GE settlement increases PCB removal, but how much stays in Housatonic River?

When the Environmental Protection Agency settled on a PCB cleanup order in 2016 for the Housatonic River, it called for the General Electric Co. to remove 46,970 pounds of PCBs from the river's banks, bottoms and flood plain.

PITTSFIELD — The Environmental Protection Agency never asked the General Electric Co. to remove all the toxins it allowed to befoul the Housatonic River.

Not even close.

This is a story is about polychlorinated biphenyls that won't be going anywhere new, even as a stalled cleanup moves ahead.

When the EPA settled on a PCB cleanup order in 2016, it called for GE to remove 46,970 pounds of PCBs from the river's banks, bottoms and flood plain.

That's the estimated weight of the substance itself, a probable carcinogen, within nearly 1 million cubic yards of dredged material.

GE fought that order, but the remedy could have been worse for the company. In a 2014 planning document, the agency considered compelling GE to remove 94,100 pounds of PCBs.

That was the most extensive remedy, but it shows that the EPA, based on evidence gathered through sampling, was later aware that its 2016 permit allowed at least 47,130 pounds of PCBs — more than 23 tons — to remain.

The Eagle asked the EPA to explain what percentage of PCBs would stay in the Housatonic under even an expanded cleanup. The agency said it is too early to say before a revised permit, based on the settlement agreement announced in February, is completed this year.

"The short answer is no," EPA spokeswoman Kelsey Dumville said in an email.

Since the agreement was announced Feb. 10, public debate has moved from what's to be left in the river to where the bad stuff that comes out will go. The pact would allow GE to dispose of nearly 1 million cubic yards of tainted soils and sediments with average PCB concentrations of 20 to 25 parts per million at a new landfill at the former Lane Construction site in Lee.

'Clear as mud'

As the EPA works up a new permit, it will review new sampling for PCBs to "refine" its understanding of what's in the river, Dumville said.

One of the agency's toughest critics, Tim Gray of the Housatonic River Initiative, says he recently looked up data on sampling for the Rest of River project from Pittsfield to the Connecticut border.

"They haven't sampled the river in 20 years, so, they don't even know what's going on," Gray said. "I've been trying to figure this out for years. This is as clear as mud."

Gray long has argued that the cleanup outlined in 2016 left too much of the toxin in the river.

"This whole cleanup is absurd," Gray told a Lenox gathering in December 2018. "Why are we cleaning the river if we are leaving all these PCBs in? There is a way to a good cleanup instead of letting GE walk all over us."

The agency has been open about what's not known. A "fast facts" bulletin it produced a few years ago said that PCBs are present "in large quantities in river sediment and floodplain soil."

By "large," the EPA wasn't kidding.

"Estimates range from between 100,000 to nearly 600,000 pounds of PCBs," the fact sheet read.

The company has said it released 70,000 pounds of PCBs into the river flood plain before the substance were banned. That's more than it fessed up to in 1982, when it paid Stewart Laboratories to research the issue. That firm's report said 40,000 pounds entered the river within Massachusetts.

But, some GE insiders felt that such estimates were off — and by a lot.

In a documentary film by Mickey Friedman, Ed Bates, former manager of the company's transformer division, says in a Sept. 12, 1990, interview that he believes GE allowed 1.5 million pounds of PCBs to enter the local environment. A colleague, Charles Fessenden, sat beside him nodding as Bates explained his rationale. He estimated that 3 percent of PCBs used by his division were spilled or intentionally disposed of into local lands and waters — perhaps 4,000 to 5,000 pounds a week.

Another way of quantifying the PCBs to be taken out, versus what's to be left in place, involves amounts of sediment.

The most intensive removal outlined in the 2014 EPA document called for 2.9 million cubic yards of soil and sediment to be removed. That's three times more than what the 2016 permit settled on. It would have occurred on 377 acres of flood plain, compared with 45 acres in the 2016 permit.

And, not surprisingly, it would have taken far longer: 52 years instead of 13.

What's acceptable

One caveat in the process of calculating a cleanup percentage, Dumville notes, is that the settlement agreement covers only PCBs "that are posing an unacceptable risk."

In other words, attempts to calculate what percentage of all PCBs will remain along the river or in its flood plain might be futile, because the agency counts only those in specific reaches and areas of the river that are "unacceptable."

That means the question isn't about PCBs that GE dumped. In the agency's view, the case revolves around concentrations that demand to be removed or capped.

One earlier figure that made it into news coverage has little value, as a result.

In 2012, the agency noted that the removal of 1 million cubic yards of material represented 25 percent of the 4 million cubic yards deemed to contain at least 1 part per million of PCBs.

"We will only be including PCBs in the denominator that create an unacceptable risk," Dumville said, referring to the arithmetic term used to indicate the whole, when calculating fractions.

In this instance that fraction would be , with the denominator of 4 representing the whole measure of contamination.

The denominator no longer is 4 million cubic yards if concentrations of 1 part per million are taken out of the equation.

"That is not the case with the totality of the 4 million cubic yards," Dumville said.

With a larger denominator, the percentage of PCBs to be removed from the whole increases, giving the impression of a more through cleanup.

The nature of the "whole" pollution problem changes because of how the agency views contamination deemed to be an acceptable risk to human health.

One part per million is considered an acceptable level of contamination, even in people's backyards, says Jane Winn, executive director of the Berkshire Environmental Action Team.

The 2016 permit lists 5 parts per million or more as the trigger for riverbank removals in certain reaches of the river. The threshold for human exposure is even higher for other areas, such as industrial zones.

The EPA permit that GE successfully has fought does not include a percentage estimate of the weight of PCBs to be left in place. Rather, the cleanup is based on acceptable concentrations for the kinds of use river areas receive.

Winn, whose group signed the settlement agreement, said that even after decades of monitoring plans for a cleanup, she doesn't know how to compare an eventual removal of PCBs to what will be left behind.

"We really don't have a way to say exactly how much is being removed," she said. "You truly can't sample everywhere, so, you come up with a methodology that's science-based."

Complicating things further is that, over time, sampling becomes outdated, as banks erode and PCBs are washed downstream toward Long Island Sound. Tons of PCBs continue to flow over dams.

Another measure

Dumville, of the EPA, suggests that one way to calculate the scale of PCBs that would be left in the river is to compare the pounds to be taken out under the 2016 permit to the top end of the proposed removal in the agency's 2014 "Statement of Basis."

That math shows the 2016 cleanup removing half of known flood plain PCBs by weight.

That takes us back to the conclusion that more than 23 tons of PCBs would remain in the flood plain — or somewhat less, given the settlement's finding that nearly 100 of the 300 acres that would be capped under terms of the 2016 permit would be excavated instead.

"The good thing is, we have more being removed," Winn said.

As the EPA conducts more sampling, Winn said, her group will be looking for results — particularly "hot spots" where PCB concentrations exceed expectations — and will press for additional PCBs to be removed.

"In that case, we don't care what the permit says, we want that out of there," she said.

Once a new permit is ready, the EPA will have a clear sense of the weight, or "mass," of PCBs to be left in the flood plain.

The cleanup design eventually will determine how much soil and sediment must be removed in the 200 acres where a cap, so far of uncertain thickness, will be installed, Dumville explained. Until that is done, the agency can't estimate the mass of PCBs that will be buried.

"That said, with less reliance on capping and additional excavation, the amount of PCB removal from the river will no doubt be larger — but we have no calculation of such an estimate at this time," she said.

Larry Parnass can be reached at, at @larryparnass on Twitter and 413-588-8341.