No axions for you!
December 7, 2006
If you were to read this U Buffalo press release or drop by Slashdot, you might be left with the impression that the axion has been found. This is not the case, unfortunately. (We’d love for it to be found, really.)
The axion is a particle postulated to solve the strong CP problem, one of the fundamental open questions in particle physics. One of the general principles that guides particle theory is that coincidences are unlikely: if a parameter is generally allowed to be large by the theory, and in the real world it turns out to be small, then you have to explain why it should be small. The parameter in this case is an angle called θ, which a priori could be anything (and has no obvious reason to be zero), but which is constrained by experiment to be less than 10-9 or so. The famous Peccei-Quinn mechanism uses an extra particle to drive θ back to zero dynamically.
I’ve heard varying reports on the accessibility of the actual paper that’s supposed to claim particle discovery (Journal of Physics G34, 129, which I can get using Cornell’s subscriptions). It’s not on arxiv, but there may be preprints floating about. You may need to do a bit of work to get it if you actually want to read it.
The experiment in question appears to be a heavy ion-on-emulsion affair, with only about a thousand actual nuclear interactions from which all conclusions are drawn. The claimed detection channel is electron-positron pairs with displaced vertices, corresponding to a lifetime within a couple of orders of magnitude of that of the D0. Only a thousand or so pairs are used in drawing conclusions; the authors claim to see two excesses of events near 7 and 19 MeV which they attribute to exotic particles, “more than three standard deviations” above background (their italics).
To this physicist, the bumps in question look like perfectly normal statistical fluctuations in a small sample: if you don’t know beforehand where to look for a peak, you take a penalty factor in how significant you can claim your result to be because you get to search in many places, and some of them are going to fluctuate. (The whole blind analysis craze has apparently passed the authors by.) Even if you take the paper’s claims of the statistical significance at face value, though, this does not count as a “discovery”; it is at best “evidence.” Discoveries require five standard deviations (more or less) and reliable, repeatable independent confirmation; that last part is critical (witness the pentaquark fiasco, as belligerently reported and retracted by the New Scientist).
The authors claim that their detection method makes them uniquely sensitive to axions decaying in this mode. This is nonsense. If they can actually pull a signal of an incredibly long-lived (and thus narrow) particle out of a thousand nuclear interactions, this should have been seen all over the place, whether or not the detectors are “electronic.”
This is a lot of text to spend on a non-discovery, but the fast pickup, extending as far as Wikipedia, makes me very annoyed. There is clearly a popular audience which is interested in particle physics results, but which isn’t able to judge the plausibility of fringe claims (as any visit to a Slashdot “science” comment section will show), and who will probably remember and repeat “axion found!” even if (ha, ha) a retraction were posted. I don’t know how one might go about amelioration, though.
As is to be expected, the PDG will tell you everything you ever wanted to know about axion searches, and then some.