Willy, Correct, with important assumptions: - that the uncertainty of each one is randomly independent of the other. - i think one also assumes the overall uncertainties have to have a Gaussian distribution. If you grab a handful of components / units, they may, or may not, show independently random distribution of values, depending on manufacturing. > I was told that the uncertainty goes down with the square root...? > That would mean 4 standards averaged together would cut the error in > half... > And 9 would cut it to one third. > Is this correct? If not, what is the formula? > > Willy >