Typical practice among race engineers is to express chassis stiffness in foot-pounds per degree of twist. Of course an F-1 tub won't twist anywhere near 1 whole degree, and they have all kind of test rigs to give them NASA accurate numbers, but the results are still expressed the same way.
It would be pretty easy to test a 914 tub. Your rig would need to secure one end, let's say the rear, to the floor by the top shock mounts since this is where the vertical loads are fed into the tub (Chassis twisting loads are the result of vertical loads being fed into the tub by the tire via the suspension, so the test needs to measure the stiffness between the points where the spring loads are fed into the tub). The other end of the tub can be supported in the centre by a single point (jackstand
) and a l-o-n-g arm attached by the crossmember mounting bolts (vertical loads in the front go through the torsion bar end caps. If you have coil-overs up front, use the shock towers.). Measure ten feet from the centre point out toward the end of the arm and add weight until you measure 1 degree of twist at the front centreline. Multiply the weight required by ten (feet) and you have a result in foot-pounds per degree.
This works, and even if the actual numbers aren't 100 percent accurate you're still able to tell whether your mods are effective as long as you are consistent with your test methods. For example, if you have 300 pounds out on the arm (times ten feet) and only see .5 of a degree, it might not (probably won't) mean you have an honest-to-God 6000 pound-foot per degree stiffness, but if it takes 400 pounds to get the same .5 of a degree after your mods you done good!
Of course, like Rick said, twisting the business out of a thirty year old tub might scare ya...
I've been thinking about building a rig like this to test the race car. Maybe I should do it sooner rather than later...