In reply to Chris Reid:
Last year while I was in Ouray, I witnessed someone testing ice screws by dropping a 50-ish (I think, I can't remember) weight onto them statically in a factor 2 fall. We spoke to the guy and his conclusions were:
The quality of the ice is the over-riding factor. Screws in crappy ice or that had been placed too close together failed regularly
Longer screws did hold more often than short screws, although I don't know what the mechanics behind this was
Screws usually failed by pulling out rather then breaking
Longer screws tended to bend, usually one or two inches along the shaft from the hanger. The did not necessarily mean failure
The second most common form of damage to the screws is failure of the hanger
The guy professed himself to be quite impressed with stubby screws, although the longer ones are better if you can place them
The guy was quick to point out that this level of unloading is highly unlikely to occur in a real scenario because you shouldn't be factor 2-ing a single unequalised screw, and you certainly shouldn't be doing it on a static line! However it was scary to watch how many screws were popping out the ice, or more frequently ripping a big chunk out the ice (the screw would remain within the placement, but a big lump would come off still attached to the screw). Most of those screws were either bent or had no visible damage (although he was taking them away for proper analysis and I don't know what the outcome was). It was particularly scary as this was Ouray in a fat year and the ice was good quality. Suffice to say there weren't many people leading in the area that day!
The over-riding message though was that it doesn't really matter what you place so much as where you place it. Make sure the ice is solid and check you aren't too near old placements.
The guy was going to publish somewhere, I believe in a scientific or engineering journal rather than a climbing mag, so the data should be available, although I can't remember where it was .....