Airborne asbestos concentrations have been reconstructed for the entire 20th century for the first time through a combination of paleolimnological methods, particle-separation techniques, and analytical transmission electron microscopy. Pb concentrations and respirable aerosol mass concentrations in air and sediments yielded collection efficiencies of approximately 3000 m3 of air per gram of lake sediment. Airborne concentrations of chrysotile, the most common type of asbestos, reconstructed from control lake sediments echoed chrysotile's usage during the 20th century, with the highest concentrations mid-century (approximately 0.1 fibers/cm3) and then decreasing in the last quarter century. Reconstructed airborne concentrations of anthophyllite asbestos, a byproduct of local talc mining and milling, increased from <0.004 to 0.022 fibers/cm3 from 1846 to 1967. These anthophyllite concentrations during the approximately 100-year period of talc mining correlated well (r2 = 0.80, p < 0.01) with annual production of local talc and were much higher (p = 0.004) than concurrent concentrations in a control lake located upwind of the mines and mills. All of the chrysotile and more than 70% of the anthophyllite asbestos fibers were too narrow to be detected by phase-contrast light microscopy, the method used to measure airborne fiber concentrations before approximately 1980.

Since 1992, the Environmental Laboratory Approval Program of the New York State Department of Health has conducted semi-annual proficiency testing (PT) of laboratories analyzing airborne asbestos by transmission electron microscopy. While these PT rounds have included magnification measurements, diffraction measurements, zone-axis calculations, and identification of unknown mineral fibers, most samples have been dispersions of asbestos and other mineral fibers on MCE filters. Participating laboratories in each PT round were required to fill out a questionnaire detailing preparation and analytical methods. When performance of laboratories was correlated with their reported preparation and analytical methods, some significant trends were uncovered. Quality of grids was best for laboratories that 1) collapsed filters using a DMF solution, 2) used gravimetric reduction of collapsed filters to calibrate etching time, 3) etched filters for three to six minutes, and 4) used a combination of Jaffe-wick and condensation washer to dissolve filters. Reported structure concentrations were highest for laboratories that 1)etched filters for longer than six minutes, 2) evaporated carbon in a tilt/rotation geometry, and 3) performed analysis at magnifications between 15,000 and 20,000.

The Environmental Laboratory Approval Program (ELAP) of the New York State Department of Health has administered a proficiency-testing (PT) program for asbestos in drinking water since 1992. While the United States Environmental Protection Agency requires that only fibers longer than 10 mu m be counted, FLAP continues to require that ail fibers longer than 0.5 mu m be analyzed. FLAP mandates the use of polycarbonate filters because short fibers are apparently lost in mixed-cellulose eater filters. Thirteen different waterborne asbestos samples have been distributed to approximately two to three dozen laboratories analyzing asbestos by transmission electron microscopy. Mean total fiber concentrations have ranged from 5.49 to 3340 million fibers per liter (MFL) while mean long (>10 mu m) fiber concentrations have ranged from 0.377 to 76.7 MFL. Failure rates have ranged from 3.5% to 21% of participating laboratories, with equivalent failure rates for total and long-fiber analyses. As laboratories gained experience in long-fiber analysis, they appeared to report results closer to consensus means.

Several analytical trends have been observed during the last 8 years of the New York State Department of Health's Environmental Laboratory Approval Program proficiency-testing program for analysis of asbestos in bulk samples. Wollastonite was the most commonly misclassified nonasbestos mineral fiber, with approximately 3–5% of participating laboratories misidentifying it as various amphibole asbestos species in any given round. Most laboratories made this mistake only once. A bias toward reporting percentages higher than formulation was noted for asbestos. This was particularly true of samples with low (<10% by weight) asbestos concentrations. Amosite asbestos consistently produced the highest bias, with the less common types (other amphiboles) yielding the most accurate results. When presumptive results for visual estimation were compared with presumptive results for point counting, point-counting results were significantly more accurate than visual-estimation results. Experience with point counting tended to improve a laboratory's accuracy, whereas experience with visual estimation did just the opposite. Gravimetric reduction of low-level asbestos samples is probably the most important step in improving accuracy and precision in quantitation.

A method has been developed for the quantitative determination of asbestos and other major components in bulk samples. This method uses concentrations of signature elements as indices of the contents of the respective components. It is based on analyzing one marker element for each component and subsequently solving a set of linear equations. This approach was applied to binary and ternary component systems of asbestos-gypsum and asbestos-gypsum-fiberglass prepared as proficiency test samples that represent real-world samples. Either chrysotile or amosite was used in these synthetic mixtures; Co and Fe, respectively, were the optimal markers for these two asbestos types. Markers for gypsum and fiberglass were Ca and Na, respectively. All these elements were measured with high precision using instrumental neutron activation analysis. Two additional markers, Sc and Fe for chrysotile and Sc and Co for amosite were used to verify our results. Homogeneous distribution of asbestos fibers in the bulk proficiency test samples was ascertained from the analysis of the three marker elements, Sc, Fe, and Co, in several replicates randomly sampled from our bulk preparations. Reproducibility in the analysis of these elements in the replicates was +/- 2-4% (1-sigma) and sample homogeneity was found to be about 93% at the 95% confidence level. Concentrations of chrysotile and amosite as well as the other two components calculated from the marker element concentrations in the proficiency test samples agreed well (within +/- 5-15%) with the formulated values at both 3% and 10% asbestos fiber contents.

Intralaboratory and interlaboratory investigations were used to evaluate the precision and accuracy of point counting versus visual estimation of asbestos in friable bulk materials. Interlaboratory analyses revealed that the commonly used "equivalent" visual-estimation methods were significantly less accurate than the Environmental Protection Agency (EPA) 400-point-count method, especially at low asbestos concentrations. In an effort to produce a method that is statistically "equivalent" yet less time consuming than the EPA method, several strategies were assessed to reduce preparation and analysis time. False negatives were produced by schemes that unconditionally reduced the number of slides/points analyzed. The best modification was a stratified scheme in which effort was inversely proportional to the Percentage of asbestos in a sample, Although precision was sacrificed at higher concentrations, that precision was deemed superfluous because such materials were categorically asbestos-containing materials (ACMs). A nonpoint scan Option was also developed for non-ACMs. Thus, the time required for preparation and analysis of ACMs by the stratified method can be less than 20% of the time required for the full EPA point count, with only ACMs in the 1% range requiring the full 400 points counted. A series of formulated- Weight bulk samples were analyzed by outside laboratories using the EPA method, the stratified method, and visual estimation. The stratified method produced results that were not significantly different than results by the EPA method. Visual-estimation results were less accurate when compared to formulation values and were usually significantly different than results from either point-count method.

The most direct evidence yet of a coal-burning source of a high-sulfate air mass reaching the Northeast has been detected in a brief episode at Whiteface Mountain, NY, on 23 June 1983. Individual microparticles were characterized by electron microscopy in three samples from this period: 12 h before the SO2−4 peak ([SO2−4] = 2.7 μg m −3), during the peak (24 μg m −3) and 12 h after the peak (2.2 μg m −3). This analysis provided the first confirmation of the association of coal fly ash and high [SO2−4] at a rural acid-stressed site. Common at the time of peak [SO 2−4] were sub-μm magnetite and glass spheres as identified by morphology, elemental composition and mineralogy. This coal fly ash was essentially absent 12 h before or 12 h after the episode. A midwestern source of this episode was indicated by Mn/V ratios and by meteorologic conditions.

Concentrations of SO42−, Al, V and Mn were determined in airborne particles during July-August 1981 at five sites in New York State: for 6-h intervals at Mayville, Brewerton and Whiteface Mountain and for 24-h intervals at Oneonta and West Haverstraw. Episodic high [SO42−] were observed during 18–21 and 24–26 July and 1–4 August. Air trajectories showed that during the 18–21 July and 1–4 August episodes peak [SO42−] coincided with slow-moving air masses which had spent 1 or more days in the midwestern states of Indiana, Michigan, Ohio or Pennsylvania prior to entering New York. During the episodes [SO42−] decreased from west to east. No significant local sources of SO42− exist at the 6-h sites. All these lines of evidence together show that during these episodes SO42− were transported from the industrial Midwest. During the 24–26 July episode, peak [SO42−] at Mayville coincided with air masses entering the state from Ohio, but at Brewerton and Whiteface Mountain the highest [SO42−] occurred when air masses came from metropolitan New York City and the mid-Atlantic states. The average [SO42−] at Mayville were 2-fold higher than at Brewerton and Whiteface Mountain. The average manganese/vanadium ratio (Mn/V) in aerosols from the Midwest (1–10) has been suggested to be ~ 10 times that in the Northeast (< 0.2). The average Mn/V during 18–21 July and 1–4 August, 7.3, 3.7 and 2.6 at Mayville, Brewerton and Whiteface Mountain, respectively, confirm the conclusion that high [SO42−] were due to transport from the Midwest. On 25–26 July the average Mn/V was high at Mayville (2.8) but low at Brewerton (0.29) and Whiteface Mountain ( < 0.24), consistent with transport of air from the East Coast at the latter two sites during this episode. Although [V] is low in rural New York State and in midwestern air masses, [V] in East Coast urban areas is at least 10-fold greater. High local [V] may effectively mask the midwestern signature in such areas. The observed 2- to 3- fold decrease in Mn/V between Mayville and Whiteface Mountain (~ 500km apart) suggests caution in applying this technique at very large distances from the emission sources.

La difference entre les taux Mn/V observes a Albany (Hollande) montre qu'il faut prendre des precautions lors de la determination des regions sources de sulfates par la methode de la signature Mn/V