The results came back in April 2022, and they covered ground Sue Hunter had never farmed.

She had been working the property in Unity, Maine, for years, organic vegetables and forage for dairy cows across about two-thirds of a hundred and fifty acres. Early that year a neighbour who bought her hay told her it had tested positive for PFAS, the fluorinated compounds that do not break down. Hunter and her son Keith went out with sample jars and worked the place methodically, dozens of samples, soil and water. A laboratory called Northern Tilth ran the analysis. Most of the farm was contaminated. So were the wetlands she had left alone, and the groundwater she used for irrigation and for drinking.

"It was devastating, it really was," she told the University of Maine the following summer. "We're farmers and we love our land. We love this place. It's home to us. I've been a farmer since I was 17 years old."

The farm then became a research site. The contamination varied from parcel to parcel, and Hunter had years of her own records, so the land turned out to be scientifically valuable in a way a funded project could not easily reproduce. Diane Rowland, who directs the Maine Agricultural and Forest Experiment Station, did the arithmetic out loud. "Setting up an experimental condition with variable PFAS levels and past data collection like Sue has on her farm would take millions of dollars and decades to create."

An experiment nobody designed had already been run. The University's account does not say how the chemicals reached her fields, and no study has traced the contamination on her land to any particular pathway; I am not going to guess. But across a great deal of American farmland there is one route that is deliberate, licensed, and performed by trucks on a schedule, and it has an official name.

The word

When a wastewater treatment plant separates the liquid from the solid, what settles out is sewage sludge. Treat that sludge to a specified standard and it acquires a different name. The Environmental Protection Agency puts the distinction in its own glossary: sewage sludge means the solids separated during the treatment of municipal wastewater, while biosolids means treated sewage sludge that meets the agency's pollutant and pathogen requirements for land application. Land application means spreading it on fields.

That second word is the hinge on which a waste-disposal problem becomes a fertiliser market. Sludge is a cost. Biosolids are a product, with nitrogen and phosphorus in them, delivered free or cheap to farmers who would otherwise buy those nutrients. The rule that draws the line was signed on 25 November 1992 and published on 19 February 1993, at 58 Federal Register 9248. It sits in the Code of Federal Regulations at Title 40, Part 503.

Part 503 is short, and the part that governs what may be spread is shorter still. Section 503.13 is called "Pollutant limits," and it consists of four tables.

The scale

From the reports filed in 2024 by roughly 2,350 treatment facilities, EPA estimates that about four million dry metric tons of sewage sludge were generated. Around 2.39 million of those tons were spread on land. Some 982,000 went to landfill and 558,000 were incinerated. Land application is the majority practice, by a wide margin, and it is ordinary: agricultural fields, grazing land, reclaimed mine sites, forestry, parks, golf courses, and bagged products sold for home lawns and gardens.

Those figures come with a caveat the agency states itself. Nine states run their own biosolids permitting under the National Pollutant Discharge Elimination System, and their sludge does not appear in the totals at all. The nine are Arizona, Idaho, Michigan, Ohio, Oklahoma, South Dakota, Texas, Utah and Wisconsin. Smaller facilities are also absent. The agency's conclusion is the last sentence anyone would expect to find on a federal programme page in its thirty-fourth year: "there is no definitive source that reports the amount of biosolids produced annually in the United States."

The tables

Table 1 sets ceiling concentrations, the levels above which sewage sludge may not be applied to land at all. It lists nine substances. Arsenic at 75 milligrams per kilogram, dry weight. Cadmium at 85. Copper at 4,300. Lead at 840. Mercury at 57. Molybdenum at 75. Nickel at 420. Selenium at 100. Zinc at 7,500.

Tables 2, 3 and 4 govern cumulative loading, monthly average concentration, and annual loading. Each lists eight. Molybdenum drops out after the ceiling.

Every one of them is a metal. For a pharmaceutical, a flame retardant, a plasticiser, a hormone, a solvent or a fluorinated compound, these tables set no figure at all.

That is the whole of the substantive chemistry of the rule, and these four tables have been amended twice since 1993, not once. In February 1994, two molybdenum producers petitioned EPA to reconsider the metal's limits, a group of eight companies petitioned a federal appeals court on the same question, and within nine months EPA agreed the limits looked overprotective and deleted molybdenum from Tables 2 through 4, pending a new calculation it promised the Federal Register "in the near future" (59 FR 9099, Feb. 25, 1994). That promise is why Table 1's ceiling is the only molybdenum figure left in the rule today, thirty-two years later; no replacement has ever been proposed. The other amendment, in October 1995, is the one the rest of this essay is mostly about. Nothing in these tables has moved since.

The limits are real, they are enforced, and metals were a sensible thing to worry about in 1993. A ceiling on nine metals is also a strange instrument for a substance that arrives at a treatment plant from every drain, sink, floor, factory, hospital, car wash and laundry connected to the sewer.

The recurring look

Congress appears to have anticipated the second problem. The authority printed at the head of Part 503 is section 405 of the Clean Water Act, 33 U.S.C. 1345, and subsection (d) does not merely authorise the rule. It requires the agency to go back and look again, identifying additional toxic pollutants in sewage sludge and setting standards for them. The review runs on a two-year cycle. It is known, unglamorously, as the biennial review.

The looking has happened, and the agency's own account of it sits on its website under the heading of risk assessment:

To date, the EPA has found over 700 chemicals in sewage sludge. This number is based on the EPA's nine biennial reviews of published literature covering 2004 through 2021 and three EPA national sewage sludge surveys conducted from 1988 to 2006.

Nine reviews. Twelve national surveys and reviews in total across three decades. Over 700 chemicals identified. Nine metals limited.

The same page states how many times the agency has carried a pollutant through the full analysis needed to regulate it: "To date, the EPA has published two refined risk assessments for pollutants found in sewage sludge." One of those was the analysis that produced Part 503 in the first place. The other covered dioxins and PCBs, and produced no new numeric limits in these tables.

Further down the same page:

Prior to 2021, there was inconsistent reporting of chemicals identified and the EPA did not cumulatively track chemicals found in sewage sludge.

For most of two decades, the statutory review ran without a running list of what it had found. Each cycle looked, and reported, and the finding was not carried forward into a form that accumulated. The agency fixed this, and published a curated list in 2022, which is how the figure of 700 exists to be quoted. The review was performed throughout. What the review was for did not happen.

What the Inspector General found

In November 2018 the EPA's own Office of Inspector General published Report 19-P-0002, under a title that leaves little to interpret: EPA Unable to Assess the Impact of Hundreds of Unregulated Pollutants in Land-Applied Biosolids on Human Health and the Environment. The summary is four lines long. The agency, it says, identified 352 pollutants in biosolids but "cannot yet consider these pollutants for further regulation due to either a lack of data or risk assessment tools." The examples the OIG offers are pharmaceuticals, steroids and flame retardants.

The agency knows the chemicals are there; it named 352 of them. The obstacle is the machinery for deciding what to do about them, and the biennial review's own history suggests why: identifying a chemical and building the tools to act on it are not the same work, and only one of them has ever come with a deadline anyone outside the agency could enforce.

The strongest case for the agency

EPA's position is better than the numbers make it look, and the agency states it in a single sentence on the same page: "The presence of a pollutant in sewage sludge alone does not necessarily mean that there is risk to human health or the environment from its use or disposal."

That is straightforwardly true, and forgetting it is how bad regulation gets made. Detection limits have improved by orders of magnitude since 1993; a great deal of what appears on a 700-chemical list appears because instruments got better, not because fields got dirtier. Risk depends on how much is present, how much contact anyone actually has with it, and how toxic it is at that dose. Establishing all three for one chemical, through fate-and-transport modelling, is slow and expensive work.

EPA has also been building the thing the Inspector General said it lacked. It proposed a prioritisation and screening framework, put it to the agency's Science Advisory Board in 2021, and designed the screen around a deliberately unlucky hypothetical: a self-sufficient farm family, growing and eating and drinking from the affected land. In Unity, Maine, that hypothetical has an address. The stated purpose of the screen is to identify with confidence the chemicals that pose no risk even to that family, so they can be set aside and the remainder pursued. That is a defensible way to spend limited analytic capacity.

And beneath all of it sits a problem with no comfortable answer. The sludge exists. Four million dry metric tons of it were generated in a single reporting year, and it does not stop arriving because it becomes inconvenient. The alternatives are landfill, which fills up and leaks, and incineration, which is expensive and moves the problem into the air. Land application returns nutrients to soil that needs them.

And there is a sharper defence still, one this essay has not yet made on the agency's behalf. In November 1994 a federal court told EPA it had gotten the chromium and selenium limits wrong and ordered the agency to fix them. EPA had no lawful option to wait. Complying with that order inside eleven months, reassessing exposure pathways and updated toxicology standards along the way, is not evidence of capture or neglect. It is what a court order looks like when an agency obeys it promptly.

The difference

I accept that defence, and I still think the record shows something it does not cover. Yes, and that is the point: nobody had to act in bad faith, or even act slowly, for the ratchet to turn only one way. There is a difference between a question that is hard to answer and a question whose hardness has become the answer. The test is what happens when the difficulty lifts, and twice it has.

The first time was in 1995. On 5 March 1993, two weeks after Part 503 was published, the Leather Industries of America challenged the chromium limits in Tables 1 through 4. On 17 June 1993 the City of Pueblo, Colorado, filed its own petition challenging the selenium limits in Tables 1 through 3. The cases were consolidated and transferred to the D.C. Circuit, and on 15 November 1994 the court remanded the chromium cumulative loading rate and the chromium and selenium concentration limits, finding the agency had not shown an adequate evidentiary basis. EPA responded eleven months later, on 25 October 1995, at 60 Federal Register 54764. It deleted the land application limits for chromium entirely, and it raised the selenium limit to the risk-based figure of 100 milligrams per kilogram.

That was the tables' second substantive change, and their last. It removed a pollutant and loosened a second, and it took eleven months, because a federal court, acting on two industry petitions, required it. Chromium is still in Part 503, in the subpart that governs incinerators, where it is monitored and limited in the sludge fed to the furnace. On land it has no limit at all.

The reasoning in that decision matters more than the outcome, and it is the strongest thing anyone can say in the agency's defence. EPA had set the Table 3 limits at whichever was lower, the risk-based figure or the concentration already achieved by the best ninety-nine per cent of the country's sludge. It gave two reasons: a margin of safety, and preventing sludge quality from deteriorating below where it already stood. The court rejected both, holding that the statute requires a demonstrated link between risk and any pollutant concentration limit the agency adopts, and that the rule, applied evenly, only ever cuts one way: sludge that is currently clean may not be required to stay clean. Only a completed demonstration of harm will do.

So the stasis is not simple inertia, and it is not really an asymmetry of evidence either. Molybdenum and chromium were both removed by an agency that judged, on reflection, the numbers were probably too strict, and deleting an overprotective limit is not a scientifically harder task than adding a protective one. What differs is who can force EPA's hand. A trade association, a mining company, or a single city can start a legal clock: file a petition, and a court can order an answer inside a fixed window. Nobody can do that for the opposite question. Section 405(d)'s biennial review is a duty EPA owes to no petitioner in particular, and no one outside the agency can sue to make a risk assessment finish on time. The rule that governs both directions is symmetrical on its face, requiring a demonstrated link to risk whichever way a limit moves. What is not symmetrical is who can compel the demonstration. The agency has completed two refined risk assessments in thirty-three years. It has deleted a pollutant's limits, under outside compulsion, twice: molybdenum in 1994, chromium in 1995.

That asymmetry was not designed by anybody. It is the shape the machinery took.

The second thing happened in the same document, and I found it by accident while reading the 1995 amendment in full. Part 403's Appendix G, the list of pollutants eligible for a pretreatment removal credit, is not new to 1995. EPA built it in 1993, alongside Part 503 itself, and its land application column already carried figures for seventeen substances the pollutant tables never regulated, fourteen of them organic compounds — aldrin and dieldrin, benzo(a)pyrene, chlordane, DDT, lindane among them. That capacity existed from the start. What is new in 1995 is what happened to chromium.

Chromium had just lost its Part 503 limit for want of an adequate evidentiary basis, in the very rule that did this. It could not simply vanish from Appendix G, because that appendix is where a treatment plant's industrial customers get relief from stricter pretreatment standards, and EPA did not want them to lose it. So the agency returned to a chromium exposure pathway it had used once already, a tractor operator inhaling dust off a treated field, and, applying an updated occupational-safety standard, decided its earlier finding of 5,000 milligrams per kilogram had been too low. Elsewhere in the same day's Federal Register, in a companion proposed rule, EPA floated the pathway's new number: 12,000 milligrams per kilogram, more than double. The public had commented on the original figure in 1993. It had no chance to comment on this one. And rather than let dischargers wait even briefly, EPA wrote an interim footnote into the final rule granting chromium removal credits on a case-by-case basis until the new number was settled.

None of that is illegal, or even unusual as agency practice. On the same day, in the same rulemaking, EPA found the evidence too thin to keep a number that protected land, and found a reevaluation, run without the comment period its own rules ordinarily require, sufficient to raise a number that protected a discharger's convenience. The direction of the doubt was not neutral.

Two documents, eighteen months apart

Which brings the argument to the present, and to the chemicals on Sue Hunter's land.

On 14 January 2025 EPA released a draft risk assessment for PFOA and PFOS in sewage sludge, docket EPA-HQ-OW-2024-0504. It followed two completed refined risk assessments in thirty-two years; this one is still a draft. The comment period closed on 14 August 2025.

On 1 July 2026, eighteen months later, the agency released what came of it: a memorandum titled Draft Guidance for Reducing Risk from PFOA and PFOS in Biosolids, docket EPA-HQ-OW-2026-2509. The comment period runs to 5 October 2026, extended from 4 September. In the agency's own description, the memorandum "provides voluntary recommendations for potential ways to mitigate risks."

Read the qualifiers in order. It is draft. It is guidance. Its recommendations are voluntary. If finalised, it would advise. Nothing in it changes Table 1, and no PFAS compound has a numeric limit on land application anywhere in Part 503. EPA has not banned PFAS in biosolids.

The next National Sewage Sludge Survey, which will gather the concentration data on which any future assessment depends, is described on the agency's site as planned.

Unity, Maine

Maine did not wait. In 2022 the legislature enacted Public Law 2021, chapter 641, which added subsection 7 to section 1306 of Title 38 of the Maine Revised Statutes. Its terms are eleven lines long. A person may not apply to or spread on any land in the State sludge generated from a municipal, commercial or industrial wastewater treatment plant, nor compost made with it, nor any product derived from it intended as fertiliser, soil amendment, topsoil replacement or mulch.

Three exceptions follow. Landfill disposal is permitted. So is the spreading of residuals from food processing, brewing, fermenting and distilling, provided they have not been mixed with sludge. So is sludge from the production of precipitated calcium carbonate.

There is no testing threshold. An earlier draft of the bill had one, a committee amendment that would have allowed spreading where the material tested below 25 parts per billion of PFOA and 50 of PFOS. That version is not the law. What Maine enacted sets no number at all.

Maine is one state with a particular history, its own contaminated farms, and a legislature that had watched families lose their land. A national rule cannot be written from a single state's experience, and a prohibition is a blunter instrument than a limit. I think Maine went further than the evidence about most farms would support, and the state is still working out what to do with the sludge it no longer spreads.

What changed in Augusta was the form of the question. For thirty years the federal question has been how much of a given substance may safely be spread, and a court has held that this must be established before a limit may be set. Establishing it requires knowing which substances, at what concentrations, reaching whom, at what dose. That question has produced nine biennial reviews, twelve surveys and reviews, over 700 identified chemicals, two completed risk assessments and a third in draft, two deleted metals, one loosened limit, and a memorandum of voluntary recommendations now open for comment.

Maine stopped asking it in that form. Whether that was wise is a live argument in the state today. That an outright prohibition was available as a choice to any of the fifty states, for thirty years, and only one of them made it, is the part that took me by surprise.

Sue Hunter's fields are being sampled again this season, by researchers who came because the damage was uneven enough to be informative. The chemicals in her wetlands were in commerce for most of a century before anyone measured them there. They were not on any table.