One of the main objectives of a ring trial is the determination of repeatability and reproducibility variances for the population of the participating laboratories. In machine smoking studies a broad spectrum of influencing factors on the test results is inherent to the measurement procedure and therefore a certain number of laboratories is necessary to estimate the mean level and the variability measures of the analyzed variable with a certain trueness and precision.

This study investigates the effects of the number of laboratories on the grand mean and the variability measures. Several simulation studies have been performed on the basis of large studies like the EUCS Studies (European Collaborative Studies) and the Collaborative Study 2010 within WG10 of ISO-TC126. In a study performed for example by 3 laboratories only the grand mean would be clearly over- or underestimated in nearly 50% of all possible study combinations. Even more extreme is the situation when the variability measures are considered. The simulation studies will give some rough ideas about acceptable number of laboratories for this type of study.