Of course you can color me skeptical. As Chad Orzel points out, the signal they are talking about is only about 3 times as strong as their noise. Now when you look at one of their runs, i.e. figure 4 of gr-qc/0603033, the peaks look pretty good, no? Well figure 4b is a little strange: the gravitomagnetic effect appears to occur before the acceleration. Okay a bit strange, but a single run proves nothing, right? Okay, what about figure 5? Ignore the temperature dependence now, but would you have picked out the peaks that they picked out? Okay so these things make me a little uneasy. Okay, so well certainly they did a lot of runs and tried to get some statistics on the effect. Indeed, they did something like this. This is figure 6. And this is what makes the paper frustrating: “Many measurements were conducted over a period from June to November 2005 to show the reproducibility of the results. Fig. 6 summarizes nearly 200 peaks of in-ring and above-ring tangential accelerations measured by the sensor and angular acceleration applied to the superconductors as identified e.g. in Fig 4 with both electric and air motor.” Why is this frustrating? Well because I have no clue how they analyzed their runs and obtained the tangential accelerations. Were the peaks extacted by hand? (And what is the angular acceleration? Is it the average acceleration?) Argh, I can’t tell from the paper.