In reply to TobyA:
I happened on this thread and think I can offer some insights without invoking too much of the underlying physics.
A rope provides variable resistance to elongation, and this gives it the ability to absorb high fall energies by becoming stiffer and stiffer as it elongates. In fact, resistance is proportional to percentage stretch, not to absolute elongation, and this translates into the fact that peak loads depend, ideally, on the fall-factor and not on just the height of the fall.
A Screamer is different in two ways. (1) When ripping, the Screamer offers a constant resistance---unlike the rope this resistance does not increase as the Screamer elongates. (2) The Screamer has a fixed length that does not increase as the climber advances up the pitch.
The consequence of these two conditions is that there is a fixed maximum amount of fall energy a screamer can absorb (equal to its full elongation times the resistance it provides). As the fall height gets bigger and bigger, the fixed amount of energy the screamer can absorb becomes a smaller and smaller portion of the total fall energy, and all the rest will still have to be absorbed by the rope. For this reason, the effect a screamer can have decreases as the height of the fall goes up; there is nothing like the relative effects of the fall factor that come from a scalable rope.
One way to view a screamer's effect on fall energy is that in terms of peak load to the gear, it essentially "shortens" falls by a fixed amount, which a back-of-the-envelope calculation suggests is a meter or so (depending of course on the weight of the faller). So if the height of your fall is two meters, the screamer absorbs half the total fall energy and your gear gets the impact of a one-meter fall, a significant lessening of the peak load. On the other hand, if your total fall is six meters, the screamer still only absorbs a meter's worth of fall energy, your gear gets the impact of a five-meter fall, and the effect on the peak load is negligible. For this reason, Screamers are only useful, in theory, for short falls.
In reality, all this may be optimistic. The CAI did a bunch of tests
http://www.caimateriali.org/index.php?id=27
that suggest that Screamers don't contribute significantly to peak load reduction in general.