The Reliability of the Dimensional Measurements of Man-Made Vitreous Fibers Used for Biopersistence Assays: A Statistical Approach

The dimensional measurements of Man-Made Vitreous Fibers (MMVF) are required in order to conduct in-vivo and in-vitro experiments: for in-vivo experiments, the geometric mean diameter must be assessed; for in-vitro experiments, the length-weighted diameter mean and standard deviation is needed in order to calculate an in-vitro solubility coefficient. Seven fiber samples were prepared for measurement purpose-using a process described in Hanton et al. (1998). On the basis of these samples, optimal estimation methods and a criteria for stopping the counting of fibers were investigated. It is shown that, unlike the diameter and length distributions, the length-weighted diameter distribution of these samples can be considered as lognormal. Therefore log-normal based estimators have been applied and appear to be more reliable than standard ones. A simple rule for stopping the counting of fibers has been defined and retrospectively applied to the samples. In several cases, log-normal based estimators decrease significantly the number of fibers which would be otherwise needed in order to meet the stopping condition.