ERROR AND UNCERTAINTY REDUCTION - CHALLENGE FOR A MEASURING SYSTEMS DESIGNER

The existing classification of errors is unpractical because of its incoherence. The consequent splitting of two notions "error" and "uncertainty" removes disorder and makes designer work more clear. The basic rules for error and uncertainty reduction are presented in the paper. It is emphasised that these rule are completely different from each other. 1. Historical background The idea that all measurement errors should be divided into two groups: systematic errors and random errors, appeared many years ago, in the early period of the measurement science establishing. The conception seems to be very simple and clear: - Errors with the values not changing or changing in a known manner belong to the systematic errors. - Errors with the values changing in an unexpected manner belong to the random errors. Otherwords: systematic errors are deterministic ones, while random errors are random ones. It is really clear from the theoretical point of view. Unfortunately, such a classification is not practical. There are two reasons for it: 1. The definition of the error is purely theoretical X X