I have a way to remove errors from experiments.

Construct your experiment such that it measures a value and the complement value (in the mathematical dual system), and such that it measures at the same time (or other parameter being the same).

We assume that you have the exact transformation (T) from the original to the dual and that T is linear and invertible.

Now for:

M_EX = your measured value

M_EX,CP = your measured complement value (in the dual system)

M_C = the exact value in the base system (not the dual)

M_E = the error in the base system

M_EX = M_C + M_E

and we construct this formula in the dual system:

M_EX,CP = T(M_C + M_E) = T(M_C) + T(M_E).

Now M_EX and M_EX,CP are known and T(M_C) and T(M_E) can be computed. Now the idea is to choose a M_C and adjust it untill:

M_E = T^(-1)(T (M_E))

using the above two formulas.

If this is satisfied you have the exact value of M_C.

Earth owes me.