Accuracy

When I use the method of standard addition should I calculate the recovery with the nominal (theoritical) value of the standard or experimentally by linearity?

Accuracy = X(Experimental)/X(theoritical = nominal?) X 100

Thank you!