BS ISO 9276-3:2008

Foreword
Introduction
1 Scope
2 Normative references
3 Symbols and abbreviated terms
4 Adjustment of an experimental curve to a reference model
  4.1 General
  4.2 Quasilinear regression method
  4.3 Non-linear regression method
5 Goodness of fit, standard deviation of residuals and
  exploratory data analysis
6 Conclusions
Annex A (informative) Influence of the model on the regression
                      goodness of fit
Annex B (informative) Influence of the type of distribution
                      quantity on the regression result
Annex C (informative) Examples for non-linear regression
Annex D (informative) X[2]-Test of number distributions of
                      known sample size
Annex E (informative) Weighted quasilinear regression
Bibliography

Describes methods for the adjustment of an experimental curve to a reference model with respect to a statistical background.

This part of ISO9276 specifies methods for the adjustment of an experimental curve to a reference model with respect to a statistical background. Furthermore, the evaluation of the residual deviations, after the adjustment, is also specified. The reference model can also serve as a target size distribution for maintaining product quality. This part of ISO9276 specifies procedures that are applicable to the following reference models: normal distribution (Laplace-Gauss): powders obtained by precipitation, condensation or natural products (pollens); log-normal distribution (Galton MacAlister): powders obtained by grinding or crushing; Gates-Gaudin-Schuhmann distribution (bilogarithmic): analysis of the extreme values of the fine particle distributions; Rosin-Rammler distribution: analysis of the extreme values of the coarse particle distributions; any other model or combination of models, if a non-linear fit method is used (see bimodal example in AnnexC). This part of ISO9276 can substantially support product quality assurance or process optimization related to particle size distribution analysis.