Warning Signs

Strip away the administrative record, the acronyms, and five years of procedural history, and the federal government’s regulation of decaBDE in the American recycling stream comes down to this: the sole company EPA identified as owning and using plastic shipping pallets containing the chemical must post warning signs for its workers and require respirators and chemically resistant gloves.

That was the agency’s considered answer after a voluntary remand, a new proposal, and another round of public comment. The substance at issue is an additive flame retardant used in electronics, appliances, wire and cable, textiles, and automotive and aerospace parts. EPA rates it high for hazard, exposure, persistence, and bioaccumulation. It can remain in soil for years and in sediment likely for decades, move up the food chain, and break down into byproducts that persist even longer. EPA has linked it to damage to the immune and reproductive systems, the brain, thyroid, and liver, as well as cancer, endocrine disruption, and altered gene expression.

Against that substance, the signs.

The Ninth Circuit confined itself to the work Congress had assigned. It compared EPA’s explanations with the statute and the administrative record, pathway by pathway. Recycling, disposal, wastewater, sewage sludge: each arrived at the same defect. EPA had chosen restraint without substantiating it.

Congress had already selected the governing question. For this small class of persistent, bioaccumulative, and toxic chemicals, the agency was to reduce exposure as far as practicable. Low estimates could help calibrate a tool. Cost could rule out a measure. Technical limits could matter. But each conclusion had to be built from the tool, the cost, the technology, and the pathway actually under review.

Congress Had Already Answered the Threshold Question

The history of the Toxic Substances Control Act explains the impatience built into Section 6(h).

Under the original 1976 statute, EPA’s authority over existing chemicals was burdened by a requirement to choose the "least burdensome" regulatory option. The monument to that regime is asbestos. In 1991, the Fifth Circuit vacated EPA’s decade-in-the-making asbestos ban in Corrosion Proof Fittings v. EPA, faulting the agency’s proof and its treatment of less restrictive alternatives. For the next quarter-century, TSCA’s core authority over existing chemicals was widely regarded as close to unusable.

One feature of the old regime survived: substantial-evidence review. It is a searching standard, generally more demanding than the arbitrary-and-capricious test most rules receive, and it was the standard the Fifth Circuit turned against the asbestos ban. Congress retained it while discarding the substantive constraint that had helped doom the rule. The symmetry is striking. The same close examination of the record that once tested whether EPA had justified an ambitious restriction now tests whether, under a statute rewritten to demand action, it has justified restraint.

Congress rewrote TSCA in 2016. For most chemicals, the modernized law created a structured process of prioritization, risk evaluation, and risk management. For a smaller group already identified in EPA’s 2014 work plan—chemicals rated high for persistence or bioaccumulation and likely to expose the public, susceptible populations, or the environment—Congress created a faster lane. DecaBDE was one of them, alongside PIP (3:1) and three other PBT chemicals.

Section 6(h) dispensed with a new risk evaluation and put EPA on a clock: a proposed rule within three years, a final rule eighteen months later. In choosing regulatory tools, the agency was directed to address the risks and "reduce exposure to the substance to the extent practicable."

That language fixes the starting point. Congress had already decided that these chemicals warranted expedited attention. The remaining work was practical and concrete: identify the exposure, examine available controls, measure their costs and limits, and explain the choice.

EPA’s first decaBDE rule arrived on January 6, 2021, shortly after the statutory deadline. It broadly prohibited manufacture, processing, and distribution of decaBDE and products or articles containing it, subject to delayed compliance dates and exclusions. But it left much of the chemical’s afterlife alone: plastics already in circulation, new articles made from recycled feedstock, disposal, releases to air, water, and soil, and concentrations in sewage sludge.

Environmental, tribal, and consumer petitioners challenged the rule. EPA sought a voluntary remand, which the Ninth Circuit granted in 2022. The agency proposed revisions, accepted another round of comments, and issued amendments in November 2024.

At the pallet operation, the amendments became a sign, respirators, and chemically resistant gloves. Wastewater received an anti-backsliding provision for releases during manufacture, processing, and distribution of decaBDE and decaBDE-containing products, while the articles dominating downstream recycling and disposal remained outside it. Disposal and sewage sludge received no new controls.

Two rules and two administrations had reduced an expedited command to negotiating room.

The Agency Priced the Wrong Problem

Judge Gould’s opinion accumulates force by refusing drama. It takes EPA’s explanations in the order offered and sets each beside the authority Congress granted and the evidence the agency had.

The first explanation was "low levels." EPA relied on low estimated concentrations to leave recyclable articles and sewage sludge largely untouched. But Section 6(h) uses exposure to shape the response, not to reopen the threshold question Congress had bypassed. A low estimate might support a concentration limit instead of a ban, targeted controls instead of universal ones, or monitoring instead of prohibition. Standing alone, it says nothing about whether those measures are practicable.

The remand leaves EPA two lawful paths. It may regulate, or it may assemble a record showing why a particular additional control cannot practicably be used. Either path begins with the actual measure and the actual pathway.

EPA called further recycling regulation prohibitively expensive. Its economic support consisted largely of two studies aimed elsewhere. One surveyed laboratory testing across many chemicals without isolating decaBDE or the industries at issue. The other, from the Consumer Product Safety Commission, concerned children’s products tested for lead and phthalates. From those materials EPA cited a range of $50 to $350 per article, while acknowledging that the figure did not represent the cost of screening recycled plastics for decaBDE.

The agency priced the wrong chemical, in the wrong products, for the wrong regulated community—and called the answer impracticability.

It also priced the most drastic option as though it exhausted the field. The record contained narrower possibilities: concentration limits for waste streams known to carry more decaBDE, including end-of-life vehicles, construction and demolition debris, and electronic waste; density separation, in which many high-bromine plastics sink while lower-density plastics float; X-ray transmission scanners positioned over a conveyor; handheld XRF instruments that can read bromine in seconds. The court noted that XRF screening cost roughly one-tenth of the average laboratory-testing figure EPA had invoked. EPA neither costed those tools in a meaningful way nor explained why they would fail.

The machinery matters because it turns an abstract dispute about “testing costs” into a sequence of physical choices. Material enters a sorting line. A tank separates by density. A scanner reads fragments as they move. A handheld instrument checks a suspect stream. None of those facts proves that every recycler can adopt every method at any price. They do, however, identify measures that can be evaluated, combined, narrowed, or rejected on evidence. EPA’s analysis largely stepped over that middle ground and moved from the cost of laboratory testing to the cost of a complete ban.

Then came the policy defense. Regulating contaminated recyclables, EPA argued, would undermine its broader goal of promoting recycling. The concern is real. Circular systems can reduce extraction, energy use, and waste. Yet a circle is not virtuous merely because it closes. When persistent chemicals travel with the material, recycling can carry yesterday’s hazard into tomorrow’s products.

Policy could influence the choice among lawful tools. It could not replace the practicability inquiry with a general preference for keeping material in circulation.

A Neighboring Statute Is Not an Alibi

For disposal, EPA pointed to the Resource Conservation and Recovery Act and to state solid-waste programs. Existing controls, the agency said, had already reduced exposure as far as practicable.

RCRA matters. It regulates waste facilities and disposal practices that overlap with the decaBDE problem. But overlap is not equivalence, and TSCA was designed in part to reach chemical risks left between statutes aimed at emissions, discharges, waste sites, workplaces, and consumer products.

Those gaps were visible in the record. Ordinary municipal incinerators can send decaBDE into the stack while burning nonhazardous trash. Construction and demolition landfills receive the crushed remains of buildings. Certain small landfills fall outside important parts of the federal scheme. Once an article has become mixed debris, the chemical no longer travels under the label of the product that introduced it. It moves with ordinary waste into incinerators and construction-and-demolition landfills, where the original product category no longer defines the exposure pathway. That physical afterlife is precisely where a statute organized around chemical exposure can see something a facility-based waste program may miss. Petitioners also identified disposal controls used abroad and technologies that might reduce exposure. European and British controls appeared as evidence, not templates; they still required an answer before EPA could declare further action impracticable.

A neighboring statute could be part of the answer. It could not serve as the answer without a map of what it actually covered.

The Zero That Existed Only on Paper

Wastewater made the weakness harder to miss.

EPA defended its limited water prohibition in part with 2021 reporting data showing "zero releases" of decaBDE to water. The number came from self-reporting by a limited set of facilities. In the same record sat monitoring from the Washington State Department of Ecology documenting substantial decaBDE discharges from categories EPA had not addressed, including wastewater treatment plants, industrial laundries, and metal finishers.

The state measurements were not a judicially adopted national discharge total. They were something more basic and more immediately relevant: contrary evidence. A zero-release finding could not carry the rule while the agency left those measurements unanswered.

The physical pathway helps explain why the paper zero was unstable. A peer-reviewed Washington study sampled house dust and laundry wastewater from twenty homes near the Columbia River. Eighteen of the twenty-one flame retardants detected in dust also appeared in the wash water. The study covered a suite of flame retardants rather than a court-certified national mass balance for decaBDE, but it showed how chemicals shed from household products can leave through a washing machine, enter a sewer, and arrive at a treatment plant.

Treatment does not make the pathway vanish. Some contamination can pass out in effluent; some can concentrate in the solids treatment leaves behind. Those solids become sewage sludge, and sludge may be applied to land. Water and biosolids are therefore not separate regulatory curiosities. They are the liquid and solid endpoints of the same downstream system.

EPA’s rule treated each missing control as an isolated exception. The record described a connected route from legacy products to sewers, treatment plants, receiving waters, and soil.

"We Will Get to It Later" Is Not Expedited Action

EPA’s fallback appealed to institutional common sense. Complex regulation often proceeds in stages. Agencies learn, adjust, and tighten standards over time. The D.C. Circuit’s Bluewater Network decision had upheld a staged Clean Air Act rule where the statute contemplated future revisions and EPA committed to stricter standards later.

Staging, however, needs a statutory foothold and an actual next stage.

Section 6(h) is titled expedited action. It sets deadlines and omits the open-ended language that supported incremental regulation in Bluewater. EPA had made no commitment to return for another round of controls. The analogy supplied a future that neither the statute nor the rule contained.

Who Gets Hidden Inside the Word "Low"

The opinion speaks in the dry vocabulary of substantial-evidence review. Tribal trust duties, treaty rights, and environmental justice were not the bases of decision. That boundary is real.

The petitioners nevertheless reveal something a national average can conceal. The Yurok Tribe and Alaska Community Action on Toxics came to the case from places where water, sediment, fish, food, economy, and culture meet. Persistent chemicals moving through those systems do not distribute exposure evenly. EPA itself identifies tribes and subsistence fishers among the populations that may be especially vulnerable to PBT chemicals. For a community relying heavily on local fish, the relevant exposure is not the national mean detached from place. It is the concentration that reaches the water, the sediment that holds it, the species that accumulate it, and the frequency with which those species become food.

That is why "low exposure" is never a complete sentence. Low where? Averaged across whom? Measured at which point in a chemical’s life cycle? A national estimate can be numerically correct and still obscure the community living where the average breaks down.

Substantial-evidence review cannot equalize those burdens by itself. It can force the agency to face contrary evidence, identify the populations inside its aggregates, and explain why the chosen rule reaches—or fails to reach—the places where exposure travels.

The Honest Case for EPA

EPA’s best defense is difficulty.

Downstream regulation of legacy chemicals is genuinely hard. Recyclers may not know which incoming articles contain decaBDE. Testing and sorting costs fall on businesses far removed from the original manufacture. Wastewater plants receive contaminants from thousands of upstream sources. A closed-loop recycling system may offer real environmental advantages over disposal or replacement. The 2024 anti-backsliding provision was modest, but it was not meaningless. An agency with finite staff and money must triage.

That case deserves to be stated fairly. Pretending every pathway can be eliminated cheaply and immediately is not serious administration.

Difficulty, though, must be demonstrated at the level where the choice is made. A cost claim needs the cost of the relevant measure. A technical objection needs an examination of the relevant technology. Reliance on another statute needs an account of the gaps it leaves. Uncertainty needs engagement with the evidence that creates it.

EPA’s record never completed that work. Its restraint was asserted more clearly than it was substantiated.

What the Remand Leaves Open

The rule remains in force. Both sides asked the panel to remand without vacating it, because vacatur would have erased protections along with defects: the existing prohibitions, the water-release provision, the warning signs, the respirators, the gloves. The rule was too weakly justified to survive in four areas and too protective to discard wholesale. That is the practical tension of regulatory minimalism. A measure can fall short of the statute and still leave the public worse off if removed. Remand without vacatur preserves the floor while requiring EPA to rebuild the reasoning beneath it; the public keeps the controls already won, and the agency loses the ability to treat those controls as the end of the inquiry.

Time remains less disciplined. Petitioners sought a 180-day deadline, but the panel imposed none. EPA returns to the familiar tempo of agency rulemaking, with the opinion as its instruction rather than a judicial calendar. The next measure of progress will be whether the agency engages the alternatives and evidence already identified.

The policy choice also remains with EPA. The agency may adopt additional controls or build a substantially better explanation for declining them. Judicial review sets the governing question and tests the reasons; administration still chooses among the lawful answers.

The decision’s immediate shadow falls across the rest of Section 6(h), including PIP (3:1), which EPA revised in the same 2024 rulemaking. PIP’s exclusions remain in place, as does the 0.1 percent trace threshold. Exposure levels may inform the selection of a tool. The transferable demand is that thresholds, exemptions, and delays be tied to a record-specific account of practicability.

The reasoning travels further than the PBT rules. Borrowed cost studies, unexamined alternatives, statutory overlap, selective data, and future action promised only in litigation are recurring weaknesses in complex rulemaking. Expertise earns deference most readily when the record makes the expert work visible.

The Sign and the Signal

The pallets are presumably still circulating: sturdy, reusable, quietly brominated, a circular economy in miniature. At the operation EPA identified, workers handle them beneath a warning placard, wearing respirators and chemically resistant gloves. The sign speaks to the people closest to the chemical.

The opinion is a warning sign of another kind, addressed to the agency itself.

Congress chose practicability as the governing question. Exposure may shape the tool; it cannot empty the toolbox. Cost must belong to the measure actually under review. Another statute matters only to the extent it closes the same gaps. Contrary evidence must be answered. A later stage must exist in the rule, not merely in the brief.

For years, EPA treated Section 6(h)’s command as an opening bid. Three judges in San Francisco have now sent it back with instructions to show the work.