Steven Krivit wrote: Let me know if that helps clarify. >
I think that is clearer. You can see that the power levels are really not that varied, and there is practically no real-time correlation to the helium. Actually, the error bars are probably even larger because the instrument is at the limits of detection. At higher levels both accuracy and precision improve. That's true of most instruments. Plus, I have no doubt the helium levels really do vary, and this is not just an instrument artifact, for the reasons given in the document: different amounts of helium are captured in the near surface layers of the metal, depending on complex factors. That is well established. You have to go to great lengths to recover all of the helium. During the run the stuff will suddenly vent at odd times, presumably when a crack forms. This is how marginal experimental data looks. If the correlation was better than this I would begin to suspect someone doctored the data. With an instrument that can detect 1 ppb minimum, the difference between 1 and 3 ppb is hardly meaningful. On the other hand the difference between 100 ppb and 101 ppb is more significant and reliable. - Jed
