systemsnull

The Chemist Who Found PFAS in Everyone's Blood

Human ThingsSD 2026.27111 min read

In 1997, a young chemist at 3M named Kris Hansen was asked to test human blood for one of the company’s own compounds. She ordered samples from blood banks across the United States and found PFOS, the chemical at the heart of 3M’s Scotchgard line, in every one of them. The only clean samples she could find were ones that predated 3M’s products or came from people far removed from them. Her supervisors’ response was to question her equipment, have her run the tests again, and tell her not to share the data outside the company.

The measurements were right. It would be years before the public was told what she had found, and decades before regulators acted on it.

I think about that story a lot, because it contains the whole problem in miniature. The knowledge existed. It was produced competently, inside the company, by someone doing exactly her job. What determined whether it became public knowledge was not whether it was true but who owned it and what it would cost them to share it.

A while ago I was talking through a map of the interpretability research landscape and realised I’d seen the same shape before: the same regulatory windows and the same research monopoly show up in PFOA and microplastics research, in food research, anywhere the people with the most to gain from an answer also pay for the work that produces it. My argument here is that this isn’t a story about a few bad actors. It’s a structural property of how knowledge gets made: when the party that profits from a substance also controls most of the research on it, the public scientific record lags the private one by decades, and the gap is closed not by regulation but by whatever forces disclosure, which for PFAS turned out to be lawsuits. PFAS is simply the most extreme and best-documented case, because here the lag wasn’t only a byproduct of who could afford to do the science; it was managed.

PFAS (per- and polyfluoroalkyl substances) are the chemicals that made nonstick pans, stain-resistant carpet, grease-proof food packaging, and firefighting foam work, and the property that made them so useful is the property that makes them dangerous: the carbon-fluorine bond is among the strongest in chemistry, so they essentially don’t break down. Contamination laid down in the 1970s is still there. The two manufacturers at the center of the story are 3M, which made them, and DuPont, which bought 3M’s PFOA to make Teflon and later made its own.

The best account of what those companies knew and when comes from a 2023 paper in the Annals of Global Health by Gaber, Bero, and Woodruff, which applied to DuPont and 3M’s internal documents the same methods researchers developed for analyzing tobacco industry documents. The timeline it reconstructs is hard to read without some anger.

In 1961, DuPont’s own toxicologists found that PFOA enlarged the livers of rats at low doses and advised that it be handled “with extreme care.” In 1970, DuPont described it internally as “highly toxic when inhaled.” By 1978, 60% of a group of exposed workers had abnormal liver-function results, which the company dismissed. In 1981, a 3M study found that PFOA caused eye defects in the fetuses of exposed rats; DuPont quietly moved women out of its Teflon unit, and a follow-up survey of its own workers’ pregnancies found two babies with eye defects. The results weren’t published and the workers weren’t told.

3M’s side of the record, released through Minnesota’s lawsuit against the company, includes monkey studies in 1978 in which PFOS killed the animals, and, as late as 1980, an internal line describing PFOA as “about as toxic as table salt.”

The paper’s conclusion is the number at the centre of this essay: the industry held toxicity data roughly forty years before the public health community established the harm, which it did around 2011 and 2012. Forty years in which the people who made the chemicals knew more about them than the people drinking them.

It’s worth being precise about what the authors found and didn’t find, because the precision is the point. They identified four of the five strategies documented in the tobacco industry’s playbook: shaping which research questions got asked, suppressing unfavorable results, distorting public messaging, and setting their own exposure limits. They found no evidence of the fifth, funding favorable published research.

The consultant pitch that best captures the posture came in 2003, from the Weinberg Group, a “product defense” firm that had also done work for Philip Morris, which wrote to DuPont that it “must shape the debate at all levels,” and proposed looking for PFOA’s “likely known health benefits.” You don’t need a conspiracy to explain this. You need an organization whose incentives run one way, which owns the only detailed data, and which gets to decide when that data becomes public.

If regulation didn’t close the gap, what did? The answer, repeatedly, is litigation, and the most important case began with cows.

In 1998, a West Virginia cattle farmer named Wilbur Tennant called Rob Bilott, a corporate defense lawyer in Cincinnati, because his herd was dying. His family had sold land to DuPont that became a landfill, the creek from it ran through his pasture, and by his account he’d lost well over a hundred cattle, with blackened teeth, tumors, and deformed calves. Bilott, who until then had represented chemical companies rather than sued them, took the case, and in the documents DuPont produced in discovery he found references to a chemical he’d never heard of: PFOA. His 2001 letter to the EPA laying out what DuPont knew led to the company’s 2005 settlement with the agency for failing to report PFOA’s risks, then the largest civil penalty ever obtained under a federal environmental statute (the story is told in Bilott’s book Exposure and the film Dark Waters).

The most consequential science in the whole saga was also a product of a lawsuit. A class action by residents of the Mid-Ohio Valley, near DuPont’s Parkersburg plant, settled in 2005, and the settlement funded the C8 Health Project, which drew blood from roughly 69,000 people in 2005 and 2006, and an independent science panel to study them. It took seven years. In 2011 and 2012 the panel found “probable links” between PFOA and six diseases: kidney cancer, testicular cancer, ulcerative colitis, thyroid disease, high cholesterol, and pregnancy-induced hypertension. That study is a large part of why we know what we know about PFOA’s health effects, and it exists because a court settlement paid for it.

I’d go further than that, although I want to mark it as argument rather than fact: civil discovery has arguably produced more actionable information about PFAS harms than decades of voluntary regulation did, because discovery is one of the very few mechanisms that can compel a company to hand over what it knows. The published scientific record probably still undercounts the harm, because suppressed internal research is only now entering it through litigation.

The same asymmetry shaped the money. When DuPont spun off its chemicals business as Chemours in 2015, taking the PFAS liabilities with it, the new company’s solvency depended on how large those liabilities were believed to be. DuPont certified to the adviser assessing that solvency a “maximum realistic exposure” of $128 million for the roughly 3,500 personal-injury cases then pending in Ohio. Nineteen months after the spin-off, those cases settled for $670.7 million, more than five times the certified maximum. Across all 87 matters, the certified maximums totaled about $1.42 billion, while Chemours took on roughly $4 billion in debt to pay DuPont a $3.91 billion dividend on the way out the door.

Chemours later sued DuPont, arguing that the estimates had been “routinely and radically understated” and that it had been insolvent from the moment it was created; the case was sent to confidential arbitration, and the companies eventually agreed in 2021 to split legacy PFAS costs, up to a combined $4 billion.

I don’t know what anyone at DuPont believed about those estimates. But the structure is the same one as in the lab: the party with the most detailed knowledge of the liability was also the party that benefited most from the liability looking small, and the number that everyone else relied on came from them.

What’s happening now is less a resolution than a slow, contested repricing. In April 2024 the EPA finally set the first enforceable federal drinking-water limits for PFAS: 4 parts per trillion each for PFOA and PFOS, 10 ppt for three others including GenX, and a combined limit for mixtures. It also listed PFOA and PFOS as hazardous substances under Superfund, which the D.C. Circuit upheld in August 2026.

But in May 2025 the new administration announced it would keep the PFOA and PFOS limits, rescind the rest, and push compliance back from 2029 to 2031, and in May 2026 it formally proposed doing so. The original rule remains legally in effect while that proposal and a challenge by water utilities work their way through; the D.C. Circuit heard arguments this month.

The four compounds on the chopping block are, notably, the replacements: GenX was introduced after PFOA was phased out, and was later found to be harmful in its own right. Regulating these chemicals one substance at a time produces exactly this cycle, which is why the European Union is weighing a restriction on the entire class of more than ten thousand PFAS; its risk committee backed a near-universal ban in March 2026, while its socio-economic committee has called a full ban “likely not proportionate,” and a final decision isn’t expected before 2027.

The gaps are as telling as the rules. There is still no federal limit on PFAS in food, no binding federal standard for PFAS in the treated sewage sludge spread on farmland (the EPA withdrew its 2025 draft risk assessment in June 2026 and issued voluntary guidance instead), and no federal PFAS air-emission standard. A few states have moved on their own after farms were contaminated. And the personal-injury side of the national firefighting-foam litigation, with more than twelve thousand cases pending, still hasn’t produced a single trial.

I don’t want to leave this as a story about chemicals, because the shape is older than PFAS and it isn’t confined to them. In 2016, researchers at UCSF published internal documents showing that in the 1960s the Sugar Research Foundation sponsored a review of diet and heart disease, written by Harvard scientists and published in the New England Journal of Medicine in 1967, and that the foundation set the review’s objective, supplied articles for it, and saw drafts before publication. The review played down sugar and blamed dietary fat, and it helped set the direction of nutrition science for decades.

Nobody had to falsify data. They only had to shape which questions got asked, and pay for the answer.

That’s the part I think matters most, because it’s the part that doesn’t require villains. Knowledge is expensive to produce, and whoever pays for it gets a strong say in what gets studied, what gets published, and when. Most of the time that influence is diffuse and nobody intends harm. PFAS shows what the same structure looks like when the stakes are high enough that someone does manage it deliberately, and it shows how long the lag can last when the only party that can see the whole picture is the one that least wants it seen.

The reason we know about PFAS at all comes down to a handful of people who forced the knowledge out: a farmer who wouldn’t accept that his cattle were dying of nothing, a lawyer who read the documents, a court settlement that paid for 69,000 blood tests.

And one chemist, in 1997, who looked at her results, found the compound in everyone’s blood, and was told to check her equipment.

References