LL2-A2 — Hindsight check: Annex 2, overview of Late lessons from early warnings vol. 1#

Late lessons from early warnings: science, precaution, innovation, EEA Report No 1/2013, Annex 2, report pp. 701–716 (PDF 703–718). Checked against developments from 2013 to late September 2026.

Working files used: the digest (working/late-lessons/digests/LL2-A2.md) and the source extract (working/text/chunks/LL2-A2.txt). All other evidence comes from external sources, listed at the end.

Overview#

Annex 2 recaps the fourteen case studies from the 2001 volume. It has three parts: a lag table, Table A2.1 (p. 702, taken from Gee 2009, a paper on electromagnetic fields); a short summary of each case; and a chronology for each case. Most chronologies end in 2001. The annex does not say that its parts date from different years.

This check therefore asks two questions:

1. Were the 2001 facts, projections and open questions that the annex repeats in 2013 borne out by later events?
2. Do the later events strengthen or weaken the general lessons the annex is used to illustrate?

Headline result. Hindsight mostly vindicates the qualitative story in the annex, and several of its lessons come out stronger than the 2013 text states them. The lessons concerned are:

• long lags between warning and action
• slow recovery after action
• regrettable substitutes
• the gap between adopting a rule and actually reducing risk
• the low-dose behaviour of carcinogens

Four qualifications apply.

• The headline lag measure is not robust. Table A2.1’s “years of substantial inaction” figures are built on inconsistent dating rules. Later events also show that “effective risk reduction action” was a process that ran for decades beyond the dates in the table, not a single moment in time (claim 1).
• Several numbers repeated from 2001 are imprecise or unverifiable. These are the 400,000 asbestos deaths (claim 2), the “more than 200 deaths” from the benzene delay (claim 3) and the EUR 160 million sanction (claim 6).
• The hormones case remains contested. International risk-assessment bodies still regard the residues from approved use as safe. The EU has kept its ban, but the trade dispute was settled with a beef quota, not by scientific agreement (claim 6).
• Some policies later went further than the annex describes. The EU’s full ban on antimicrobial growth promoters (2006) and its rules for imports (from September 2026) go beyond it (claim 7). So do the US 2024 ban on chrysotile asbestos and the EU’s 2022–2026 cut in the benzene exposure limit (claims 2 and 3). All of these came long after the “effective action” dates in Table A2.1.

Scope note. The 2013 report updates several vol. 1 cases in other parts of the volume. Under the project’s file rules, those updates were not opened for this check. Where this check reaches similar conclusions, it does so from external sources.

Access note. A few primary documents could not be retrieved: OSHA’s 1987 benzene rule preamble, the minutes of ECHA’s Risk Assessment Committee meeting RAC-77, and some national news releases. Where a point depends on a secondary source, or on a document that was not seen, the text says so.

Summary of verdicts#

Claim 1 — Table A2.1: “years of substantial inaction” (p. 702)#

Original claim (p. 702). Table A2.1 gives, for each of the 14 vol. 1 cases, a date of first early warning, a date of “effective risk reduction action” and the “years of substantial inaction” between them. The figures range from 5–30 years (TBT) and 10–17 (BSE) up to 65 (radiation), 81 (benzene), 101 (asbestos), c. 100 (PCBs) and “hundreds” (fisheries). The source given is Gee (2009).

Internal consistency (what the table itself implies)#

Recomputing each lag from the table’s own dates shows that the table does not use one dating rule.

• Radiation, asbestos, antimicrobials, MTBE and hormones follow simply from the dates given: 1896→1961 = 65; 1898→1999 = 101; 1969→1999 = 30; 1960→2000 = 40; 1972→1988 = 16.
• Benzene (81 years) is measured to the 1978 withdrawal from consumer products. The case, however, is about occupational exposure. Measured to the 1987 1 ppm occupational standard, the lag is about 90 years.
• PCBs (c. 100 years). The table’s own action entry (“1970–1980s: EU and US restrictions”) gives about 71–90 years from 1899. “c. 100” only works if the end point is the 1996 EU elimination directive (p. 707), or later.
• Great Lakes (45+ years). The action entry (“1970s DDT banned”) gives about 7–10 years from 1962/63. “45+” is measured to roughly the time the table was compiled, 2008–09, on the grounds that “debates continue”. This is a different kind of end point: continuing inaction rather than a date of action.
• BSE (10–17 years) starts both ends of the range at the earliest warning (1979). Measured from the 1986 official recognition, the lag would be 3–10 years.
• SO₂ (25–55 years) and halocarbons (10–30 years) do not exactly match any single pairing of the dates shown. For halocarbons, 1974→1987 = 13 and 1974→2010 = 36.
• Fisheries (“hundreds”) runs from a 1376 petition to the 1995–2008 period of “responsible” management. The “action” in that entry is one the table itself calls “not very effective”.

The digest’s caveat is therefore confirmed: the PCB and Great Lakes figures do not follow from the table’s own entries, and the table mixes lags to action with lags to the date the table was written. Gee (2009) is a discussion paper about definitions and about electromagnetic fields; it is not a systematic method for dating warnings and actions (PubMed 19467848, published online 21 May 2009).

What happened after 2013: “effective action” kept arriving#

For several cases, substantive new risk-reduction action came well after the “effective action” date in the table.

• Asbestos (table date 1999).
• In the EU, Directive (EU) 2023/2668 (22 Nov 2023) cuts the occupational exposure limit tenfold, from 0.1 to 0.01 fibres/cm³. From 21 December 2029 it requires either 0.01 f/cm³ (also counting the thinnest fibres) or 0.002 f/cm³ (EUR-Lex CELEX 32023L2668).
• In the US, EPA’s final TSCA rule banning ongoing uses of chrysotile was published on 28 March 2024 and took effect on 28 May 2024. It gives some users multi-year transition periods, and chlor-alkali plants a phase-out that runs past five years (Federal Register 2024-05972; EPA rule page, last updated 5 May 2026).
• Measured to the US ban, the asbestos lag in the largest economy that had not banned it becomes about 126 years.
• Benzene (table date 1978). The EU cut its binding limit from 1 ppm to 0.2 ppm, which applies from 5 April 2026 (Directive (EU) 2022/431, 9 March 2022) (claim 3).
• PCBs (table: “phase out by 2010”).
• The Stockholm Convention asks Parties to remove PCB equipment from use by 2025 and to manage PCB wastes in an environmentally sound way by 2028.
• Reported figures as of 2020 put total PCB production at 1.05–1.51 million tonnes. Only 581,240 t had been disposed of (local destruction plus exports), and 639,057 t was held in inventories. That is well under half ahead of the 2025/2028 deadlines (Stockholm Convention PCB progress page, accessed Sept 2026; decision SC-11/3).
• In the North-East Atlantic, OSPAR’s Quality Status Report 2023 finds that most sub-regions (10 of 12) are in poor status for hazardous substances in marine animals. The main causes are mercury and PCB118 (OSPAR QSR 2023, thematic assessment on hazardous substances).
• Great Lakes.
• Areas of Concern (AOCs) are the polluted harbours and river mouths designated for clean-up under the 1987 protocol. As of September 2026, EPA’s AOC page lists eight US AOCs as delisted: Ashtabula River, Deer Lake, Lower Menominee River, Muskegon Lake, Oswego River, Presque Isle Bay, Rochester Embayment and White Lake. Clean-up of the rest is still under way.
• This is consistent with the annex’s statement that “Decades later the efforts to clean up the area continue” (p. 713) (EPA Great Lakes AOCs pages, accessed Sept 2026).
• Antimicrobial growth promoters (table date 1999). The EU’s full ban came in 2006, and import conditions follow from 3 September 2026 (claim 7).
• TBT (table: “2008 global ban”). The global convention entered into force only on 17 September 2008 (claim 8).

Verdict: partly held up#

The qualitative claim holds and is strengthened: lags of decades between credible warning and effective protection were common, and in several cases protection took even longer than the table implies. The specific numbers do not hold up as a measure. They depend on inconsistent choices of “first warning” and “effective action”, and at least two (PCBs, Great Lakes) do not follow from the table’s own entries.

Implications for weight#

• The lesson “delays of decades are common” (digest insight 1, p. 702) deserves moderate weight. It should be stated as a range, with the dating rule made explicit.
• Hindsight adds a sharper and better-supported point: effective action is usually a sequence of measures over decades, not a single moment. Asbestos is the clearest example: a 1999 EU ban, a 2023 EU exposure-limit cut and a 2024 US ban. Any use of the table as a lens should compare like with like: warning to first binding measure, and warning to measurable reduction in harm.
• The table was built only from cases where harm occurred (the selection caveat in the digest). It still cannot say how often early warnings turned out to be false alarms.

Claim 2 — Mesothelioma as “the bulk” of asbestos harm; “some 400 000 expected deaths in Europe over the next three decades” (p. 706)#

Original claim (p. 706). “It is this cancer [mesothelioma] that is now causing the bulk of current and future harm from asbestos, including some 400 000 expected deaths in Europe over the next three decades.” This is a 2001-era projection repeated without change in 2013.

Subsequent developments#

Origin of the figure. The best-known late-1990s projection is Peto et al. (1999). It projected that male mesothelioma deaths in Western Europe would rise from about 5,000 in 1998 to about 9,000 around 2018 and then decline, giving about 250,000 deaths over 1995–2029 (Br J Cancer 79:666–72, Feb 1999). The annex does not say whether its 400,000 means mesothelioma alone, or all asbestos cancers, or a wider “Europe”. It is larger than Peto’s mesothelioma-only figure.

Observed mesothelioma mortality.

• Great Britain (the most detailed series). HSE’s July 2026 statistics give:
• 2,146 mesothelioma deaths in 2024
• 2,255 deaths in 2023
• an average of 2,508 deaths a year over 2011–2020

Deaths “remained broadly level during the 2010s at around 2,500 deaths per year”. Updated projections show male deaths falling during the 2020s and total deaths “continu[ing] to decline during the 2020s and 2030s”. The decline among women may not be clear until the end of the 2020s (HSE, Mesothelioma statistics for Great Britain, 2026, July 2026). This matches the late-1990s picture of a peak around the late 2010s followed by decline. - EU-27. The Commission’s 2022 asbestos communication cites Global Burden of Disease 2019 estimates for 2016. Of an estimated 66,808 EU-27 deaths attributable to past occupational asbestos exposure, mesothelioma accounted for 7,510, and trachea, bronchus and lung cancer for 61,035. It also states that “in 2019, occupational exposure to asbestos claimed over 70 000 lives in the EU-27”, and that 78% of cancers recognised as occupational in the EU are asbestos-related (COM(2022) 488, 28 Sept 2022). The Commission expects asbestos deaths from pre-2005 exposure “to occur until the late 2020s and 2030s”. - Europe and the US more broadly. A 2020 review finds that mesothelioma incidence is declining modestly in countries that restricted asbestos early. Rates are still rising worldwide, and the disease is increasingly affecting women (Alpert et al., Transl Lung Cancer Res, Feb 2020).

Checking the annex’s two statements against these data.

• “About 400,000 deaths over three decades” (roughly 13,000 a year). If this means mesothelioma alone, it is the right order of magnitude but on the high side. Observed mesothelioma deaths across the EU-27 plus Great Britain were about 10,000 a year around the late 2010s (about 7,500 plus about 2,500), and are now falling. If it means all asbestos deaths, it is far too low: later burden estimates of roughly 67,000–70,000 a year in the EU-27 alone imply about 2 million deaths over 30 years.
• Mesothelioma is “the bulk” of current and future harm. Later burden estimates do not support this. In the GBD-based figures cited by the Commission, asbestos-attributable lung cancer deaths outnumber mesothelioma deaths by about eight to one. One caveat: lung-cancer attribution is modelled, while mesothelioma is an unusually specific marker of asbestos exposure. That specificity may be why the 2001 authors treated mesothelioma as the measure of harm.

Later regulatory action.

• The EU tightened the occupational limit tenfold in 2023 (Directive (EU) 2023/2668).
• The US banned ongoing chrysotile uses in 2024 (EPA; Federal Register 2024-05972).
• Global production was still about 1.2 million tonnes in 2021 (COM(2022) 488). The harm in Europe is falling, but it continues elsewhere.

Verdict: partly held up#

The expected course of the European mesothelioma epidemic (a peak around the late 2010s, then a slow decline) has been borne out. The claim that mesothelioma is the bulk of asbestos harm has not. Later burden estimates attribute far more deaths to asbestos-related lung cancer, and the total asbestos toll in Europe is several times larger than the annex’s figure.

Implications for weight#

• The lessons about latency and lock-in are strengthened: harm continues for decades after a ban (digest insight 15; pp. 702, 706). Deaths in Great Britain are only now declining, a quarter-century after the 1999 ban.
• The case also strengthens the digest’s insight 3, that the first harm recognised is often not the largest. The most specific, recognisable harm (mesothelioma) acted as the signal, but the larger harm (lung cancer) is harder to attribute and so less visible.
• The 400,000 figure should not be quoted as a measure of total asbestos harm. It is not sourced in the annex and is ambiguous in scope.

Claim 3 — Benzene: the 1977–87 delay “caused more than 200 deaths”; disease at 1 ppm by 1996 (p. 705)#

Original claim (p. 705). The US Department of Labor sought a 1 ppm standard in 1977. It was challenged by the American Petroleum Institute (API), and the 1980 Supreme Court “Benzene Decision” severely limited regulatory action. A 1 ppm standard came only in 1987, and “this 10-year delay caused more than 200 deaths in the United States”. The chronology adds: “1996 Studies showing benzene-related diseases from 1 ppm level of exposure” and “2001 Petrol contains benzene, giving public exposure risk”.

Subsequent developments#

The legal doctrine. Industrial Union Dept. v. American Petroleum Institute, 448 U.S. 607 (decided 2 July 1980) requires OSHA to make “a threshold finding that the place of employment is unsafe in the sense that significant risks are present” before it issues a standard (Cornell LII text). This “significant risk” requirement still governs US occupational health standards. The US limit remains 1 ppm as an 8-hour time-weighted average (29 CFR 1910.1028, OSHA, accessed Sept 2026). The US limit has not changed since 1987.

Low-dose evidence after 2001. The evidence on low doses has broadly supported the annex, though not uniformly.

• Lan et al. (2004) found lower white-cell and platelet counts in exposed workers “even for exposure below 1 ppm in air”, with susceptibility modified by genetic variants in metabolising enzymes (Science 306:1774–6, 3 Dec 2004). Letters later published in Science challenged the study.
• An industry-led pooled analysis of petroleum-distribution workers found a monotonic dose–response for myelodysplastic syndrome at low cumulative exposures (highest tertile >2.93 ppm-years) and with peak exposures above 3 ppm. It found “little evidence of dose-response relationships for AML, CLL, CML, or MPD” (Schnatter et al., JNCI, Nov 2012; the first author was affiliated with ExxonMobil). This cuts against a simple reading that low exposure raises AML risk.
• IARC Monograph 120 (evaluated in October 2017, published in 2018) reconfirmed benzene as a Group 1 carcinogen and a cause of AML in adults. It also reported positive associations with other blood and lymphatic cancers (Loomis et al., Lancet Oncol, online 26 Oct 2017; IARC Monographs vol. 120).
• A systematic review and meta-analysis found a 33% raised risk of non-Hodgkin lymphoma in highly exposed groups (meta-RR 1.33, 95% CI 1.13–1.57) (Rana et al., Lancet Planet Health, 25 Aug 2021).

Limit reductions. The EU’s binding limit fell from 1 ppm to 0.2 ppm (0.66 mg/m³). It was set at 0.5 ppm from 5 April 2024 to 5 April 2026, and 0.2 ppm applies from 5 April 2026. The Commission is to “assess the feasibility of a further reduction”, taking into account the 2018 opinion of ECHA’s Risk Assessment Committee (RAC), which recommended a lower value of 0.05 ppm (Directive (EU) 2022/431, 9 March 2022). The EU limit is now one-fifth of the US limit.

Public and consumer exposure.

• The chronology’s 2001 entry on public exposure through petrol has been reinforced. In March 2025 IARC classified automotive gasoline as carcinogenic to humans (Group 1), causing AML and bladder cancer (IARC news, 21 March 2025).
• The 1978 withdrawal of benzene from US consumer products did not close off consumer exposure. FDA reports recalls of hand sanitisers, benzoyl peroxide products and aerosol drug products because of benzene contamination. It treats levels above 2 ppm as grounds for recall (FDA alert, content current as of 24 July 2025).

“More than 200 deaths”. No independent source consulted for this check reproduces or tests this figure. It is an estimate by the chapter author (Infante), who also wrote the 1977 cohort study at the centre of the dispute. OSHA’s 1987 final rule (52 FR 34460, 11 September 1987) contains the agency’s own quantitative leukaemia risk estimates, but its preamble could not be retrieved for this check. The figure should be treated as an estimate from an interested party that has not been verified.

Verdict: strengthened (for the low-dose and regulatory-trajectory claims); unverified (for the death toll)#

The annex’s direction of travel was right. Effects at or below 1 ppm have been reported, IARC has broadened the evidence of harm, and the EU has cut its limit five-fold with a further cut under consideration. The one industry-linked pooled analysis did not find a low-dose AML dose–response, but it did find one for myelodysplastic syndrome.

Implications for weight#

• The lesson that “safe” thresholds tend to recede as evidence accumulates (digest insight 4; pp. 704–705) is strengthened by benzene.
• The standard of proof governs the pace of action (insight 5, p. 705). The US–EU split since 2022 illustrates it: under the “significant risk” doctrine the US limit has stayed at 1 ppm, while the EU acted on a scientific committee’s recommendation.
• The “>200 deaths” figure should be quoted, if at all, as the author’s estimate.

Claim 4 — CFCs replaced by HCFCs and HFCs; “the political process failed to stimulate more radical changes”; chlorine restored by 2050–2060 (p. 708)#

Original claim (p. 708). CFCs were replaced largely by HCFCs, which still deplete ozone though less strongly, and by HFCs, which do not deplete ozone but are “powerful greenhouse gases”. “The political process failed to stimulate more radical changes towards halocarbon-free and energy-efficient technology.” The 1997 Montreal amendments aimed “to restore levels of chlorine by 2050–2060”.

Subsequent developments#

The substitute problem was formally recognised.

• The Kigali Amendment to the Montreal Protocol, adopted on 15 October 2016, added a phase-down of HFCs. It entered into force on 1 January 2019. It has 174 parties, including China (17 June 2021), India (27 September 2021) and the United States (31 October 2022) (UN Treaty Collection, status accessed Sept 2026).
• The 2022 Scientific Assessment Panel report estimates that Kigali will avoid 0.3–0.5 °C of warming by 2100, not counting HFC-23 (WMO/UNEP press release, 9 Jan 2023).
• The EU went further. Its 2024 F-gas Regulation reduces the HFC quota to zero from 2050, with a review of remaining needs before 2040 (Regulation (EU) 2024/573, 7 Feb 2024).

Recovery timeline. The 2022 assessment found that nearly 99% of banned ozone-depleting substances had been phased out. If current policies stay in place, ozone is expected to return to 1980 values:

• around 2040 for most of the world
• around 2045 over the Arctic
• around 2066 over the Antarctic

(WMO/UNEP, 9 Jan 2023). The 2050–2060 horizon in the annex refers to chlorine levels, not ozone. On current assessments, the Antarctic ozone hole recovers somewhat later than that horizon suggests. The next quadrennial assessment had not been published by the time of this check.

Compliance and monitoring.

• Atmospheric measurements revealed an unexpected rise in global CFC-11 emissions after 2012, “despite reported production being close to zero since 2006”. This pointed to unreported new production (Montzka et al., Nature, 16 May 2018).
• Emissions then fell sharply after enforcement action (Park et al., Nature, 10 Feb 2021).
• This is a post-2013 example of three lessons: a rule is not the same as a reduction in risk, independent monitoring is essential, and illegal trade matters. The annex’s 1999 “Beijing Declaration” entry anticipated the last point.

A second round of regrettable substitution.

• The low-global-warming-potential hydrofluoroolefins (HFOs) now replacing HFCs break down in the atmosphere to trifluoroacetic acid (TFA). TFA is highly persistent and mobile in water. Germany’s Environment Agency expects the switch to R1234yf and similar gases to “lead to continuous and even increasing input of persistent TFA into the environment” (Umweltbundesamt, TFA as degradation product of fluorinated greenhouse gases).
• On 22 July 2026 EFSA lowered the acceptable daily intake for TFA from 0.05 to 0.014 mg/kg bw/day, citing new evidence including effects on thyroxine (EFSA news, 22 July 2026). EFSA framed this mainly around pesticide sources.
• Secondary sources report that ECHA’s Risk Assessment Committee agreed in June 2026 that TFA meets the criteria for classification as toxic to reproduction, category 1B (RAC-77 minutes, https://echa.europa.eu/documents/d/guest/rac77_final_minutes_en). The ECHA document could not be accessed for this check, so this point is unverified.

Verdict: strengthened#

Both halves of the annex’s assessment look better with hindsight. The Montreal Protocol is on track, which supports “action works, but recovery is slow”. The critique of substitutes has been confirmed twice: by the need for Kigali to deal with HFCs, and by the emerging concern about TFA from the next generation of substitutes. The recovery timeline is roughly as stated, with the Antarctic running later.

Implications for weight#

• The digest’s insight 8, that substitutes can carry hazards that could have been known in advance (p. 708), carries strong weight. It is also recurrent: each generation of substitutes has raised a new persistence or climate problem.
• Insight 15 (action works, but recovery takes decades) is strengthened.
• Insight 12 (monitoring is a precondition for any warning) is strengthened by the CFC-11 episode.
• The claim that “the political process failed” is fair for 1987–2001. After 2013, politics did push for more radical change (Kigali, the EU’s 2050 end to HFCs), but a generation later than the annex implies was needed.

Claim 5 — MTBE: groundwater contamination “rendered large water reserves useless”; permitted except in Denmark and California; “the debate goes on” (pp. 702, 712)#

Original claim (pp. 702, 712).

• MTBE contamination of groundwater “rendered large water reserves useless”, and Santa Monica “lost 71 % of its local water supply”.
• In 2001 MTBE was “undesirable in Denmark and California, permitted elsewhere”.
• The European Chemicals Bureau decided not to classify it as a carcinogen.
• “The debate goes on”.

Subsequent developments#

US phase-out.

• The Energy Policy Act of 2005 removed the oxygenate requirement for reformulated gasoline, and refiners “made a wholesale switch” to ethanol. MTBE “has not been used in significant quantities in RFG areas since 2005” (EPA, MTBE in Fuels, 2013 status update).
• Between 2000 and 2007, 23 states banned MTBE in whole or in part (US EIA, Today in Energy, 13 July 2018).

Harm moved elsewhere. US producers kept making MTBE for export: 38,000 barrels a day in 2017, mainly to Mexico, Chile and Venezuela, where it is still used as an oxygenate (EIA, 13 July 2018).

Remediation and liability costs.

• Santa Monica. The Charnock sub-basin wells, about half of the city’s daily water demand, were shut in 1996. Under EPA and regional-board orders, oil companies were expected to spend “in excess of $200 million” on clean-up. They also had to supply replacement water at more than $3 million a year and, under a 2003 agreement, build treatment systems (Environment News Service, 17 Feb 2005, archived; secondary source quoting EPA).
• New Hampshire. The state’s Supreme Court affirmed a jury award of about $236 million against ExxonMobil for MTBE groundwater contamination (State of New Hampshire v. Exxon Mobil Corp., 168 N.H. 211, decided 2 Oct 2015).
• Other cases. Litigation continued in other states, for example Rhode Island v. Atlantic Richfield Co. (2018). The City of New York won a $105 million verdict in 2009 (reported by the New York Times, 20 Oct 2009; secondary source).

EU status. MTBE is still permitted. The EU petrol specification allows “ethers containing five or more carbon atoms per molecule” (a group that includes MTBE) up to 22% by volume (Directive 2009/30/EC, 23 April 2009, Annex I). The chronology shows Denmark’s action. No EU-wide restriction was found in this check.

Carcinogenicity. In March 2025 IARC classified MTBE and ETBE as possibly carcinogenic to humans (Group 2B), based on sufficient evidence in experimental animals. It found the human evidence “inadequate” (IARC news, 21 March 2025; the full assessment is Monographs vol. 138). IARC had rated MTBE Group 3 in 1999. The suspicions of asthma and endocrine disruption noted in the annex (p. 712) were not found to be established in the sources consulted.

Verdict: strengthened for the water-resource harm and its costs; health suspicions only partly borne out#

• The groundwater harm and its costs have been confirmed and priced by courts in hundreds of millions of dollars.
• The US phase-out that was expected in 2001 took place.
• “The debate goes on” was accurate: the EU still allows MTBE, and the carcinogenicity rating moved only to “possible”, on animal evidence alone.

Implications for weight#

• The case strongly supports the digest’s insight 8. The hazards of the substitute (persistence, taste and odour in water) were documented in the 1950s–60s (p. 712) and could have been known before adoption.
• It supports insight 10 (knowledge sits unused) as far as the facts cited in the annex go.
• Hindsight adds a lesson the annex only hints at: a domestic phase-out can move production and exposure elsewhere. Exports of MTBE continued. Separately, a clean-air law (the 1990 Clean Air Act amendments) created the market for it, which is a case of one environmental goal creating a different hazard.

Claim 6 — The EU hormone growth-promoter ban as “an early example of application of the precautionary principle”; EUR 160 million a year in sanctions (p. 715)#

Original claim (p. 715).

• The EU banned several hormones as growth promoters in 1988 and extended the ban to imports in 1989.
• It did so despite the Lamming Committee (1982) and the FAO/WHO Joint Expert Committee on Food Additives (JECFA, 1988) finding them safe.
• The annex presents this as “an early example of application of the precautionary principle”.
• The EU “still suffers from sanction to its exports of around EUR 160 million per year” (2001).

Subsequent developments#

WTO rulings.

• Original dispute (1998). The Appellate Body upheld the finding that the EU import ban breached Article 5.1 of the SPS Agreement because it was “not based on a risk assessment”. It reversed the finding that the ban was arbitrary discrimination or a disguised restriction on trade, noting “genuine anxieties concerning the safety of the hormones” (AB report circulated 16 Jan 1998, adopted 13 Feb 1998) (WTO DS26 summary; one-page summary). The Appellate Body also considered the precautionary principle and declined to let it override the SPS requirement for a risk assessment (listed among “other issues” in the WTO summary).
• Sanctions. The authorised retaliation was US$116.8 million a year for the United States and C$11.3 million for Canada (Congressional Research Service, R40449, updated Jan 2017). The annex’s EUR 160 million therefore looks like a different measure, or a rounded figure. It is not the authorised level.
• EU response (2003). Directive 2003/74/EC made the ban on oestradiol-17β permanent and the bans on five other hormones provisional. It relied on the opinions of the EU’s Scientific Committee on Veterinary Measures relating to Public Health (1999, 2000 and 2002).
• EFSA (2007). EFSA concluded that newer data “do not call for a revision of previous risk assessments” (EFSA opinion, 12 June 2007; cited by CRS).
• Scientific disagreement persisted. The UK Veterinary Products Committee (1999, 2006), JECFA and Codex continued to judge the residues from approved use safe. Codex decided that maximum residue limits were not needed for the three natural hormones and adopted limits for the synthetic ones (CRS R40449).
• Follow-up dispute (2008). The Appellate Body’s October 2008 ruling was mixed. It found that the panel’s consultation of experts had compromised the EU’s due-process rights, and it could not complete the analysis of whether the 2003 Directive brought the EU into compliance. Both the EU measures and the retaliation stayed in place (WTO DS320 summary).

How it was settled: trade compensation, not scientific agreement.

• A 2009 US–EU memorandum of understanding (revised in 2013) opened a duty-free EU quota for hormone-free “high-quality beef” in exchange for lower US retaliation (WTO DS26 summary).
• In December 2016 the US moved to reinstate its retaliatory tariffs (CRS; 81 FR 95724).
• The 2 August 2019 agreement set aside a US-only share of the 45,000 t quota, rising from 18,500 t to 35,000 t over seven years (USTR press release, 2 Aug 2019).
• The EU ban is still in force.

Verdict: contested#

The annex describes the case fairly as precaution exercised against the advice of expert committees. Its framing, however, implies that precaution was vindicated, and the post-2001 record does not settle that. The international risk-assessment bodies (JECFA, Codex) and some national committees still regard residues from approved use as safe. The EU’s scientific bodies (the Scientific Committee on Veterinary Measures relating to Public Health, then EFSA) maintain that no threshold can be defined for some of these hormones, and no decisive new evidence has emerged either way. The WTO found the original ban non-compliant because it was not based on a risk assessment. The dispute ended with a market-access arrangement rather than scientific convergence.

Implications for weight#

• This case is the main counterweight in the annex. It supports the digest’s insight 16: acting against expert advice carries costs to legitimacy and to trade (p. 715).
• It does not support a general claim that precautionary action is later vindicated. Hindsight has neither confirmed nor refuted the underlying risk.
• For the lens, the useful lessons are procedural:
• Precaution taken without a risk assessment done openly on its own terms is legally fragile.
• Withholding an expert committee’s results (the Lamming Committee, p. 715) damages credibility.
• Disputes about risk can end up settled by trade deals that leave the scientific disagreement where it was.

Claim 7 — EU bans on antimicrobial growth promoters (1998–99) survived an industry legal challenge; Sweden banned them in 1985 at farmers’ request (p. 710)#

Original claim (p. 710).

• The EU banned four antimicrobials in feed in 1998 as a “precautionary” measure.
• Industry challenged the ban in court, and the annex’s note says “EU won the case against Pfizer Animal Health”.
• Sweden banned growth promoters in 1985 after farmers asked it to.

Subsequent developments#

The court case. The Court of First Instance dismissed Pfizer’s action on 11 September 2002 in Case T-13/99, about virginiamycin. The annex’s 2001 text had expected judgment in late 2001. The judgment became a key statement of the EU precautionary principle:

“where there is scientific uncertainty as to the existence or extent of risks to human health, the Community institutions may… take protective measures without having to wait until the reality and seriousness of those risks become fully apparent”

The judgment also held that a measure “cannot properly be based on a purely hypothetical approach to the risk” (judgment, EUR-Lex 61999TJ0013, paras 139–152).

The judgment records that the Council acted against the opinion of its own scientific committee. The Scientific Committee for Animal Nutrition (SCAN) had concluded on 10 July 1998 that using virginiamycin “does not constitute an immediate risk to public health in Denmark” (para 53). This parallels the hormones case (claim 6), with the opposite legal result.

The full EU ban and later extensions.

• The last four permitted growth-promoting antibiotics were removed on 1 January 2006 under Regulation 1831/2003 (Commission press release IP/05/1687, 22 Dec 2005).
• The Veterinary Medicinal Products Regulation (EU) 2019/6 then banned antimicrobials for growth promotion and restricted preventive group use.
• From 3 September 2026, animals and animal products imported into the EU must not have been treated with antimicrobials for growth promotion or with antimicrobials reserved for humans (Delegated Regulation (EU) 2023/905, 27 Feb 2023; Implementing Regulation (EU) 2024/2598, 4 Oct 2024). Whether this has been enforced in practice could not be assessed only weeks after it began to apply.

Other countries followed.

• WHO’s 2017 guidelines recommend stopping routine use for growth promotion and disease prevention in healthy animals (WHO, 7 Nov 2017).
• FDA’s Guidance for Industry #213 (final December 2013) led to the withdrawal of growth-promotion claims for medically important antimicrobials in the US, a process completed in January 2017 (FDA GFI #213).
• China withdrew colistin as a growth promoter in 2017.

Evidence on resistance. On balance the evidence is supportive, but the effect on human health is less certain than the effect on animals.

• In Denmark, glycopeptide-resistant E. faecium in broilers fell from 72.7% in 1995 to 5.8% in 2000 after the avoparcin ban (Aarestrup et al., AAC, July 2001).
• A WHO-commissioned meta-analysis found that restricting antibiotic use lowered resistance in animals by “commonly… 10 and 15%” in absolute terms, and was associated with 24% lower resistance prevalence in humans, mainly those in direct contact with animals. “The implications for the general human population are less clear” (Tang et al., Lancet Planet Health, Nov 2017).
• A follow-up review linked the avoparcin ban to less of the vanA resistance gene in samples from both animals and humans (Nobrega et al., JAC, Feb 2021).
• After China’s colistin withdrawal, resistance and mcr-1 carriage fell in both animals and humans. For example, human carriage fell from 14.3% (2016) to 6.3% (2019) (Wang et al., Lancet Infect Dis, online 4 June 2020).
• Critics’ view. An early critique argued that the bans had worsened animal health (post-weaning diarrhoea in pigs, necrotic enteritis in broilers) and raised the use of therapeutic antibiotics that also matter in human medicine. It found the only attributable human effect to be less resistance in enterococci carried in the gut (Casewell et al., JAC, online 1 July 2003).
• Later sales data. Veterinary antimicrobial sales in the 25 countries reporting continuously fell by 53.0% between 2011 and 2022 (from 161.2 to 75.8 mg per population correction unit, mg/PCU). The EU-27 had achieved just over half of the Farm to Fork target of a 50% cut by 2030 (ESVAC, 13th and final report, 20 Nov 2023). This suggests that the early rise in therapeutic use, where it happened, was not permanent at EU level.

Verdict: strengthened#

• The legal outcome the annex expected came in 2002.
• Policy went well beyond the 1998–99 bans: a full EU ban in 2006, an extension to imports in 2026, and adoption by WHO, the US and China.
• Resistance in animals fell measurably after the bans.
• The size of the benefit to human health is still less certain, and there were transitional costs to animal health. Both points qualify the annex’s framing of “resistance to common sense”.

Implications for weight#

• This case supports two lessons: precaution can be legally robust when the underlying risk is “adequately backed up by the scientific data” (insight 5), and warnings often come from outside expert circles (insight 11). Swedish farmers asking for their own ban (p. 710) is an unusually clear example.
• The judgment’s formula is a durable way to state the standard of proof:
• no need to wait for full proof, but
• no action on “purely hypothetical” risk.
• The lens should also carry the qualification that the benefits of such bans can be smaller and slower than advocates expect, while their costs fall on particular groups (here, the health of livestock).

Claim 8 — TBT: global phase-out timetable (no new application from 2003, all coatings replaced by 2008); OSPAR aim of ending organotin releases by 2020 (p. 714)#

Original claim (p. 714).

• The 2001 convention text on harmful anti-fouling systems would bring a worldwide prohibition on new application of organotin antifoulants in 2003.
• Existing organotin coatings would be replaced on all vessels by 2008.
• OSPAR aimed to end all releases of organotins to the marine environment by 2020.

Subsequent developments#

Treaty timing.

• The International Convention on the Control of Harmful Anti-fouling Systems on Ships was adopted on 5 October 2001 but entered into force only on 17 September 2008 (IMO AFS Convention page).
• The 2003 date for prohibiting new application was therefore not globally binding. The EU enforced it for ships flying EU flags from 1 July 2003, and barred ships with active TBT hulls from EU ports from 1 January 2008 (Regulation (EC) No 782/2003, 14 April 2003).
• OSPAR itself notes that the convention came into force “26 years” after France’s 1982 restriction.
• The convention’s mechanism for adding new substances has worked: controls on the biocide cybutryne were adopted in 2021 and took effect on 1 January 2023 (IMO).

Ecological recovery.

• OSPAR’s QSR 2023 imposex assessment finds that the share of monitoring sites above the environmental assessment criterion (EAC, the level below which no chronic effects are expected) fell from 81% in 2008 to about 21–22% in 2020. Sites near background levels rose from 0.5% to about 22%.
• Imposex levels “decreased significantly in all subregions” with enough data.
• Two sub-regions (Skagerrak and Kattegat, and the Iberian Sea) remain above the EAC. About three-quarters of sites still show more imposex than normal.
• OSPAR also notes that “many local TBT hotspots… are not routinely monitored”, that “TBT is still manufactured in the USA and is available on the market”, and that the replacement copper-based paints need scrutiny.

(OSPAR QSR 2023, Status and trends in the levels of imposex in marine gastropods) - In English Channel estuaries, recovery of dogwhelk imposex “is evident at many sites but, near ports, is prolonged by shipping impacts, including sediment legacy” (Langston et al., Mar Pollut Bull, 2015). - Taken together: releases from new applications have largely stopped in the OSPAR area, but OSPAR’s 2020 aim of concentrations “close to zero” for synthetic substances has not been met everywhere.

Verdict: held up (with timetable slippage)#

The phase-out took place, and by 2008 the hulls were due to be clear, as planned. The ban has been followed by one of the clearest documented ecological recoveries among the vol. 1 cases. The 2003 global application ban slipped to 2008 as a binding rule, and the legacy in sediments and hotspots remains.

Implications for weight#

• TBT strongly supports insight 15: action works, but recovery is slow and uneven.
• It supports insight 12: long-term monitoring (here, imposex surveys) made both the warning and the proof of recovery possible.
• National or regional action led global action by decades (France 1982, the UK 1987, the EU 2003, the world 2008). This is a recurring pattern in the annex, and it held up here.
• OSPAR’s note on the replacement paints repeats the substitution caution in insight 8.

Claim 9 — Fisheries: 1995–2008 “responsible” management “which is not very effective”; “Positive changes are occurring, but are they happening fast enough to avoid further collapses?” (pp. 702–703)#

Original claim (pp. 702–703). Table A2.1 describes 1995–2008 management as “‘responsible’ management (which is not very effective)”. The 2001 chronology ends with a question: are positive changes “happening fast enough to avoid further collapses?”

Subsequent developments#

EU reform and its deadlines.

• The 2013 reform of the Common Fisheries Policy (CFP) set a legal deadline: the fishing rate that gives maximum sustainable yield (MSY) was to be reached “by 2015 where possible and… at the latest by 2020 for all stocks” (Regulation (EU) No 1380/2013, 11 Dec 2013, Art. 2(2)).
• The deadline was missed. The European Court of Auditors found that in the Atlantic “the majority of fish stocks were fished sustainably. However many stocks were still overfished”, while in the Mediterranean “fishing rates were at twice sustainable levels” (ECA Special Report 26/2020, Nov 2020).
• The Commission’s 2023 assessment reported “particularly good results in the Northeast Atlantic”. It found the Mediterranean situation “still concerning and improvements too slow”, and said the Baltic “saw a reversal of recent improvements”. It also reported that the landing obligation (the ban on discarding catch) had not been effectively controlled or enforced (COM(2023) 103, 21 Feb 2023).

Further collapses.

• Eastern Baltic cod was closed by emergency measures in July 2019. The measures cited an “unsustainably low biomass” driven by declining recruitment, high natural mortality and excessive fishing mortality (Commission Implementing Regulation (EU) 2019/1248, 22 July 2019).
• Pacific sardine, the annex’s Californian example (collapse in 1942, recovery from the mid-1980s), collapsed again. The fishery has been closed since 2015, the stock is classed as overfished, and the rebuilding target date is 2035 (NOAA Fisheries, Pacific sardine, accessed Sept 2026).

Northern cod.

• In 2024 Canada ended the 1992 commercial moratorium on northern cod and set a total allowable catch of 18,000 t.
• The decision followed a revised stock-assessment model. The model added inshore and juvenile survey data back to 1954, concluded that historical productivity had been lower than thought, and so lowered the population size at which the stock counts as recovered. The stock moved from the “critical” to the “cautious” zone.
• Oceana Canada called the decision “short-sighted”. The inshore fishers’ union (FFAW-Unifor) asked for it to be reversed.
• Memorial University scientists estimated a 62–76% risk that the stock would decline over 2024–2027.

(All from a secondary source: Responsible Seafood Advocate, 8 July 2024. The DFO news release could not be retrieved.)

Global picture.

• FAO reports that the share of marine stocks fished within biologically sustainable levels fell to 62.3% in 2021, 2.3 points lower than in 2019 (FAO, SOFIA 2024 news release, 7 June 2024).
• The WTO Agreement on Fisheries Subsidies entered into force on 15 September 2025 (WTO). This is a first binding global constraint on one of the drivers of overfishing.

Verdict: held up#

The annex’s scepticism was justified. Management improved where catch limits were set on scientific advice and enforced (the North-East Atlantic), but further collapses followed (Eastern Baltic cod, Pacific sardine). The EU’s own 2020 deadline was missed. Northern cod shows how a recovery can come to be declared partly by redefining the reference points against which it is measured.

Implications for weight#

• The case supports insight 13: precaution can be claimed without being practised (pp. 702–703). Legal targets with deadlines (CFP 2013) did not guarantee results.
• It supports insight 7: adopting a rule is not the same as reducing risk. The landing-obligation enforcement gap is a clear example.
• It adds a lesson the annex does not state. How reference points and models are defined can change what counts as “recovered”. Where the thresholds that trigger protection are set by modelling choices, those choices need the same scrutiny as the data.
• Environmental change (Baltic eutrophication, climate) now interacts with fishing pressure. The Commission itself says “fisheries management measures alone are not enough” (COM(2023) 103). This fits the annex’s emphasis on complexity.

Claim 10 — “Similar concerns are now being expressed at the frequent over-exposure to radiation from CT scans”; NRPB found 20% of X-rays clinically unhelpful and 50% of collective dose avoidable (p. 704)#

Original claim (p. 704). The radiation case ends by noting that concerns like those Alice Stewart raised about prenatal X-rays “are now being expressed at the frequent over-exposure to radiation from CT scans”. Its chronology cites UK National Radiological Protection Board (NRPB) findings (1990–97): 20% of medical X-rays probably clinically unhelpful, 50% of collective patient dose avoidable, and 100-fold variation in dose between hospitals.

Subsequent developments#

Direct epidemiological evidence has appeared since 2012. Earlier concern rested on risks projected from other exposures.

• UK cohort. Among children and young adults scanned on NHS CT, cumulative doses of about 50 mGy “might almost triple the risk of leukaemia”, and doses of about 60 mGy “might triple the risk of brain cancer”. The absolute risks are small: about one excess leukaemia and one excess brain tumour per 10,000 head CTs in children under 10, within 10 years (Pearce et al., Lancet, online 7 June 2012). Letters challenged the study on grounds of confounding by indication and reverse causation.
• EPI-CT (a nine-country European cohort of 948,174 people scanned before age 22).
• It found a significant linear dose–response for brain cancer (excess relative risk per 100 mGy 1.27, 95% CI 0.51–2.69). This held when follow-up was delayed and when people with possibly unreported earlier cancers were excluded (Hauptmann et al., Lancet Oncol, online 6 Dec 2022).
• For all haematological malignancies, the excess relative risk was 1.96 per 100 mGy (95% CI 1.10–3.12). That means “for every 10,000 children examined today… 1–2 persons are expected to develop a haematological malignancy attributable to radiation exposure in the subsequent 12 years” (Bosch de Basea Gomez et al., Nat Med, online 9 Nov 2023; author correction 2025).
• US projection. An estimated 93 million CT examinations in the US in 2023 were projected to cause about 103,000 future cancers (90% uncertainty limits 96,400–109,500). “If current practices persist, CT-associated cancer could eventually account for 5% of all new cancer diagnoses annually” (Smith-Bindman et al., JAMA Intern Med, online April 2025, issue June 2025).
• This estimate is modelled, using the BEIR VII risk models, not observed.
• The lead author co-founded a company that collects CT dose data (a disclosed interest).
• An accompanying commentary stressed the need to balance these risks against the benefits of CT (Richman & Katz, JAMA Intern Med, 2025).
• Radiology response. Radiology leaders called for moving “from knowing to acting” on justification and dose optimisation (Frush et al., Pediatr Radiol, online 16 May 2024).

Regulation. The EU’s revised Basic Safety Standards Directive notes that “important technological and scientific developments have led to a notable increase in the exposure of patients”. It requires justification of medical exposures and the use of diagnostic reference levels, dose benchmarks for common examinations (Council Directive 2013/59/Euratom, 5 Dec 2013; transposition deadline 6 February 2018).

The low-dose model is contested. In the other direction, a US executive order of 23 May 2025 directs the Nuclear Regulatory Commission to “reconsider reliance on the linear no-threshold (LNT) model… and the ‘as low as reasonably achievable’ standard”, calling them “flawed” (White House, Ordering the Reform of the Nuclear Regulatory Commission, 23 May 2025). The order concerns nuclear regulation, not medical imaging. It shows that the assumption underlying the radiation-protection principles the annex celebrates (the ICRP’s 1949 conclusion of “no dose threshold”, p. 704) remains politically contested, even as the EPI-CT results give it direct support at CT doses.

NRPB audit figures. The 1990–97 figures were not re-audited in this check. EPI-CT’s call for “continued justification… and optimization of doses” implies that unnecessary or excessive exposure is still seen as a problem.

Verdict: strengthened#

The annex’s one-line concern was ahead of the direct evidence. Large cohort studies since 2012, above all EPI-CT, now show dose–response relationships for cancer at CT doses in young people. The absolute individual risks are small, the population-level projections depend on models, and the confounding critiques have been addressed in part but not fully.

Implications for weight#

• This case supports insight 9 (scrutinise claimed benefits, not just risks; p. 704), because the case for justifying each scan rests on its clinical benefit.
• It supports insight 4 (risk estimates tend to rise and “safe” thresholds recede).
• It also shows a recurring dynamic: a technology whose benefits are real and whose per-use risk is tiny can still add up to a large population burden as use grows. The key variables are how often it is used and how exposure varies between providers. The 100-fold variation the NRPB found (p. 704) is a variable of that kind.
• The political challenge to the low-dose model in 2025 is a reminder that the scientific basis for precaution can be reopened long after it seems settled.

Cross-cutting implications for the annex’s lessons (technology-neutral)#

The mechanism-level lessons in the digest are restated below in the light of hindsight. Page references are to the annex.

1. The lag from warning to protection is long, and “action” is a process (pp. 702, 705–706, 714). This is strengthened. Several cases saw major new protective measures years or decades after the table’s “effective action” date: asbestos (2023–24, against a table date of 1999), benzene (2022–26, against 1978), antimicrobial growth promoters (2006 and 2026, against 1999), PCBs (Stockholm deadlines of 2025/2028 at risk of being missed, against a table date of “phase out by 2010”) and TBT (binding worldwide only from 2008). Headline lag numbers should be treated as illustrative and as depending on how “warning” and “action” are defined.
2. The standard of proof sets the pace (pp. 705, 710, 715). This is strengthened, with a legal refinement. The US “significant risk” doctrine (1980) kept the benzene limit at 1 ppm, while the EU cut to 0.2 ppm. EU courts (Pfizer, 2002) upheld precaution where risk is “adequately backed up” but not where it is “purely hypothetical”. The WTO required a risk assessment (hormones, 1998). Who sets the standard (national regulator, regional court, trade tribunal) matters as much as the standard itself.
3. Substitutes can carry hazards that could have been foreseen (pp. 708, 712). This is strengthened and recurrent. CFCs were replaced by HCFCs and HFCs, which needed Kigali; HFCs are being replaced by HFOs, which produce persistent TFA. MTBE was adopted despite 1950s–60s evidence that it persists in water. OSPAR asks whether the replacements for TBT are sustainable.
4. Harm migrates (digest mechanism). This is strengthened. MTBE production moved to exports after the US phase-out. Asbestos use continued globally at about 1.2 Mt in 2021. TBT is still manufactured and available in places that have not ratified the convention.
5. Adopting a rule is not the same as reducing risk (pp. 706, 708, 710, 713, 715, 716). This is strengthened. Examples: unreported CFC-11 production after the ban; the CFP 2020 MSY deadline missed; the landing obligation not enforced; the Stockholm PCB deadlines at risk. The EU’s 2026 import conditions for antimicrobials are the latest attempt to enforce a domestic rule on products made elsewhere.
6. Action works, but recovery is slow (pp. 702, 706, 708, 713, 714). This is strengthened. Ozone recovers around 2040–2066. Mesothelioma in Great Britain is only now declining, 25 years after the ban. Imposex fell sharply but with local hotspots. Great Lakes AOCs are still being delisted one at a time.
7. Risk estimates rise and “safe” thresholds recede (pp. 704–705). This is strengthened for benzene and for low-dose radiation. It is not supported for hormone residues, where no new decisive evidence emerged.
8. Acting against expert advice has costs to legitimacy (p. 715). This holds. The hormones case remains contested after 25 years. The antimicrobial case (acting against SCAN in 1998) shows that such action can nonetheless be upheld in court and followed by broader adoption. The difference seems to lie in whether the underlying risk hypothesis was later supported: yes for resistance, still disputed for hormone residues.
9. Models and reference points shape what counts as harm or recovery. This is a new emphasis from hindsight (northern cod, 2024; the modelled CT projections, 2025). The annex attributes delays to missing measurement units and weak analytical methods (pp. 704, 714). The later cases add that the definitions and models chosen to interpret data can move decisions as much as the data do.

Overall weight. Taken as a whole, the annex’s case recaps are more reliable as qualitative history than as quantitative measurement. Its core lessons hold up and in most cases are strengthened by post-2013 events: long lags, slow recovery, regrettable substitution, and rules that do not by themselves reduce risk. The lessons about precaution against expert advice deserve more caution. Hormones remain contested, and the annex’s framing tends to present precautionary action as vindicated where later evidence has not settled the matter.

Sources#

Section source and project files#

• EEA (2013), Late lessons from early warnings: science, precaution, innovation, EEA Report No 1/2013, Annex 2, pp. 701–716 (project extract working/text/chunks/LL2-A2.txt; digest working/late-lessons/digests/LL2-A2.md).
• Gee, D. (2009), “Late Lessons from Early Warnings: Towards realism and precaution with EMF?”, Pathophysiology 16:217–231, online 21 May 2009. https://doi.org/10.1016/j.pathophys.2009.01.004 (PubMed https://pubmed.ncbi.nlm.nih.gov/19467848/)

Asbestos and PCBs (claims 1–2)#

• Peto, J. et al. (1999), “The European mesothelioma epidemic”, Br J Cancer 79:666–72, Feb 1999. https://doi.org/10.1038/sj.bjc.6690105
• Alpert, N., van Gerwen, M., Taioli, E. (2020), “Epidemiology of mesothelioma in the 21st century in Europe and the United States, 40 years after restricted/banned asbestos use”, Transl Lung Cancer Res 9(Suppl 1):S28–S38, Feb 2020. https://doi.org/10.21037/tlcr.2019.11.11
• Health and Safety Executive (2026), Mesothelioma statistics for Great Britain, 2026, July 2026. https://www.hse.gov.uk/statistics/assets/docs/mesothelioma.pdf
• European Commission (2022), Communication on working towards an asbestos-free future, COM(2022) 488 final, 28 Sept 2022. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52022DC0488
• Directive (EU) 2023/2668 amending Directive 2009/148/EC (asbestos at work), 22 Nov 2023, OJ L 30.11.2023. https://eur-lex.europa.eu/eli/dir/2023/2668/oj
• US EPA (2024), “Asbestos Part 1; Chrysotile Asbestos; Regulation of Certain Conditions of Use Under TSCA”, final rule, Federal Register 2024-05972, published 28 March 2024, effective 28 May 2024. https://www.federalregister.gov/documents/2024/03/28/2024-05972/asbestos-part-1-chrysotile-asbestos-regulation-of-certain-conditions-of-use-under-the-toxic
• US EPA, “Risk Management for Asbestos Part 1: Chrysotile Asbestos”, last updated 5 May 2026. https://www.epa.gov/assessing-and-managing-chemicals-under-tsca/risk-management-asbestos-part-1-chrysotile-asbestos
• Stockholm Convention Secretariat, “PCB overview” and “Progress towards PCB elimination” (decision SC-11/3; data as of 2020), accessed Sept 2026. https://chm.pops.int/Implementation/IndustrialPOPs/PCBs/Overview/tabid/273/Default.aspx ; https://chm.pops.int/Implementation/IndustrialPOPs/PCB/ProgresstowardsPCBelimination/tabid/8304/Default.aspx
• OSPAR Commission (2023), Quality Status Report 2023, thematic assessment “Hazardous substances”. https://oap.ospar.org/en/ospar-assessments/quality-status-reports/qsr-2023/thematic-assessments/hazardous-substances/
• US EPA, “Great Lakes Areas of Concern” and “List of Great Lakes AOCs”, accessed Sept 2026. https://www.epa.gov/great-lakes-aocs ; https://www.epa.gov/great-lakes-aocs/list-great-lakes-aocs

Benzene (claim 3)#

• Industrial Union Dept., AFL-CIO v. American Petroleum Institute, 448 U.S. 607 (decided 2 July 1980). https://www.law.cornell.edu/supremecourt/text/448/607
• OSHA, 29 CFR 1910.1028 Benzene, accessed Sept 2026. https://www.osha.gov/laws-regs/regulations/standardnumber/1910/1910.1028
• Lan, Q. et al. (2004), “Hematotoxicity in workers exposed to low levels of benzene”, Science 306:1774–6, 3 Dec 2004. https://doi.org/10.1126/science.1102443
• Schnatter, A.R. et al. (2012), “Myelodysplastic syndrome and benzene exposure among petroleum workers: an international pooled analysis”, JNCI 104:1724–37, online 30 Oct 2012. https://doi.org/10.1093/jnci/djs411
• Loomis, D. et al. (2017), “Carcinogenicity of benzene”, Lancet Oncol 18:1574–5, online 26 Oct 2017. https://doi.org/10.1016/S1470-2045(17)30832-X
• IARC (2018), IARC Monographs vol. 120, Benzene. https://publications.iarc.who.int/576
• Rana, I. et al. (2021), “Benzene exposure and non-Hodgkin lymphoma: a systematic review and meta-analysis of human studies”, Lancet Planet Health 5:e633–43, online 25 Aug 2021. https://doi.org/10.1016/S2542-5196(21)00149-2
• Directive (EU) 2022/431 amending Directive 2004/37/EC (carcinogens and mutagens), 9 March 2022. https://eur-lex.europa.eu/eli/dir/2022/431/oj
• IARC (2025), “IARC Monographs evaluation of the carcinogenicity of automotive gasoline and some oxygenated gasoline additives”, news release, 21 March 2025. https://www.iarc.who.int/news-events/iarc-monographs-evaluation-of-the-carcinogenicity-of-automotive-gasoline-and-some-oxygenated-gasoline-additives/
• US FDA, “FDA alerts drug manufacturers to the risk of benzene contamination in certain drugs”, content current as of 24 July 2025. https://www.fda.gov/drugs/pharmaceutical-quality-resources/fda-alerts-drug-manufacturers-risk-benzene-contamination-certain-drugs
• OSHA (1987), Occupational Exposure to Benzene, final rule, 52 FR 34460, 11 Sept 1987. Cited but not retrieved.

Halocarbons (claim 4)#

• UN Treaty Collection, Amendment to the Montreal Protocol (Kigali, 15 October 2016), status accessed Sept 2026. https://treaties.un.org/Pages/ViewDetails.aspx?src=TREATY&mtdsg_no=XXVII-2-f&chapter=27&clang=_en
• WMO/UNEP (2023), “Ozone layer recovery is on track, helping avoid global warming by 0.5°C”, press release on the 2022 Scientific Assessment of Ozone Depletion, 9 Jan 2023. https://wmo.int/news/media-centre/ozone-layer-recovery-track-helping-avoid-global-warming-05degc
• Regulation (EU) 2024/573 on fluorinated greenhouse gases, 7 Feb 2024. https://eur-lex.europa.eu/eli/reg/2024/573/oj
• Montzka, S.A. et al. (2018), “An unexpected and persistent increase in global emissions of ozone-depleting CFC-11”, Nature 557:413–417, online 16 May 2018. https://doi.org/10.1038/s41586-018-0106-2
• Park, S. et al. (2021), “A decline in emissions of CFC-11 and related chemicals from eastern China”, Nature 590:433–437, online 10 Feb 2021. https://doi.org/10.1038/s41586-021-03277-w
• Umweltbundesamt, “TFA as degradation product of fluorinated greenhouse gases”, accessed Sept 2026. https://www.umweltbundesamt.de/en/topics/climate-energy/fluorinated-greenhouse-gases-fully-halogenated-cfcs/emissions-degradation-products/degradation-products-of-fluorinated-greenhouse
• EFSA (2026), “EFSA lowers safe level for exposure to TFA”, news, 22 July 2026. https://www.efsa.europa.eu/en/news/efsa-lowers-safe-level-exposure-tfa
• ECHA RAC-77 minutes (June 2026), https://echa.europa.eu/documents/d/guest/rac77_final_minutes_en. Reported in secondary sources; not accessed (HTTP 403).

MTBE (claim 5)#

• US EPA, “Methyl Tertiary Butyl Ether (MTBE): Gasoline” (archived; 2013 status update). https://archive.epa.gov/mtbe/web/html/gas.html
• US EIA (2018), “The United States continues to export MTBE, mainly to Mexico, Chile, and Venezuela”, Today in Energy, 13 July 2018. https://www.eia.gov/todayinenergy/detail.php?id=36614
• Environment News Service (2005), “Oil Companies Pay Santa Monica MTBE Cleanup Costs”, 17 Feb 2005 (archived; secondary source). http://web.archive.org/web/20101228032804/http://www.ens-newswire.com/ens/feb2005/2005-02-17-09.html
• State of New Hampshire v. Exxon Mobil Corp., 168 N.H. 211, 126 A.3d 266 (N.H. Supreme Court, 2 Oct 2015). https://www.courtlistener.com/opinion/3006720/state-of-new-hampshire-v-exxon-mobil-corporation-a/
• Rhode Island v. Atlantic Richfield Co. (2018). https://www.courtlistener.com/opinion/8344333/rhode-island-v-atl-richfield-co/
• Directive 2009/30/EC amending Directive 98/70/EC (fuel quality), 23 April 2009, Annex I. https://eur-lex.europa.eu/eli/dir/2009/30/oj
• IARC (2025), news release on gasoline and oxygenated additives, 21 March 2025 (see Benzene above).

Hormones (claim 6)#

• WTO, DS26 EC — Measures Concerning Meat and Meat Products (Hormones), case summary. https://www.wto.org/english/tratop_e/dispu_e/cases_e/ds26_e.htm ; one-page summary: https://www.wto.org/english/tratop_e/dispu_e/cases_e/1pagesum_e/ds26sum_e.pdf
• WTO, DS320 US — Continued Suspension of Obligations in the EC — Hormones Dispute, case summary (AB report 16 Oct 2008). https://www.wto.org/english/tratop_e/dispu_e/cases_e/ds320_e.htm
• Johnson, R. / Congressional Research Service, The U.S.-EU Beef Hormone Dispute, R40449 (updated Jan 2017). https://www.everycrsreport.com/reports/R40449.html
• USTR (2019), “United States and European Union Sign Breakthrough Agreement on U.S. Beef Access to EU”, 2 Aug 2019. https://ustr.gov/about-us/policy-offices/press-office/press-releases/2019/august/united-states-and-european-union
• EFSA (2007), Opinion of the CONTAM Panel related to hormone residues in bovine meat and meat products, EFSA Journal 510, 12 June 2007. https://doi.org/10.2903/j.efsa.2007.510 (conclusion as reported by CRS R40449; full text not retrieved)

Antimicrobial growth promoters (claim 7)#

• Judgment of the Court of First Instance, Case T-13/99 Pfizer Animal Health SA v Council, 11 Sept 2002. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:61999TJ0013
• European Commission (2005), “Ban on antibiotics as growth promoters in animal feed enters into effect”, IP/05/1687, 22 Dec 2005. https://ec.europa.eu/commission/presscorner/detail/en/IP_05_1687
• Commission Delegated Regulation (EU) 2023/905, 27 Feb 2023. https://eur-lex.europa.eu/eli/reg_del/2023/905/oj
• Commission Implementing Regulation (EU) 2024/2598, 4 Oct 2024 (applies from 3 Sept 2026). https://eur-lex.europa.eu/eli/reg_impl/2024/2598/oj
• WHO (2017), WHO guidelines on use of medically important antimicrobials in food-producing animals, 7 Nov 2017. https://www.who.int/publications/i/item/9789241550130
• US FDA, CVM Guidance for Industry #213 (final, December 2013). https://www.fda.gov/regulatory-information/search-fda-guidance-documents/cvm-gfi-213-new-animal-drugs-and-new-animal-drug-combination-products-administered-or-medicated-feed
• Aarestrup, F.M. et al. (2001), Antimicrob Agents Chemother 45:2054–9, July 2001. https://doi.org/10.1128/AAC.45.7.2054-2059.2001
• Casewell, M. et al. (2003), “The European ban on growth-promoting antibiotics and emerging consequences for human and animal health”, J Antimicrob Chemother 52:159–61, online 1 July 2003. https://doi.org/10.1093/jac/dkg313
• Tang, K.L. et al. (2017), Lancet Planet Health 1:e316–27, Nov 2017. https://doi.org/10.1016/S2542-5196(17)30141-9
• Nobrega, D.B. et al. (2021), J Antimicrob Chemother 76:561–75, Feb 2021. https://doi.org/10.1093/jac/dkaa443
• Wang, Y. et al. (2020), Lancet Infect Dis 20:1161–71, online 4 June 2020. https://doi.org/10.1016/S1473-3099(20)30149-3
• EMA (2023), Sales of veterinary antimicrobial agents in 31 European countries in 2022 (13th and final ESVAC report), 20 Nov 2023. https://www.ema.europa.eu/en/documents/report/sales-veterinary-antimicrobial-agents-31-european-countries-2022-trends-2010-2022-thirteenth-esvac-report_en.pdf

TBT (claim 8)#

• IMO, International Convention on the Control of Harmful Anti-fouling Systems on Ships (AFS), adopted 5 Oct 2001, in force 17 Sept 2008; cybutryne amendments in force 1 Jan 2023. https://www.imo.org/en/about/Conventions/Pages/International-Convention-on-the-Control-of-Harmful-Anti-fouling-Systems-on-Ships-(AFS).aspx
• Regulation (EC) No 782/2003 on the prohibition of organotin compounds on ships, 14 April 2003. https://eur-lex.europa.eu/eli/reg/2003/782/oj
• OSPAR Commission (2023), QSR 2023 indicator assessment, “Status and Trends in the Levels of Imposex in Marine Gastropods (TBT in Shellfish)”. https://oap.ospar.org/en/ospar-assessments/quality-status-reports/qsr-2023/indicator-assessments/tbt-shellfish/
• Langston, W.J. et al. (2015), “Recovery from TBT pollution in English Channel environments: A problem solved?”, Mar Pollut Bull 95:551–64, online 27 Feb 2015. https://doi.org/10.1016/j.marpolbul.2014.12.011

Fisheries (claim 9)#

• Regulation (EU) No 1380/2013 on the Common Fisheries Policy, 11 Dec 2013. https://eur-lex.europa.eu/eli/reg/2013/1380/oj
• European Court of Auditors (2020), Special Report 26/2020, Marine environment: EU protection is wide but not deep, Nov 2020. https://www.eca.europa.eu/Lists/ECADocuments/SR20_26/SR_Marine_environment_EN.pdf
• European Commission (2023), The common fisheries policy today and tomorrow: a Fisheries and Oceans Pact, COM(2023) 103 final, 21 Feb 2023. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52023DC0103
• Commission Implementing Regulation (EU) 2019/1248 (Eastern Baltic cod emergency measures), 22 July 2019. https://eur-lex.europa.eu/eli/reg_impl/2019/1248/oj
• NOAA Fisheries, “Pacific Sardine” species page, accessed Sept 2026. https://www.fisheries.noaa.gov/species/pacific-sardine
• Responsible Seafood Advocate / Global Seafood Alliance (2024), “Canada lifts 30-year cod fishing ban off Newfoundland and Labrador to mixed reactions”, 8 July 2024 (secondary source). https://www.globalseafood.org/advocate/canada-lifts-30-year-cod-fishing-ban-off-newfoundland-and-labrador-to-mixed-reactions/
• FAO (2024), “FAO Report: Global fisheries and aquaculture production reaches a new record high” (SOFIA 2024), 7 June 2024. https://www.fao.org/newsroom/detail/fao-report-global-fisheries-and-aquaculture-production-reaches-a-new-record-high/en
• WTO, Agreement on Fisheries Subsidies (entry into force 15 Sept 2025). https://www.wto.org/english/tratop_e/rulesneg_e/fish_e/fish_e.htm

Radiation and CT (claim 10)#

• Pearce, M.S. et al. (2012), “Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours”, Lancet 380:499–505, online 7 June 2012. https://doi.org/10.1016/S0140-6736(12)60815-0
• Hauptmann, M. et al. (2023), “Brain cancer after radiation exposure from CT examinations of children and young adults: results from the EPI-CT cohort study”, Lancet Oncol 24:45–53, online 6 Dec 2022. https://doi.org/10.1016/S1470-2045(22)00655-6
• Bosch de Basea Gomez, M. et al. (2023), “Risk of hematological malignancies from CT radiation exposure in children, adolescents and young adults”, Nat Med 29:3111–3119, online 9 Nov 2023 (author correction 2025). https://doi.org/10.1038/s41591-023-02620-0
• Smith-Bindman, R. et al. (2025), “Projected Lifetime Cancer Risks From Current Computed Tomography Imaging”, JAMA Intern Med 185:710–719 (online April 2025; issue 1 June 2025). https://doi.org/10.1001/jamainternmed.2025.0505
• Richman, I.B., Katz, M.H. (2025), “Balancing Computed Tomography’s Benefits With Radiation Risks”, JAMA Intern Med 185:719. https://doi.org/10.1001/jamainternmed.2025.0514
• Frush, D.P. et al. (2024), “CT radiation exposure and cancer risk: from knowing to acting”, Pediatr Radiol 54:1407–9, online 16 May 2024. https://doi.org/10.1007/s00247-024-05949-x
• Council Directive 2013/59/Euratom (Basic Safety Standards), 5 Dec 2013. https://eur-lex.europa.eu/eli/dir/2013/59/oj
• The White House (2025), Executive Order “Ordering the Reform of the Nuclear Regulatory Commission”, 23 May 2025. https://www.whitehouse.gov/presidential-actions/2025/05/ordering-the-reform-of-the-nuclear-regulatory-commission/