My dwarves have been doing some Science lately, trying to figure out how high gibs can bounce. (This is pure theoretical research; probably no practical applications.) First approach involved a 25 Z-level diving board over a roof-top waterless swimming pool. Test runs with puppies show it working okay, but the most interesting result was that the wind that's providing 40 power/windmill doesn't seem to perturb the trajectory of the test puppies. Shockingly, I think one test-goblin managed to survive, but I suspect that Urist McCorpsedragger was just on the wrong tile at the wrong time and managed to break his fall.
Built another (solid gold) diving board off to the side that's 27 Z-levels down to a 1x1 hole in the ground, which gradually widens over 10-15 more z-levels to a 1x1 constructed steel floor-plate in a roughly 15x15 chamber. Added bonus: the bottom level is pure smoothed diorite, and the petrified wood crosshairs on the ground give the chamber a grim aesthetic perfect to the task. Form and function in perfect harmony.
Gibs are bouncing at least two, maybe three or more z-levels up. Doing more hollowing out in some of the z-levels above to give the giblets more freedom to arc, and perhaps reduce experimental error. Considering trapping the kill-tile, in case that serves to tenderize or "loosen up" the test subject and perhaps yield better results.
One interesting early result was that a test-puppy's head seemingly went shooting off on a diagonal along the round, while the legs did the more parabolic arc the dwarves were expecting. This has been noted as a possible avenue for future work.