I remember someone posted something here about a machine they made to roll their dice. Using robotics and a video scanner to input the result into excel.
So he was able to roll each die 5,000+ times. He found the dice were actually fairly decent as a whole, some where better than others. But as a set they rolled about as expected.
Edit: The thing about the saltwater test. That's great for a D20, because in theory each side should come up about the same. But for D&D, Pathfinder, ect... A 20 is always good. So if you find your D20 is weighted in such a way that the 20 is more likely to come up, that's a 'good' die.
The same doesn't really work with X-Wing dice, since they effectively have 3 results vs 20, and each die has multiple sides with a good result.
Edited by VanorDM