There is an odd paradox at work here:
there seems to be a lot of concern for precisely positioning the dip stick, concern that one tank would be different from another, concern that one airplane might be different from another.
And yet, we then treat this measurement as an approximation with a lot of conservatism about using the information.
If it is going to be treated as an approximation, then why work so hard to get precise data?
I took the readings off of Bruce's photo of his dipstick ( thanks Bruce for updating the picture
) and plotted it. I was interested to see how non-linear the curve would be, which would be expected because of the odd shape of the tank. The curve is surprisingly linear, although it is convex upward everywhere. The curve being fairly close to linear, it is also possible to extrapolate Bruce's curve to the zero point. His curve shows 2 gallons at zero on the stick.
So how linear is it? Well, if instead of using the measurements and data, you were to simply mark a stick with linear graduations from 0" =2 gallons to 6 3/4" = 21 gallons (0.35" per gallon), you would have a dip stick that reads within ONE GALLON of correct at all tank levels. And because the actual curve is convex, the linear curve always reads low (conservative).
Depending on how you set your mixture, this allows you to estimate your endurance to within 6--8 minutes. No one is going to flight plan to an expected precision of 6--8 minutes of endurance--we keep more than 30 minutes reserve.
I would be satisfied to just make a stick with that linear graduation and use it straight away.
Caveat: Bruce's data is from an RV-8. I have an RV8. The stick as I describe may not be within one gallon of correct on an RV-8A, or on an RV-7, etc. because the zero-point fuel quantity may be different. But if you can establish what your zero-point fuel quantity is, you can then make a linear stick for your model. We know what the full-point value is.