I've got a backlog of papers I should try and read that's so long it's preventing me from reading any at all, but at long last I managed to knuckle down and read one.
Today's paper sees the return of our old friend, the Radial Acceleration Relation. Very briefly, this is the observed strong correlation between the observed and theoretical prediction for acceleration of material in galaxies. What's supposedly odd about this is that we can predict the acceleration very well without knowing anything about the dark matter at all, which other observations show is usually dominant over the ordinary matter we have to observe. It's a bit like being able to predict the acceleration of a horse without knowing that it's on a train : the fact that the horse has four legs ought to tell you nothing about the power of the train's engine, but apparently it does.
As shown many times, however, this initially surprising result doesn't look to be anything very interesting, as it falls out quite naturally from bog-standard simulations of galaxies. It seems it's just a perfectly normal scaling relation based on the dark matter; in my silly (and flawed) analogy, it's as though there are actually very good reasons to expect a certain kind of horse to always be found on a certain type of train. True, theories without dark matter (most notable MOdified Newtonian Dynamics, MOND) predicted this in advance, whereas the standard model didn't, but that doesn't change the result that there appears to be no way to distinguish the expected behaviour from the two competing theories. Hence the RAR is not much use to anyone, really.
But hold on, say the authors of this paper*, maybe there is a way to tell them apart after all - in which case, I recant, and the RAR does potentially become very interesting again. And ironically, this may come from that much-maligned feature of MOND, the External Field Effect.
* I know two of them and indeed we collaborate. I was not involved in this research at all, which is something on which we usually disagree.
This is far beyond my level of specialist understanding, but the basic deal is that the MONDian behaviour of a gravitational system breaks down if it's present in a sufficiently strong external gravitational field. So MOND gives different predictions for identical system in a way that's much more strongly dependent on its environment than in the standard model of gravity.
Consider a star in the outskirts of a galaxy, orbiting at a very small acceleration. If we neglect dark matter, Newton's theory says the star should be orbiting at a much lower velocity than MOND predicts.
At least, that is, if the galaxy is isolated. If instead our galaxy is also close to a second, massive galaxy, then the speed of its most extended material becomes greatly reduced - everything fades back into ordinary Newtonian behaviour. So for MOND, the rotation curve depends not just on the individual system but also its environment. And this just doesn't happen (at least, not to anything like this level of strength) with Newtonian gravity.
This potentially gives a neat way to discriminate between MOND and the standard model. MOND implies that we should see declining rotation curves if systems are not sufficiently isolated, whereas dark matter says their environment is basically irrelevant, except for secondary effects like galaxy evolution. On the other hand, the dark matter paradigm doesn't say you can't ever have a dark-matter-free galaxy, it just doesn't say there should be any correlation with something as neat, say, a cluster-centric distance.
In short, the speed of the orbiting material of an isolated galaxy can have flat rotation curves (high speeds at high distances) under MOND, but they ought to have declining rotation curves (lower speeds at higher distances) if the galaxy isn't sufficiently isolated. In the standard dark matter paradigm, by contrast, they should almost always have flat rotation curves, simplifying slightly.
This paper looks at the velocity dispersion of Ultra Diffuse Galaxies in clusters to see if any influence of this External Field Effect can be detected. Being extremely low density, UDGs are good tests since their stars should be orbiting at low accelerations around their galactic centres. Being inside clusters, the EFE should be strong, so they should be a good test case where the EFE can most easily overcome the internal gravitational fields of the galaxies.
What they find, after some very extensive analytical modelling, is... unclear.
Or rather, it isn't. It's just that what they say they find doesn't look much like what I think the figures suggest they actually do find.
They have a sample of ten UDGs. Of these, four have velocity measurements at many different distances, i.e. with proper rotation curves. The rest have only one data point each, and I would tend to neglect these as they don't seem to show anything very much. But the ones with multiple points do look useful. For example, here's one case, comparing the observations (points with error bars) with MOND predictions that neglect the EFE (sold lines, under a few different assumptions) :
Which is I think very clearly a much better agreement with the data. It's the same story for the other three galaxies : MOND with the EFE works better than without. As you would hope, really, at least if you're a MOND supporter. Pretty neat.
The problem is that the authors say the exact opposite, that the EFE makes things worse ! If that were so then this would be a big problem for MOND. As a MOND skeptic, I'd have a field day (pun intended, oh hah hah). So I find myself in the strange position of preferring dark matter but defending these results against the MOND-advocating authors...
It gets even stranger as they suggest several possibilities to save MOND, not least of which is that the EFE could be screened in clusters by some kind of "dust" (hence His Dark Materials). How this is supposed to work is unclear to me.
This all feels very weird. A more natural narrative would be that they've shown that the EFE can quantitatively explain the unexpected dynamics of specific galaxies, hence the popular opinion that MOND can't allow declining rotation curves is falsified. Had they done that, I'd say, "Yep, you've got a point. I still don't like MOND, but this allows a testable prediction, so good job."
That they claim the opposite is very confusing. If the EFE could be somehow be screened, this would seem to be a powerfully anti-Occam argument against the whole idea of modified gravity : if you're going to allow it to work so inconsistently, that would seem to make it unphysical. Yet there doesn't seem any need to invoke anything like this, here, rather the data seems to show that MOND's EFE might not be such a get-out-jail-free, fit-anything-to-anything panacea fudge factor as it sometimes appears.
In conclusion, I don't know.
