A model for predicting crocidolite fiber size distributions

To minimize the problems of sample mass inconsistency associated with limited fiber counts, a model has been developed to predict the particle size distribution for crocidolite asbestos fibers with lengths exceeding the median log length value. With this model, it is possible to predict the dimensions of many more fibers than it is practical to measure by typical electron microscopic techniques. Thus, sample mass extrapolations, required for animal implantation studies, can be based on a much larger number of particles, reducing error that could occur, for example, from the presence of one or two very large particles. The model assumes a linear relationship between length and width. The number of fibers predicted in each length category is determined from the measured data. Particle lengths were randomly chosen within size classes, and the corresponding widths were predicted using the model. The similarity of the predicted population to the actual population was tested by comparing the width distributions using the chi-square test at the 90% confidence level. Model populations ranging from 250 to 10,000 particles were predicted. Those populations with 2500 or more fibers were similar to the actual sample in their size characteristics. Predicted populations in excess of 7500 fibers were required to produce consistent sample masses.