The processing of observed data for the same physical quantities are often met in nuclear data evaluations or measurements. When the data are statistical consistency

they can be easy processed with statistical method

but when the systematic error exists

the processing is quite difficult. For this reason an universally useful method has been developed. According to the method

a prior covariance matrix of systematic error must be given

and then the average value and its error can be calculated. The weighted average value

arithmatical average value and internal error

external error and arithamatical average error are a specific case of this method. A program has been made

the results show that the method is reasonable.