Standardisierung quantitativer Analysenverfahren

By standardised investigations in a chosen working range a set of 24 data blocks will be obtained with given working amountsn _A and measured blank valuesI _B, rough valuesI _x and eventually reference valuesI _R. An analysis of variancy, applicated to the 6 groups of blank values and in a given case also to the 6 groups of reference values and the differencesI _RB=I _R-I _B, leads to an objective decision about homogeneity of the respective data material. This gives the background for the kind of evaluation of the composed information valuesI. The connexion between given amountsn _A and found valueI, called analytical functionsn _A=F(I), is determined by comparative calculations with the method of least squares of deviationsΔn _A=n _A-F(I). In each case the standardised calculation process is started with the hypothesis of the existence of a complete parabolic functionn _A=U+ V ·I +W·I². Statistical and heuristical tests decide about the practicable neglecting of one of the constants. In a general scheme of logical decisions and following new calculations for reduced functions only with 2 or 1 constants the relevant function can be obtained. If the calculation steps are finished with a lineary functionn _A=V ·I, a standard procedure exists. The characteristic data evaluated by this standard method are the constants of the analytical function, the mean errors of the constants and the standard deviations _n of the analytical procedure in the given working range. By outlier tests for each of the found deviationsΔn _A the reliability of these data is to be guarded. A new defined determination limitn _G has a plausible connexion to the standard deviations _n of an analytical procedure.