REDUCING THE VARIABILITY OF NOA LABORATORY ANALYSIS TEST RESULTS THROUGH PROPER LABORATORY SUBSAMPLING DURING BULK SOIL AND ROCK PROCESSING

“The seemingly simple task of taking a small amount of material out of a laboratory sample bottle could possibly be the largest source of error in the whole measurement process.” – “Guidance for Obtaining Representative Laboratory Analytical Samples from Particulate Laboratory Samples”, EPA/600/R-03/027. It is a well known fact in the asbestos testing and consulting industry that lack of reproducibility of asbestos laboratory test results on soil and rock samples is a major problem. Most samples collected in field environments exhibit considerable heterogeneity of size, shape, density, and composition/rock type (e.g. large 1-10 kg samples with silt to rock size fragments of ≤3” diameter of unknown origin). These samples must be reduced in mass and volume on the order of 10^-5 for polarized light microscopy and 10^-10 for transmission electron microscopy analysis, respectively. Sample reduction is performed through a series of subsampling and comminution steps. Significant errors can result at every step of the sample reduction process if proper care is not taken to insure representative subsampling techniques are followed. Unfortunately, little, if any, guidance is given in asbestos soil testing methods in this regard. However, geochemists/geostatisticians, Dr. Pierre Gy and Dr. Francis Pitard, whose work has focused primarily on the mining industry, have developed a theory of soil and rock sampling, and practical sampling handling methods, which can be applied to environmental soil and rock samples which contain asbestos. Their theories are the basis for EPA/600/R-03/027, and have been developed to a point that, with only a modicum of extra effort, the relative variance of the total subsampling error during sample preparation can be reduced by orders of magnitude. A summary of Dr. Gy’s & Dr. Pitard’s theories on, and approaches to, proper sampling techniques will be presented along with laboratory test data on samples prepared by both proper and improper methods.