What is the reason arXiv gives for rejecting the papers?Originally Posted by JMB
OK, but It becomes more and more difficult to publish on the CREIL. Even arXiv rejects the papers now.
Order of Kilopi
What is the reason arXiv gives for rejecting the papers?OK, but It becomes more and more difficult to publish on the CREIL. Even arXiv rejects the papers now.
Established Member
Not enough papers published in "good" reviews. (I publish about 1 paper each year in such reviews: most "good" reviews reject the papers without referee's comment.)
Order of Kilopi
There is a breakdown in logic here somewhere: How can it be determined whether a paper is important or not if it is not publicly available? Is it necessary to get papers published first, which do not contribute to the body of knowledge before someone can write a paper that introduces new concepts? Does this make the author more credible?Not enough papers published in "good" reviews. (I publish about 1 paper each year in such reviews: most "good" reviews reject the papers without referee's comment.)
This is a double tragedy because Jacques' best paper, on the nature of atomic particles as solitons, is not, to the best of my knowledge, publically available.
It is an honor to have Jacques sharing this concept with us on the BA board. (He addresses this on the "wave or particle" thread) Welcome aboard!
Established Member
cyrek comment
JMB wrote (Jerry's post above)
What is the reason arXiv gives for rejecting the papers?
Not enough papers published in "good" reviews. (I publish about 1 paper each year in such reviews: most "good" reviews reject the papers without referee's comment
reply
The key words here are 'without referee's comment'.
Since no reason is given, this tells me this is plain censorship.
Established Member
I'm not sure I would trust this. Distances for the Virgo cluster have inordinately large uncertainty, for one. The author doesn't seem to adequately address this fact, but is rather convinced that he knows what the distances are -- sometimes even omitting error bars. I don't know that I would even trust Cepheid measurements in cluster galaxies, but I'm not well versed on the details of this part of the distance ladder.
Order of Kilopi
Welcome to BABB Astronomy!
I'm the author of the paper.I'm not sure I would trust this. Distances for the Virgo cluster have inordinately large uncertainty, for one. The author doesn't seem to adequately address this fact, but is rather convinced that he knows what the distances are -- sometimes even omitting error bars.
The Virgo cluster has a large depth effect, and interactions may create some additional uncertainty in individual galaxy distances. The point of that section had very little to do with the actual mean distances to galaxies in Virgo. As you can see in Table II, morphologically divided subsamples all agree within +/-1.3 Mpc of the overall mean cluster distance.
The important point RE Virgo and the other clusters discussed is that late type spirals have significantly larger mean redshift than early type spirals - while at the same time having nearly the same mean distance.
Now in section 5 I discussed galaxies with extreme deviations between Hubble and Tully-Fisher distances. Table VII gives the uncertainty in distance modulus derived from data errors in HyperLeda. The point there being that the uncertainty in data is insufficient to account for the discrepancy between individual redshift and TF distances.
The potential importance of this result is that in the intrinsic redshift debate, the mainstream has always been able to fall back on the fact that distances to the objects in question (mostly quasars) had not been determined. The mainstream has also been able to claim that there is tight Hubble flow for normal galaxies. My paper tests the situation because galaxy distances are determined by a method independent of the Hubble relation. Section 7 of the paper discusses the failings of various potential mainstream explanations and Section 5 and Table VII clearly demonstrate that huge TF errors cannot be the reason for the results either.
I also have two more papers I'm nearly finished with that extend these results.
There is no reason to trust Cepheid distances in cluster galaxies any less than in field galaxies.I don't know that I would even trust Cepheid measurements in cluster galaxies, but I'm not well versed on the details of this part of the distance ladder.
Established Member
I think, if I were refereeing the paper, I would have had you omit them, then.
When you're dealing with nearby, noisy measurements of the Hubble expansion, an error bar like that makes a huge difference.As you can see in Table II, morphologically divided subsamples all agree within +/-1.3 Mpc of the overall mean cluster distance.
"Nearly the same" doesn't seem to be really addressing the issue well. I haven't looked carefully at the data, but since I'm not convinced we have the distance to Virgo down even to a factor of 2, I'm not convinced that your relative redshift with respect to the Hubble Flow is adequately measured.The important point RE Virgo and the other clusters discussed is that late type spirals have significantly larger mean redshift than early type spirals - while at the same time having nearly the same mean distance.
I'm not aware of people using T-F to get distances to the nearest cluster to us. The spread is just too big.Now in section 5 I discussed galaxies with extreme deviations between Hubble and Tully-Fisher distances.
I respectfully disagree. The data errors in HyperLeda are not including unknown systematics.Table VII gives the uncertainty in distance modulus derived from data errors in HyperLeda. The point there being that the uncertainty in data is insufficient to account for the discrepancy between individual redshift and TF distances.
Well, yes and no. There have been gestures in that direction, but the real controversy was that people thought such small objects couldn't be that bright. Accretion has been found to work well to explain this, so that orthodox argument no longer exists.The potential importance of this result is that in the intrinsic redshift debate, the mainstream has always been able to fall back on the fact that distances to the objects in question (mostly quasars) had not been determined.
The fact that we see host galaxies around a fair number of nearby quasars and the fact that we have Seyfert analogs means that there are very few in the community that still believe in the local quasar models, and their number is fast dwindling.
Only at medium distances. Nearby, the Hubble Flow is a notoriously poor distance measure.The mainstream has also been able to claim that there is tight Hubble flow for normal galaxies. My paper tests the situation because galaxy distances are determined by a method independent of the Hubble relation.
TF should not be used as a local distance measure because of unexplored systematics. Normal errors on this are not as tight as you claim them to be.Section 7 of the paper discusses the failings of various potential mainstream explanations and Section 5 and Table VII clearly demonstrate that huge TF errors cannot be the reason for the results either.
No, I'm saying in general, Cepheid distances are okay, but there is still some significant unexplained error on them, despite the claims of Hubble Key Project. The instablity strip is simply too unstable.There is no reason to trust Cepheid distances in cluster galaxies any less than in field galaxies.
Order of Kilopi
I would argue that is exactly why its important to look at the observed scatter in the TF distances derived from that data.Astronomy: I respectfully disagree. The data errors in HyperLeda are not including unknown systematics.
My sample reaches large enough distances that noise in the Hubble Flow is not the problem. With some of those galaxies you're talking about required peculiar motions from 3000 to 5000 km s-1 and distance modulus errors greater than 1.00 magnitudes. There is no indication from observed scatter of my samples for TF distance modulus errors that large.Only at medium distances. Nearby, the Hubble Flow is a notoriously poor distance measure.
Typical observed TFR scatter for cluster galaxies and the calibrators is ~+/-0.35 mag (Tully&Pierce, Giovanelli et al, Sakai et al). However, I corrected for morphological type which results in observed scatter of +/-0.22 mag or less. You can see that in part with the companions to the galaxies in question. The scatter between the companion distance moduli and the galaxies in question is less than 0.20 mag. You also have to keep in mind that I restricted the sample to galaxies with rotational velocities > 100 km s-1. TF scatter is significantly larger when you start including the slow rotators.TF should not be used as a local distance measure because of unexplored systematics. Normal errors on this are not as tight as you claim them to be.
I certainly have no objection to someone evaluating the distances to these galaxies. But I'd say its a difficult task to show that the results are caused by TFR errors. You really can't expect to see observed scatter less than 0.25 mag if the actual error from intrinsic TF scatter and data errors is 0.70 mag or larger. The HKP showed that with their samples.
Thanks for your input!
Established Member
I don't think that's correct. You cannot get a decent error bar simply by looking at an observed scatter from distances that have systematic errors.
Noise in the Hubble Flow is a problem well past Coma Cluster and probably past the Great Attractor as well.My sample reaches large enough distances that noise in the Hubble Flow is not the problem.
Again, there is no analysis of TF systematics. To do the analysis properly, one should appeal to the distance ladder directly.With some of those galaxies you're talking about required peculiar motions from 3000 to 5000 km s-1 and distance modulus errors greater than 1.00 magnitudes. There is no indication from observed scatter of my samples for TF distance modulus errors that large.
Including slow rotators is equivalent to including smaller mass galaxies which would be necessarily closer for a standard bias. I'm not convinced that you have an analysis that stands up to scrutiny. If I were a referee, I would tell you to use another distance modulus as well. Simply relying on Tully-Fischer isn't good enough.Typical observed TFR scatter for cluster galaxies and the calibrators is ~+/-0.35 mag (Tully&Pierce, Giovanelli et al, Sakai et al). However, I corrected for morphological type which results in observed scatter of +/-0.22 mag or less. You can see that in part with the companions to the galaxies in question. The scatter between the companion distance moduli and the galaxies in question is less than 0.20 mag. You also have to keep in mind that I restricted the sample to galaxies with rotational velocities > 100 km s-1. TF scatter is significantly larger when you start including the slow rotators.
I also contend that the morphological-type corrections are incredibly rough and unreliable. There is obviously a trend, but the quantifiability of morphological type is still a ways away. For now, Tully-Fisher is best used in consort with the entire distance ladder.
It's not so much caused by TFR errors but rather by morphology errors. It is difficult to quantify what the expected luminosity of a galaxy should be from first principles.I certainly have no objection to someone evaluating the distances to these galaxies. But I'd say its a difficult task to show that the results are caused by TFR errors.
If you, for example, were to use starburst galaxies or active galaxies (which can be of any morphological type) the systematics are enormous.
Yes you can. You mention the Malmquist bias which certainly is one consideration. Of other considerations are fingers-of-god effects and other redshift distortion effects which bring apparent scatter down. Read this and this and this and this article for more information.You really can't expect to see observed scatter less than 0.25 mag if the actual error from intrinsic TF scatter and data errors is 0.70 mag or larger. The HKP showed that with their samples.
Thanks for your input!
Order of Kilopi
My questions are: (1) How do you know there are any significant systematic errors? (2) What are the source of these hypothetical systematic errors?Astronomy: I don't think that's correct. You cannot get a decent error bar simply by looking at an observed scatter from distances that have systematic errors.
The observed scatter relative to the cepheid distances is ~ 0.35 mag for standard calibrations and less than 0.20 mag with my Type dependent calibrations. For the cluster sample, there was an observed scatter of 0.22 mag relative to mean cluster distances. Clearly the intrinsic scatter in the TF relationship is small. Why that should be is a matter for theorists.
At the distance of Coma, which is the TF distance for some of the galaxies in Table 7, a peculiar motion of 1000 km s-1 would introduce a distance modulus error of ~ +/-0.35 mag. The distance modulus differences exceed 1.00 mag in numerous cases. However, the galaxies in Table 7 are for the most part in lower density environments and should not be experiencing large peculiar motions - which means the Hubble noise should be even less than the 0.35 mag.Noise in the Hubble Flow is a problem well past Coma Cluster and probably past the Great Attractor as well.
The Cepheid calibrators were the basis of the TF calibration I used. This is exactly what was done with the major TF studies - cepheid calibrators were utilized to determine th zero point of the relationship.Again, there is no analysis of TF systematics. To do the analysis properly, one should appeal to the distance ladder directly.
I've compared the distances I get for clusters with distances from the Fundamental Plane and SBF studies - there are no systematic differences there.
As I said, my cluster TF distances are consistent with the FP and SBF distances. But independent of that - what other distance modulus would you suggest? The Tully-Fisher relation is it for determining distances to large numbers of spiral galaxies at larger distances.If I were a referee, I would tell you to use another distance modulus as well. Simply relying on Tully-Fischer isn't good enough.
I did not find it to be a significant problem in my May ApJ paper. About 5% of the galaxies turned out to have Sb/ScIII group morphology, but fit the cluster distance better with the ScI group equation. There were not galaxies that were misclassified the other direction. ScI morphology is pretty easy to spot because of the narrow arms. Also the morphological T-type is not quantitatively built into my calculations as some (such as Sandage) have tried. It boils down to two different zero points. Look at figure 3 in my May paper. If one divides the sample by morphological type, then ScI distances are systematically underestimated. It is present both in the calibrators and in the clusters.I also contend that the morphological-type corrections are incredibly rough and unreliable. There is obviously a trend, but the quantifiability of morphological type is still a ways away.
Local starbursts and Seyferts fit right on the ScI group calibration. Same with the cluster samples - Seyferts in those samples fit the cluster distance when the ScI group equation was applied.It's not so much caused by TFR errors but rather by morphology errors. It is difficult to quantify what the expected luminosity of a galaxy should be from first principles.
If you, for example, were to use starburst galaxies or active galaxies (which can be of any morphological type) the systematics are enormous.
Anyway, there are of course ways to check for evidence of this. For example, if there is large scatter and huge unknown systematics in the TF distances, then you should see large observed scatter when you apply the TF relation to the calibrators and cluster samples. You don't see that. As I said, most researchers find ~0.35 mag, but naturally if you correct for a real systematic effect like morphological type effect, then that observed scatter gets smaller.
You can also compare other distance determination methods with the TF distances for clusters. I didn't include that in my paper, but I have compared the Tonry et al SBF distances to clusters and the HKP FP distances to clusters - with my cluster distances and the distance modulus differences are very small (~0.15 mag or less).
You can look at additional characteristics such as diameters. That's in a paper I'm working on. The diameter results match the TF results - at the redshift distances, these same galaxies in Table VII would have ridiculously large diameters for their rotational velocity.
You can also look at the images of these galaxies and get a qualitative sense from them. The images in the Appendix clearly show that the TF distances are much more consistent than the Hubble distances. Those qualitative comparisons are in fact backed up quantitatively with additional diameter analysis.
In order for Malmquist bias to have a significant effect, the distance determination method must have a very large intrinsic scatter - which there is no evidence for in the TFR - unless you try to calibrate it from Hubble distances - which is a point I make in the other paper I'm working on.You mention the Malmquist bias which certainly is one consideration. Of other considerations are fingers-of-god effects and other redshift distortion effects which bring apparent scatter down.
But we're looking at field or low density environment galaxies for most of the galaxies in Table VII so fingers of god are not the issue.
Thank you again for the feedback! I'll take a look at the articles.
Established Member
Galaxy systematics (especially TF), for the most part, and a density modulus that isn't well captured for the local universe.
You've said this many times, and what it seems you are missing is that there are confinement processes that lead to lower scatter than the true scatter. In particular, your statistics indicate a bad fit. A tighter scatter for a smaller sample that is confined for no model reasons is generally not analytically what you want to do.The observed scatter relative to the cepheid distances is ~ 0.35 mag for standard calibrations and less than 0.20 mag with my Type dependent calibrations. For the cluster sample, there was an observed scatter of 0.22 mag relative to mean cluster distances. Clearly the intrinsic scatter in the TF relationship is small. Why that should be is a matter for theorists.
To understand this, just think of likelihood analysis. If you have a fit that falls directly on the points you propose are true and consider the scatter to be due to the error bars, you have a bad chi-squared fit, for example. Eliminating points based on "type" (which is not a quantifiable disentanglement, but one based on eye-identification) brings in a huge systematic error that is inappropriate for a neutral investigation. I have seen this done a lot in science, so you're in good company, but it doesn't make it right.
Lower density environments do not necessarily indicate a lower peculiar velocity with respect to the Hubble Flow. The Hubble Flow is an aggergate average, and while the spread is greater in clusters there converse statement (that you expect for an individual galaxy in the field a smaller peculiar velocity) is not true.At the distance of Coma, which is the TF distance for some of the galaxies in Table 7, a peculiar motion of 1000 km s-1 would introduce a distance modulus error of ~ +/-0.35 mag. The distance modulus differences exceed 1.00 mag in numerous cases. However, the galaxies in Table 7 are for the most part in lower density environments and should not be experiencing large peculiar motions - which means the Hubble noise should be even less than the 0.35 mag.
TF calibration is fine, the systematics of galaxies is not, so separating out based on type is highly dubious.The Cepheid calibrators were the basis of the TF calibration I used. This is exactly what was done with the major TF studies - cepheid calibrators were utilized to determine th zero point of the relationship.
Well, that just means you're in line with those models. However, the cosmic distance ladder is still an error-propagation nightmare. I notice that your error analysis is based solely on looking at data scatter which is the trap a lot of observationalists fall into. That's not a good estimate for error as discussed previously.I've compared the distances I get for clusters with distances from the Fundamental Plane and SBF studies - there are no systematic differences there.
All of them, actually. That way you'd get more believalbe error bars.As I said, my cluster TF distances are consistent with the FP and SBF distances. But independent of that - what other distance modulus would you suggest?
Key being larger distances.The Tully-Fisher relation is it for determining distances to large numbers of spiral galaxies at larger distances.
Low red-shift galaxies are notoriously more difficult to measure surface brightness from because you have noise associated with a larger extended feature, for example. At higher distances (beyond coma) the angular size is a weakly dependent function on redshift and lends itself quite nicely to TF.
What you should do is turn your analysis to all redshift regimes. Sloan data is publically available. Why have you ignored large surveys?
You miss the point. Quantifying morphology is just not done well and there isn't any reason for me to expect a marked differend in what the kind of galaxy is between, say, ScI and Sb/ScIII. These designations are poor holdovers from a time when people identified objects by eye. It's a highly subjective and crude focus to base any analysis on.I did not find it to be a significant problem in my May ApJ paper. About 5% of the galaxies turned out to have Sb/ScIII group morphology, but fit the cluster distance better with the ScI group equation.
Dealing with the number counts you do, it hardly surprising you can confine them in parameter space arbitrarily. This isn't good science, unfortunately. It's playing with data.There were not galaxies that were misclassified the other direction.
Narrow arms in the visible is a selection effect that can be due to at least three different reasons: dust, tidal stripping, and disk instabilities. These three reasons would be three different kinds of galaxies that would all be ScI. Morphology is not a good measurement.ScI morphology is pretty easy to spot because of the narrow arms.
Picking morphology is a bad move. Since it is the only discriminant you rely on, it strikes me as actually extremely biased as well. If I were the referee of the paper, I would also ask you to find tests other than morphology.Also the morphological T-type is not quantitatively built into my calculations as some (such as Sandage) have tried. It boils down to two different zero points. Look at figure 3 in my May paper. If one divides the sample by morphological type, then ScI distances are systematically underestimated. It is present both in the calibrators and in the clusters.
So, that confirms my suspicion that your analysis is completely arbitrary. Are all ScIs Seyferts? Are all starbursts? If not, then there seems to be a major flaw in your undertaking.Local starbursts and Seyferts fit right on the ScI group calibration. Same with the cluster samples - Seyferts in those samples fit the cluster distance when the ScI group equation was applied.
You keep repeating yourself, but you don't realize the fundamental point that I'm making: a small scatter does not mathematically indicate low errors.Anyway, there are of course ways to check for evidence of this. For example, if there is large scatter and huge unknown systematics in the TF distances, then you should see large observed scatter when you apply the TF relation to the calibrators and cluster samples.
I don't think it reasonable that you claim "real systematic effect" like morphological type is easy to pare down with respect to this kind of data analysis. I understand that people try to correct for morphology in TF, but what they don't do is the reverse (correct TF for morphology) which is basically what you are doing because of the trap-door function problem. The confinement in the dataset is a reasonable expectation from the converse, but the analysis of peculiar velocities with respect to the field is highly suspect.You don't see that. As I said, most researchers find ~0.35 mag, but naturally if you correct for a real systematic effect like morphological type effect, then that observed scatter gets smaller.
Again, you are repeating yourself. I addressed this point above.You can also compare other distance determination methods with the TF distances for clusters. I didn't include that in my paper, but I have compared the Tonry et al SBF distances to clusters and the HKP FP distances to clusters - with my cluster distances and the distance modulus differences are very small (~0.15 mag or less).
Nearly impossible measurement. Optical diameter, h-alpha diameter, 21 cm diameter? What diameter do you use?You can look at additional characteristics such as diameters.
Have you read Binney & Tremaine recently? I think you might find it useful in your endeavors.That's in a paper I'm working on. The diameter results match the TF results - at the redshift distances, these same galaxies in Table VII would have ridiculously large diameters for their rotational velocity.
Never a good idea.You can also look at the images of these galaxies and get a qualitative sense from them.
This is very poor science. One should never rely on gut instincts in observation. That's Arp's undoing, for example.The images in the Appendix clearly show that the TF distances are much more consistent than the Hubble distances. Those qualitative comparisons are in fact backed up quantitatively with additional diameter analysis.
Other way around. The luminosity function must have the large intrinsic scatter, not the distance.In order for Malmquist bias to have a significant effect, the distance determination method must have a very large intrinsic scatter - which there is no evidence for in the TFR -
No, but redshift distortions for the field are nearly impossible to measure.But we're looking at field or low density environment galaxies for most of the galaxies in Table VII so fingers of god are not the issue.
Scattering effects can pump velocities out there to enormous degrees. Bulk motion on large scales and with high statistics may indicate a good fit to Hubble Law, but there aren't enough galaxies to do it. That's why WMAP was so important. It fixed H_0 independent of the local effects.
Order of Kilopi
On what basis do you calibrate the TFR if not the actual calibrator sample for which you have independent distances? Its an empirical relationship between rotational velocity and absolute magnitude. That's it. Without having independent distances you don't know what that relationship is.You've said this many times, and what it seems you are missing is that there are confinement processes that lead to lower scatter than the true scatter. In particular, your statistics indicate a bad fit. A tighter scatter for a smaller sample that is confined for no model reasons is generally not analytically what you want to do.
It is what it is - empirically. Now there may be a subjective element to classifying the galaxies, but I'm not the one being subjective because I didn't classify them. I'm adopting the LEDA classifications. If there is nothing to this type effect, then it is quite remarkable that in a sample of 27 calibrators and 152 cluster galaxies the exact same type offset is seen that depends upon a very specific morphological breakdown.Eliminating points based on "type" (which is not a quantifiable disentanglement, but one based on eye-identification) brings in a huge systematic error that is inappropriate for a neutral investigation. I have seen this done a lot in science, so you're in good company, but it doesn't make it right.
Frankly, I'm surprised that you aren't more curious about that.
Then you're going to need to let Allan Sandage and Sheila Kannappan know that because its the position they adopt in their papers dealing with TF scatter. Could you put a number on how large those peculiar motions can get for field galaxies?Lower density environments do not necessarily indicate a lower peculiar velocity with respect to the Hubble Flow. The Hubble Flow is an aggergate average, and while the spread is greater in clusters there converse statement (that you expect for an individual galaxy in the field a smaller peculiar velocity) is not true.
Is there an upper limit?
I'm calculating Tully-Fisher distances - that requires rotational velocities.What you should do is turn your analysis to all redshift regimes. Sloan data is publically available. Why have you ignored large surveys?
And it works just fine. Can you explain why the TF distances of the ScI group galaxies in the cluster samples are systematically underestimated relative to the non ScI group galaxies if a single calibration is utilized? You keep objecting to the subjective nature of morphological classification - but the same subjective classification keeps leading to a specific TF result.You miss the point. Quantifying morphology is just not done well and there isn't any reason for me to expect a marked differend in what the kind of galaxy is between, say, ScI and Sb/ScIII. These designations are poor holdovers from a time when people identified objects by eye. It's a highly subjective and crude focus to base any analysis on.
Empirically it works.Narrow arms in the visible is a selection effect that can be due to at least three different reasons: dust, tidal stripping, and disk instabilities. These three reasons would be three different kinds of galaxies that would all be ScI. Morphology is not a good measurement.
Kannappan looked at a series of possible causes of scatter. I'm mystified by your lack of curiosity about the fact that morphology actually works extremely well. Surface brightness has absolutely nothing significant to do with the luminosity TFR scatter. Color Tully-Fisher relations haven't gone far either.Picking morphology is a bad move. Since it is the only discriminant you rely on, it strikes me as actually extremely biased as well. If I were the referee of the paper, I would also ask you to find tests other than morphology.
And what exactly is the "bias"? Either it has ScI morphology and its zero point is 0.55-0.60 mag larger - or it doesn't. The bias occurs when you don't account for this effect. I showed with pairs and with clusters that if you don't account for this the galaxies of ScI and Seyfert/STarburst end up with systematically underestimated distances. That's the empirical result whether you like it or not.
Why is that a flaw? That's where things fit empirically. You can't envision any circumstance under which Seyferts and starbursts could follow the same TF scale as ScI's?So, that confirms my suspicion that your analysis is completely arbitrary. Are all ScIs Seyferts? Are all starbursts? If not, then there seems to be a major flaw in your undertaking.
I realize that many think a small scatter is impossible in the TFR. And I'm aware about your concerns about small number statistics. If all I had looked at was the calibrators I would understand your complaint. But when you start calculating distances to clusters of galaxies and small groups of galaxies, and pairs of galaxies - and you find that the galaxies within those clusters give the same distance with a scatter less than 0.25 mag. - I must ask exactly where these giant unaccounted for errors are hiding out.You keep repeating yourself, but you don't realize the fundamental point that I'm making: a small scatter does not mathematically indicate low errors.
From your Bertshinger&Dekel paper linked to earlier:
That is exactly what I did with the galaxies in Table VII.Bertshinger&Dekel: A more direct approach to confronting theory with observations makes use of distance estimates based on intrinsic relations between observed strcutural parameters of galaxies, such as the Tully-Fisher relation for spirals and the improved Faber-Jackson relation for ellipticals. Here one obtains a direct estimate of the radial peculiar velocity of each galaxy by subtracting the Hubble velocity at the distance of the galaxy from the redshift velocity.
What scattering effects? If that's the explanation, then where are the extremely blueshifted galaxies in my sample? Funny that these galaxies are nearly all being scattered away from us.No, but redshift distortions for the field are nearly impossible to measure. Scattering effects can pump velocities out there to enormous degrees.
Established Member
Binney and Tremaine deal with this. It's necessarily calibrated to independent distances. Nothing more can be done other than this because it's a phenomenon that is non-trivially related to a large number of related effects (including such things as mass to light ratios and obscuring patterns).
No, but just because you didn't classify them doesn't mean that you're right in using the classifications as a discriminant.It is what it is - empirically. Now there may be a subjective element to classifying the galaxies, but I'm not the one being subjective because I didn't classify them.
Your morphological breakdown seems to bring your statistics way down from your sample. Small number statistics allows for all sorts of coincidences. Just look at Arp.I'm adopting the LEDA classifications. If there is nothing to this type effect, then it is quite remarkable that in a sample of 27 calibrators and 152 cluster galaxies the exact same type offset is seen that depends upon a very specific morphological breakdown.
Curiousity doesn't come into it when people fiddle with data at this level. I'm interested in overall trends, not local phenomena. In a neutral sample of 100 galaxies, only 23 will be beyond the statistical deviation and only five will be beyond two statistical deviations. In a bimodal cut, you should have half those numbers again: but given all the possible combinations of data analysis you could look at, conceivably you'd only need look at 5 different discriminants before you found an effect similar to yours. It seems you landed on one of these 20% of all encounters for such a sample size. A bit luck? Maybe, but the human eye is characteristically drawn to looking at outliers. It seems to me that that is all your analysis is.Frankly, I'm surprised that you aren't more curious about that.
No, you can't put a number on it.Then you're going to need to let Allan Sandage and Sheila Kannappan know that because its the position they adopt in their papers dealing with TF scatter. Could you put a number on how large those peculiar motions can get for field galaxies?
Sandage has always been a stick-in-the-mud with respect to the analysis of redshift distortions and the Hubble Flow. He had to be carried kicking and screaming to the Hubble Constant number found by WMAP.
Is there an upper limit?
So why not use spectral data from Sloan?I'm calculating Tully-Fisher distances - that requires rotational velocities.
Yes, above I put the chance calibration that you see at about 20% for any discriminant you care to use.And it works just fine. Can you explain why the TF distances of the ScI group galaxies in the cluster samples are systematically underestimated relative to the non ScI group galaxies if a single calibration is utilized?
Just because something leads to similar results doesn't mean that the subjective nature of the something is moot. I'm surprised you even think this is a reasonable argument.You keep objecting to the subjective nature of morphological classification - but the same subjective classification keeps leading to a specific TF result.
Poorly argued. Correlation does not imply causation.Empirically it works.
So now we're getting somewhere. Here's two discriminants that are discarded because they "don't work". So out of three possible discriminants then chance that one would work is about 60%.Kannappan looked at a series of possible causes of scatter. I'm mystified by your lack of curiosity about the fact that morphology actually works extremely well. Surface brightness has absolutely nothing significant to do with the luminosity TFR scatter. Color Tully-Fisher relations haven't gone far either.
Just because you get smaller scatter does not mean you've improved your understanding.And what exactly is the "bias"? Either it has ScI morphology and its zero point is 0.55-0.60 mag larger - or it doesn't. The bias occurs when you don't account for this effect.
As long as you're within error bars (which we are when we don't apply your correction) a correction that moves closer to a smaller error bar claiming to make the first error bars smaller is a bad move. That's the way analysis works.I showed with pairs and with clusters that if you don't account for this the galaxies of ScI and Seyfert/STarburst end up with systematically underestimated distances. That's the empirical result whether you like it or not.
I'm saying that TF has enormous scatter for these galaxies since central engines are not necessarily indicative of rotational velocities, yet Starbursts and Seyferts have remarkably well-determined luminosities due to these central engines.Why is that a flaw? That's where things fit empirically. You can't envision any circumstance under which Seyferts and starbursts could follow the same TF scale as ScI's?
Not only impossible, unreasonable. There is a lot of features that are understood to cause TFR, the biggest being a rough mass-to-light ratio. However, the absolute magnitudes of objects are highly environmental and have scatters that are enormous. It's not unreasonable to have geometric effects, luminosity function dependencies, etc. that drive scatter way, way up.I realize that many think a small scatter is impossible in the TFR.
To claim that all these well-known and well-understood effects are simply pushed under the carpet by looking at the morphology and taking cuts solely based on this isn't a matter of empiricism, it's a matter of convenient model fitting.
See above. The fact is that simply because you get a smaller scatter with a new calibration to small statistics doesn't mean you've done the right thing.And I'm aware about your concerns about small number statistics. If all I had looked at was the calibrators I would understand your complaint. But when you start calculating distances to clusters of galaxies and small groups of galaxies, and pairs of galaxies - and you find that the galaxies within those clusters give the same distance with a scatter less than 0.25 mag. - I must ask exactly where these giant unaccounted for errors are hiding out.
Think of the analogy to the shooting method. Take a calibrated measurement at a nearby point and then shoot with an error. Now take two calibrated measurements at a nearby point that are arbitrarily discriminated and shoot twice with another error. You should get a larger systematic error.
If you don't believe that morphology is arbitrary, just look at the fact that you named two other discriminants that plain don't work.
Again, with the small numbers, I'm not surprised you get none scattering towards us. I'm not convinced you've got the right distances for many of these field galaxies anyway, which would call into question your Hubble Flow numbers for them.What scattering effects? If that's the explanation, then where are the extremely blueshifted galaxies in my sample? Funny that these galaxies are nearly all being scattered away from us.
There is no reason to think that over a Hubble time any given field galaxy has a velocity determined uniquely by a close approximation to the Hubble Flow. That's the scattering effect.
Order of Kilopi
I would be very interested to hear your take on the data reduction techniques used by the supernova research teams. The light curve templates, the spectral image matching, the subjective elemental criteria used for rejection of Ib/c supernova, limited analysis on such systemics as the decreasing rise time with increasing redshift, the 'stretch factor' application, and the rejection of promising Ia spectra on the bases of rapid high redshift attenuation rates:
These highly subjective analytical tools are found in the methodologies Riess and Perlmutter & Co. use to tighten and constrain supernova Ia magnitudes, and they don't have the secondary or tertiary distance indicators Russell has been able to use to confirm his Tully-Fisher morphology corrections. Even so, cosmologist are using the numbers cranked out to make extraodinary astronomical predictions.
Welcome to the board, by the way, I overlook your arrival in prior postings. Your expertise is readily apparent and greatly appreciated.
Order of Kilopi
Exactly … that’s what I did! I used the independent distances of the calibrator sample to calibrate the TF relation. This how the TF relation is calibrated – unless Hubble distances are used. As you say it’s the necessary way to do it. There are no other options.Astronomy: Binney and Tremaine deal with this. It's necessarily calibrated to independent distances. Nothing more can be done other than this because it's a phenomenon that is non-trivially related to a large number of related effects (including such things as mass to light ratios and obscuring patterns).
What evidence is there that I was wrong in using it? That the distances strongly disagree with Hubble distances? But Hubble distances are what was being tested!No, but just because you didn't classify them doesn't mean that you're right in using the classifications as a discriminant.
You realize of course that if I used absolutely no morphological correction that the discrepancy between Hubble and TF distances would be even more extreme. If I just went ahead and used the plain old method of calibrating the TFR where morphology is ignored I could’ve claimed even larger discrepancies between Hubble and TF distances. Don’t you think if I was really trying to play with numbers that I would have just went ahead and done that? I could have avoided all the scrutiny of a new approach to the Tully-Fisher relation and just went ahead and made an even stronger case for large excess redshift.
You’re welcome to your skepticism about my interpretations. I understand that’s part of the process. But don’t think for a minute I’m massaging (intentionally or otherwise)data to make it fit preconceived notions.
Early on in this I was interested in Arp’s work, but I felt it was important to test by comparing actual distances to objects with Hubble distances. Since QSO’s have no independent means of distance determination I turned to galaxies. Arp was claiming ScI’s have the largest excess redshift. Well when I looked at the TF calibrators it quickly became apparent that ScI’s and Seyferts had underestimated distances with a single calibration while Sb/ScIII galaxies had overestimated distances. So I utilized the TD-TFR to correct for that. Its not a radical change from what is traditionally done. The only significant difference is that you need a different zero point for the two different TF groups. That’s the empirical reality – like it or not.
I could’ve said “You know, I’ll just stick with the previously published calibrations because with those the ScI distances will be even less and that will make my arguments for excess redshifts even stronger.” But I didn’t do that. Instead, I used the method that gives the best TF distances – even though it leads to a smaller excess redshift for ScI’s.
Not exactly – the ScI sample shows a clear trend of excess redshift relative to predictions for H0=72. The point of the “outliers” is that they are so extreme they require TF errors or alternatively peculiar motions well beyond what is accepted as possible. You’ve suggested my TF scatter is impossibly small … but what I’m telling you is that the required TF errors to explain the redshift discrepancies are impossibly large.Maybe, but the human eye is characteristically drawn to looking at outliers. It seems to me that that is all your analysis is.
And another point on the scatter. If I calibrate the TFR with the old-fashioned morphologically blind TFR, I recover the exact same scatter as Tully&Pierce (+/- 0.30 mag). But I identified a systematic pattern within that scatter associated with morphology. If I correct for that – the scatter cannot get larger – it can only get smaller!
Well then I guess I should just revert to the morphologically blind TFR and tack on that extra redshift resulting when the ScI distances become even less!As long as you're within error bars (which we are when we don't apply your correction) a correction that moves closer to a smaller error bar claiming to make the first error bars smaller is a bad move. That's the way analysis works.
So then you’re suggesting that the large redshift deviations I’ve identified could be real peculiar motions? Peculiar motions from 3000 km s-1 to 5000 km s-1 are possible despite the fact that every source I’ve ever seen says ~1500 km s-1 is an upper limit in clusters and a few hundred km s-1 is an upper limit for field galaxies?No, you can't put a number on it.
Look, if someone wants to take the position that these large excess redshifts are real peculiar motions that’s fine – its something new either way.
There is absolutely nothing wrong with the sources I’m using! The catalogs I have used were created for variations on the purpose I’ve put them to! You’re complaining about morphology as it is – and the galaxies I’m utilizing are close enough to reasonably interpret the morphology. The other problem is that if you push the distances too far it becomes difficult to get a reasonable estimate of peculiar motion because even the small observed scatter I’ve found with the TFR will lead to spurious large peculiar motions.So why not use spectral data from Sloan?
I’m not trying to assign causation. Its strictly empirical. When galaxies are ScI morphology they have a larger TF zero point. That is what is observed with the data in my samples. Why? Not my issue. I’m using the TFR as a distance indicator.Correlation does not imply causation.
So now we're getting somewhere. Here's two discriminants that are discarded because they "don't work". So out of three possible discriminants then chance that one would work is about 60%.Unbelievable. You’re actually arguing that because SB and color are not related to TF scatter that morphology too is unrelated. Morphology is not arbitrary. Its that simple! Literally, you’re raising hypotheticals that there is absolutely no evidence in any of the samples have any importance.If you don't believe that morphology is arbitrary, just look at the fact that you named two other discriminants that plain don't work.
Funny that nobody sees this large scatter unless they use the Hubble relation to calibrate the TFR.There is a lot of features that are understood to cause TFR, the biggest being a rough mass-to-light ratio. However, the absolute magnitudes of objects are highly environmental and have scatters that are enormous. It's not unreasonable to have geometric effects, luminosity function dependencies, etc. that drive scatter way, way up.
And would you be convinced if the sample gave H0=72 and none of the galaxies showed evidence for large excess redshifts? It amazes me that my sample of 100+ galaxies is “small” and yet we have papers like this and this that use one or two galaxies to “verify” H0=71 km s-1 Mpc-1.Again, with the small numbers, I'm not surprised you get none scattering towards us. I'm not convinced you've got the right distances for many of these field galaxies anyway, which would call into question your Hubble Flow numbers for them.
Established Member
I can understand the detractors, but the interesting thing is that there is no distance modulus required for the SNIa fits to be correct. The template fitting seems uncomfortable for similar reasons I outlined above, but it ends up really working well because it is a neutral fit as opposed to a subjective fit. In other words, there isn't any imposition of the data other than a template rather than a particular template.
In other words, the fits themselves are actually empirical while the fits dgruss has described are arbitrary.
Order of Kilopi
Astronomy: Not only impossible, unreasonable. There is a lot of features that are understood to cause TFR, the biggest being a rough mass-to-light ratio. However, the absolute magnitudes of objects are highly environmental and have scatters that are enormous. It's not unreasonable to have geometric effects, luminosity function dependencies, etc. that drive scatter way, way up.
To claim that all these well-known and well-understood effects are simply pushed under the carpet by looking at the morphology and taking cuts solely based on this isn't a matter of empiricism, it's a matter of convenient model fitting.Alright, I just can't let this go any further. All your complaints about the morphological type dependence corrections in my analysis are flat out inconsistent with the Tully-Fisher research. I outlined all of that in the introduction to my May ApJ paper, but I'll do it again here so that everybody interested in this is up to speed.Astronomy: If you don't believe that morphology is arbitrary, just look at the fact that you named two other discriminants that plain don't work.
First, as early as 1978 Roberts found evidence that late type spirals rotate more slowly at a given luminosity than early type spirals. Rubin et al 1985 added evidence confirming this trend.
Theureau et al 1997 made corrections for the type effect as did Giovanelli et al 1997 and Sandage 1999 . The corrections of Giovanelli et al are not that different from my TD-TFR. Sandage included luminosity class in his corrections which is consistent with my Type corrections in which luminosity class I/I-II galaxies are those that have the larger zero point.
Finally, as recently as last year Courteau et al made the following conclusion:
a conclusion drawn from this result in their paper:Courteau et al: While Hubble-type classification depends on the bandpass of selection, a morphological type dependence of the TFR, even at near-infrared bands, is undeniable.
and while we're at it on the matter of scatter they said the following:By selection, the mean TF fit is dominated by Sb-Sc galaxies. For
the Shellflow sample, early and late-type spirals di_er by 0.08 dex and -0.25 dex from the mean TFR respectively. This is accentuated in the SCII sample (Fig. 7) with TF luminosity residuals of +0.17 dex and -0.33 dex for early and late-types, respectively. Dale et al. (1999) applied a morphological-type correction to their galaxy magnitudes for the construction of their TF template. Their corrections to total I-band magnitudes for T=Sab, Sb, and T=Sbc are -0.27, -0.11, and 0.00 mag, respectively, which is somewhat larger than the ones we find for their data.
Astronomy, your critique of my type corrections does not stand up to scrutiny and is at complete odds with the evidence derived from much larger TF studies than mine.One of the most firmly established empirical scaling relation of disk galaxies is the Tully-Fisher relation (TFR; Tully & Fisher 1977); a tight correlation between the total luminosity and the rotation speed of a disk galaxy.
While I'm at it, any TF study you choose that does not use Hubble distances to calibrate it will verify the small intrinsic scatter of the relation. And your claims about peculiar velocities are at odds with prevailing views as well. Local deviations from the Hubble flow are at most 75 km s-1. See Karachentsev & Markarov 1996 , Karachentsev & Markarov 2001 and Ekholm et al . If you want to move to clusters, then there is clear agreement that peculiar motions do note exceed 1500 km s-1. See Jensen et al 2003 and Karachentsev et al 2003 . Frankly I'm mystified as to how you can present such nebulous arguments regarding limits on the size of peculiar motions when there is universal agreement among distance scale researchers as to what those limits are. Same goes with the TF scatter and morphological type dependence issue.
I'm strongly reminded of some of the arguments made by JS Princeton last year. The arguments didn't hold then and they don't hold now.
Order of Kilopi
Correct, and the supernova Ia do make excellent standard candles. When the Hubble scaling factor is applied to the SNIa magnitude curve, the 'evidence' of accelerating expansion emerges. But if Russell's estimate of the distance modulus is correct, all hell breaks loose. The supernovae are then, on average, more distance, and brighter than supernova Ia should be at these greater distances. They would have to be of the same magnitude of the brightest supernova type Ib/c observed in the local universe.
There are several other studies that indicate the 'true' Ho value may be significantly smaller than the Freedman consensus value (72km/sec/MPC). This is why the template matching, curve fitting and stretch factor manipulations suffer from the same potential inconsistances that are found in alternative solutions.
In astrophysics there must always be parametric assumptions. This is where the role of statistical logic must come into play, and this is why I must always pose the question: If the supernova type Ib/c are the brightest supernova type in the local population, why do we think the most distance supernova observed are Ia and not Ib/c?
Is this really true? The researchers who are doing the fitting know which templates will provide distance estimates that are most consistent with the Freedman distance scale. There are a half dozen subjective steps in this process – everything from the host galaxy reddening correction, (which is usually much less in distant supernova than local ones), to the amount of stretch factor correction that must be applied.The template fitting seems uncomfortable for similar reasons I outlined above, but it ends up really working well because it is a neutral fit as opposed to a subjective fit. In other words, there isn't any imposition of the data other than a template rather than a particular template.
This is why secondary sanity checks should always be applied. Is the (apparent) high supernova type distribution the same as the local distribution? NO. Are the rise times the same as low redshift risetimes? NO. Is the mean stretch factor remaining consistent across all redshifts? NO, it is getting smaller. Is the distribution of supernova Ia at increasing redshifts consistent what we would predict, based upon the local population density ? NO, it is less. All of these systemic failures are big red flags: Either supernova are evolving or something else is wrong.
Established Member
There is obviously a connection between TF and Morphology. This is because there is a physical connection between morphology and luminosity effects in galaxies. However, the relationship is very noisy. This is why you cannot use it to gain any information, which is essentially what you are arguing that is different from the mainstream TF research.
Sandage is right to include luminosity class since this is certainly related to mass-light ratios of galaxies. However, using a new zero point is a bad way of trying to account for this.Sandage included luminosity class in his corrections which is consistent with my Type corrections in which luminosity class I/I-II galaxies are those that have the larger zero point.
It would be like noting that people who have cancer and who smoke live less long than people who have cancer and don't smoke, but instead of plotting longevity curves based on both phenomena independently and then calculating a covariance matrix you simply started with a "zero point" for each data set and extrapolated from there. Sure, you're likely to get a lower scatter than before, but you've sacrifised any measure of goodness of fit.
[quote]
Hardly a quantified statement here.One of the most firmly established empirical scaling relation of disk galaxies is the Tully-Fisher relation (TFR; Tully & Fisher 1977); a tight correlation between the total luminosity and the rotation speed of a disk galaxy.
No, I have criticized very particularly your attempt to state that there is a way to get more accurate measures of distance by grouping morphology classes to different modulus for TF. This is not a good analysis for reasons I have been very clear about.Astronomy, your critique of my type corrections does not stand up to scrutiny and is at complete odds with the evidence derived from much larger TF studies than mine.
Don't confuse local volume measurements with peculiar velocities in the voids.While I'm at it, any TF study you choose that does not use Hubble distances to calibrate it will verify the small intrinsic scatter of the relation. And your claims about peculiar velocities are at odds with prevailing views as well. Local deviations from the Hubble flow are at most 75 km s-1.
I have looked at peculiar velocities in the most local void that exceed well over 400 km/sec.
I would advise against making proclamations such as this when it doesn't seem you have researched your claims that carefully. Clearly you have done a lot of homework on the matter, but there is a lot of things lacking that I would hope you would look at.Frankly I'm mystified as to how you can present such nebulous arguments regarding limits on the size of peculiar motions when there is universal agreement among distance scale researchers as to what those limits are.
You might start with calculating the covariance matrix between morphology and TF.
Don't get me wrong. There has to be a relationship between them. I'm not sure you've done a very good job with your work in trying to cull this.Same goes with the TF scatter and morphological type dependence issue.
If someone else was also making this case, why doesn't it give you pause to consider it?I'm strongly reminded of some of the arguments made by JS Princeton last year. The arguments didn't hold then and they don't hold now.
Established Member
Agreed. I don't, however, believe Russell.
WMAP could be trucked out here again, but I'll spare you the lecture. I think that what we'll have to do is take this other value and try to manipulate Max Tegmark's page to get a model you like and see if it works. Tell me what you come up with.There are several other studies that indicate the 'true' Ho value may be significantly smaller than the Freedman consensus value (72km/sec/MPC). This is why the template matching, curve fitting and stretch factor manipulations suffer from the same potential inconsistances that are found in alternative solutions.
Number statistics. There has only been a literal handful of Ib/cs observed.In astrophysics there must always be parametric assumptions. This is where the role of statistical logic must come into play, and this is why I must always pose the question: If the supernova type Ib/c are the brightest supernova type in the local population, why do we think the most distance supernova observed are Ia and not Ib/c?
If we could find more, then we might have a better shot at seeing more distant ones.
Yes they do. What is remarkable is that you don't have to invoke this to get the fits to work.Is this really true? The researchers who are doing the fitting know which templates will provide distance estimates that are most consistent with the Freedman distance scale.
Believe me, I was skeptical at first too, but Kirschner can be very convincing!
A half dozen is an exaggeration by two, I think. However, these are all included in Riess's work very explicitly. That it works is an amazing thing, but it's not subjective.There are a half dozen subjective steps in this process – everything from the host galaxy reddening correction, (which is usually much less in distant supernova than local ones), to the amount of stretch factor correction that must be applied.
To be expected! The environments are, after all, epoch-dependent.This is why secondary sanity checks should always be applied. Is the (apparent) high supernova type distribution the same as the local distribution? NO.
Also to be expected since they are definitely due to metallicity which is epoch dependent.Are the rise times the same as low redshift risetimes? NO.
Well if you follow the logic of the first two, then you can guess my answer for this one.Is the mean stretch factor remaining consistent across all redshifts? NO, it is getting smaller.
I think you mentioned this above.Is the distribution of supernova Ia at increasing redshifts consistent what we would predict, based upon the local population density ? NO, it is less.
Failures?All of these systemic failures are big red flags: Either supernova are evolving or something else is wrong.
We know supernovae environments are evolving, if that's your drift.
Order of Kilopi
Wait earlier you said this:There is obviously a connection between TF and Morphology. This is because there is a physical connection between morphology and luminosity effects in galaxies.
But now after I note the 25+ years of research that confirms the type effect you suddenly state that the connection is obvious. That's what I've been saying from the start and that's what I said in my paper. That's what others have said in their papers.You miss the point. Quantifying morphology is just not done well and there isn't any reason for me to expect a marked differend in what the kind of galaxy is between, say, ScI and Sb/ScIII. These designations are poor holdovers from a time when people identified objects by eye. It's a highly subjective and crude focus to base any analysis on.
Astronomy, you are wrong here. I've shown that. Look at the TF research. You're trying to categorize my approach to dealing with the type effect as something completely different from other approaches. Giovanelli et al applied zero point offsets to reduce scatter. That's what I've done. Sandage accounted for luminosity class. That's what I've done. Your claim that this morphological connection is noisy is not backed up by the volumes of TF research. You still haven't justified where all this scatter is hiding given the lack of large scatter observed by any TF researchers.However, the relationship is very noisy. This is why you cannot use it to gain any information, which is essentially what you are arguing that is different from the mainstream TF research.
The referee of my May ApJ paper was a TF expert and had some outstanding suggestions for improving the earlier draft of the paper. The point is that TF researchers have tried to quantify the type effect. That's what I did in a slightly different way - and it passed the review of a TF expert. Your issue with my efforts is an issue with Theureau et al, Sandage, Giovanelli et al. The type effect can be quantified - and it has been by all of them. To separate my results out and say I can't do that, but they can is ridiculous!
That's what is done Astronomy! New zero points. Look at the Giovanelli et al type corrections - its a modificatioin of zero point based upon type. The Sandage corrections modify zero point based upon type and luminosity class. I've modified zero point based upon type and luminosity class.Sandage is right to include luminosity class since this is certainly related to mass-light ratios of galaxies. However, using a new zero point is a bad way of trying to account for this.
It doesn't need to be - you can go to any of the TF papers to get quantified examples of what is meant by "tight correlation". And to those following this discussion note that I have provided quantified values of scatter from my own paper and the results of other studies. Meanwhile Astronomy is tossing out qualitative statements like:Hardly a quantified statement here.
However, the relationship is very noisy.Why don't you quantify this tremendous true large scatter for us since you are claiming that its been underestimated by TF researchers? I've provided the quantified observed scatter from my results and the results of others. You've responded with qualitative retorts about "noise" and "enormous scatter". Your claims about large scatter are backed up by none of the TF studies that calibrate the TFR from cluster samples and Cepheid calibrators!I'm saying that TF has enormous scatter for these galaxies since central engines are not necessarily indicative of rotational velocities, yet Starbursts and Seyferts have remarkably well-determined luminosities due to these central engines.
Yes, you've been clearly incorrect as I've shown above by citing 25+ years of TF research. You're faced with taking the position that a systematic effect identified in the TF sample cannot be used as a way to reduce scatter. That is wrong. If you find a factor that correlates with scatter. You correct for that effect and scatter is reduced. Rotational velocites are corrected for inclination (a systematic effect) - that reduces scatter relative to the raw measurements. Magnitudes are corrected for absorption due to inclination. Scatter is reduced. Morphology is simply another effect that allows reduction of scatter. Its as basic as correcting for inclination and absorption.No, I have criticized very particularly your attempt to state that there is a way to get more accurate measures of distance by grouping morphology classes to different modulus for TF. This is not a good analysis for reasons I have been very clear about.
Well I've never seen anybody besides you and JS Princeton actually suggest that real peculiar motions could exceed 1500 km s-1 in clusters and field galaxies. You're wrong on this count too. I cited the articles above the demonstrate this is the case and at any rate your 400 km s-1 is insignificant compared to the excess redshifts I identified. Nobody is proposing that real peculiar motions can be as large as 5800 km s-1.I have looked at peculiar velocities in the most local void that exceed well over 400 km/sec.
Not to mention you've really offered no answer to the point about seeing blueshifted galaxies. If my results are from peculiar motions, then where are the galaxies with large peculiar motions toward us? Your answer earlier doesn't cut it.
And if someone else joins cyrek in proclaiming that conservation laws disprove the Big Bang, should I suddenly take that incorrect argument seriously. Besides, I'm not convinced that we are talking about "someone else" here.If someone else was also making this case, why doesn't it give you pause to consider it?
Order of Kilopi
I need a reference, or a better explanation. Tegmark’s page?....I think that what we'll have to do is take this other value and try to manipulate Max Tegmark's page to get a model you like and see if it works. Tell me what you come up with.
Check out Homeier. A handful of local events that have greater magnitude should be much easier to find at redshift distances and we should be able to find them.Number statistics. There has only been a literal handful of Ib/cs observed.
If we could find more, then we might have a better shot at seeing more distant ones.
…Then we should expect the spectra of supernova Ia to evolve as well, but they do not. If the rise times and metal content are assumed to evolve with distance, why not the the magnitude and lightcurves? If supernova Ia evolve, comparison with local events is shaky at best, and the claim "expansion of the universe" is increasing even shakier. Ironically I don’t think they are evolving – but I do think distant supernovae Ib/c are being categorically miss-classified as supernovae Ia.To be expected! The environments are, after all, epoch-dependent.
…Also to be expected since they are definitely due to metallicity which is epoch dependent.
Riess covers a lot of systemics, but he did not justify discarding apparent Ia because the redshifted lightcurves were much too small. What are these high redshift events that look like supernovae Ia? If the Doppler interpretation of redshift is wrong, that is EXACTLY what they are…this would also make the majority of what we are characterizing as supernovae Ia supernovae Ib/c, so they haven’t gone missing in action at all!
edit - lost verb
Order of Kilopi
In addition to the points about morphological classification at large distance and false peculiar motions introduced by distance modulus uncertainty at large distance, I would think being a person who uses SDSS data that you must already know why.Astronomy: So why not use spectral data from Sloan?
Order of Kilopi
I let this one go earlier too, but now that it has become abundantly clear that your statements about the distance scale are subject to significant errors, I'd like to return to this and explain to everyone.
Distances to galaxies can be determined by the following methods:
Cepheid's: Spiral galaxies out to about 35 Mpc.
Surface Brightness Fluctuation(SBF): Elliptical and early type spirals out to about 100 Mpc
Fundamental Plane(FP): Ellipticals out to about 130 Mpc
Tully-Fisher(TF): Spiral galaxies reliably out to at least a few hundred Mpc
Planetary and Globular cluster luminosity functions: Generally larger spirals or ellipticals out to about 20-25 Mpc.
Type Ia supernova: Any type of galaxy out to distances well beyond all the other methods.
If I want to check my TF results I could compare different distance methods for individual galaxies, but there is a problem. Notice that SBF and FP methods are primarily utilized with ellipticals. So you can't directly compare those distances with TF distances. Cepheids are used as the calibrators, so that throws that option out. PNLF and GCLF are limited to within only ~ 25 Mpc, have few sampled galaxies and are therefore of little value in such a comparison. Type Ia supernova: You get the data when one happens. You cannot select a sample of galaxies with TF distances and say "Ok, now I'm going to get the Type Ia distances and see how I'm doing." You don't just have a cluster of galaxies with available Type Ia data. Doesn't work that way.
So how does when test their TF distances against other methods? One must compare cluster distances derived from the TF relation against cluster distances derived from SBF and FP plane distances using the ellipticals in those clusters.
I've already done that and there is no evidence of signficant or systematic differences in mean cluster distance moduli for those methods.
When Astronomy says that is not enough and I should use "All of them", - meaning all of the distance determination methods - Astronomy is asking for something that is not possible. There are no other suitable methods and data available for such a comparison.
And if you don't know that Astronomy, then I have to ask upon what basis you were tossing out statements like "If I was the referee of your paper,..." Referee's are selected based upon their expertise in the subject matter of a paper. You have not demonstrated that you have the expertise to evaluate the subject matter of my papers in the role of a referee. I'm being direct, not rude here. You'r the one that came on and said things like this:
Refereeing is serious business for everyone involved. The editor has a responsibility to select an appropriate refereee. The referee has the responsibility to make the call as to whether or not he/she feels qualified to evaluate the paper. Both of my papers we've discussed here were significantly improved by suggestions of the referee. That's not possible when the referee doesn't know what they're talking about. You've made significant mistatements regarding Type dependence in the Tully-Fisher relation, Tully-Fisher scatter, Peculiar motions, and distance scale comparisons - all the fundamental subjects of my paper!I think, if I were refereeing the paper, I would have had you omit them, then.
And just to further underscore the point, here’s another one that I left alone initially:
This is wrong too: Yasuda et al 1997 ; Gavazzi et al 1999 ; Federspiel et al 1998 ; Ekholm et al 2000 .
Established Member
cyrek comment
All the distance indicators mentioned above by dgruss are not suitable for 'cosmological indicators'.
Look at the HDFN for more reliable DI.
There are several ellipticals in that field. Comparing them to M87 data in Virgo, could be a way of estabishing a cosmological DI.
The redshift of these ellipticals can be correlated with 'angular size' and magnitudes (luminosity) because space is considered to be 'flat'.
The most reliable distance to the Virgo cluster is about 54^6 light years or 16.8 m/pcs. The angular size of M87 is given as 9 minutes of arc. Its absolute magnitude is given to be -21 plus a fraction.
Established Member
Where your paper is wrong is in assuming that the error bars can be brought down because of the relationship.
No, what you have shown is that you can eliminate scatter in your dataset. That's not the same as eliminating systematic errors associated with the morphological type.Astronomy, you are wrong here. I've shown that.
No, your approach is very different. You do not use Giovanelli's approach as you state in your paper. When you describe them, it may sound like they are similar, but they are different.Giovanelli et al applied zero point offsets to reduce scatter. That's what I've done.
No, you used a separate zero point for luminosity class.Sandage accounted for luminosity class. That's what I've done.
You have continued to ignore my beef, which is not with a connection but with the statistics. Your argumentative tone and refusal to look at what I am saying -- your refusal to even approach looking at a covariance matrix makes me think that you simply want your result to stand as showing that there is an erroneous velocity-distance relation in the nearby Hubble Flow.The referee of my May ApJ paper was a TF expert and had some outstanding suggestions for improving the earlier draft of the paper. The point is that TF researchers have tried to quantify the type effect. That's what I did in a slightly different way - and it passed the review of a TF expert. Your issue with my efforts is an issue with Theureau et al, Sandage, Giovanelli et al. The type effect can be quantified - and it has been by all of them. To separate my results out and say I can't do that, but they can is ridiculous!
No, Sandage uses luminosity class as a covariance measure, not as a new initial measure.That's what is done Astronomy! New zero points. Look at the Giovanelli et al type corrections - its a modificatioin of zero point based upon type. The Sandage corrections modify zero point based upon type and luminosity class. I've modified zero point based upon type and luminosity class.
You cannot simply claim something is tight without looking at a neutral comparison. Just giving numbers of the scatter does not indicate how tight a correlation is. That's simple statistics.It doesn't need to be - you can go to any of the TF papers to get quantified examples of what is meant by "tight correlation". And to those following this discussion note that I have provided quantified values of scatter from my own paper and the results of other studies. Meanwhile Astronomy is tossing out qualitative statements like:
No, I'm saying it is underestimated by you. If you apply the Sandage model I think you'll find it a better approximation to what I'm saying.Why don't you quantify this tremendous true large scatter for us since you are claiming that its been underestimated by TF researchers?
That's not what I'm criticizing. I'm criticizing your attempt to claim that this spread is anamolous with respect to physical quantities which, indeed, have admittedly high systematics. You cannot live on TF alone.Yes, you've been clearly incorrect as I've shown above by citing 25+ years of TF research.
No, that is right. Just read any book on data analysis.You're faced with taking the position that a systematic effect identified in the TF sample cannot be used as a way to reduce scatter. That is wrong.
That doesn't indicate that the measurement is better because reducing scatter isn't the same thing as getting more accurate: only getting more precise.If you find a factor that correlates with scatter. You correct for that effect and scatter is reduced.
Precision is needed, but accuracy is more important.
Nobody is suggesting that you actually measured peculiar motions, either.Well I've never seen anybody besides you and JS Princeton actually suggest that real peculiar motions could exceed 1500 km s-1 in clusters and field galaxies. You're wrong on this count too. I cited the articles above the demonstrate this is the case and at any rate your 400 km s-1 is insignificant compared to the excess redshifts I identified. Nobody is proposing that real peculiar motions can be as large as 5800 km s-1.
Here's one of many papers that suggest large peculiar velocities in the local void: http://adsabs.harvard.edu/cgi-bin/np...18d30661d10539
Please, explain why it doesn't cut it?Not to mention you've really offered no answer to the point about seeing blueshifted galaxies. If my results are from peculiar motions, then where are the galaxies with large peculiar motions toward us? Your answer earlier doesn't cut it.
I don't think you've measured actual peculiar velocities.
Established Member
[/quote]This is wrong too: Yasuda et al 1997 ; Gavazzi et al 1999 ; Federspiel et al 1998 ; Ekholm et al 2000 .
I should clarify. TF relations occur over many bands for many empirial relationships. Of course, optical is always the first appeal, but in an absolute calibration sense, it is better to do the analysis across many bands.
Order of Kilopi
Nope. I didn't assume anything. I demonstrated that the observed error is smaller when you make the type correction.Astronomy: Where your paper is wrong is in assuming that the error bars can be brought down because of the relationship.
We've already been over this. Intrinsic TF scatter is small. You can consult any of the numerous TF papers to find this for yourself. Your hypothetical systematics is just that - an unjustified hypothetical with no evidence grounded in the observational results.No, what you have shown is that you can eliminate scatter in your dataset. That's not the same as eliminating systematic errors associated with the morphological type.
I really don't know what tone you're talking about. You come on here making severely incorrect statements while making claims about what you would do if you were the referee. But that's been dealt with.You have continued to ignore my beef, which is not with a connection but with the statistics. Your argumentative tone and refusal to look at what I am saying
The test I've made is a straightforward and perfectly valid approach as indicated by one of the articles you linked to. I'm not buying into your attempt to obfuscate the matter.-- your refusal to even approach looking at a covariance matrix makes me think that you simply want your result to stand as showing that there is an erroneous velocity-distance relation in the nearby Hubble Flow.
And they are valid. Giovanelli et al use additive corrections based upon morphological type. Sandage used a correction based upon Hubble T-type and luminosity class. I put the sample into one of two morphological groups. Each group has a different zero point. Since the slope of the relation is the same in the Ap&SS paper, the net result of the different zero points is an additive type correction.No, your approach is very different. You do not use Giovanelli's approach as you state in your paper. When you describe them, it may sound like they are similar, but they are different.
Try as you might, you're simply not succeeding in your effort to claim that my approach is a dramatic break from other TF research.
No, Sandage uses luminosity class as a covariance measure, not as a new initial measure.Sandage: As in Paper V, the data can be freed from the T and L dependencies by applying a correction of 0.23T + 0.5L.
Of course the Sandage papers indicate larger scatter. He used Hubble distances to determine absolute magnitudes. Studies that use cluster samples and the calibrators consistently provide much smaller scatter than those that use Hubble distances.No, I'm saying it is underestimated by you. If you apply the Sandage model I think you'll find it a better approximation to what I'm saying.
And what evidence can you bring forward that the distances are not accurate? The cluster distances match with FP and SBF distances. The cluster TF distances with my equations and samples are consistent with the distances to the same clusters found with other TF studies.That doesn't indicate that the measurement is better because reducing scatter isn't the same thing as getting more accurate: only getting more precise.
Precision is needed, but accuracy is more important.
You've managed to talk an awful lot about accuracy and precision while bringing forward no actual evidence that there is an accuracy problem here.
I'm not suggesting I'm measuring peculiar motions either. What I'm suggesting is that the peculiar motions that would be required to account for the discrepancies between Hubble and TF distances are too large to be accounted for as real peculiar motions. We've already discussed that.Nobody is suggesting that you actually measured peculiar motions, either.
For everybody that's following this, let me cut through all this attempt to complicate the matter and get to the point. I've calculated the distances to a sample of galaxies using the TF relation. The scatter of this relation is observed to be 0.35 mag by most researchers. Because of my type corrections I'm observing scatter less than 0.25 mag. In either case, the intrinsic scatter in the TF relation is very small - much smaller than Astronomy seems to realize. Any claims about significantly larger scatter are at odds with the TF research.
Now here's what it boils down to: Either my TF distances are accurate within the data uncertainties and typical observed scatter or they are not. Let me give everybody one example. ESO 445-27 has a TF distance of 89.5 Mpc. According to the Hubble relation its true distance would have to be 162.8 Mpc - indicating a velocity discrepancy of almost 5300 km s-1. In terms of magnitude errors that requires an error of 1.30 mag. As I've noted observed scatter in TF is ~0.22 mag for my study and ~0.35 mag for other studies.
Despite attempts here to confound this simple matter with "unknown systematics" and "covariance matrix", it boils down to whether the distances are accurate or not. An error of 1.30 magnitudes is ridiculously large. If such large errors were common it would increase the observed scatter when comparing TF distances within clusters, small groups, and pairs. That is not seen. In fact, in my paper, I calculated TF distances to companions of these ScI galaxies that had close redshifts. Those companions (anywhere from 2 to 8 companions), gave exactly the same TF distances. Perhaps you can argue that one galaxy might have a wildly inaccurate distance, but when you start finding companions having the same distance, it really weakens that possibility. The referee seemed to understand that.
So, if the TF distance of ESO 445-27 (for example) is accurate, then there is a 5300 km s-1 discrepancy between the expected redshift and the measured redshift. Peculiar motions cannot be that large which leaves one possible interpretation being that there is a contribution from intrinsic redshift.
Its not even an answer.Please, explain why it doesn't cut it?
Order of Kilopi
I should clarify. TF relations occur over many bands for many empirial relationships. Of course, optical is always the first appeal, but in an absolute calibration sense, it is better to do the analysis across many bands.[/quote]This is wrong too: Yasuda et al 1997 ; Gavazzi et al 1999 ; Federspiel et al 1998 ; Ekholm et al 2000 .
That doesn't have anything to do with your statement about the Virgo cluster. B,R,I, and K' TF relations have been shown to be consistent with each other. See Tully&Pierce 2000 for example.
Established Member
65 MILLION (LIGHT) YEARS AWAY (AGO), the Dino Galaxy, and why the Big Bang is reeaaally a STRETTCHH!
Here's a brain teaser: If we look back 65 million light years away, we're looking back into the final days of the dinosaurs, 65 million years ago. Now, pick a galaxy visible there, let's call it the "Dino Galaxy", and imagine yourself looking from it towards us. What do you see?
Well, it makes sense that what you're seeing is Earth's time 65 million years ago, the final age of the Dinosaurs. But look in any other direction, and that is the same thing you see, that Dino Age for any galaxy. Of course, what you see is no longer there, time has gone by, but also if space is truly expanding, then what you see is no longer there because it has moved away. So when we look back into time at cosmic distances, we never see what's there, if space is expanding, or even what was there, since it was no longer there almost immediately. Yet, from either perspective, whether looking out 65 million light years from Earth, or looking from Dino Galaxy towards Earth, you are on the expanding edge of the universe, if it is expanding. What's wrong with this picture?
First of all, in which ever direction you look, if far enough, you are looking into the origin of the Big Bang, no matter from where you look. Then, since you are on the expanding edge of the universe, you can only see backwards in time, and never laterally into your present universe, since light does not travel fast enough to allow you to do this. We never-ever-ever see the present in space, only the past, if the universe is static. But if the universe is expanding, then we are double deceived, since now what we see is not even there anymore. What does that make of astronomical observations? They become pretty much meaningless. Now, imagine that simultaneously, someone from Dino Galaxy, 65 million light years away, is watching you through a telescope as you are watching them. Is there some point midway where your vision meets, so that you are both watching the same space at the same time? That depends.
If the universe is not expanding, static, then somewhere around 32.5 million light years away, you're both looking at the same midway point, call it the Midway Galaxy. But if the universe is expanding, then you're both short of the mark, since what was Midway Galaxy there at the time no longer is. This means both observers will get different images on the same point. Yet, if someone was there at the midway point looking in both directions, he/she/it would be looking towards Earth's (post-dinosaur) period, and Dino Galaxy, as it was 32.5 million years ago. However, that midway point is now the expanding edge of the universe, if it is a Big Bang generated expansion. So from their point of view, today, they see the universe as it was for us 32.5 million years ago, except that neither the dinosaurs here, nor Dino Galaxy there, are still there. So in looking in these two simultaneous directions, if Big Bang expanding, they only see an illusion, since neither is there anymore.
Why is this significant? On a very large scale, we never see the present, only the past. On a very small scale, such as within the Milky Way, we have access to a more recent past, about 40,000 years ago, if looking into the black hole at the center, for example, but there is no evidence of expansion within a galaxy. This space-expansion is reserved for intergalactic space, which in itself makes it suspect, though theoretically possible, even necessary to prove Einstein's cosmological constant right. But also, on a very large scale, we never see what was actually there at the time, since it already moved away by the time we trained our scopes on it. Remember that each observer, from wherever this observation is taking place in the present, is on the expanding edge of the universe, the present. But what each observer sees is only a fallacious image of what it may have been long ago, if the universe is expanding. And if so, how can we make justifiable projections of what the universe is at any point in space-time since we never know what that is? It's always changing. If not expanding, then at least we have the certitude that what we see is what was once there. But if expanding, then what we see is no longer there, and was there only for a brief moment, so that next time we train our scopes in that direction, it should no longer be there, but moved away. But! And this is a major BUT: Where did it go? Did it go towards us? Did it move away from us? Or did it go sideways? If it is expanding, where is it, and which way is it going?
This is the fundamental problem with the Big Bang: If Dino Galaxy is moving away from us, then when they train their scopes on us, are we still here, or did we move away? But which way? We don't know which way, because the universe has no center, so no direction is guaranteed. If I pick Dino Galaxy 2, at 180 degrees away from the first one, but at the same distance of 65 million light years, which way is that galaxy moving? Can it be moving towards us, rather than away? Or is space expansion so consistent that it has to move away from us. Then if observers from Dino 2 are looking towards Dino 1, can they still see it, if it's moving away? And if they do see it, how can they judge how far it is? And if they both train their scopes towards each other, and we are now the midway Milky Way galaxy, are they seeing us 65 million years ago? Are we still here, or did we expand in space in some direction or other? We cannot know from here in which direction because the universe has no center, so all directions are valid: Which way did the Milky Way go?
So here is a problem, where we're all looking at each other, but cannot see each other because we are all in motion away from each other. By the time we train our scopes at distances in any direction billions of light years away, we should not blink, for we will miss what was there, what cannot be there moments later, as they speed away from us at multiples of lightspeed. In fact, it is something of a stretch that space can expand at above lightspeed, and we can still see it. How is light reaching us from sources that are moving away from us above lightspeed? Or are we merely seeing an illusory shadow of what was there, very long ago, but no longer there? We know Hubble Deep Space gave us image from over 12 billion light years away, but at what point was light so stretched by an 'expanding' universe that it should no longer have been visible for us?
I suspect that it makes much more sense that if we can see billions of light years away, that we are actually seeing into the past of what was there at the time, but it had not moved away. Rather, whatever happened to it over the past billions of years is still there, in its current form. True the universe has no center, but nor does it have an expanding edge. There is no border to reality, because there is no expansion taking place. Things move around because of internal forces manifest within a universe that affects each and every part of itself, ad infinitum, of which we are but one small infinitesimal piece. We are not on the expanding edge of reality, anymore than we are at the center of all existence.
Finally, if space is expanding, why should light reach us at all? Why doesn't it get lost into some other dimension of space instead? Why would stretched-space allow light to reach us in this dimension of unstretched-space at all? Or is this too much of a strettchh?
Just bean' silly with a silly notion!
Posting Permissions
