Fourth post updated with some thoughts. D90 Users try this and tell me what you get.
OK, I haven't played with this yet but I see an issue.
The D90 doesn't "know" what the aperture is on an AI (fully manual) lens. Some Nikons mate with that funny metal foot on the lens, but the D90 (and lesser cameras) don't. So the values you are estimating are based entirely upon the amount of light entering the camera for you at those apertures. This will obviously vary depending on the lighting and surface.
I think it would be helpful to measure the actual amount of light... using a meter, or by metering with the camera itself off a photographic grey card. This would be a fairly simple control, and then we could work up a chart where we know a certain amount of light (Exposure Value) on that grey card equals a certain known ISO / Shutter if we enter LV with that grey card filling the frame. Removing lighting and surface variables should = consistancy.
ISO could, maybe, be gaged by measuring the noise in a video still and comparing it to the noise in a small low quality jpg taken at known ISO... again all of the grey card.
Speed could be measured with the tv trick mentioned in the mother of all threads.
I'll see if I can play with this some later tonight.