On the estimation of one - and multi-dimensional uncertainties

摘要

In regard to the propagation of random errors, it is proposed to choose the same number of repeated measurements for each of the variables, so that confidence intervals may be established by means of the Student distribution. The propagation of so-called unkown systematic errors is based on the idea of "non- randomization". These concepts of error handling are suitable for introduction into any linear model for the evaluation of measurements, allowing the defintion of one- and multi-dimensional uncertainties, depending on the number of variables considered.