Comparison of environmental chemical results for split samples analyzed in different laboratories

Data comparisons were made for split or co-located samples analyzed in contract laboratories and quality assurance (QA) laboratories during environmental studies directed by the U.S. Army Corps of Engineers. Archived results were analyzed statistically as concentration ratios (contract laboratory/QA laboratory). Concentration ratios were found to be lognormally distributed, and this was the model used for comparisons. For metals in soils and volatile organic compounds (VOCs) in groundwater, 10.2% of metal ratios in soils and 5.6% of VOC ratios in groundwater exceeded limits of 0.40-2.50. Considering that both methods are multianalyte, we find that only 4.0% of the metal samples and 2.0% of the VOC samples had more than one outlier ratio per sample. More recent data produced very similar results. For VOCs, total petroleum hydrocarbons (TPHs), and explosives in soils, limits of 0.25-4.00 are suggested with the understanding that large improvements are badly needed. Even with these wide limits, approximately 42% of VOCs, 14% of TPHs, and 11% of explosives contract laboratory/QA laboratory ratios were outside these limits. Here, too, the most recent data yielded very comparable results. Sampling and preparation procedures for VOCs in soils requires immediate attention, but all methods can and should be capable of producing improved agreement between laboratories.