Why do astronomers reject 72 km/s redshift quantization /periodicity?

[rtomes]

I stated my wish to give astronomers a grilling on why they reject some of Tifft's findings that have been replicated several times, in the same way that I was grilled on why I support them. Nereid suggested that I start a thread in Q&A on the matter. I am very happy to do that.

First a couple of distinctions that I think it is important to make.

A. When looking at distance periodicity or redshift periodicity it is important to distinguish between the two cases of periodicity in redshift and the various correlation functions that use all pairs of galaxies and compute distances between them based on 3 dimensions. If redshifts are quantized then the correlation functions destroy the periodicity at small scales. I can explain this further if necessary. It is my suspicion that astronomers have not understood this point and that is why Tifft and replicators have been ignored. I would like to see evidence of whether this is right or not.

B. There are differences between large and small scale periodicity. At large scales, such as the reasonably commonly reported 128/h Mpc, the periodicity is detected in both redshift taken alone and in correlation functions. This is not true of the 72 km/s periodicity which is only detected in redshift in certain preferred frames.

From the recent thread Quantized Redshift Revisited I thought that I found the following reasonable conclusions in this summary:

1. That large scale redshift periodicity and correlation function distance periodicity seems to be well established and accepted by most astronomers as real (i.e. Broadhurst etc 128/h Mpc). Please tell me if this is not so.

2. That small scale distance periodicity as measured by correlation functions is not found. I agree with this.

3. That small scale (72 km/s etc) redshift periodicity is found provided certain constraints are followed:

a. That the data is sufficiently accurate. Tifft has shown that +/-18 km/s is the maximum tolerable uncertainty as larger errors will wash out the periods. I think more accurate is desirable.

b. That the periodicity is only true in certain reference frames as found by Tifft and others. These include one near the galaxy centre frame and one near the CMBR frame. I primarily am interested in the galaxy centre frame, because if galaxy centre redshifts are co-ordinated then we certainly would expect that to be the right frame to choose.

c. That the reference frame can be relaxed when comparing pairs of galaxies or tight groups which will have a common adjustment needed from heliocentric to galactocentric velocities. If this were not true then the 72 km/s quanta would never have been discovered in the first place.

4. That the galaxies included in the sample be restricted to spiral galaxies and not dwarf or irregular galaxies. Different quant may apply to these but the results are not replicated as far as I know.

I will mention here the original and two replications that I know of.

APJ 268:56-59 1983 May 1, Redshift Quantization in compact groups of galaxies, W J Cocke and W G Tifft (actually there is a better Tifft paper than this one which I am having difficulty finding).

APJ 345:72-83 1989 Oct 1, Periodicities in Galaxy Redshifts, Martin R Croasdale

J.Astrophys.Astr. 1997 18:455-463, Quantized Redshifts: A Status Report, W M Napier & B N G Guthrie

These papers find the results replicated in various ways at levels of p<.001 and p<.0001 on totally new samples and so are seriously difficult to dismiss as chance.

Additional replications of part of the results.

APJ 385:32-48 1992 Jan 20, Velocity Differences in Binary Galaxies I...., Stephen E Schneider and Edwin E Salpeter

From the Napier and Guthrie paper:

[01101001]

From the Department of Amusing Statistics (via Unix wc):

Question (from title):

1 line
9 words
70 characters

Statements:

39 lines
617 words
3699 characters
plus graphic image, 52561 bytes

[mugaliens]

rtomes, I think I see your point. Mine would be that one can average symmetric wave functions, but trying to apply that to nonsymmetrical functions induces error, without accounting for the nonsymmetry. ie, a sinusoidal wave verses a sawtooth wave.

[Nereid]

I stated my wish to give astronomers a grilling on why they reject some of Tifft's findings that have been replicated several times, in the same way that I was grilled on why I support them. Nereid suggested that I start a thread in Q&A on the matter. I am very happy to do that.

First a couple of distinctions that I think it is important to make.

A. When looking at distance periodicity or redshift periodicity it is important to distinguish between the two cases of periodicity in redshift and the various correlation functions that use all pairs of galaxies and compute distances between them based on 3 dimensions. If redshifts are quantized then the correlation functions destroy the periodicity at small scales. I can explain this further if necessary. It is my suspicion that astronomers have not understood this point and that is why Tifft and replicators have been ignored. I would like to see evidence of whether this is right or not.

B. There are differences between large and small scale periodicity. At large scales, such as the reasonably commonly reported 128/h Mpc, the periodicity is detected in both redshift taken alone and in correlation functions. This is not true of the 72 km/s periodicity which is only detected in redshift in certain preferred frames.

From the recent thread Quantized Redshift Revisited I thought that I found the following reasonable conclusions in this summary:

1. That large scale redshift periodicity and correlation function distance periodicity seems to be well established and accepted by most astronomers as real (i.e. Broadhurst etc 128/h Mpc). Please tell me if this is not so.

2. That small scale distance periodicity as measured by correlation functions is not found. I agree with this.

3. That small scale (72 km/s etc) redshift periodicity is found provided certain constraints are followed:

a. That the data is sufficiently accurate. Tifft has shown that +/-18 km/s is the maximum tolerable uncertainty as larger errors will wash out the periods. I think more accurate is desirable.

b. That the periodicity is only true in certain reference frames as found by Tifft and others. These include one near the galaxy centre frame and one near the CMBR frame. I primarily am interested in the galaxy centre frame, because if galaxy centre redshifts are co-ordinated then we certainly would expect that to be the right frame to choose.

c. That the reference frame can be relaxed when comparing pairs of galaxies or tight groups which will have a common adjustment needed from heliocentric to galactocentric velocities. If this were not true then the 72 km/s quanta would never have been discovered in the first place.

4. That the galaxies included in the sample be restricted to spiral galaxies and not dwarf or irregular galaxies. Different quant may apply to these but the results are not replicated as far as I know.

I will mention here the original and two replications that I know of.

APJ 268:56-59 1983 May 1, Redshift Quantization in compact groups of galaxies, W J Cocke and W G Tifft (actually there is a better Tifft paper than this one which I am having difficulty finding).

APJ 345:72-83 1989 Oct 1, Periodicities in Galaxy Redshifts, Martin R Croasdale

J.Astrophys.Astr. 1997 18:455-463, Quantized Redshifts: A Status Report, W M Napier & B N G Guthrie

These papers find the results replicated in various ways at levels of p<.001 and p<.0001 on totally new samples and so are seriously difficult to dismiss as chance.

Additional replications of part of the results.

APJ 385:32-48 1992 Jan 20, Velocity Differences in Binary Galaxies I...., Stephen E Schneider and Edwin E Salpeter

From the Napier and Guthrie paper:

I'd like to address an implied question in the thread title: "Are galaxy or quasar redshifts quantized?"¹

It is extremely easy to show that they are not: for any redshift range that encompasses 3 times the quoted uncertainty/error of any observed redshift of any quasar or galaxy (or extra-galactic PNe, SNe, GRB, globular cluster, ICL, intra-cluster gas, ....), can you find at least one object which has at least one component with a redshift in that range? If you can, then the only 'redshift quantization' must be occur at a finer scale, or among objects not yet observed.

AFAIK, no one has ever claimed there is such a quantization², so that's one part of the question answered³.

¹To be quite clear: this was NOT a question in the OP, only in the thread title.
²With the possible except of Tifft; as I understand his (ATM) idea, the quantization of redshifts can only be extracted from observations by analysis using a framework that incorporates features for which there is no proposed mechanism and for which there is no independent corroboration. IOW, a late 20th century variation on N-rays.
³Until such a thing were to be predicted from a serious new theory, or if someone published a paper showing that extra-galactic objects do have such observed quantized redshifts.

[rtomes]

I'd like to address an implied question in the thread title: "Are galaxy or quasar redshifts quantized?"¹

It is extremely easy to show that they are not: for any redshift range that encompasses 3 times the quoted uncertainty/error of any observed redshift of any quasar or galaxy (or extra-galactic PNe, SNe, GRB, globular cluster, ICL, intra-cluster gas, ....), can you find at least one object which has at least one component with a redshift in that range? If you can, then the only 'redshift quantization' must be occur at a finer scale, or among objects not yet observed.

You have lost me. Can we just deal with the case of spiral galaxies and ignore all the other types? If I try to follow what you say it goes like this - please correct my understanding.
I find a galaxy with redshift quoted as 6834+/-14 km/s.
Three times the quoted range is +/-42 km/s.
So are we now looking in the range 6834+/-42 km/s?
You say that if any other object is in that range then it proves there is no quantization?
I say that if another object is at 6841 km/s as an example, it would be totally consistent with quantization wouldn't it?

But I would suggest that periodicity is the correct criterion, by which I mean a tendency for being nearer a regular 72 km/s interval. The Napier and Guthrie graph shows such a case. Every peak has about twice as many cases as the troughs.

AFAIK, no one has ever claimed there is such a quantization², so that's one part of the question answered³.

¹To be quite clear: this was NOT a question in the OP, only in the thread title.
²With the possible except of Tifft; as I understand his (ATM) idea, the quantization of redshifts can only be extracted from observations by analysis using a framework that incorporates features for which there is no proposed mechanism and for which there is no independent corroboration. IOW, a late 20th century variation on N-rays.
³Until such a thing were to be predicted from a serious new theory, or if someone published a paper showing that extra-galactic objects do have such observed quantized redshifts.

I think that raising N-rays is an insult to the fine work of the people that I quoted. Some of them started out not believing Tifft and did a study with the intention of proving him wrong.

[Nereid]

I'd like to address an implied question in the thread title: "Are galaxy or quasar redshifts quantized?"1

It is extremely easy to show that they are not: for any redshift range that encompasses 3 times the quoted uncertainty/error of any observed redshift of any quasar or galaxy (or extra-galactic PNe, SNe, GRB, globular cluster, ICL, intra-cluster gas, ....), can you find at least one object which has at least one component with a redshift in that range? If you can, then the only 'redshift quantization' must be occur at a finer scale, or among objects not yet observed.

You have lost me. Can we just deal with the case of spiral galaxies and ignore all the other types? If I try to follow what you say it goes like this - please correct my understanding.
I find a galaxy with redshift quoted as 6834+/-14 km/s.
Three times the quoted range is +/-42 km/s.
So are we now looking in the range 6834+/-42 km/s?
You say that if any other object is in that range then it proves there is no quantization?
I say that if another object is at 6841 km/s as an example, it would be totally consistent with quantization wouldn't it?

Sorry that wasn't clear; let me try again.

Pick a number, any number, between 0 and ~3*. Let's use km/s, and choose 6834. Let's take the uncertainty as 14 (km/s). Our redshift range is thus [6813, 6855]. Is there an extra-galactic object, or component of such an object, with an observed redshift in this range?

Repeat.

...

If there are no gaps, then there is no quantization.

But I would suggest that periodicity is the correct criterion, by which I mean a tendency for being nearer a regular 72 km/s interval. The Napier and Guthrie graph shows such a case. Every peak has about twice as many cases as the troughs.

[snip]

Perhaps some time spent getting a common understanding of key terms would be helpful?

First of all, what is being measured? The OP is quite narrow in scope - only spiral galaxies with redshifts that are sufficiently accurately determined. What is the redshift of a spiral galaxy? How can it be reliably estimated? How many different ways can it be estimated? Perhaps we should look at these questions in some detail.

Second, what test (or tests) should be used to assess the existence of a tendency towards nearness?

Third, and perhaps most important, how should selection effects be estimated? and, having been estimated, how should they be incorporated into the tests of tendency?

*This is z; km/sec will do too. Note that a more precise statement needs to account for the paucity of objects with observed redshifts >~3.

[Nereid]

[snip]

First a couple of distinctions that I think it is important to make.

A. When looking at distance periodicity or redshift periodicity it is important to distinguish between the two cases of periodicity in redshift and the various correlation functions that use all pairs of galaxies and compute distances between them based on 3 dimensions. If redshifts are quantized then the correlation functions destroy the periodicity at small scales.

[snip]

(my bold)

As there are no quantized redshifts, for extra-galactic objects, there is nothing to destroy.

[Nereid]

[snip]

1. That large scale redshift periodicity and correlation function distance periodicity seems to be well established and accepted by most astronomers as real (i.e. Broadhurst etc 128/h Mpc). Please tell me if this is not so.

[snip]

Today, the consensus understanding is much richer and deeper.

The so-called 'large scale periodicity' is, today, well understood within the concordance cosmological model (LCDM).

Given the tiny, pre-inflation, fluctuations (that the WMAP team were able to characterise so well, from observations of the CMB), and GR, the universe evolved structure that is seen today as filaments, sheets, voids, clusters and super-clusters. Quantitative descriptions of this large-scale structure (P(k), the matter power spectrum, is how it is summarised by researchers today) are explained well by LCDM models with quite tightly constrained parameters.

Further, large simulations - such as the Millennium Simulation - reproduce this observed large-scale structure very nicely.

There is thus no need for any 'large scale periodicity' in modern cosmology.

[rtomes]

....
There is thus no need for any 'large scale periodicity' in modern cosmology.

So you are saying that large scale structure at some scale is accepted but it isn't accepted as periodic?

[rtomes]

...
First of all, what is being measured? The OP is quite narrow in scope - only spiral galaxies with redshifts that are sufficiently accurately determined. What is the redshift of a spiral galaxy? How can it be reliably estimated? How many different ways can it be estimated? Perhaps we should look at these questions in some detail.

I don't feel that I have a lot to contribute in this matter. I would say simply that the correct redshift should be the value that is found at the core of the galaxy and/or the middle point between the two extremes. No doubt there will be some galaxies that are weird and where the appropriate redshift is harder to determine. My answer to that is simple - do not include such cases in the analysis but use only determinations that can be made confidently to some accuracy.

Second, what test (or tests) should be used to assess the existence of a tendency towards nearness?

The Napier an Guthrie case shows a graph that makes clear what is going on. You can divide the phase of the period in half and see what proportion of the galaxies fall in each half.

Third, and perhaps most important, how should selection effects be estimated? and, having been estimated, how should they be incorporated into the tests of tendency?

*This is z; km/sec will do too. Note that a more precise statement needs to account for the paucity of objects with observed redshifts >~3.

I know that astronomers worry a lot about selection effects, but periodicity is not going to be affected by selection effects unless they are deliberate selection of cases that satisfy the periodicity. If this sort of selection affected whether periodicity existed or not it would simply prove that peculiar galaxies are periodically located. Selection effects relating to magnitude or colour will not cause periodicity to appear.

[Nereid]

....
There is thus no need for any 'large scale periodicity' in modern cosmology.

So you are saying that large scale structure at some scale is accepted but it isn't accepted as periodic?

I don't think modern cosmological research can be reduced to this minimal soundbite level.

If you regard modern (observational) cosmology as a scientific research program, then you can ask what are the questions the program is seeking to answer, and what is the conceptual framework within which it is operating (the 'paradigm', to use the overworked Kuhnian term).

From this perspective, an important question is something like 'to what extent is the (observed) large-scale structure of the universe consistent with {insert name of cosmological model here}?' P(k) is a powerful tool for mediating between observation and model, but is not, in principle, the only such. However, P(k) is certainly much more powerful than 'periodicity analysis' (or similar) in that it is more general (any periodic structure will show up in P(k) anyway).

OTOH, if the cosmological model being investigated predicts a strongly periodic structure, then P(k) might be too blunt, and some kind of periodicity analysis more appropriate. However, as there are no such (mainstream) cosmological models, it should not be surprising that this approach is not used.

Finally, re 'accepted': the observational basis for the existence of large-scale structure (LSS) is pretty overwhelming, and as new surveys reach completion, the granularity of our understanding of that structure just keeps getting better. Would you be interested in looking at some of these surveys, and how they shed new light on LSS?

[Nereid]

[snip]

I will mention here the original and two replications that I know of.

one) APJ 268:56-59 1983 May 1, Redshift Quantization in compact groups of galaxies, W J Cocke and W G Tifft (actually there is a better Tifft paper than this one which I am having difficulty finding).

two) APJ 345:72-83 1989 Oct 1, Periodicities in Galaxy Redshifts, Martin R Croasdale

three) J.Astrophys.Astr. 1997 18:455-463, Quantized Redshifts: A Status Report, W M Napier & B N G Guthrie

[snip]

Additional replications of part of the results.

four) APJ 385:32-48 1992 Jan 20, Velocity Differences in Binary Galaxies I...., Stephen E Schneider and Edwin E Salpeter

[snip]

It would be helpful to include a link to the relevant ADS entry.

I've added numbers to the above; here are what I think are the corresponding ADS entries: one, two (the only paper he wrote, apparently), three, four.

[parejkoj]

I would say simply that the correct redshift should be the value that is found at the core of the galaxy and/or the middle point between the two extremes.

Oh rly? I can has sum details?

Specifically, can you verify that the central velocity is the correct one? And can you tell me which would have worse errors: the core or the edge of a spiral galaxy (assuming long-slit or narrow fiber optical spectroscopy)? And can you tell me why?

I know that astronomers worry a lot about selection effects, but periodicity is not going to be affected by selection effects unless they are deliberate selection of cases that satisfy the periodicity. ... Selection effects relating to magnitude or colour will not cause periodicity to appear.

Oh rly? I can has detailed analysis?

Suffice it to say, you are absolutely, 100% wrong here. Selecting anything based on magnitude does cause clumping. I put up a plot of SDSS galaxies in some previous thread that shows "obvious periodicity" due to selection. I can't find the thread, and won't bother to reproduce the plot--if someone else finds it, please link it. I can think of a number of different ways "periodicity" can appear in a magnitude selected sample. And even more ways in a color selected sample.

Explain to me the difference between a magnitude limited sample and a volume limited sample, and why it is vitally important to use one instead of the other, and then we'll talk.

[Nereid]

[snip]

I know that astronomers worry a lot about selection effects, but periodicity is not going to be affected by selection effects unless they are deliberate selection of cases that satisfy the periodicity. If this sort of selection affected whether periodicity existed or not it would simply prove that peculiar galaxies are periodically located. Selection effects relating to magnitude or colour will not cause periodicity to appear.

Just quickly, on this (and to add to what parejkoj wrote): not only is this bald assertion easily shown to be wrong with synthetic examples, but the complicated history of quasars, especially the first ~20-30 years, provides a spectacular, real-world example of just how easily selection effects (etc) can give rise to spurious periodicity.

rtomes, you have stated that you are no astronomer, and have not studied extra-galactic astronomy. If so, then please do not post this kind of sweeping, bald, ill-informed assertion; far better to ask a question, such as "are there any examples, from recent astronomical history, of selection effects giving rise to the appearance of periodicity (which later turned out to be artifacts)?"

[rtomes]

rtomes, you have stated that you are no astronomer, and have not studied extra-galactic astronomy. If so, then please do not post this kind of sweeping, bald, ill-informed assertion; far better to ask a question, such as "are there any examples, from recent astronomical history, of selection effects giving rise to the appearance of periodicity (which later turned out to be artifacts)?"

Well perhaps I misunderstand what you mean by selection effects. I assume that you refer to unintended selection effects due to such things as brightness, colour, and the like. The probability of selection due such effects will be a generally smooth curve that has few inflexion points over the range of redshifts. That means that it will have only low frequency components. If a sample of galaxies has z in the range 0 to 1 say, then in terms of a 72 km/s period that represents 4000 cycles of the period. I do know about cycle data analysis and can safely say that selection effects of the sort I just mentioned will not produce even tiny amounts periodicity at such high frequencies.

If you have mean something different by "selection effects" or have some evidence of selection effects that do affect results when there are thousands of cycles present in the data then I would be interested to know about it. Or stated another way "are there any examples, from recent astronomical history, of selection effects giving rise to the appearance of periodicity (which later turned out to be artifacts)?"

[Nereid]

Well perhaps I misunderstand what you mean by selection effects. I assume that you refer to unintended selection effects due to such things as brightness, colour, and the like. The probability of selection due such effects will be a generally smooth curve that has few inflexion points over the range of redshifts.

And you have determined this "probability of selection" how?

Was it, perhaps, an assumption, made without any attempt to check any of the astronomical literature?

Or, perhaps, a robust conclusion arrived at after a detailed study of thousands of papers?

I suspect the former.

It bears repeating that selection effects (defined broadly) are the bane of many astronomical research programs, and the history of astronomy is full of examples of how tricky they are to account for (and, in some cases, to even recognise). It also bears repeating that there is a difference between recognising the existence of a selection effect and a robust method for measuring, and accounting for, it.

If you are interested in exploring this very broad, but important, topic in more depth, I recommend that you start another thread. In this one, from now on, I will be focussing on only those selection effects which have relevance to the remaining two items in the OP not yet covered (3. and 4.).

That means that it will have only low frequency components. If a sample of galaxies has z in the range 0 to 1 say, then in terms of a 72 km/s period that represents 4000 cycles of the period. I do know about cycle data analysis and can safely say that selection effects of the sort I just mentioned will not produce even tiny amounts periodicity at such high frequencies.

When you have completed your analysis (of HIPASS data? more??), written it up, and are ready to submit it for publication, I for one will be very interested to see how you addressed selection effects in "a sample of galaxies [that] has z in the range 0 to 1".

Until then, how about we see no more of these kinds of claims, here in the Q&A section of BAUT?

If you have mean something different by "selection effects" or have some evidence of selection effects that do affect results when there are thousands of cycles present in the data then I would be interested to know about it. Or stated another way "are there any examples, from recent astronomical history, of selection effects giving rise to the appearance of periodicity (which later turned out to be artifacts)?"

There's a big difference between the first part ("when there are thousands of cycles present in the data") and the last ("are there any examples, from recent astronomical history, of selection effects giving rise to the appearance of periodicity (which later turned out to be artifacts)?").

Off the top of my head, I can't think of any examples* of the former, in extra-galactic studies - can you give any?

Of course, there are many examples of 'multi-k cycles' studied by astronomers, with many different research objectives. Perhaps the best known is PSR 1913 + 16, the study of which resulted in a Nobel for Hulse and Taylor. Another example is a tight constraint on the acceleration of the solar system barycentre.

In any case, such a question is rather OT for this thread, so why not start a new one, here in the Q&A section?

*Other than pulsars in the LMC, M31, etc. I guess there might also be some other variables (in the LMC, M31, etc) that have been observed for long enough to have a few thousand cycles under their belts.

[Nereid]

Turning to the two remaining questions in the OP.

If I may rephrase them as:

a) What's the status, within the mainstream astronomical community, of the existence of a ~72 km/s periodicity in the redshift differences of (close) pairs of (isolated, spiral) galaxies*?

b) What's the status, within the mainstream astronomical community, of the existence of a ~72 km/s periodicity in the redshifts of spiral galaxies?

This post examines a).

Paper four is the only one (of the four mentioned in the OP) that we need consider wrt a), because it is more recent, examines the reported conclusion/results of paper one robustly, and includes analysis of new observations. Note that the scopes of papers two and three are different than that of question a).

Paper four finds a relatively weak ~72 km/s period in the Δredshifts for the pairs (a weaker signal than reported in paper one^). In terms of standard astrophysics, such a signal is not expected ... or is it? Paper four concludes "One can show that f(Δv) must be a monotonically decreasing function of Δv just by making the normal geometric assumption that orbits be oriented at random to a line of sight towards Earth. We find, however, that this assumption breaks down when standard selection effects are included."

This conclusion, concerning the monotonic nature of f(Δv), is explored in subsequent papers. I will look at these in later posts; for now a one line summary: when selection effects are understood and taken account of, f(Δv) is observed to be monotonic, and the weak ~72 km/s peak disappears.

*Strictly speaking there is a third question, concerning small groups, rather than pairs. However a ~72 km/s Δredshift periodicity for small groups is not, AFAIK, something covered in other papers.
^"On the other hand, our data is compatible with Tifft’s other suggestion of a nonmonotonic behavior of f(Δv) for velocity differences near 72 km s^-1. Our results is best described as a depression of in f(Δv) over the range of approximately 30-60 km s^-1, which appears moderately significant with the present data."

[Nereid]

"Galaxy redshift", or "systemic velocity": as ngc3314 mentioned in another Q&A thread, it is likely that the precision of a particular observational technique (HI, optical emission lines, absorption lines, ...) will be greater than the accuracy of an estimate of galaxy redshift or systemic velocity, where the latter is intended to measure a physical property of the galaxy as a whole.

Rotation Curves and Velocity Measures for Spiral Galaxies in Pairs (Keel) explains why this is so, in more detail, using a sample of 76 galaxy pairs.

Redshift data and statistical inference, a 1994 paper by Newman, Haynes, and Terzian, explores some traps in the statistical methods used to establish periodicities in redshift data. Among the examples they give, in the Introduction, is quantization of the redshifts of the Coma cluster galaxies. Tifft claimed to have found such quantization; later, Sharp looked at the techniques used, and concluded that "the evidence for anomalous behavior was not conclusive"; and later still, Newman, Haynes, and Terzian "explored the statistics of a refined set of Coma Cluster redshift data without finding any evidence of quantization".

This 1994 paper aims to "show that the derived random process preserves a subtle imprint of the original distribution, rendering the derived process nonstationary and producing a small but systematic bias in the usual estimate of the mean and variance." What does this paper mean, wrt the questions in the OP? Having alerted the astronomy community to the existence of these potential traps in statistical techniques used to analyse the data, those who reported finding redshift periodicities, after 1994, should have included the cautions in their analyses. Only one of the four papers cited in the OP (three) was published after 1994; however, it did not reference the 1994 Newman, Haynes, and Terzian paper. The authors of the three other papers seem to have been unaware of the potential shortcoming.

Of course, neither of these cautions renders findings of a ~72 km/s redshift periodicity completely invalid; however, they do suggest that establishing the existence of such a periodicity would require detailed investigation of several potential biases and systematic effects, and robust methods to show these have been accurately estimated or otherwise controlled.

[rtomes]

...
Rotation Curves and Velocity Measures for Spiral Galaxies in Pairs (Keel) explains why this is so, in more detail, using a sample of 76 galaxy pairs.
...

This paper abstract states "Individual scatter between the velocity indicators taken pairwise ranges from σ = 20 to 52 km s^- 1^."

Obviously you cannot tell whether there is a 72 km/s periodicity or not if redshifts are uncertain at such a level. However Tifft has made claims that HI measurements can be accurate to around 1 km/s. I am not certain which of his papers this is in. Perhaps he is using a specific one of the different measurements referred to by these authors or by you Nereid???

Sorry I have not been keeping up lately with things here. I have a very sore shoulder and neck and it means spending a lot of time at the computer is quite uncomfortable.

[Thanatos]

Consider this article on Wiki, which a fairly good review of the issue:
http://en.wikipedia.org/wiki/Redshift_quantization

[Spaceman Spiff]

The last paragraph of that Wiki review is especially noteworthy as to where things stand at present:

In 2006, Martin Bell and D. McDiarmid, reported: "Six Peaks Visible in the Redshift Distribution of 46,400 SDSS Quasars Agree with the Preferred Redshifts Predicted by the Decreasing Intrinsic Redshift Model". [25] The pair acknowledged that selection effects were already reported to cause the most prominent of the peaks[23]. Nevertheless, these peaks were included in their analysis anyway with Bell and McDiarmid questioning whether selection effects could account for the periodicity, but not including any analysis of this beyond cursory cross-survey comparisons in the discussion section of their paper. There is a brief response to this paper in a comment in section 5 of Schneider et al. (2007) [26] where they note that all "periodic" structure disappears after the previously known selection effects are accounted for.

See Section 5 and Figure 3 of the Schneider et al. (2007) paper, linked here.