Figure 45 : Statistics for analysts is quite easy. Under standard conditions each analytical value MUST be repeated. The optimum number N of repetitions is 4. Too many data are produced with N = 2 or N = 3. Normally a single run (N=1) is useless, as there is no information about the size of a possible uncertainty. Only in 2008 we found one exception. That is µ-PLC with partially overlapped compare analysis of an authentic with a questionable sample, which is said it is EQUAL. See HERE for the N=1 situation and HERE for the N=4 optimum.
