Accidental errors entail a dispersion of results during measures repeated; however their statistical processing allows: to know the most probable value of the measured magnitude, and to fix them limit the uncertainty. When the measurement of a same value (strange of the @@mesurande has been repeated n time, giving results: m1,m2,...mn, the average value is by definition:

An indication of these results is given by the typical gap:

When accidental errors allocating the different measurements are independant, the probability of appearance of the different results satisfies habitually to the told normal law again

The fidelity is the quality of an equipment of measure whose accidental errors are weak : it translated into results of measurement grouped around their average value.