Dark Matter Exists

The great accomplishment of late-twentieth-century cosmology was putting together a complete inventory of the universe. We can tell a story that fits all the known data, in which ordinary matter (every particle ever detected in any experiment) constitutes only about 5% of the energy of the universe, with 25% being dark matter and 70% being dark energy. The challenge for early-twenty-first-century cosmology will actually be to understand the nature of these mysterious dark components. A beautiful new result illuminating (if you will) the dark matter in galaxy cluster 1E 0657-56 is an important step in this direction. (Here’s the press release, and an article in the Chandra Chronicles.)

A prerequisite to understanding the dark sector is to make sure we are on the right track. Can we be sure that we haven’t been fooled into believing in dark matter and dark energy? After all, we only infer their existence from detecting their gravitational fields; stronger-than-expected gravity in galaxies and clusters leads us to posit dark matter, while the acceleration of the universe (and the overall geometry of space) leads us to posit dark energy. Could it perhaps be that gravity is modified on the enormous distance scales characteristic of these phenomena? Einstein’s general theory of relativity does a great job of accounting for the behavior of gravity in the Solar System and astrophysical systems like the binary pulsar, but might it be breaking down over larger distances?

A departure from general relativity on very large scales isn’t what one would expect on general principles. In most physical theories that we know and love, modifications are expected to arise on small scales (higher energies), while larger scales should behave themselves. But, we have to keep an open mind — in principle, it’s absolutely possible that gravity could be modified, and it’s worth taking seriously.

Furthermore, it would be really cool. Personally, I would prefer to explain cosmological dynamics using modified gravity instead of dark matter and dark energy, just because it would tell us something qualitatively different about how physics works. (And Vera Rubin agrees.) We would all love to out-Einstein Einstein by coming up with a better theory of gravity. But our job isn’t to express preferences, it’s to suggest hypotheses and then go out and test them.

The problem is, how do you test an idea as vague as “modifying general relativity”? You can imagine testing specific proposals for how gravity should be modified, like Milgrom’s MOND, but in more general terms we might worry that any observations could be explained by some modification of gravity.

But it’s not quite so bad — there are reasonable features that any respectable modification of general relativity ought to have. Specifically, we expect that the gravitational force should point in the direction of its source, not off at some bizarrely skewed angle. So if we imagine doing away with dark matter, we can safely predict that gravity always be pointing in the direction of the ordinary matter. That’s interesting but not immediately helpful, since it’s natural to expect that the ordinary matter and dark matter cluster in the same locations; even if there is dark matter, it’s no surprise to find the gravitational field pointing toward the visible matter as well.

What we really want is to take a big cluster of galaxies and simply sweep away all of the ordinary matter. Dark matter, by hypothesis, doesn’t interact directly with ordinary matter, so we can imagine moving the ordinary stuff while leaving the dark stuff behind. If we then check back and determine where the gravity is, it should be pointing either at the left-behind dark matter (if there is such a thing) or still at the ordinary matter (if not).

Happily, the universe has done exactly this for us. In the Bullet Cluster, more formally known as 1E 0657-56, we actually find two clusters of galaxies that have (relatively) recently passed right through each other. It turns out that the large majority (about 90%) of ordinary matter in a cluster is not in the galaxies themselves, but in hot X-ray emitting intergalactic gas. As the two clusters passed through each other, the hot gas in each smacked into the gas in the other, while the individual galaxies and the dark matter (presumed to be collisionless) passed right through. Here’s an mpeg animation of what we think happened. As hinted at in last week’s NASA media advisory, astrophysicists led by Doug Clowe (Arizona) and Maxim Markevitch (CfA) have now compared images of the gas obtained by the Chandra X-ray telescope to “maps” of the gravitational field deduced from weak lensing observations. Their short paper is astro-ph/0608407, and a longer one on lensing is astro-ph/0608408. And the answer is: there’s definitely dark matter there!

Despite the super-secret embargoed nature of this result, enough hints were given in the media advisory and elsewhere on the web that certain scientific sleuths were basically able to figure out what was going on. But they didn’t have access to the best part: pictures!

Here is 1E 0657-56 in all its glory, or at least some of it’s glory — this is the optical image, in which you can see the actual galaxies.

1e0657 optical

With some imagination it shouldn’t be too hard to make out the two separate concentrations of galaxies, a larger one on the left and a smaller one on the right. These are pretty clearly clusters, but you can take redshifts to verify that they’re all really at the same location in the universe, not just a random superposition of galaxies at very different distances. Even better, you can map out the gravitational fields of the clusters, using weak gravitational lensing. That is, you take very precise pictures of galaxies that are in the background of these clusters. The images of the background galaxies are gently distorted by the gravitational field of the clusters. The distortion is so gentle that you could never tell it was there if you only looked at one galaxy; but with more than a hundred galaxies, you begin to notice that the images are systematically aligned, characteristic of passing through a coherent gravitational lens. From these distortions it’s possible to work backwards and ask “what kind of mass concentration could have created such a gravitational lens?” Here’s the answer, superimposed on the optical image.

1e0657 optical and dark matter

It’s about what you would expect: the dark matter is concentrated in the same regions as the galaxies themselves. But we can separately make X-ray observations to map out the hot gas, which constitutes most of the ordinary (baryonic) matter in the cluster. Here’s what we see.

1e6057 optical and x-ray

This is why it’s the “Bullet” cluster — the bullet-shaped region on the right is a shock front. These two clusters have passed right through each other, creating an incredibly energetic collision between the gas in each of them. The fact that the “bullet” is so sharply defined indicates that the clusters are moving essentially perpendicular to our line of sight.

This collision has done exactly what we want — it’s swept out the ordinary matter from the clusters, displacing it with respect to the dark matter (and the galaxies, which act as collisionless particles for these purposes). You can see it directly by superimposing the weak-lensing map and the Chandra X-ray image.

1e6057 optical, dark matter, and x-ray

Clicking on each of these images leads to a higher-resolution version. If you have a tabbed browser, the real fun is opening each of the images in a separate tab and clicking back and forth. The gravitational field, as reconstructed from lensing observations, is not pointing toward the ordinary matter. That’s exactly what you’d expect if you believed in dark matter, but makes no sense from the perspective of modified gravity. If these pictures don’t convince you that dark matter exists, I don’t know what will.

So is this the long-anticipated (in certain circles) end of MOND? What need do we have for modified gravity if there clearly is dark matter? Truth is, it was already very difficult to explain the dynamics of clusters (as opposed to individual galaxies) in terms of MOND without invoking anything but ordinary matter. Even MOND partisans generally agree that some form of dark matter is necessary to account for cluster dynamics and cosmology. It’s certainly conceivable that we are faced with both modified gravity and dark matter. If the dark matter is sufficiently “warm,” it might fail to accumulate in galaxies, but still be important for clusters. Needless to say, the picture begins to become somewhat baroque and unattractive. But the point is not whether or not MOND remains interesting; after all, someone else might come up with a different theory of modified gravity tomorrow that can fit both galaxies and clusters. The point is that, independently of any specific model of modified gravity, we now know that there definitely is dark matter out there. It will always be possible that some sort of modification of gravity lurks just below our threshold of detection; but now we have established beyond reasonable doubt that we need a substantial amount of dark matter to explain cosmological dynamics.

That’s huge news for physicists. Theorists now know what to think about (particle-physics models of dark matter) and experimentalists know what to look for (direct and indirect detection of dark matter particles, production of dark matter candidates at accelerators). The dark matter isn’t just ordinary matter that’s not shining; limits from primordial nucleosynthesis and the cosmic microwave background imply a strict upper bound on the amount of ordinary matter, and it’s not nearly enough to account for all the matter we need. This new result doesn’t tell us which particle the new dark matter is, but it confirms that there is such a particle. We’re definitely making progress on the crucial project of understanding the inventory of the universe.

What about dark energy? The characteristic features of dark energy are that it is smooth (spread evenly throughout space) and persistent (evolving slowly, if at all, with time). In particular, dark energy doesn’t accumulate in dense regions such as galaxies or clusters — it’s the same everywhere. So these observations don’t tell us anything directly about the nature of the 70% of the universe that is purportedly in this ultra-exotic component. In fact we know rather less about dark energy than we do about dark matter, so we have more freedom to speculate. It’s still quite possible that the acceleration of the universe can be explained by modifying gravity rather than invoking a mysterious new dark component. One of our next tasks, then, is obviously to come up with experiments that might distinguish between dark energy and modified gravity — and some of us are doing our best. Stay tuned, as darkness gradually encroaches upon our universe, and Einstein continues to have the last laugh.

This entry was posted in Science. Bookmark the permalink.

197 Responses to Dark Matter Exists

  1. Nicholas says:

    Incredible stuff!

    As always, explained very intelligibly.

    Do you think this will be the final work on MOND? 🙂


  2. Pingback: Crooked Timber » » Dark matter

  3. JD says:

    So here’s something that’s been bothering me for literally years: if dark matter is dark because it doesn’t interact with “normal” matter, does it interact with itself? If so, using what forces? If not, how does it ever clump? Wouldn’t it need some repulsive — or at least scattering — force to keep the constituents of a clump from simply passing “through each other” as they collapse? Or would a clump be truly dynamic, like a three-dimensional fountain? Or is this all left as an exercise for the reader?



  4. Belizean says:

    Bravo, Sean! Thanks again for another excellent post.

  5. Sean says:

    JD– the dark matter probably interacts with itself, but only weakly (thus “weakly interacting massive particles”). And, of course, through gravity. It’s the gravitational force that causes it to clump together, and the accumulated effect of small perturbations causes the distribution to “relax” rather than just having the particles zip through and out the other side.

  6. Pingback: MOND Laid to Rest? - Asymptotia

  7. Navneeth says:

    That’s a very nice post!

    With all the extensions, my Firefox browser is a bit slow. I think I’ll use an “empty” profile to do that tabs thingy. 🙂

  8. Cecil Kirksey says:

    If dark matter exceeds visible matter by a factor of 3-5 why doesn’t the dark matter clump due to gravity to form large structures such as stars and galaxies? BTW I realy enjoyed reading your review paper on the cosmological constant a few years back. As a layman, but retired engineer very much interested in theoritical physics, I am interested in the latest ideas. Regarding the CC: if the QFT (and ST) prediction of the CC value is off by 60 to 120 orders of magnitude what does this say about our view of the validity of these theoritical ideas vis-a-vis the vaccuum energy density?

    Really enjoy your blog.

  9. Aaron Bergman says:

    Hey Sean,

    What proportion of matter in the cluster is in the stars and the like versus in the gas?


  10. Sean says:

    Aaron– it’s about 90% gas, depending on the details.

    Cecil– the dark matter does clump into galaxies and clusters. Even into star-sized things, just not nearly as efficiently as ordinary matter, which can collide and dissipate its energy. But for galaxies and clusters, most of the matter is dark, not ordinary.

    About the cosmological constant (which is the same idea as “vacuum energy,” just with a different name), it’s clear there’s something important we don’t know. Wish I could tell you what it is.

  11. Pingback: Cycle Quark » How to See Dark Matter

  12. Pingback: Dynamics of Cats

  13. Eugene says:

    MOND is dead. Long live MOND!

  14. Count Iblis says:

    Sean, any news on constraints on DM-DM cross section as a function of the assumed DM particle mass?

  15. B says:

    Hi Sean,

    great post, thanks for the explanation!

    the dark matter does clump into galaxies and clusters. Even into star-sized things, just not nearly as efficiently as ordinary matter, which can collide and dissipate its energy.

    How do we know dark matter clumps into star-sized things? Is there evidence for that?

    Best, B.

  16. Sean says:

    Count Iblis, the actual papers did calculate a limit on the DM-DM scattering cross-section, but I don’t know the number; you should have a look at the papers directly.

    B, the DM should clump on all scales, no reason why not. But we don’t have any direct evidence (and a one-solar-mass DM overdensity would be much more diffuse than a real star). Furthemore, on those scales you have to worry about being disrupted by tides etc — so I really don’t know, again you’d have to ask an expert.

  17. Aaron says:

    Sweeeet! Loving the third picture. 🙂 I do have a question about something you mentioned:

    In most physical theories that we know and love, modifications are expected to arise on small scales (higher energies)

    What’s the connection between scale and energy? Are you just saying that at smaller scales, forces are stronger, and potential energies are higher, or is there something more going on?

  18. Scott Dodelson says:

    The pictures are beautiful, but I couldn’t find a link on the Chandra site to any paper.

    Several questions: how close are the redshifts of the clusters? The mass determination presumably was done with weak lensing since no arcs are visible. How many background galaxies were there for each cluster. The key result is that the mass distributions are offset from the gas. So the centers of the mass distributions must have been taken as free parameters in the lensing fits. What were the other free parameters in the fits? The key evidence if you think about it is that the lensing distribution is not circularly symmetric around the center of mass. [Recall that Bekenstein’s theory was constructed to get lensing right, so we expect lensing even far from the mass distribution.] So, how exactly did they do the fits? Was their mass model robust enough to allow for a circularly symmetric profile? And if so, with what confidence is such a profile ruled out?

    I know this sounds like raining on a parade, but, as perhaps Eugene was suggesting, dark matter theories themselves have died many deaths. And some of the most dramatic have been delivered with pictures [think of the early APM data of the galaxy distribution or the Moore et al. simulations of the number of satellites expected in a galaxy].

    Bottom line for me: Dark matter theories are way ahead of Mond and its extensions, and this latest result is yet another hurdle. But I still think it is interesting to explore the alternatives.

  19. Levi says:

    A wealth of links, four nice pictures to look at closely, and a very clear explanation. This is what blogs are for!

    MOND is dead! MOND will stay dead! (probably)

  20. Sean says:

    Scott, man, such a wet blanket. The paper is here. It’s a Letter, so I don’t think they go into much detail about your questions, which are good ones.

    But I don’t think it’s quite right to imply that Bekenstein’s theory could match the lensing observations. His theory is supposed to get the magnitude of the lensing right in a circularly-symmetric profile, but I don’t see how it could give rise to two big blobs of gravitational field offset from the mass distribution. Unless there was substantial energy density in the independent excitations of the new Bekenstein fields themselves — in which case it would really just be an especially messy and unmotivated version of dark matter.

    Also: even in Bekenstein’s theory it’s necessary to have dark matter (e.g. neutrinos) to simultaneously fit galaxies and clusters. The point of this new result is not “MOND can’t do it alone,” since we already knew that. It’s “no modified-gravity theory can do it alone, we need dark matter.”

    Aaron, in quantum mechanics, higher energies correspond to shorter distances. So as we probe smaller and smaller scales, new high-energy phenomena come into play. Nothing analogous usually occurs when we go to larger scales, but it’s good to be open-minded.

  21. Count Iblis says:

    Sean, I’ll have a look.

    Scott the papers will appear 2½ hours from now on arXiv:



    The first one can already be downloaded via the link Sean gave:


    I guess the DM-DM scattering cross-section is on the other paper. 🙂

  22. Pingback: Galactic Interactions » Blog Archive » Colliding Galaxy Clusters, the Chandra X-Ray Space Telescope, and “Exotic” Dark Matter

  23. Pingback: Life on the Lattice

  24. Acolyte says:

    Very nicely written and explained.

    A question: My (basic) understanding of gravitational lensing is that you measure the amount (in degrees) that angle is bent, and you infer how much mass is causing the bending. Stronger masses bend light more. But the relationship between “angle of bending” and “mass” is governed by an application of general relativity. If MOND were true, would this not imply that a smaller amount of matter (or dark matter) would be able to bend light more strongly than predicted by general relativity if it passed by at a greater distance to the matter causing the bending?

  25. Sean says:

    Acolyte, it’s not just the angle; it’s the direction in which the light is bent. The important thing here is that the gravitational field is pointing somewhere other than where the ordinary matter is located.