Now that's not a bad attempt at one way of testing how bearing seals will resist the ingress of water. I think it's deeply flawed but credit to kaesae he's making an effort to do something scientific, i.e. objectively repeatable and verifiable.
I think the major flaw is that it doesn't adequately recreate the sort of conditions bearings tend to be subjected to in real life. In fact the only way (well, the cheapest way) I can think of to test bearings alongside each other in real life situations would be to fit two different sets to a bike.
A symmetrical suspension design would be a good start; I know Specialized supply their dealers with frame prep and bearing pulling/seating kits so something like an Epic would be good, the mechanic could ream and face each mount point in the frame to ensure all fittings are clean and the angles are good. Then, on the drive side of the frame fit Katec bearings and on the non-drive side fit another brand. Granted, this only allows you to compare two kinds at a time, and you would probably have to repeat the test with new bearings and mounted on the opposite sides of the frame in order to ensure that an asymmetrical load through the frame didn't skew the results, but with time and effort this would probably be the best way to actually compare the bearings. It would simply be a matter of seeing which side developed play first, how fast that play developed and what state the bearings were in after prolonged testing.
Just saying, like.
: P