Anyone reading this blog has doubtless heard about the results announced a few weeks ago, observations of the “bullet cluster” claimed (in the title of the paper) to be “A direct empirical proof of the existence of dark matter.” (The basic idea is recounted better, and with prettier pictures, than I can do here by Sean in Cosmic Variance, and in their own press release.)

The bullet cluster is actually a pair of galaxy clusters that have recently slammed into one another. We call them “galaxy clusters” but in fact the galaxies themselves are a relatively small fraction of their mass. The rest is hot gas — shining in the x-rays — and, we think, dark matter. When the two clusters plowed into each other, the galaxies themselves, and the dark matter, just passed through, interacting only through gravity, but the gas actually collides, heating up in what is called a shock front. So we can easily observe that the galaxies, observed with optical telescopes, aren’t quite aligned with the gas, observed with the Chandra X-Ray satellite. Over the last decade it’s become possible to observe the mass distribution of clusters directly, using the technique of weak lensing. And it seems that the mass is aligned with the galaxies, not with the gas — much more mass, and more smoothly distributed, than the galaxies themselves. The argument rests on the idea that alternatives to dark matter, such as Modified Newtonian Gravity (MOND) would have the mass exactly tracing the light, specifically the x-ray-emitting gas. So: it must be dark matter. Case closed.

Well, sort of.

The simplest versions of MOND were always known to be too simple to apply to the largest scales such as clusters and the Universe as a whole (i.e., cosmology). But more recently, Bekenstein has created a “relativistic” version of the theory which, Skordis and collaborators have shown, reproduces at least some cosmological observations. This theory, known as TeVeS (for Tensor, Vector, Scalar gravity) is hardly simpler than a theory with Dark Matter; as the name implies to physicists, it requires a vector and a scalar field in addition to the metric tensor that characterizes Einstein’s relativity. These fields don’t weigh much — they’re not the dark matter — but the forces that they implicitly exert mimic its effects.

The vector field, in particular, can have unexpected repercussions beyond this, however: it’s a vector, an arrow, which means it has a direction. It breaks the symmetry of the situation, and could, in specific circumstances displace some gravitational effects from others (the lensing from the light, for example, perhaps in exactly the same way as dark matter — a distinction without a difference?). No one has performed the required calculations yet, but other groups have argued that other possibilities, such as Moffat’s MOG and massive neutrinos combined with these MOND-like theories could in principle explain offsets between lensing and light. Indeed, they point out that MOND-like theories have already had a problem with explaining observations of clusters in which the details of the mass distribution had rarely lined up with the light. (See this Cosmocoffee discussion for more, from the authors of these papers themselves.)

These responses point to the real worry here: any single observation can be refuted. (Worse, of course, is the simple fact that lots of observations are wrong, and it’s impossible to know at the time which ones.) Moreover, such an argument from a single case is the same tactic used by those offering weak evidence against the current paradigms (not to mention the real crackpots). Despite claims to the contrary, science does not progress by simple falsification, by the single case that brings down the old paradigm. Extraordinary claims require extraordinary evidence. Back when it was first suggested by Zwicky in the 1930s, the existence of dark matter was an extraordinary claim. Now, dark matter, especially the weakly interactive massive particles that seem to be a natural corollary to supersymmetric theories in particle physics, is a much more parsimonious explanation of the various observations of galaxies, clusters and the cosmos as a whole than these baroque theories. Refuting its existence would certainly be extraordinary — but so would declaring the issue completely finished, at least until its definitive non-gravitational detection.