Comments on draft 1 Relation between ISO 5725 and GUM

Leslie Pendrill

Swedish National Testing & Research Institute, SP

Quoting from the 1st paragraph of section 1.2 Uncertainty:

"... quantifies the magnitude of any possible discrepancy between the value obtained from a measurement of a well-defined quantity and the value which would be obtained at the highest level reference laboratory for the same quantity."

This fact, I feel, emphasises the importance of interlaboratory errors (as treated in ISO 5725 pt. 2) and associated uncertainties. It is true to say that rarely do measurements involve direct participation in an interlaboratory accuracy experiment (as the draft mentions in sections 4.3 and in the Gap Analysis). But behind every measurement is the requirement of being able to relate the results to those of others (through the use of a coherent measurement system such as the SI). And uncertainties associated with this traceability need to be dealt with.

In this connection, the question arises: How does one best tackle measurement uncertainties w r t trueness? The hypothetical "true value" can be introduced as follows:

Since relations between measurement results are of the essence, the concept of measurement compatibility (which is missing from GUM, for example) is important:

Two measurement results are compatible if their difference is less than their combined uncertainties.

One measurement result may have much smaller uncertainties than the other in a comparison - the former then plays the role of a de facto true value. A traceability hierarchy can of course then be established where different measurement results are ranked according to their uncertainties - the so-called traceability level of each measurement.

In summary, the question of uncertainties in traceable measurements seems to be tractable by extending the ISO 5725 Accuracy experiments in combination with GUM.