Statistical Reliability of Particle Size Distributions Determined by Microscopic Techniques

Abstract

The particle size distribution obtained from microscopic evaluation of dust samples is of little practical value, and is usually transformed to one representing particle projected area or weight. Errors of measurement and errors of sampling, or statistical errors, contribute to the unreliability of these distributions. In this article, the present practice of assuming adherence of size distributions to the log-normal distribution is discussed. Procedures are recommended for estimating the statistical error associated with count, area and weight distributions, and tabular and graphical forms are suggested for presentation of sizing data. The practice of presenting distribution data by citing Mg and σ_g is discouraged, unless tests for “goodness of fit” are performed.