DAMA explanation?

You may have heard of the DAMA dark matter search experiment. In the Grand Sasso lab they put 100kg of extremely pure scintillator, and looked for annual variations in the rate of signals.

The basic idea is: assume the dark matter in the galaxy is more or less like a fog that the orbiting stars sweep through. We know the direction our Sun is going, and we know the Earth's orbit. At one point in the orbit we're heading in the same direction as the Sun's motion, and at the opposite point we're heading away. Dark matter particles scattering off nuclei when we're going faster should involve a bigger kick than those when we're going slower--and that bigger kick should appear as more light. (Of course, if the dark matter is orbiting the galactic center at about the same speed as the Sun, you won't see much energy from the collisions--but you might still see a little variation.)

So, if on the average they see more light (energy deposited) at the times of the year when the Earth is moving faster wrt the galactic center, that might be evidence for dark matter.

They claim to see that. They've been more than a little reluctant to show their raw data, though, and the backgrounds must be huge (cosmic rays, radioactivity in the rock, residual impurities, etc).

The plots look great though. The problem is that they suggest interaction rates so high that other experiments should be able to see it too--and they don't.

There's an interesting paper from INFN (Buttazzo et al) that suggests that the annual variation (this story has some nice explanatory plots) is an artifact of the way they analyze their data. The background rate is too high to see anything clearly, so they average the background year by year, subtract that from the data on a every-few-months binning, and accumulate those residuals for a bunch of different years.

That seems OK, but what happens if the background rate grows with time? (e.g. impurities on the surface of the crystals migrate into the bulk volume, or helium migrating into the phototubes causing more after-pulsing)

Answer: you get residuals that look like a sawtooth: negative at the start of the reference period and positive at the end. If you use the DAMA start and end points and fit with a sinusoid, you get a peak at about the same place DAMA does.

Since we've never seen their raw data, or their analysis chain in any detail, we've no idea whether their background rate does grow over time.

There should be a simple check that lets them keep their data secret--use different start and end reference months, or accumulate over 2 years instead of 1 (you'll get worse statistics that way, but the sawtooth would then have a period of 2 years instead of 1. If the sinusoid remained with a period of 1 year and the same peak, even if it looked muddier than it does now, they might have something. If it had a 2-year variation (in the second case) or the peak appeared in a different month (for the first cross-check), they have zip.

I look forward to hearing what DAMA has to say.