5. “But the Bullet Cluster …” – Proof of Cold or Warm Dark Matter in galaxy clusters is but a myth

Whenever a discussion about the problems of the Cold Dark Matter Hypothesis and possible alternatives like Modified Newtonian Gravity (MOND) emerges, one argument you can be sure to hear soon is “But the Bullet Cluster …”. It is the same whether you discuss with scientists or other people interested in astronomy. But can the Bullet Cluster be considered as a proof of Cold or Warm Dark Matter? No, because that conclusion rests on further assumptions and is in itself not logically valid. Furthermore, the problems of the Dark Matter Hypothesis are independent of the Bullet Cluster, making it a false argument in many discussions. Even worse, the collision velocity of the Bullet Cluster seems to be incompatible with the concordance cosmology. At the same time, alternative gravity theories, while often said to fail in explaining galaxy clusters, can account for them rather naturally.

In this first post in a series of three we will discuss the Bullet Custer as a “smoking gun” for Dark Matter. The other two parts about “The Bullet Cluster and galaxy clusters in modified-gravity theories” and “The Train Wreck Cluster – an ‘anti-Bullet-Cluster’: disproof of Cold or Warm Dark Matter?” will follow in a few days.


What is the “Bullet Cluster”?

It all started in 2006 with a paper titled “A direct empirical proof of the existence of dark matter” and a press release with the no-less lurid headline “NASA Finds Direct Proof of Dark Matter”. Right in the title the authors claimed that their discovery would immediately settle the question whether there is Dark Matter (DM) or not. Naturally, the spectacular announcement was adopted by the majority of the media and could well be one of the most successful press releases in astronomy. Since then, the picture of the Bullet Cluster (BC) has been shown countless times:

Credit: X-ray: NASA/CXC/CfA/M.Markevitch et al.; Optical: NASA/STScI; Magellan/U.Arizona/D.Clowe et al.; Lensing Map: NASA/STScI; ESO WFI; Magellan/U.Arizona/D.Clowe et al.  


What do we see in this picture? There are two clusters of galaxies (left and right). Overlaid are blue and pink colors. The two pink clumps in the middle shows where x-ray observations find the hot gas, the Bullet Cluster got it’s name from the bullet-like shape of the gas on the right. The gas usually sits in the center of a galaxy cluster, but is here shifted to the point between the clusters. Some time ago the two galaxy clusters have passed through each other, making their gas collide. As gas interacts electromagnetically, it is slowed down when it collides, like two streams of air that can not pass through each other unhindered in opposite directions. That is why the gas is a bit behind the galaxies. Those do only interact through gravity and therefore pass each other unhindered without being slowed-down like the gas. This is maybe similar to two swarms of flies that can fly through each other.

The blue blobs were derived in a bit more complicated manner using the gravitational lens effect. To put it simple, since Einstein we know that matter deforms space-time. This leads to a bending of light rays when they come close to a large amount of matter, like a galaxy cluster. Thus, when the light of distant galaxies passes a massive cluster of galaxies before it reaches us, we will see a deformed image of the distant galaxies. The effect can be calculated and astronomers are able to trace it back. From the distorted shapes of distant galaxies behind a galaxy cluster they can infer the distribution of mass in that galaxy cluster. The heavier a galaxy cluster is, the more it bends the light and the more it distorts background galaxies. For the Bullet Cluster, the blue blobs in the picture above show the distribution of mass as inferred from the gravitational lensing effect assuming General Relativity to be valid. One can see that it follows the distribution of galaxies, not the gas.

If the hot gas would be the most massive part of a galaxy cluster, the mass found through gravitational lensing would have to be centered on it. But as this is not the case, it is said that the majority of matter in the galaxy cluster has to be close to the galaxies. Because the visible mass in the galaxies is not enough to account for the velocity dispersion of a galaxy cluster (assuming Newtonian Dynamics, i.e. General Relativity), it is conjectured that there is Dark Matter, which by definition only interacts through gravity, too. Thus distributions of DM can pass through each other just like the galaxies and the majority of mass should be found close to the galaxies in such a cluster collision.

This is why the Bullet cluster is often said to be the “smoking gun” of the Standard Cosmological Model. It behaves just like it is expected. But is it really that simple? Does this proof the existence of Dark Matter? No, it doesn’t.


Observations and Interpretations

In the most-often given description of the Bullet Cluster, what is observation and what is interpretation get mixed up. The observations tell us that the hot gas component and the lensing mass have an offset. One good conclusion from this is: The visible, hot gas can not make up the majority of mass in the system. But a wrong conclusion is: The Cold Dark Matter Hypothesis is right.

While we can argue that the majority of mass has to be close to the galaxies, we can not immediately conclude that it has to be in the form of Dark Matter as a new type of particles. Actually, we can only say that the majority of gravity, or even more specific, the major bending of space-time, happens close to the galaxies. Whether the reason is missing mass or a different law of gravity is not that easy to distinguish (there is gravitational lensing in modified gravities, too). The whole, most-mentioned conclusion is therefore based on one important, but never mentioned assumption: that gravity is best described by Newtons law. In addition to that, it supposes that other, known forms of dark matter (e.g. neutrinos) can not be the reason. Without those assumptions, the case of the Bullet-Cluster is not decided at all.

We see that the “direct proof for the existence of Dark Matter”, is an indirect hint at best, in that it is based on untested assumptions and does not even look at other possible predictions of alternative gravities. But there are more problems to come.


A Proof of Dark Matter?

Even if the cluster can be explained in the standard or concordance cosmological LCDM framework, this does not proof the theory. Because there can be no proof of a scientific theory. For the BC to be a proof of a scientific theory, it would have to rule out each and every alternative explanation, even those of which we can not even think of today. This, of course, is impossible. This fact is well known in the philosophy of science and I guess most scientists know this. Scientific inference does not function without this elementary fact.

Furthermore, there are different possible explanations for the BC. There even are different possible forms of Dark Matter. Not only the currently favored Cold Dark Matter, on which the Concordance Cosmology Model rests, but a model with Hot Dark Matter (where the DM-particles are fast/relativistic because they would be of low mass, like neutrinos) could explain the BC as well. So, please don’t state that the Bullet Cluster has proven the LCDM-model right. It has not. And it can not.


The Bullet Cluster, a problem for Dark Matter?

In fact, the Bullet Cluster might not only not be a proof of the DM hypothesis but it actually appears to be a major problem for the concordance model. Mastropietro and Burkert (2008) have found that the two colliding clusters need to have a relative velocity of about 3000 km/s to produce the observed X-ray gas properties. This result was compared to a cosmological simulation named MICE. Such cosmological large-volume simulations show the formation of structure in the universe and are often said to be another important success of the Concordance Cosmological Model. In the MICE simulation, Lee and Komatsu (2010) determined the probability that the Bullet Cluster’s velocity could be found in the concordance cosmological model. It is roughly one in ten billion! They …

“… conclude that the existence of [the Bullet Cluster] is incompatible with the prediction of the ΛCDM model …”.

This is a paradoxical situation: While the structure formation simulations are used to argue in favor of Dark Matter because they fit so well, and the Bullet Cluster is used in favor of Dark Matter as a “direct proof” or “smoking gun”, putting them both together leads to an incompatibility.


Does the Bullet Cluster matter at all?

So far we have shown that the Bullet Cluster can not be understood as proof for the Dark Matter Hypothesis. We took the argument seriously. But in doing so, we have repeated the same mistake as many people who bring up the BC when they try to dismiss our work on testing the Dark Matter Hypothesis. Why that? Well, usually the discussion follows these lines:

  • “Testing the predictions of the Cold Dark Matter Hypothesis on galaxy scales, we have found several serious problems.”
  • “But don’t you know about the Bullet Cluster? It is the proof that there is Dark Matter!”

Put that way, it is easy to spot the mistake: Even if the Bullet Cluster could only be explained with Dark Matter, the problems on small scales persist. The BC does not tell us anything about the Local Group of galaxies, the two arguments are completely independent. There is a serious problem with the DM Hypothesis and even a thousand Bullet Clusters would not make it go away. In fact this is similar to the hypothetical scenario that someone, for example at the LHC, would find “The Dark Matter Particle”. Even in that case, the current problems of the model would not go away. Rather, this would point at a much more serious issue with our understanding of physics.

In trying to understand the universe, we as good scientists should thus look for alternative explanations that account for all independent observations and discuss these without ideological pre-conceptions.

Solutions to the Bullet cluster in modified Newtonian dynamics (MOND) have indeed been shown to exist (Angus, Famaey & Zhao 2006). The authors conclude

In multicentred models, the convergence map does not always reflect the projected matter in the lens plane in MOND. This cautions simple interpretations of the analysis of weak lensing in the bullet cluster 1E 0657−56 (Clowe et al. 2004; see fig. 7).

Similarly for Modified Gravity (MOG):  Brownstein and Moffat (2007) write

The MOG prediction of the isothermal temperature of the main cluster is T = 15.5 +/- 3.9keV, in good agreement with the experimental value T = 14.8+2.0-1.7keV. Excellent fits to the 2D convergence κ-map data are obtained without non-baryonic dark matter…” and they uncover a significant disagremenet with the dark-matter based analysis (the baryon fraction is to high in a dark-matter model).


Stay tuned, there is more to be said about galaxy cluster and the Bullet Cluster in modified gravity theories in our next post.

by Anton Ippendorf, Pavel Kroupa and Marcel Pawlowski (30.07.2010): “But the Bullet Cluster … – Proof of Cold or Warm Dark Matter in galaxy clusters is but a myth” in “The Dark Matter Crisis – the rise and fall of a cosmological hypothesis” on SciLogs. See the overview of topics in  The Dark Matter Crisis.

13 thoughts on “5. “But the Bullet Cluster …” – Proof of Cold or Warm Dark Matter in galaxy clusters is but a myth

  1. I willi am excited to read more. Interesting article. And its so true, that each and everyone is shooting the “Silver” bullet (Cluster) wen it comes to DM.


  2. If gas is a major part of the mass of
    clusters and if this collision have stripped the gas off the clusters then
    the clusters should go unstable into
    a different rotation profile. Depending
    on how long ago the collision happened
    we should see some effects. I have published an all together different view of
    the origin of MOND.


  3. no “proof”, just “evidence”I was not aware that the authors of the first publication actually used the word “proof”. I fully agree that this is inappropriate.
    eagerly awaiting your next posts…


  4. Poor argumentYour whole article is bases around the straw man argument “Because there can be no proof of a scientific theory”. I accept this as true, but this is not the point of science. The point is that we accept a theory if it models the data correctly. In the case where multiple theories can model the data, Occum’s razor is used.
    No, I do not agree that the bullet cluster alone is sufficient to accept the DM hypothesis. What we can then do is look at all the other evidence and weigh it up. The lambda-CDM model provides an incredible description of a wide range of data see: http://www.sdss.org/…20031028.powerspectrum.html
    If you insist on pushing the MOND theories please point me to a publication that provides MOND predictions of: The power spectrum of the CMBR, large scale structure formation, intergalactic H clumping and rotation curves at least as good as those in the SDSS link above – AND it must be done with the same parameters, no tweaking on a case by case basis.
    This is why CDM is taken seriously and why MOND continues to lie on the fringe. Please stop playing the alternative scientist ‘victim’.


  5. @Jay “poor argument’Point 9 of
    contains a link to MOND and the CMB.
    Structure formation in a MOND-based cosmology needs to be worked out (by whom though?), but the results that already exist (by Llinares et al., e.g. http://adsabs.harvard.edu/abs/2011EAS….44…57L) show that structure formation is quicker, as is indeed required by the distribution of matter in the Local Volume (see Peebles & Nusser, 2010, Nature). Doing numerical structure formation in MOND is a much greater computational challenge though than in the comparatively simpler LCDM. And yes, locking up resources with LCDM does not help in probing alternative models (such as e.g. based on MOND).
    To claim that simplicity is the argument according to which theories are selected is silly: a nice counterexample is the necessity to go away from classical mechanics to quantum mechanics on small scales. Doing numerical computations in Newtonian mechanics is much simpler than in quantum mechanics (even today this is barely possible for all but the simplest molecules), but this is no argument against quantum mechanics (one would hope that Jay agrees here).
    LCDM is actually a bad theory: modelling the universe with LCDM is like trying to model a star based to more than 95 per cent unknown physics and with energy not being conserved. If it would work then fine, then we could argue that the 95 per cent unknown phyics is new physics. But on scales of the Local Volume and below LCDM fails on every test performed. If you do not believe Kroupa et al. (2010) then check out Peebels & Nusser (2010, Nature).
    “The lambda-CDM model provides an incredible description of a wide range of data” is true, but does not mean that another theory may not be better. It is clear that any new theory will have to agree with the observational constraints.
    The success of MOND is pointing towards a deeper understanding of gravity and mass and this is similar to the early pointers towards quantum mechanics before it was formulated properly much later: early pointers away from classical mechanics were the existence of spectral lines, Planck’s introduction of an unknown number h, and other effects. But 100 years ago Jay would probably have argued that the success of classical mechanics is so overwhelming that it’s apparent failures on the small scales are irrelevant. Luckily there were other very bright physicists who thought otherwise.


  6. Come Pavel, there is no need to stoop to mud slinging. Comments such as “But 100 years ago Jay would probably have argued
    that the success of classical mechanics is so overwhelming that
    it’s apparent failures on the small scales are irrelevant. ” do nothing to support your position and mearly detract from your otherwise well reasoned post. In attempting to point out my closed-mindedness you are only highlighting your own.
    Science is emperical, therefore burden of proof rests on those who assert the claim. As scientists we are required to change our minds, one need only supply adequate proof. Here, of course, the devil lies in the details. Until we can agree that adequate evidence has been supplied in favour of either theory, we are wont to disagree. This is a great attribute of science, the arguments are tested by the veracity of their opponents and only the fittest may survive.
    I did not claim that simplicity is an argument for dark matter, mearly that given multiple explanations – both with triumphs and pitfalls – the simplest explanation should be selected. Your following argument is therefore just knocking down another straw man. I am sure that here we disagree on which is the simplest theory, it is hard to qualify which theory makes the least assumptions.


  7. @JayJay, you are certainly correct that one needs to stay objective. But I have heard the statement you have made yourself (“The lambda-CDM model provides an incredible description of a wide range of data see: http://www.sdss.org/…031028.powerspectrum.html”) many times as an argument against MOND in particular. I fail to see this as being a relevant argument, because the LCDM fits rely on inventing a plethora of unknown and hitherto unproven physics, and LCDM fails on nearly every test that has been designed on scales of the Local Volume and below. At the same time it has been emerging that a MOND-based cosmology also seems to be able to account for important large-scale constraints, which implies that LCDM is definitely not unique in describing the large-scale data. This is why I am writing the comment which you have quoted.
    It is not “mud-slinging”, but a pointed statement to what I see as one of the central issues concerning progress in cosmological science: namely the unfortunate fact that resources are distributed according to what people believe (e.g. such as you that LCDM is the far superior model) rather than the actual evidence where we have the best data (namely Local Volume).


  8. JayPavel is not “mud slinging” when he points out that one hundred years ago people would have been arguing as you are doing now, except that then the issue was classical mechanics vs evidence for its break down in a certain regime, as Pavel points out. One may or may not agree with this, but it seems to be often forgotten that consistency of a model with data does not mean the model is right. In fact, one falsification suffices to bring down a model. The blog here appears to list many falsifications of LCDM, and so this should be taken seriously.

    [The end of the comment has been removed by the blog owners because it is not directly related to the scientific topic]


  9. simplicitythe issue of simplicity is something I would like to bring up again, since it was mentioned here, but also on other blogs (e.g. http://blogs.discovermagazine.com/…-fine-thanks/ ) First of all, LCDM is by no means a simple model, as it is based on a number of assumptions (usually tagged with a more positive term, namely “new physics”), but nevertheless, there is still a number of problems with the LCDM-model (see Question C.I in this blog).
    Admittedly, also MOND invokes “new physics”. But the “new physics” in MOND is very different from the “new physics” in LCDM, and the set of assumptions in the more successful model are more likely to represent useful formulations of natural laws. It therefore is worthwhile to check which set of assumptions is more successful in describing the observed universe.
    It is important to understand that simplicity by itself is not a good indicator for whether a model is good or bad. Sure, it is always wise to choose the simpler model, if two models are equally successful. But a simple model should be discarded for a more complicated one, if the complicated model is also significantly more successful.
    Right now, the status regarding successes and failures for MOND vs. LCDM seems to be following. MOND does a very fine job on the scale of galaxies, with NO case-by-case tweaking of parameters, making MOND superior to LCDM in describing galaxies. MOND does not allow a case-by-case tweaking for every galaxy and this makes the successes of MOND so impressive. My impression is that lately the LCDM community reacts to that by retreating to the description of the universe at even larger scales. I recall Simon White saying in a debate that galaxies are too complicated to describe them with LCDM. This, however, stands in contrast to the huge number of scientific papers that have been written on galaxies in the context of the LCDM-model. The attempts to describe the large-scale structure formation with MOND are preliminary compared to the efforts for LCDM. Thus it would be a bit unfair to make a comparison there.
    A popular example for the success of LCDM on large scales is the acoustic peaks in the micro-wave background. But let me emphasize that consistency with the acoustic peaks is no proof for LCDM. It simply is a test that every model has to pass. And indeed, there are also alternative models that pass this test, see e.g. this paper: http://de.arxiv.org/abs/0706.2443 . There is even a model where MOND is incorporated into a fit to the acoustic peaks, see this paper: http://arxiv.org/abs/0805.4014 . This does of course not prove any of those model, just as it does not prove the LCDM-model. My point is that the situation is less clear than some people claim. Many people simply tend to loudly advocate their favorite theory (which is LCDM in most cases).
    Let us stay open minded in our search for a good cosmological model. I suppose that the Universe will keep on surprising us.


  10. another perspectiveAs I understand, the blue ‘dark matter’ is derived from the difference between two derived quantities: The mass represented by the ‘ordinary’ galactic matter, estimated based on observed average luminosity; the estimated mass necessary to produce the weak gravitational lensing effects derived from statistical analysis of minute optical distortions identified in large number of distant background objects.
    The estimated dark matter represents the estimated additional mass necessary to produce the derived lensing effects.
    IMO, if the highly derived estimate of collective galactic mass has been underestimated or the highly derived lensing effects have been overestimated estimated, the dark matter identified will be invalid.
    I’m not a physicist but I’d expect that, since the clustered galaxies and proposed dark matter are regionally collocated, the collective curvature of spacetime produced by a cluster of sufficiently massive proximal galaxies could produce any detected weak gravitational lensing effects within the region without any dark matter…


  11. Observations and InterpretationsWhile I generally agree with this article’s analysis, I think there is another potential cause for the ‘missing’ galactic mass.
    Obviously, any estimate of the actual mass contained within a cluster of galaxies is at the very best a rough approximation with very large potentials for error. The observable luminosity of a spiral galaxy, for example, depends largely on the observer’s viewing angle. Viewed edge-on, the stars in the galaxy’s disc will appear to be largely non-luminous. Viewed face-on, apparent disc luminosity is greatly increased even though much of the disc’s massive stellar luminosity is obscured in visible light spectra. Viewed full-on in infrared spectrum it’s clear that much of the actual luminosity is produced by disc masses.
    However, the viewing angle obviously does not change actual galactic mass – only its apparent luminosity!
    While luminosity is a direct indicator of main sequence stellar mass, estimations of galaxy mass based on their observed luminosity is an imprecise approximation.
    For that reason, any conclusions based on the rough approximations of mass contained within a clusters’ galaxies are unreliable.
    IMO, far better evidence should be required for any conclusions regarding the presence of dark matter or modifications to gravitational theories!


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s