The BPA-Free Trap

In 1936, in a London laboratory, the biochemist Edward Charles Dodds was hunting for a synthetic oestrogen he could turn into a medicine. One of the compounds he tested stimulated the reproductive tract of ovariectomised rats — in the body, it behaved like the hormone. It was tens of thousands of times weaker than natural oestrogen, so Dodds set it aside and moved on to a stronger candidate, diethylstilbestrol, which a later generation would learn to regret.¹ The compound he put down was bisphenol A.

At that point bisphenol A was not a plastic. It would not become one for another seventeen years; polycarbonate arrived in the 1950s, and epoxy can linings followed.² So the first recorded property of the molecule that now lines tins and moulds water bottles was that it acted like a hormone. The estrogenicity that frightened parents in the 2000s was not a discovery that science made and consumers reacted to. It was written down in 1936, before the plastic existed.

That sequence — known, forgotten, re-encountered, relabelled — is the whole of this report, and it is calmer than it sounds. The question a parent in Reading is actually asking when she reaches past the ordinary tin for the one stamped BPA-free is did I avoid the hormone-active thing? The badge is answering a different, narrower question: is one named molecule, CAS number 80-05-7, absent? Both are real questions. They are not the same question. And the gap between them is where this story lives.

The bridge, not the rings

To see why removing bisphenol A is not the same as removing the hazard, you have to look at the shape of the molecule, because the shape is what a hormone receptor reads.

Every bisphenol is built the same way: two phenol rings — six-carbon rings each carrying a hydroxyl group — joined in the middle by a bridge. The oestrogen receptor reads the two hydroxyls. They sit at roughly the spacing of the hydroxyls on the body's own oestrogen, and that resemblance is what lets the molecule slip into the receptor and switch it on. The bridge in the middle is structural scaffolding; the receptor barely notices which bridge it is.³

Which is the catch. Bisphenol A's bridge is a carbon flanked by two methyl groups. Swap that bridge for a sulphur-and-oxygen group and you have bisphenol S. Swap it for a single carbon and you have bisphenol F. The name on the safety data sheet changes. The two hydroxyl "teeth" that the receptor reads do not. That is why a manufacturer can remove bisphenol A and drop in a structural near-relative without the receptor being any the wiser — and it is exactly why the published literature calls the move a "regrettable substitution," a term now established across peer-reviewed toxicology rather than coined by any campaign.⁴

Here the honesty has to be exact, because this is the point at which the topic is most easily overstated. "Structurally similar" does not mean "equally potent." When researchers pooled the laboratory studies, bisphenol F came out roughly as strong as bisphenol A on oestrogen activity — a relative potency of about 1.07. Bisphenol S came out weaker, around a third as potent, near 0.32.³ Some other relatives in the family — bisphenols AF and B — test stronger than bisphenol A.⁴ So no single sentence covers the family. "The substitutes are just as toxic" is false for bisphenol S; "the substitutes are safer" is false for bisphenol F and several others. The honest statement is that they are the same kind of molecule, acting on the same receptor, in the same order of magnitude — and that what they do to a human body over a lifetime, at the tiny doses real people get, is genuinely not yet established. The laboratory and animal evidence is consistent; the human disease evidence is immature. That distinction — between an intrinsic hazard, which is well-evidenced, and a real-world risk, which is not yet proven in either direction — has to be held the entire way through.

The receipt that names what the badge won't

There is one everyday surface where you can actually see which molecule replaced bisphenol A. It is not the tin, which carries a badge and makes a promise. It is the till receipt, which carries no badge and promises nothing.

Thermal receipt paper develops its print not with ink but with a colour-forming chemical spread loose across the surface — a free, un-polymerised bisphenol, sitting at around twenty milligrams per gram of paper, with no plastic to hold it in.⁵ For years that developer was bisphenol A. Then Europe restricted bisphenol A in thermal paper, in force from January 2020, and the market moved. In 2022 a team collected 571 receipts from shops across two dozen US states and ran each through an infrared spectrometer to identify the developer. Bisphenol S was the developer in about 85% of them. A genuinely non-bisphenol alternative accounted for around 12%. Bisphenol A: about 1%.⁶ (That survey is American; no equivalent receipt-by-receipt breakdown has been published for Britain, though the same European restriction drove the same switch here.)

Sit with what that means in the hand. The shopper who heard the warning about bisphenol A and acted on it — choosing the bottle with the badge, declining nothing, doing the responsible thing — has very probably been handed its successor across the counter, on a slip of paper that makes no claim at all. The badge on the bottle is true: there is no bisphenol A in it. The receipt is honest in a different way: it shows you the molecule that took its place. And the only place that successor's identity is written down is a spectroscopy dataset the shopper will never be handed. No rule requires it on the receipt. No rule requires it on the tin. The badge answers the narrower question — is the named molecule gone? — and falls silent on the one she was actually asking.

It is worth saying the uncomfortable part out loud, because the report's own best evidence points at it: where the swap can be measured, the market went overwhelmingly to bisphenol S — the relative that tests weaker on oestrogen activity, not stronger. On that one axis, the dominant substitution may well have been a partial step down in hazard rather than a sideways shuffle. That does not close the gap; it shows what kind of gap it is. The injury here is not "you were given something worse." It is that you cannot tell which way the trade went — better, worse, or neither — because the thing you traded into is unnamed, less studied, and, on the available human data, of unestablished consequence. There is even a wrinkle that cuts the other way on bisphenol S specifically: it crosses skin less readily than bisphenol A, but the body conjugates it less efficiently, so a higher fraction of what does get in circulates in the active, unbound form — about 6.9% of the total, against 2.7% measured for bisphenol A in the same volunteers in a concurrent study, in research on just five people.⁷ A weaker molecule that the body clears less cleanly is not obviously a better or worse bargain. It is an unanswered one. And an unanswered question you didn't know you were holding is the precise shape of the thing the badge was trusted to settle.

Why the badge can't be read straight — and why that's a fact about the claim, not the silence

A fair objection arrives here. No food-packaging rule anywhere makes a manufacturer list what the package is made of. A tin is not required to name its lining chemistry, its can-coating resin, or the trace substances that migrate from it; under the European food-contact framework the consumer-facing label carries only the "for food contact" mark, a traceability code and the maker's name and address.⁸ So why single out the missing bisphenol? Isn't "the substitute isn't named" simply true of every package on the shelf?

The answer is that the injury does not come from the silence. It comes from the claim that breaks the silence. A blank tin makes no promise; you read nothing into it. "BPA-free" makes one — and because it names a single famous villain and announces its absence, it is heard well past its literal four characters, as "bisphenol-free," as "the hazard is gone," as "safe." Harvard's Petrie-Flom Center, reviewing the labelling in late 2024, put it plainly: "BPA-free" is a voluntary, self-declared claim with no regulated definition, it does not mean bisphenol-free, and the substitutes need not be disclosed.⁹ The defect is not that food packaging is quiet about its chemistry. It is that an active, trusted, voluntary signal resolves a specific anxiety without delivering the thing the anxiety was about. The badge says something true and small, and is heard as something larger that no one ever certified.

The dose is something you already control

None of this means a worried reader is at the mercy of the molecule, because how much bisphenol actually leaves a container and reaches a body is not a fixed property of the plastic. It is a rate, and the rate is set mostly in the cannery and the kitchen.

For canned food, the great majority of bisphenol A migration, by the review evidence, happens during the heat-sterilisation step in manufacture, when the filled can is cooked at around 121°C to make it shelf-stable.¹⁰ Heat, fat and acid are the accelerants. In one classic test, bisphenol A migrating from a can coating into a fatty food simulant after that 121°C sterilisation reached about 646 micrograms per kilogram — roughly thirteen times the 0.05 milligram-per-kilogram migration limit that applies to food-contact bisphenol A in Britain today.¹¹ (That was a worst-case laboratory measurement — an older can coating, a fatty simulant, full sterilisation heat — not a typical reading from a modern tin. The point is not the single number but the size of the lever: heat and fat move how much bisphenol leaves a lining by a large factor.) Polycarbonate tells the same story from the other direction: the plastic is broken down by hot water, and one study found boiling-water exposure raised bisphenol A release from polycarbonate bottles as much as fifty-five-fold, with the bottle's age making almost no difference — temperature was the driver.¹²

That is the most useful thing in this report, because it is the part the shopper holds the controls to. The conditions that turn chemistry into exposure — heat, fat, acid, time — are the conditions a person decides. Don't microwave food in plastic; don't keep boiling liquid or the dishwasher's heat cycling through a polycarbonate bottle; don't store hot, oily or acidic food in a lined tin longer than it needs. None of that depends on the hazard being proven. It is low-cost and low-regret either way.

The system is moving — unevenly

The honest counterweight to all of this is that the regulators are not asleep, and the strongest version of the case for the status quo deserves to be stated before it is qualified.

Removing bisphenol A is genuine risk reduction, not theatre. It is the single most studied member of the family — the one with a harmonised hazard classification, and the one whose level in human urine has been measurably falling as regulation bit, even as its substitutes' levels rise.¹³ In 2023 the European Food Safety Authority cut its tolerable daily intake for bisphenol A by roughly twenty-thousand-fold, to 0.2 nanograms per kilogram of body weight, on the strength of an immune effect seen in mice.¹⁴ That figure is not unchallenged. Germany's own risk-assessment body, the BfR, published a detailed rejection of the derivation, calling the chosen endpoint insufficiently justified; and the long-running US CLARITY-BPA programme split, with the government's core study read by the FDA as showing little concern at typical exposures while the academic arms reported low-dose effects.¹⁵ The headline number, in other words, is contested at the scientific margin, and this report does not present it as settled.

But the regulatory direction is clear, and it has not stopped at bisphenol A. When the European Commission acted, in Regulation (EU) 2024/3190 — adopted in December 2024 — it did not merely ban bisphenol A in food-contact materials. It deleted the authorisations for both bisphenol A and bisphenol S from the list of substances permitted in food-contact plastics; under a system where only listed substances may be used, striking bisphenol S off the list removes its licence too.¹⁶ Bisphenol S had already been formally identified as a substance of very high concern, listed by the European Chemicals Agency on 17 January 2023 and classified as toxic for reproduction.¹⁷ The drop-in twin did not slip through the new European rule. That is the loophole closing.

Which makes Britain's position the live one. The Food Standards Agency consulted on aligning Great Britain with the European restriction, then postponed: there would be no enforcement until new domestic legislation is introduced, and no such legislation is yet in place — leaving no firm British date. The Foodservice Equipment Association had pressed the case, in the consultation process, that matching the European timetable would cause significant economic and environmental damage.¹⁸

It is important to read what that does and does not leave behind. Great Britain is not a vacuum. The pre-Brexit European migration limit for bisphenol A — 0.05 milligrams per kilogram, with non-detection required for food contact in baby food — was retained in GB law and is in force and enforceable today.¹⁹ What has been postponed is the update: the move to post-2023 science and the de-authorisation of the substitute. So the honest description of the British position is not "no protection." It is that the British backstop is frozen on the 2015 science that Europe has since abandoned, and silent on the substitute that Europe has since struck off — while across the Irish Sea, Northern Ireland stays on the European rules. The number that protects a shopper in Belfast and the number that protects a shopper in Birmingham are, for now, different numbers.

There is a pattern under all of this worth naming carefully, because it is easy to inflate into a conspiracy. When a regulator bans a single named substance, and a structurally similar relative is already registered and ready to drop into the same factory line, the fast and cheap compliance move is the near-twin — and under deadline pressure, fast and cheap tends to win. That is not an iron law; it is a documented tendency under specific conditions, and the conditions matter.²⁰ The proof that it is escapable is in this very story: Europe's 2024 rule went after the whole hazardous class rather than one molecule, and genuinely non-bisphenol can linings — polyester, acrylic and the old plant-resin oleoresin enamels — already exist and, by the US can-makers' own trade figures, account for the great majority of food cans there.²¹ A clean lining is not a fantasy; it is in mass production. The substitution to a bisphenol was a choice among available chemistries, not a forced move. (A caveat travels with that hope: "non-bisphenol" is not the same as "proven safe" — these systems are simply newer and less studied — and can linings exist for a real reason, to stop metal corroding into the food. The issue is disclosure and dose, not that cans are bad.)

Reading past the badge

The instinct the badge rewards is to scan for one word and relax. The steadier move is to trust the material over the claim, because a claim is a promise earned by the smallest possible change, while a material is just the thing itself.

The single most reliable signal at the shelf is not a badge at all — it is the small recycling triangle. A polycarbonate item carries a 7 (or no clear code); that is the plastic that hydrolyses under heat into bisphenol A, and for hot or repeated use it is the one to set down in favour of glass or stainless steel. For storage and reheating, the glass dish or jar already in the cupboard does the job a "BPA-free" container is bought to do, at no cost and with nothing to swap out underneath the label. Where you can choose the format, fresh or glass-jarred over a lined tin removes the question rather than relabelling it — and for the receipt, declining the paper slip or taking it digitally removes the one exposure route in this whole story that is pure surface and pure free monomer.

If you do want a sealed, lined product, the honest position is that no consumer label on a British shelf will currently tell you whether the lining is a genuinely non-bisphenol enamel or a bisphenol-S epoxy — so the question to put to a brand's customer line is specific: which coating chemistry is this lining, and is it a bisphenol system or not? The makers who have moved to polyester, acrylic or oleoresin tend to be able to answer; the answer, not the badge, is the thing worth trusting.

What would change this analysis is concrete and not far-fetched. If a British or European rule began requiring the substitute bisphenol to be named on consumer packaging — or if a shelf-readable term reliably separated a non-bisphenol lining from a bisphenol-S one — the informational gap this report rests on would close, and the badge would start answering the question it is trusted to answer. If Great Britain enacted its postponed alignment on a firm date, the frozen British backstop would catch up to the science. And if the human-outcome evidence on bisphenol S and F matured enough to settle whether these low, near-ubiquitous exposures matter to health, the hazard-versus-risk question that this report deliberately leaves open could be closed in either direction. None of those has happened yet. Until they do, the proportionate reading is the calm one: "BPA-free" means one named molecule is gone — which is real, if narrower, risk reduction — not that the hazard is gone and not that the container is safe. Treat the badge as the small true thing it is, prefer glass and steel where it is easy, control the heat and the time, and put the precise question to the people who can answer it.