August 19, 2002

Mr. Alcamo:

Your letter of April 22, 2002 is an inadequate response to the question I raised to EPA in my letter of March 20, 2002 concerning the use of the Aroclor method for PCB analyses. Recent EPA actions and reports continue to reveal errors in the EPA's analysis for PCBs. Your consultants have again warned you that there are significant problems. To persist in this error, then to use these erroneous results in important health risk calculations, and to present them to the public as definitive statements of safe conditions is at least questionable.

To repeat my assertion in the March 20 letter, the Aroclor analytical method is incorrect for virtually all of the environmental samples collected in the Bloomington Project. These environmental samples containing PCBs rarely contain Aroclors. Aroclors are commercial mixtures used in industrial processes. Although what may have been discarded at the landfills was an Aroclor waste, what shows up in environmental samples will not have the same PCB pattern as when disposed. Because the various PCB congenors have different water solubilities, air volatilities, soil sorption characteristics, and biological activity (including both, destruction by microbes and accumulation in biota), the patterns differ. To use the Aroclor analytical technique introduces errors that can be extremely large. EPA scientists understand this problem and recommend the congenor technique as the correct protocol (SW-846, Method 8082). For you to persist in using the less accurate method to measure PCB levels and health impacts is disturbing.

That you have persisted is evident in the work the EPA has assigned to Tetra Tech EM Inc. in assessing air emissions from the Illinois Central Spring Treatment Facility. Three reports are pertinent. These show the collection of data (Tetra Tech TDD No. S05-0105-013), the data validation report (Tetra Tech TDD No.S05-0105-016) and the risk assessment prepared by Dr. Milton Clark for you dated June 6, 2002. The collection of data involved air sampling at a number of locations inside the facility, outside at nearby residences, a location at a distance in Ellettsville, as well as monitoring of one worker.

These methods use particulate filters and polyurethane foam (PUF) absorbers to collect both solid and vapor-phase PCBs from the air. The first question arises as to whether the laboratory analyzed both components. The document continually refers to the "PUF this" and the "PUF that". This indicates that only part of the sample was analyzed. Please confirm that the data presented was from both components. Second, it is apparent that any PCBs collected have gone through severe compositional distortion in transport from the site in water and partitioning to the air. They have undergone solution, interaction with sediment and vaporization, and they are definitely no longer classifiable as Aroclors. Yet, the laboratory used the Aroclor analytical method.

And, the validation report by Tetra Tech warned you that there was a severe problem: "None of the environmental samples chromatograms were good matches with the Aroclor standards." Reports on storm event samplings submitted to you from Tetra Tech also warn you of the error (4/9/02, 4/26/02, 5/15/02, 5/30/02, 7/15/02). Of course, because you required the laboratory to use the Aroclor method they "forced" an estimated result. As I have pointed out in the past and your own scientists have warned, for environmental the congenor method is the correct method. Whether you have always done it the Aroclor way, does not make it correct, only easier and cheaper. Considering that the Aroclors used by Westinghouse were of the 1248 and 1260 type, forcing an analysis using 1242 illustrates the shift to lower-chlorinated PCBs resulting from environmental partitioning through solution in water. The huge differences in PCB pattern make the results differ by large factors as your consultants warn you in their reports.

Then, to compound the problem, your risk assessor, Dr. Milton Clark uses the results to assess Bloomington's health risks from the leaky Illinois Central Treatment System and assures us that we are safe. The actual dangers are much greater as a result of the miscalibration of the PCB analyses. In addition, you persist in assessing only the cancer risk without concern for the multitude of other health risks presented by these sites.

Further, considering that the personnel exposure levels were much higher than other air samples, analytical errors there may be even more significant for that individual. Also questionable are the levels that the individual experienced over an eight-hour period, which were many times greater than the longer exposed, ostensibly higher-concentration locations closer to the source. Where and how much higher are the concentrations in those locations where the worker went than what Tetra Tech measured in the plant?

Not only does the facility leak PCBs to the atmosphere as an "incidental" part of the process, it intentionally disperses them in its "passive" treatment mode in the storage tanks. Then it releases untreated water to the stream, and reports an inflated treatment volume. It could be called "Enron" accounting. It is very disheartening to hear of this system and the EPA's efforts touted as a solution to the PCB problem when it actually is an expensive, continuing release of PCBs to the streams and air (which would have occurred without its installation). Your unwillingness to meaningfully validate the system's performance indicates that you also realize its ineffectiveness and want to avoid the consequences.

An assessment of performance requires an accurate "mass balance" measurement of the partitioning of PCBs at all treatment points. Wherever PCBs can separate from the spring water, including the spring itself, volumes and concentration measurements (using the congenor method) would track the capture and the escape of PCBs. To determine that the PCB problem is resolved, you must accurately determine what is exiting the site via all pathways and show that your treatment facility captures all of it. If it is not capturing it all, then you must correct it.

Sincerely,

Joseph G. Hailer