Notes on the use of propagation of error formulas