field have differences of opinions. You've found a few things you question, but two referee's that ApJL sent the paper too didn't choose to question those matters - perhaps because of other things they focussed on.


It's funny because it reminds me of the saying in research - the only ones that trust the data are the people that didn't collect it because the people that collected it know all of the potential pitfalls - or something like that.

Bell chooses to trust that the AGN in the VCV are what they are classified to be. Your issues with his understanding of the SDSS where quantization is concerned are a different issue - an issue that give you justifiable reason to be cautious. I can't tell you what to believe or not to believe. I don't care whether you believe or not. That's never been a huge concern of my where people on this board are concerned. Generally, I don't say much unless I see a serious error or mischaracteriztion.

What I can tell you is that if you care enough you'll carefully evaluate. If you don't care you won't. Either option is fine, but if someone claims that the paper should never have been published, then there is a certain amount of obligation to make sure the evaluation that led to that claim is careful, accurate, and based upon demonstrated flaws that are critical to the chief claims in the paper.


Since I'm not an AGN or radio galaxy researcher, I would have to rely on those that specialize in that part of the research field. Bell obviously was doing that when he chose the Stickel et al paper. This goes back to points I've made above.

Does the Stickel paper get into that? I'd have to go back and look. What would seem logical to me- since you read about QSO's that are radio sources and Seyferts that are radio sources, is that a radio galaxy should not be any other type of AGN - just a strong radio emitter - and of course the strength of the radio emission would probably be part of the definition. Whatever the experts choose as criteria will be adopted by the non-specialists if the specialists are explicit and consistent.

Yes and no. I do my best to treat all results as suspect until I've had a chance to find out what the potential pitfalls are. In this case, you've raised a couple issues that should be investigated, but I don't think that those issues would warrant outright rejection of the paper. It's not Bell's fault that you would like to know more about the VCV. Perhaps Bell could've looked into using the FIRST survey, but again that would be a follow-up type result. Does the FIRST data change figure 2 in any significant way other than adding data points?

__________________
"The scientist who asks the right question reconnoiters a new patch of the unknown, and may, with luck, bring it within the constricted but expanding boundaries of the known."

~Timothy Ferris (The Red Limit) 1982

Thank you. I try to be both contamination free, and complete...

Heh... That'd be great. Though I've never heard of type 3 Seyferts...

There have been "working definitions" and various authors have used their own definitions, which sometimes overlap. I just got ahold of a copy of the VCV paper, which includes a decent summary of some of the definitions in the Catalog Description section. I'd recommend giving it a look, if you are really curious.

The main point here, is that they are all accreting supermassive black holes, just with different accretion rates, viewing angles, intervening dust, wind strengths, jet strengths, etc. For example, the typical distinction between quasar and Seyfert 1 is an absolute magnitude cut (generally -23, though I think I've seen -22.5 somewhere), which is relatively arbitrary. The distribution is essentially continuous, unlike the distribution of emission-line widths, which has a trough around 1000km/s (roughly the dividing line between type 1 and type 2 objects).

But that's mostly talking about optical selection, which potentially misses a lot of sources. You need to include hard X-ray, far IR and radio selection as well (and possibly sub-mm, though that field is too young to be of much use, yet), if you want to build up a "truly complete" catalog of AGN.

Someone like Bell, who wants to completely overturn our current understanding of what AGN are, absolutely needs to understand these issues and take them into account. He does neither. This is why I basically dismiss him out of hand. The onus is on him to understand how these different effects affect his results, but he isn't even aware of most of them.

I don't know if you are aware, but intrinsic redshift stuff is pretty much the only thing that Bell does these days... If he can't spend time understanding the above issues, then I can't be bothered to do more than shake my head at his papers. Heck, the quick reading I did of this paper was probably more than it deserves.

Let me just quote from the VCV paper, and let's leave it at that:

That is the entirety of that paragraph. I'd say that Bell violates that comment pretty thoroughly, no?

As to your slew of other questions, I don't have time to fill them in right now, though they are good questions. Suffice it to say, there is good evidence that our current catalogs of nearby quasars (defined as the high end of the AGN luminosity function) are not particularly complete due to survey design, and the low end is much more difficult to identify (being that the galaxy light dominates, even in the spectroscopy). So just taking a literature search catalog like VCV and trying to derive results that depend fundamentally on both the high and low ends of the luminosity function is completely wrong.

Yes, and just grabbing a catalog and plotting things is not the place to start. I hope I've convinced you with what I've written above that the correct starting point is understanding the definitions and what goes into the catalogs. Bell does not do this.

I'd bet good money that it is due to the way the "radio galaxy" sample was created: sources with a radio flux > 1 Jy that match to a galaxy visible in the optical. Thus, they are all relatively nearby, relatively large, relatively bright galaxies. And, as I point out, many of the Stickel galaxies are probably hosting AGN. Bell didn't bother to check. Radio galaxy is not exclusive with AGN, especially because it is a purely observational definition.

As to the second point, AGN have a very broad range of luminosities, while often having very similar properties across luminosity bins. This generally makes for a very messy selection function.

I agree, in some cases (the same colleague that I complained about the Bell paper to is currently struggling with a "review done badly"). In some cases however, I see no problems with an outright rejection. Should a paper describing the "magical unicorn theory" of star formation be given any more than the shortest of readings? That's basically the level of the intrinsic redshift proponents, as viewed by experts in the field (and just so you know: I'm not one of them.... yet...).

They are welcome to do so. But they shouldn't expect anyone else to treat them as more than a comedy routine, at best. Until they bother to understand the data they are working with, astronomers will continue to (rightly) ignore them.

No idea. It depends on what you define as a "radio galaxy." Most are various types of AGN, or recently "expired" AGN. Some are due to strong nuclear star formation regions. I don't fully understand what the term "radio galaxy" referred to historically, besides an extended object with strong radio emission, and/or noticeable jets/lobes.

But that isn't a very useful definition these days: as I said, FIRST+SDSS produces >20,000 galaxies with spectra and radio emission > 1mJy. Those galaxies span a wide range of spectroscopic classes (star forming, LINER, Seyfert, passive), and thus the radio sources have a wide range of production mechanisms.

__________________
"What do you care what other people think?" -- Richard Feynman
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Feynman, at the conclusion of his Challenger report

Sounds like a backwards chap. ;-)

It is always desirable to put forward the exact basis of a test before it is done. In this case, I think it is desirable to get to a sensible method that both sides of the cosmological argument agree is fair before it is done. Who knows, it might even convince someone to change their mind!
As long as the method of selecting samples is not to hand-pick them, but to define such things as some combination of redshift ranges, brightness ranges and closeness of pairs (possibly as a relationship to redshift of the galaxy) then it is OK to make variations to these things when testing. These variations should be recognized as degrees of freedom available to the researcher and are properly allowed for in statistical tests.

It is also useful to do the tests on smaller samples to sort out the genuine issues of trying to exclude line of sight cases in the Arp theory. Then a test without freedoms can be done on a larger independent sample. That is the real test.
It must be fully publicly documented or it isn't science. It is desirable that it be done in advance as much as possible, but if it follows the outline of properties that I mentioned above then some variation in tuning is perfectly valid afterwards.
I don't see how you would double blind on this, can you explain. I thought that normal practice is a selection from a public catalog with specified selection parameters so that anyone can replicate the calculations.
The idea is to use a sample of thousands of quasars. I think that if Arp is right you will get a significant result with a hundred. If he is wrong it will be random how ever many you use.

In proposing the modified test with randomized sample as a control, I think that this addresses many of the potential problems with sample selection. If brightness or redshift limits are causing some bias, then the bias will be exactly the same with the randomized pairing. This control makes for a test that is very robust against selection effects.

How about a first draft of an actionable proposal then, rtomes?

At this stage, I expect several aspects would be constrained to ranges only, but several others should be quite explicit.

For example, what statistic(s) to use? What criteria to use for rejection of outliers (if any), or masking, or ...? How to construct mock input catalogues (should such be developed?)?

These - or at least a significant subset of them - can be worked on independent of the (likely very, very thorny) issues to do with selection, sampling, biases, etc.

Of course, we also need to be mindful of well-established* astronomy and physics. For example, if at least a subset of quasars are at cosmological distances, how to handle strong lensing (multiple images, only one real quasar)? weak lensing ('on the sky' displacements, magnification)? how to address variability? extinction? reddening?

And how about trying to come up with some tests, using the real universe, that are independent of quasars, 'controls' if you will?

This is your ATM idea rtomes, what's your timetable for rolling up your sleeves and actually doing some hard yakka?

* For the purposes of this thread, an operations definition might be whatever Burbidge does not explicitly reject, of what can be found in standard textbooks

Yes, it is an arbitrary brightness limit that depends also on the redshift being really a cosmological distance indicator.
If it was haphazard then it should not lead to such a structure. You should not get a good relationship between galaxy redshift and galaxy-quasar angular separation from haphazard data.

As far as possible I want a definition of quasar that is compatible with what was used then. By all measn be more refined, but I wouldn't want a sample that overlaps what quasars were considered to be then by only 20%.
OK, you also need the number of galaxies to do the statistics. If you have 10 quasars and 100 galaxies per square degree for example, the chance of a given quasar having a galaxy (at a different redshift) within 6' is 1/10. If we look at closest galaxy with a true 50 KPc separation in mind for the Arp model, then 50KPc is 6' at 30 MPc. That is around just z=.04 so we are considering galaxies in the range z=.04 to just several times that, no need to include galaxies at higher than z=.2 to test Arp as if he is right we will not see quasar companions for them. Even at z=.2 we are down to 1.2' or arc separation for true companions and 1/2500 probability of being that close. This is all rough, but gives an idea of the thinking for the statistics.
They should be picked as nearly as possible to what were picked as quasars in the earlier days when the mentioned work was done, consistent with getting unbiased samples.