March 20, 2002

Mr. Alcamo:

I have received your letter of March 8, 2002 concerning my questions on the calculation method to estimate releases from the landfill springs. It was very detailed and showed a far greater understanding of the complexity of flow and PCB releases than that of the questioned Tetra Tech report. I wonder why a report of the caliber of your letter could not have been given to the public rather than what was supplied. I think this problem applies to the same contractor's reports on split samples. Why expose the EPA to such obvious questions when you have the data at hand to make a reasonable case? Hopefully, the public can expect that this estimation technique that you are using are temporary, and will be kept in mind as "conservative" that is meaning underestimating, and will be replaced with more accurate methods as soon as the data are gathered.

This whole issue of the PCB levels in spring water analyses and split samples has raised another question for me. I went through the EPA's Method 8082 looking for the misquoted recovery data for split samples discussed in your response to my letter of February 22, 2002. It occurred to me that when the data from Viacom/PSARA on the other split are reviewed, it could possibly show that both laboratories working on the split samples did their work correctly. So, how might that occur?

I came back to your revelation that the laboratories are discarding the sediment portion of some water samples sent to them. The split sample discrepancy could be caused if the laboratory that got the higher PCB concentrations didn't do a complete job of removing the PCB-rich sediment. Not only would not analyzing the whole sample alter PCB levels for split samples, but the practice would impact assessment of what you are trying to do on the spring flows. The correlation of suspended solids with spring flow and associated PCB release would be problematic if variable portions of the sediment are discarded.

And then, re-reading the Method and thinking about it further, I realized that in addition to the recurring distortion of water results from discarding sediments, another problem might be the cause. Even when the laboratories weren't discarding the most PCB-laden portion of the samples, they were analyzing virtually all of the Bloomington Project samples with the wrong technique! Almost every result over the past 20 years using the Aroclor method on environmental samples contains serious error. All results generated as Aroclors need to be re-evaluated to determine the extent of the error and its impact on the decision process. In the following I will try to explain my concerns for both practices.

Discarding of Sediment and Its Impacts - Because of the low water solubility of PCBs and strong sorption on sedimentary particles, huge differences in concentrations occur between what's in the water and on the sediment of a spring, stream, or well water sample. By letting a small amount of the mud in a water sample get into the extraction process, I think PSARA's laboratory has opened the door to an example of what I think is a significant error in the analytical procedure. As you pointed out in your response to COPA on data quality issues (relevant response section attached), your laboratory procedures have discarded the sediment component of the water samples sent to them unless the analytical request calls for a "total" analysis. Since the analytical call is not for a "total" analysis, the laboratory separates the water from the sediment and analyzes only the water, or the sediment if the sample is described as a soil.

Potentially, different results for split samples come about because either the samples were somehow actually different, or something was done differently inside the laboratories. I believe that your difference is not a case of either-or, but of both. They were not homogenous samples, and the laboratory treated them differently. I believe that what has occurred in this particular split sample event and with probably with every mixed-phase sample sent in for the Bloomington Project is that the samples have been "un-mixed". Hard to imagine, except if one considers that most natural water samples are not just water, containing only dissolved PCBs. Water coming out of a spring, especially in a rainstorm, or coming from a drill hole, contains a suspended sediment load. That mix of components is actually what is coming out of the spring and flowing downstream. Sediments can contain PCBs at orders-of-magnitude greater levels than the associated water. The suspended sediment is often the greatest carrier of PCBs from these sites. That mix should be maintained all the way through the sampling and analysis process. And, to understand the quantity of PCBs moving into the environment from springs, it is the total sample, water and sediment, that is the target of analysis.

The recurring cause of discrepancies is a serious dereliction in assurance of sample integrity that has been frequently ignored in this project. Project management has long concurred in this discarding of the more contaminated component of the samples resulting in a distortion of the water data. Additionally, there are other manipulations that also have also distorted the data for water samples. By ignoring the Henry's Law-driven partitioning of PCBs from water to the air through exposure of water samples to the air, volatilization of PCBs will significantly reduce the level originally present. Such actions as taking spring samples downstream from the exit point after they tumble over the streambed, or allowing water to cascade into a sump before sampling, or to sit in an open holding tank, all will cause losses and incorrect results. Air sample results also appear to be seriously affected. The quality assurance plan for air sampling indicates that only the vapor form of the air samples is analyzed, ignoring the PCB content in the dust, which was apparently the target of all the abatement procedures used in the cleanup. The problems go on to include a failure to analyze pure PCBs found in the subsurface and aggressively pursue their removal. And the list of data quality concerns includes the following discussion focusing on the use of the wrong method for much of the work.

Incorrect Analysis Method and Its Impacts - In addition to the errors in handling and analysis of water sample, which has been routine, I also question the accuracy of virtually all of the analyses done in the Bloomington Project using the Aroclor approach. EPA has specifically warned in SW-846, Method 8082 (and earlier versions) for PCB analysis that project managers and laboratories modify the method if other than Aroclors from industrial equipment are the objects of the investigation. I have attached pg. 2 of EPA Method 8082 from the EPA's analytical manual SW-846 used for the vast majority of the Bloomington Project samples. I have looked at much of the data generated in the Project, and few appear to be for PCBs in electrical equipment. The vast majority are clearly for environmental samples of soil, water, sediment, air, landfill debris, or biological material. The EPA method directs a different calibration and measurement technique for environmental samples than that for Aroclors, if accurate results are desired. Every environmental sample done by the Aroclor method will likely undermeasure the PCB content. I have attached a few of the hundreds of examples of data results showing Aroclor results rather than PCBs.

PCBs were supplied commercially as Aroclors, which are mixtures of the 209 different PCB isomers or congenors. Environmental exposure of these Aroclors changes the congenor composition of the Aroclor through loss of components. Because virtually all of the samples have undergone some alteration, many to an extreme degree, analytical methods must be modified to account for this alteration. Alteration of the Aroclors includes partitioning due to differential solubility in water, sorption on sedimentary solids, bioaccumulation, vaporization, and degradation by biological or chemical reaction of the congenors. The accurate analysis as an Aroclor requires the unaltered presence of specific PCB congenors. Loss of any one of these congenors (5 or fewer congenors out of 209 possible are used in the Aroclor approach) from the target samples results in undermeasurement of the PCB presence.

For accurate results on environmental samples the congenor technique should be applied. The congenor technique uses all of the congenors in its standards to calibrate the instrument and to measure all of the PCBs in an environmental sample. The result is a total PCB content unaffected by the weathering loss of a few congenors from the samples. The Aroclor method error could range from 20 to possibly even 100% of what is actually present depending on the loss of PCBs through alteration.

Every analytical result for the Bloomington Project should be reviewed to assess the errors introduced by using the Aroclor method, and to evaluate the impact of this error on cleanup goals, compliance levels, risk assessments, and monitoring of remediation effectiveness. Further, EPA should direct EPA and IDEM contractors to institute use of the correct analytical procedures for all future work.

These problems add to the list of data quality questions that the community has raised over the years and which the EPA has not resolved. The poor application of the Superfund process began with inadequate site characterizations, incomplete removal of contamination, and now, a flawed water treatment program are only "tips of the iceberg" of questionable workmanship. When we add in discarding the PCB-rich portion of samples, and the use of an incorrect analytical method, confidence in the EPA as a protector of human health and the environment of the Bloomington community declines further. EPA has rejected advice from knowledgeable community individuals in pursuit of closure without regard for a truly effective solution based on a quality work performance. Further resistance to openness in these matters will reinforce the community's concerns that the work has been done with little regard for the health impact to the community itself. Please respond to these issues and hopefully, begin to correct them as soon as possible.

Sincerely,
Joseph G. Hailer