LL1-05 hindsight check: Asbestos, from “magic” to malevolent mineral#

Late lessons from early warnings: the precautionary principle 1896–2000 (EEA Environmental Issue Report No 22, 2001), Chapter 5, pp. 52–63. Authors: David Gee and Morris Greenberg.

The check was prepared on 25 September 2026 and covers developments from 2001 to September 2026. All web sources were accessed on 25 September 2026. Page references in the form “p. NN” point to the 2001 report.

Overview#

Twenty-five years on, the chapter’s core case has held up. Where the claims were quantitative, the record is more mixed.

What has been vindicated or strengthened - Chrysotile. Every major authority now treats all asbestos types, chrysotile included, as carcinogenic with no identified safe threshold: IARC (2009/2012), WHO (fact sheet of August 2026), ECHA’s Risk Assessment Committee (2021), the EU legislature (2023) and the US EPA (2024). - Bans. Bans work, though with the multi-decade lag the chapter predicted. Italian modelling estimates that the 1992 ban avoided roughly 8,000–22,000 mesothelioma deaths by 2020. In Sweden, the cohorts who started work after asbestos use fell have about one-sixth of the pleural mesothelioma risk of earlier cohorts. - Trade law. The WTO asbestos ruling was described accurately and was later relied on. Canada, the complainant, banned asbestos itself in 2018, and the United States banned chrysotile in 2024. - Substitutes. The warning about substitutes of the same physical form (Lesson 10, p. 61) has been built into law. EU hazard classification now exempts man-made fibres that are shown to be non-biopersistent. Non-asbestos minerals with asbestos-like fibre form have been classified as human carcinogens. - Continued use. Use did continue “largely in developing countries” (pp. 52–53). About 930,000 tonnes were consumed in 2025, mainly for construction materials in Asia. The same pattern of weak surveillance and “reassuring” short-term studies can be seen there.

What was wrong or overstated - The British projection. The projections in Figure 5.1 (p. 52) overstated the height of the British mesothelioma peak. Male deaths peaked at 2,205 in 2016, against the roughly 2,700–3,300 read off the chart. The timing was about right. - The European total. The European total of “250,000–400,000” deaths (p. 52) was internally inconsistent. On every later method it looks too low for all asbestos diseases combined. How much too low is contested, because estimates of asbestos-related lung cancer vary by an order of magnitude between methods. - Some figures in the chapter are unreliable: - The Dutch “34,000 victims” estimate (p. 58) came from a Greenpeace-commissioned study built on projections that Dutch epidemiologists soon revised down by 44%. - The “USD 2 billion” in US settlements (p. 58) is far too low. RAND counted USD 70 billion spent on asbestos litigation by 2002. - “Grew to 2 million tonnes in 1998” (p. 52) misdescribes a market that had peaked at about 4.8 Mt around 1977. - Compensation (Lesson 8, pp. 60–61). Faster compensation schemes did spread: France, the Netherlands, the UK, Korea and elsewhere. But there is no evidence that they sharpened prevention incentives, which is the mechanism Lesson 8 relies on. A French Senate inquiry (2005) found the opposite: mutualised funding diluted employer accountability. - Error rebalancing (Lesson 6, p. 60). The “rebalance false positives and false negatives” argument remains a normative position. It has not been tested. As applied to asbestos, hindsight supports it strongly. As a general rule it is contested. The asbestos story itself has since produced a live risk–risk dispute: the UK Government refused a removal deadline on the grounds that removal can create exposure.

What the EEA said in 2013. The EEA’s own 2013 update (Late lessons from early warnings: science, precaution, innovation, EEA Report 1/2013, Annex 3, pp. 724–726, “Asbestos – the campaign for a global ban”, by Barry Castleman and David Gee) covered: - the continuing global trade; - Canada’s obstruction of the Rotterdam Convention and the closure of its mines (2011–2012); - a 2013 European Parliament call to remove asbestos from public buildings by 2028. That call was not enacted.

Chapter 24 of the same report (Panel 24.2, pp. 591–592, by Owen McIntyre) described how UK courts relaxed causation rules for mesothelioma. Both are noted under the relevant claims below. Several developments since 2013 go further than Annex 3, some in the section’s favour and some against it.

How to read the lessons now. The lessons about mechanisms are the most durable: - latency hides harm; - short-follow-up “negative” studies mislead; - survivors reassure falsely; - exposure studies miss the worst-exposed users; - producers shape the evidence; - externalised costs slow substitution.

Each is independently corroborated by later evidence. The chapter’s specific numbers should be cited only with the corrections below. Its prescriptive claims (Lessons 6 and 8) should be presented as arguments, not findings.

Claim-by-claim assessment#

Claim 1: About 250,000 EU mesothelioma deaths over 35 years; 250,000–400,000 total asbestos deaths (p. 52, citing Peto 1999)#

Original claim (p. 52). “Some 250 000 cases of mesothelioma … will occur in the European Union (EU) over the next 35 years (Peto, 1999)”, and the “total disease burden could be around 250 000–400 000 deaths”, including lung cancer and asbestosis. On p. 58 the upper figure is restated as “the estimated 400 000 European asbestos cancer deaths expected over the next few decades”.

Subsequent developments

What the source actually said. Peto et al. (1999) projected male deaths only, for Western Europe, over 1995–2029: “from 5000 in 1998 to about 9000 around 2018 … with a total of about a quarter of a million deaths over the next 35 years”. The projection was built on six countries and assumed a 1:1 ratio of mesothelioma to recorded pleural cancer outside Britain, where the ratio is 1.6:1 (Peto et al., Br J Cancer 1999). The chapter therefore turned a projection for men in Western Europe into one for “the EU”.

There is also an internal problem. If mesothelioma alone is 250,000, a total of 250,000–400,000 that includes lung cancer and asbestosis leaves the lower bound with zero lung-cancer and asbestosis deaths.

Later re-estimates of mesothelioma lowered national peaks: - Peto’s own group, 2004. It found that male pleural-cancer deaths in France, Germany and Italy had risen only modestly (about 8,750 in 1990–94 to 9,550 in 1995–99). It concluded that mesothelioma deaths “may be levelling off in most of Western Europe” (Pelucchi et al., Br J Cancer 2004). - Netherlands, 2003. Adding five years of data cut the predicted Dutch death toll by 44% (Segura et al., Occup Environ Med 2003). - Britain. The British peak estimate was revised down (see Claim 2).

Where the order of magnitude now stands. - The European Commission, using Global Burden of Disease (GBD) 2019 data, puts mesothelioma deaths attributable to occupational asbestos in the EU-27 at 7,510 in 2019 (COM(2022) 488, 28 Sept 2022, fn. 27). - Great Britain, which is no longer in the EU, recorded about 2,500 mesothelioma deaths a year through the 2010s (HSE, July 2026). - The WHO European Region reported 106,180 deaths from mesothelioma and asbestosis in 1994–2010, with heavy under-reporting in countries that had not banned asbestos (Kameda et al., Bull WHO 2014).

These figures are consistent with a Western European mesothelioma toll in the low hundreds of thousands over 35 years. The order of magnitude held, even though national peaks came in lower than first projected.

The “total burden” is where later estimates diverge. - The Commission (GBD method). It attributes 71,750 EU-27 deaths in 2019 to past occupational asbestos exposure, of which 61,035 are lung cancer. It describes this as “over 70 000 lives in the EU-27” (COM(2022) 488). - The British HSE (national epidemiology). It uses about one asbestos-related lung cancer per mesothelioma, and possibly as few as 0.55 for men born in the 1940s. It stresses that lung-cancer-to-mesothelioma ratios depend heavily on fibre type (HSE Asbestos-related disease statistics, July 2026).

On the GBD method the 35-year European total would be in the millions. On the HSE ratio it would be at least double the mesothelioma total, so over 500,000. Either way, 250,000–400,000 now looks too low for all asbestos diseases, but the size of the gap depends on a contested attribution method.

Verdict: partly held up. The mesothelioma order of magnitude and the approximate timing held. National peaks were over-projected. The total-burden range was internally inconsistent and is probably an underestimate.

Implication for weight. The mechanism, that harm peaks decades after use (p. 52), is strong. The specific numbers should be cited as “Peto’s 1999 projection for men in Western Europe”, not as an EU total. The lesson generalises in a technology-neutral way. Early cohort-model projections of long-latency harm are uncertain in height even when they are right about shape and timing. Total-burden figures are only as good as their attribution method, and on lung cancer the methods disagree by roughly eight-fold.

Claim 2: UK male mesothelioma deaths peaking 50–60 years after peak imports, at about 2,700 (about 2015) or about 3,300 (about 2020) (Figure 5.1, p. 52)#

Original claim (Fig. 5.1, p. 52; values read off the chart). Two projected curves, both sourced to Peto (1999): - men born before 1953, peaking at about 2,700 around 2015; - all men, assuming the risk to men born after 1953 is 50% of the 1943–48 cohort’s, peaking at about 3,300 around 2020.

The text says the import peak is “followed some 50–60 years later by the estimated peak of mesotheliomas” (p. 52).

Subsequent developments

The projection’s origin. Peto et al. (1995) projected “a peak of annual male mesothelioma deaths in about the year 2020 of between 2700 and 3300”. They also noted a scenario, which they called “extreme but arguable”, in which diagnostic trends would bring the peak down to 1,300 around 2010 (Peto et al., Lancet 1995).

The 2005 revision. Hodgson, McElvenny, Darnton, Price and Peto (2005) revised the peak to “around 1950 to 2450 deaths per year between 2011 and 2015” (Br J Cancer 2005).

What actually happened. HSE register data, published 1 July 2026 (Table MESO01; Mesothelioma statistics for Great Britain, 2026): - Male deaths in Great Britain peaked at 2,205 in 2016, after a plateau of about 2,100–2,200 from 2012 to 2020. - They fell to 1,771 in 2024 (provisional). - Total deaths (both sexes) peaked at 2,606 in 2016 and averaged 2,508 a year in 2011–2020. There were 2,146 deaths in 2024.

Against the chart, the actual male peak was about 18% below the “born before 1953” curve and about 33–36% below the “all men” curve. The 2020 figure of 2,108 compares with about 3,300 projected. The 2005 revised range was right.

Why the projection overshot. Peto (1995) found risk “increased from mid-1893 to mid-1948” by birth cohort. HSE’s 2026 cohort analysis finds that the 1935–39 male cohort has “among the highest male mesothelioma rates at any given age”, with successively lower rates for later cohorts.

HSE explains the earlier male decline as the result of the heaviest exposures, in industries such as shipbuilding, “being eliminated first”. Lung-burden measurements confirm that exposure fell rapidly in the 1970s (Gilham et al., Int J Epidemiol 2018). HSE now expects male deaths to fall through the 2020s. Female deaths are not expected to decline until the late 2020s.

The lag. UK imports in Figure 5.1 peaked roughly in the 1960s to early 1970s, and mortality peaked in the mid-2010s. A lag of about 45–55 years is consistent with the chapter’s 50–60-year description.

Verdict: partly held up. The timing and the lag between imports and deaths held. The peak height was overestimated by roughly a fifth to a third.

Implication for weight. This supports the chapter’s structural point that harm lags use by decades (pp. 52, 55, 59–60). It also shows an instructive irony: a projection made with incomplete cohort data mis-estimated where the highest-risk cohort would fall. That is the “latency lacuna” (p. 55) working in the other direction. The technology-neutral lesson is that projections of delayed harm made early in an epidemic are strongly sensitive to assumptions about exposure in cohorts not yet observed. Ranges and later revisions should travel with the numbers.

Claim 3: UK asbestos-related mesothelioma and lung cancer deaths “around 3 000 … and rising” (p. 58)#

Original claim (p. 58). “The annual UK cancer death rate from mesothelioma and lung cancer from asbestos is estimated by the Health and Safety Commission (Health and Safety Commission, 1994–95) to be around 3 000 deaths per year and rising.”

Subsequent developments. - Mesothelioma. Deaths in Great Britain rose from 1,317 in 1995 to a plateau of about 2,500 a year in the 2010s, peaking at 2,606 in 2016. They have fallen since 2021, to 2,146 in 2024 (HSE MESO01, 1 July 2026). - All asbestos-related disease. HSE now summarises the burden as “Around 5,000 asbestos-related disease deaths per year, including mesothelioma, lung cancer and asbestosis”. It estimates about as many asbestos-related lung cancers as mesotheliomas, while noting that the ratio is uncertain. In 2024, 503 death certificates mentioned asbestosis, excluding those that also mentioned mesothelioma (HSE, July 2026). - Where deaths now fall. HSE also notes that “over 70% of annual deaths … now occur in those aged over 75”, with deaths at younger ages falling (HSE, July 2026).

Verdict: held up. The “rising” trend continued for about two decades. The combined toll of mesothelioma and lung cancer rose well above 3,000 before plateauing and starting to decline.

Implication for weight. The claim supports the chapter’s point about harm lagging use (p. 52). The shift of deaths to people over 75 also bears on the “pensioners’ party fallacy” (Lesson 7, p. 60). A large share of the harm appears only after working life ends, where workplace health surveillance cannot see it. In technology-neutral terms, monitoring that stops at retirement or at the end of a programme systematically under-counts delayed harm.

Claim 4: Asbestos use continuing “largely in developing countries”; histories “being repeated” in Asia, Africa and South America (pp. 52–53)#

Original claim. “Asbestos use is continuing, now largely in developing countries” (p. 52). The histories “are now being repeated, albeit with some differences, in Asia, Africa and South America” (p. 53).

Subsequent developments

Global use (USGS Mineral Commodity Summaries, February 2026). World consumption was “an estimated 930,000 tons in 2025, a decrease of nearly 55% from approximately 2 million tons in 2000”. Demand “was expected to continue for the foreseeable future, particularly for cement pipe, roofing sheets, and other construction materials in Asia”. Production in 2025 (USGS):

The EEA’s 2013 update had described consumption as stabilised at about 2 Mt a year since 2000, and said bans in over 50 countries had been “offset by dramatically increasing use in China, India and other emerging economies” (EEA 2013, Annex 3, p. 724). The fall since then is real, but it is not complete.

Bans. - The International Ban Asbestos Secretariat (IBAS), an advocacy NGO and therefore a secondary source, listed about 70 national bans as of 23 June 2026. The list includes Brazil, Canada, Colombia, Korea, Japan, Ukraine, the UK and the US, but not China, India, Russia, Kazakhstan, Indonesia or Vietnam (IBAS, rev. 23 June 2026). - WHO reports prohibitions in “more than 50 WHO Member States” and estimates that more than 200,000 deaths a year are caused by occupational asbestos exposure worldwide (WHO fact sheet, 11 Aug 2026). - Brazil’s Supreme Federal Court upheld its 2017 ruling in February 2023, and Goiás state legislated in August 2024 to end mining. As of 30 September 2025 there was no agreed closure plan (USGS 2026).

Features of the earlier history that are now recurring: - Surveillance gaps (compare Lesson 2, p. 59). In India, cancer registries cover about 16% of the population. Registries recorded 54 mesothelioma cases in 2012–16, while 83 hospitals reported 1,126 over the same period (Singh & Frank, Public Health Action 2024). In the WHO European Region, only 6 of 19 countries without bans reported mesothelioma data to WHO (Kameda et al. 2014). A systematic review found the knowledge base for Central Asia “substantially” limited (Kurzhunbaeva et al., Med Lav 2024). - Early, reassuring surveys (compare p. 55). A nationwide Vietnamese survey of asbestos-cement plants found mean exposures of 0.19 f/ml. That is well above the EU limit, which is now 0.01 f/cm³. It reported a “low” share of pleural lesions among workers (Le et al., Int J Occup Saf Ergon 2023). This is the kind of short-latency reassurance the chapter warned against. - Rising burden where use continues. GBD analyses show asbestos-attributable mortality rising in China, especially among men, while age-standardised rates fall globally (Chen et al., BMC Public Health 2022). A case-control study in south-eastern China found strong dose-response relationships between hand-spinning chrysotile and mesothelioma (Jiang et al., Int J Cancer 2018).

The “differences” the chapter allowed for. Several South American countries banned asbestos within two decades of the chapter (Argentina, Chile, Uruguay, Brazil, Colombia), as did South Africa. That is faster than the European and North American timelines.

A 2022 global analysis offers a sobering qualification, however. Countries tended to ban asbestos when their own mesothelioma burden became identifiable. The odds of a ban rose 14.1-fold when a country identified mesothelioma, plus 26% for each additional death per million a year in the five years before a ban (Chimed-Ochir et al., BMJ Glob Health 2022). The authors’ framing is “Must countries shoulder the burden of mesothelioma to ban asbestos?”. On this evidence, many countries learned from domestic harm rather than from other countries’ early warnings.

Verdict: held up. Use continued, concentrated in Asian construction markets. Several recurring features of the earlier history (weak surveillance, reassuring short-term surveys, production-country advocacy) are documented. Some regions did act faster than Europe had.

Implication for weight. This supports treating the chapter’s historical pattern as recurrent, not unique. It also qualifies the premise that “late lessons” transfer readily between jurisdictions: the best predictor of action was a country’s own visible body count. In technology-neutral terms, a hazard can be displaced geographically instead of eliminated. Where monitoring is weakest, the early-warning cycle restarts, and other places’ lessons are discounted until local harm becomes visible.

Claim 5: EU ban by 2005; the WTO Appellate Body upholding the French and EU bans (qualitative evidence suffices, minority opinion may be relied on, “controlled use” is not a reasonable alternative); implications “for other hazardous agents” (p. 57, Box 5.1)#

Original claim (p. 57, Box 5.1). An EU ban “which is to be implemented by Member States by 2005”. The Appellate Body’s 2001 report yielded points with “implications for other hazardous agents”: - all forms of asbestos are carcinogenic, with no known threshold; - qualitative evidence is sufficient; - Members may rely on “qualified and respected” minority scientific opinion; - “controlled use” is not a reasonable alternative; - glass fibre products are not “like” products.

Subsequent developments

The EU ban took effect in 2005, but full completion took 20 more years. Commission Directive 1999/77/EC banned the remaining chrysotile uses, with a longer derogation for chlor-alkali diaphragms (Directive 1999/77/EC, 26 July 1999). The Commission confirms “the EU asbestos ban taking effect in 2005” (COM(2022) 488). The diaphragm exemption was ended only by REACH amendment, with a final cut-off of 1 July 2025 (Commission Regulation (EU) 2016/1005, 22 June 2016).

The Box’s summary of the WTO report is accurate on the main points, checked against the Appellate Body report (WT/DS135/AB/R, circulated 12 March 2001, adopted 5 April 2001; WTO case summary): - Qualitative evidence (para. 167). “There is no requirement under Article XX(b) of the GATT 1994 to quantify, as such, the risk to human life or health. A risk may be evaluated either in quantitative or qualitative terms.” - Minority opinion (para. 178). “A Member is not obliged, in setting health policy, automatically to follow what, at a given time, may constitute a majority scientific opinion.” - No threshold. The Appellate Body upheld the panel’s finding that “no minimum threshold of level of exposure or duration of exposure has been identified” for chrysotile-related pathologies other than asbestosis. - Controlled use. It upheld the panel’s rejection of “controlled use” as a reasonably available alternative. - Likeness. It reversed the panel’s likeness finding because the panel had excluded health risks from the comparison, and held that Canada had not proved that PVA, cellulose and glass fibres were “like” chrysotile.

Two corrections to the Box: - The minority-opinion principle was not new in 2001. The Appellate Body imported it from EC – Hormones (1998, para. 194) under the SPS Agreement and extended it to Article XX(b). - The United States was a third participant, not a cross-appellant (AB report, cover page).

Later reach of the ruling. - Later disputes. In Brazil – Retreaded Tyres, the Appellate Body and the parties relied on EC – Asbestos paras. 167–174 when analysing whether a health measure was “necessary” under Article XX(b) (WT/DS332/AB/R, 2007). - Canada. The complainant banned asbestos itself: the Prohibition of Asbestos and Products Containing Asbestos Regulations came into force on 30 December 2018, with limited exemptions, including chlor-alkali use until 2030 (SOR/2018-196). The history of that reversal is set out in Ruff, IJERPH 2017. - United States. The US EPA banned ongoing uses of chrysotile in March 2024 (EPA, page updated 5 May 2026; final rule, 89 FR 21970, 28 March 2024). This effectively rejected “controlled use” for most applications, but it allowed transitions of up to 12 years for chlor-alkali plants. USGS notes that deadlines “may be modified … as a result of legal proceedings in 2025” (USGS 2026). A search of the Federal Register API (EPA documents mentioning asbestos, June 2024 to September 2026) found no published amendment to the rule.

Limits on the reach of the ruling. - The Appellate Body no longer functions. It “is unable to review appeals given its ongoing vacancies”, and the last member’s term expired on 30 November 2020 (WTO Appellate Body page). There has therefore been no further appellate development of these principles. - The ruling concerned GATT Article XX(b), with strong underlying science. It did not endorse the precautionary principle as such. SPS-based measures still require risk assessment, as EC – Hormones shows (discussed in the report’s hormones chapter). - “Controlled use” survives outside the WTO. At the Rotterdam Convention, the Chemical Review Committee’s recommendation to list chrysotile has been before the Conference of the Parties since its third meeting (decisions RC-3/3 in 2006 and RC-4/4 in 2008), and it remains on the list of chemicals “recommended for listing”. The twelfth meeting (2025) listed other chemicals (RC-12/3 and RC-12/4) but not chrysotile (PIC: recommended for listing; PIC: COP reports and decisions). The EEA’s 2013 update records that Canada led the opposition until 2011–2012, after which Ottawa said it would no longer oppose listing (EEA 2013, Annex 3, pp. 725–726). Listing has remained blocked since.

Verdict: held up as a description of the law and of the EU ban, with small corrections. The claimed “implications for other hazardous agents” partly held up. The reasoning was reused in later GATT Article XX(b) cases and vindicated by later national bans. Its broader influence was limited by the scope of the ruling, by the collapse of the Appellate Body, and by the stalemate at the Rotterdam Convention.

Implication for weight. The legal lesson (pp. 57–58) is sound but narrower than the Box implies. In technology-neutral terms: trade law can accommodate protective measures based on qualitative risk evidence and credible minority science, when the hazard evidence is strong. Consensus-based international instruments can still be held up indefinitely by producer states. Derogations for “essential” uses can outlast the headline ban by decades.

Claim 6: All three asbestos types, including chrysotile, carcinogenic with no known safe level (IARC by 1986); the potency controversy “was, and still is” live (p. 57)#

Original claim (p. 57). “There was, and still is, scientific controversy about the relative cancer and asbestosis potencies of the three types of asbestos … By 1986 [IARC] had concluded that all three types were carcinogenic and, as with other carcinogens, there was no known safe level of exposure to any of them.”

Subsequent developments

Carcinogenicity has been reaffirmed and extended: - IARC. Monograph 100C (meeting 2009, published 2012) evaluated “Asbestos (Chrysotile, Amosite, Crocidolite, Tremolite, Actinolite and Anthophyllite)” (IARC 100C; summary in Straif et al., Lancet Oncol 2009). IARC’s current list by cancer site, covering Volumes 1–141 (2026), gives “Asbestos (all forms)” sufficient evidence for the lung, larynx, ovary and mesothelium, and limited evidence for the pharynx, stomach and colorectum (IARC list by cancer site). - WHO. The fact sheet (11 August 2026) states: “All forms of asbestos, including chrysotile, are carcinogenic to humans” (WHO). - ECHA. Its Risk Assessment Committee opinion (June 2021) “confirmed that asbestos does not have a safe exposure level”. It estimated that at the former EU limit of 0.1 f/cm³, 125 in 100,000 exposed workers could develop lung cancer or mesothelioma (COM(2022) 489). - EU law. Directive (EU) 2023/2668 codifies the point: “Because asbestos is a non-threshold carcinogen, it is not scientifically possible to identify a level below which exposure would not lead to adverse health effects” (recital 7) (Directive (EU) 2023/2668). - United States. The EPA’s 2020 risk evaluation found “unreasonable risks … for all ongoing uses of chrysotile asbestos” (EPA).

Chrysotile-specific epidemiology since 2001: - The Balangero (Italy) chrysotile miners cohort showed dose-related pleural mesothelioma (Ferrante et al., Am J Ind Med 2020). - A US case-control reanalysis found OR 3.8 for exposure to chrysotile only (Wong et al., Occup Environ Med 2021). - Chinese hand-spinning data showed strong dose-response (Jiang et al. 2018).

The potency controversy is still live, but it is now about magnitude, not causation: - Hodgson & Darnton (2000) put mesothelioma potency “broadly in the ratio 1:100:500 for chrysotile, amosite and crocidolite” (Ann Occup Hyg 2000). - A 2018 re-analysis obtained 1:83:376 (Garabrant & Pastula, Toxicol Appl Pharmacol 2018). It was criticised for omitting later cohort data (Finkelstein 2019). - Darnton’s 2023 update, used by HSE, found “significant heterogeneity” among chrysotile cohorts. Lung-cancer potency spanned “two orders of magnitude” (Darnton, Environ Res 2023). - A minority literature has argued for chrysotile no-effect levels (Pierce et al., Crit Rev Toxicol 2008). The authors themselves acknowledge “numerous potential biases”. No regulator cited here has adopted a chrysotile threshold. - Industry sponsorship. Critiques documented how industry-funded research sustained the idea that Quebec chrysotile was “essentially innocuous” (Egilman et al., Am J Ind Med 2003). They also documented industry control of UK research agendas (Tweedale, Am J Ind Med 2000). In both cases, sponsorship shaped the potency debate.

A historical correction. The chapter’s “by 1986 … IARC” wording is loose: - IARC’s evaluations of asbestos date from the 1970s (Monograph Volumes 2 and 14, 1973 and 1977), and asbestos was listed in Group 1 in the 1987 Supplement 7. These dates come from IARC’s standard record and were not re-checked against the volumes in this session. - The explicit formulation “No threshold has been identified for carcinogenic risks” appears in WHO/IPCS Environmental Health Criteria 203 on chrysotile (1998) (EHC 203, section 10). - The broader 1986 WHO/IPCS review (EHC 53) is also referenced there.

Verdict: strengthened on carcinogenicity and the absence of a threshold. The claim that the potency controversy is “still live” has held up: the debate continues, but it now concerns the size of the risk, not whether chrysotile causes cancer.

Implication for weight. The chapter’s scientific core is on firm ground. For the lens this project is building, the potency dispute is the more useful part. It shows how a real scientific uncertainty (relative potency) can be widened by sponsored research into a policy claim (safe “controlled use”) that regulators eventually rejected. In technology-neutral terms: when interested parties fund research on a genuine uncertainty, expect the uncertainty to be stretched toward the conclusion that suits them. Judge policy on the robust core finding, not on the contested margin.

Claim 7: Substitutes of the same physical form are likely carcinogenic, but mineral wool and glass fibre appear much less hazardous and can be engineered to be non-carcinogenic (p. 61, Lesson 10)#

Original claim (p. 61). “If asbestos substitutes reproduce the same physical form as asbestos – long, respirable (< 3 microns in diameter) and durable fibres – it is likely that they too will be carcinogenic … However, mineral wool and glass fibre appear to be much less hazardous … and they can be manufactured to be … not as thin, or durable enough in human tissue, to be carcinogenic.”

Subsequent developments

Insulation wools downgraded. IARC’s 2001 re-evaluation (Volume 81, 2002) separated insulation wools from special-purpose fibres. It reclassified insulation glass wool, rock/stone wool, slag wool and continuous glass filament as Group 3, while refractory ceramic fibres and special-purpose glass fibres remained Group 2B. The change followed better epidemiology and more attention to biopersistence (Baan & Grosse, Mutat Res 2004).

Biopersistence written into EU law. EU hazard classification under the CLP Regulation lists mineral wool as Carc. 2 but applies Note Q. Under Note Q the carcinogen classification “need not apply” if, for example, “a short term biopersistence test by inhalation has shown that the fibres longer than 20 μm have a weighted half-life less than 10 days”. Refractory ceramic fibres are classified Carc. 1B (Regulation (EC) No 1272/2008, Annex VI; entries 650-016-00-2 and 650-017-00-8). This is the chapter’s “engineer out durability” idea in legal form.

The physical-form paradigm is supported beyond asbestos: - Erionite and “fluoro-edenite fibrous amphibole”, both non-asbestos minerals with asbestos-like fibre form, carry sufficient evidence for mesothelioma (IARC list by site). - IARC Volume 111 evaluated fibrous silicon carbide and fluoro-edenite (Grosse et al., Lancet Oncol 2014). - WHO’s 2026 fact sheet states: “Many fibre substitutes for chrysotile asbestos assessed by WHO pose a relatively low hazard … though the carcinogenic hazard of some fibre substitutes was found to be high” (WHO).

Substitution has proved feasible and affordable: - The WTO Appellate Body treated PVA, cellulose and glass fibres as not “like” chrysotile, because of their health risk profile (see Claim 5). - The World Bank’s guidance, quoted in the EEA’s 2013 update, estimated that non-asbestos fibre-cement roofing costs 10–15% more to manufacture, partly offset by lower handling and disposal costs (EEA 2013, Annex 3, p. 725). - USGS lists numerous fibre and non-fibre substitutes (USGS 2026). - The US and EU chlor-alkali transitions show that even the last “safety-critical” uses had alternatives, though the transitions took years.

Verdict: strengthened.

Implication for weight. Lesson 10 is one of the chapter’s best-supported forward-looking claims. In technology-neutral terms: when a hazard follows from a physical property rather than a chemical identity, substitutes that share the property should be presumed to share the hazard until shown otherwise. Designing out the property (here, durability in tissue) can be verified and regulated. The chapter’s companion advice, to minimise exposure with “whatever materials are being used” (p. 61), remains prudent given WHO’s finding that some substitutes are highly hazardous.

Claim 8: Delay costs of EUR 400 billion for 400,000 deaths at EUR 1m each, plus “further billions” for building removal; a Dutch estimate that a 1965 rather than 1993 ban would have saved about 34,000 victims and NLG 41 billion (p. 58)#

Original claim (p. 58). “If lives are valued at EUR 1 million each … the costs of the estimated 400 000 European asbestos cancer deaths … is EUR 400 billion … Removing asbestos from buildings safely … will cost further billions.” A Dutch illustration (Heerings, 1999, a study done for Greenpeace Netherlands, p. 62) estimated that a 1965 ban “would have saved the country some 34 000 victims and NLG 41 billion”. The same page reports that “in the United States, asbestos compensation settlements reached USD 2 billion”.

Subsequent developments

The direction of the counterfactual is now directly supported: - Italy. Modelling of the 1992 ban estimates 8,341–21,981 mesothelioma deaths avoided in 1992–2020, depending on the scenario (Marinaccio et al., J Epidemiol Community Health 2026). - Sweden. Men born in 1955–79 had a relative risk of pleural mesothelioma of 0.16 compared with men born in 1940–49 (Järvholm & Burdorf, Scand J Public Health 2015). - Nordic countries. Male incidence peaked and began to fall, first in Sweden around 1990, after an early ban (Hemminki et al., BMC Cancer 2021). - Global analyses. Countries with bans show falling age-standardised rates, but the effect “take[s] decades to manifest” (Li et al., BMJ Public Health 2025; Jin et al., Lung Cancer 2025).

Magnitudes: the cost framing was probably conservative, but specific inputs are weak: - Value of a life. EUR 1m per life was low even by 2001 standards. OECD guidance (2012) recommends base values of roughly USD 3 million per statistical life (2005 prices) for OECD countries (OECD 2012); this figure was not re-verified against the full text in this session. - Death count. The 400,000 deaths are probably too few (see Claim 1). - Buildings. The Commission estimates that EU asbestos-containing materials “run to tens of millions of tonnes and [are] likely to exceed 100 million tonnes”. It also estimates that 4.1–7.3 million workers are currently exposed (COM(2022) 488). - France. A 2005 Senate inquiry projected EUR 27–37 billion in victim-support spending over 20 years for France alone (Sénat, Rapport d’information n° 37 (2005–2006), 26 Oct 2005). - United States. The “USD 2 billion” figure is badly understated. RAND found that “defendants and insurers have spent a total of $70 billion on asbestos litigation through 2002, more than half of which was consumed by … litigation expenses” (RAND MG-162, 9 May 2005). GAO counted 60 asbestos bankruptcy trusts with about USD 37 billion in assets, and about 100 companies bankrupt partly because of asbestos liability (GAO-11-819, 2011). These figures strongly support the chapter’s point that delay costs rebounded on industry and insurers (pp. 56, 58).

The Dutch figures should not be relied on. They came from a Greenpeace-commissioned study (p. 62) built on Dutch Ministry projections for 1969–2030. Re-modelling in 2003 found that “adding five recent years of observed mortality … resulted in a 44% lower prediction” of the future pleural-mesothelioma toll (Segura et al. 2003). The Dutch counterfactual is therefore probably overstated in magnitude, although its logic is sound.

Smaller slip. World production did not “grow to” 2 Mt in 1998 (p. 52). It had peaked at “almost 4.8 Mt per year” around 1977 (Virta, USGS Circular 1298, 2006).

Verdict: partly held up. The direction, that earlier action would have avoided large losses, is strengthened by ban-effect studies. The aggregate cost claim was probably conservative. The specific supporting figures are unreliable: the Dutch estimate is overstated, the US settlements figure is far too low, and the production trend is misdescribed.

Implication for weight. Lesson 4 (earlier action would have avoided loss, p. 59) now rests on good quasi-experimental evidence, not only assertion. The cost numbers in the chapter should not be quoted without correction. In technology-neutral terms: the costs of delay compound through long-lived installed stock (the “legacy” in buildings) and through liability systems whose transaction costs can absorb more than half of what is spent. Both are invisible in the prices paid at the time of use (p. 58).

Claim 9: Recommendation for “speedy, affordable and transparent” anticipatory compensation, to strengthen prevention incentives and preserve exposure histories (pp. 60–61, Lesson 8)#

Original claim (pp. 60–61). “It seems necessary to establish speedy, affordable and transparent compensation arrangements, based on agreed liabilities, as soon as any harmful effects become known, so as to both increase the incentives to prevent further harm and to improve the chances of recording accurate exposure histories.”

Subsequent developments

Fund-based and no-fault schemes spread, mostly reactively, after harm had become obvious: - France. FIVA (created by the social security financing law of December 2000) offers “réparation intégrale et rapide” (Sénat 2005). - Netherlands. The Institute for Asbestos Victims began handling mesothelioma claims in January 2000, with a fast expert review of disputed diagnoses taking a median of 34 days (Baas et al., Ann Oncol 2006). - UK. - Parliament restored full (joint and several) liability for mesothelioma (Compensation Act 2006, s. 3). This followed the House of Lords’ relaxation of causation in Fairchild (2002), which the EEA’s 2013 report discusses (EEA 2013, Panel 24.2, pp. 591–592). - The Mesothelioma Act 2014 established a Diffuse Mesothelioma Payment Scheme for victims who cannot trace an employer or insurer (Mesothelioma Act 2014, 30 Jan 2014). - Korea. An asbestos injury relief system has operated since 2011 (Kwon et al., IJERPH 2021). - Japan. It enacted a relief law for non-occupational victims in 2006; this was not verified against the primary statute in this session. - Switzerland. The European Court of Human Rights held in Howald Moor v. Switzerland (11 March 2014) that applying an absolute limitation period that expired before mesothelioma could be diagnosed violated the right of access to a court (Article 6). The Committee of Ministers closed supervision in 2019 after Switzerland reported general measures (judgment, HUDOC 001-141952; CM/ResDH(2019)232, 25 Sept 2019). This is a clean legal confirmation that latency and conventional limitation rules conflict. - European Union. At EU level, recognition of occupational disease remains a non-binding Recommendation. The 2022 update keeps asbestosis, mesothelioma, asbestos-related lung cancer and pleural disease on the schedule (Commission Recommendation (EU) 2022/2337). The Commission notes that the Treaty does not allow a binding instrument in this field (COM(2022) 488).

The claimed prevention-incentive mechanism is not supported, and there is some evidence against it: - France. The French Senate concluded that funding victim support mainly through the social-security occupational-injury branch was “privilégiant ainsi la mutualisation du risque au détriment de la nécessaire prévention et de la responsabilisation des employeurs” (privileging the pooling of risk at the expense of prevention and employer accountability) (Sénat 2005). - United States. The US tort system was neither speedy nor affordable: more than half of USD 70 billion went on litigation expenses (RAND 2005). - Coverage gaps persist. In Britain, 1,680 new mesothelioma cases were assessed for the no-fault Industrial Injuries Disablement Benefit in 2024, against 2,146 deaths, because “not everyone … is eligible and those that are may not claim” (HSE 2026).

We found no study showing that a compensation scheme improved the recording of exposure histories. National mesothelioma registers and lung-burden studies did that work (Gilham et al. 2018).

The EEA itself carried the argument forward in 2013, proposing “no-fault” administrative schemes and “anticipatory assurance bonds” for emerging technologies (EEA 2013, Chapter 24; Summary, p. 34).

Verdict: partly held up. The need for speedy, accessible compensation, and the tension between latency and legal process, are vindicated. The claimed effects on prevention incentives and record-keeping remain asserted, and pooled funding may blunt them.

Implication for weight. Treat Lesson 8 as a well-motivated design proposal, not an evidence-backed finding. In technology-neutral terms: compensation arrangements for delayed harms must survive long latency (limitation rules, insolvent or untraceable defendants). Whether they deter depends on who pays. Pooled or state funding compensates victims but can weaken the link between the party creating the risk and the cost of the harm.

Claim 10: For long-latency, no-threshold carcinogens, assume exposures are unsafe absent good evidence of safety, subject to proportionality; rebalance the bias from avoiding false positives toward balance with false negatives; claimed net social gain (p. 60, Lesson 6)#

Original claim (p. 60). “In the absence of good evidence that today’s exposures to carcinogens are safe, it is wiser to apply the precautionary principle, and assume they are unsafe, especially if the disease … [has] no known threshold”. Measures should depend “on the proportionality principle”. Science should switch “away from avoiding ‘false positives’ … towards producing a better balance”. “The asbestos case strongly suggests that society would gain overall.”

Subsequent developments

Regulators adopted the non-threshold presumption, with proportionality built in: - EU exposure limit. The EU cut the occupational limit tenfold, from 0.1 to 0.01 f/cm³. From 21 December 2029 it applies either 0.01 f/cm³ counting thin fibres by electron microscopy, or 0.002 f/cm³ for fibres 0.2–3 µm wide. The legislation explicitly calls asbestos a “non-threshold carcinogen” and sets limits by “an acceptable level of excess risk” (Directive (EU) 2023/2668, arts. 7–8, recitals 7, 15–18). - Measurement. The move to electron microscopy answers the chapter’s observation that invisible fibrils escape optical monitoring (p. 57). - National limits. Germany combines a binding limit with a lower “acceptable concentration” (COM(2022) 488). The Health Council of the Netherlands (2010) recommended risk-based limits well below the Dutch occupational limit (as summarised in EEA 2013, Annex 3, p. 725).

The low-dose risk is real but small, so proportionality matters: - Lung-burden work by Peto’s group estimates that the “average lifetime mesothelioma risk caused by recent environmental asbestos exposure in Britain will be about 1 in 10 000”. The risk is “an order of magnitude higher” in some exposed workers and probably in occupants of the most contaminated buildings (Gilham et al., Int J Epidemiol 2018). - The Commission’s impact assessment for the EU limit estimated 663 cancers prevented over 40 years, with monetised health benefits of EUR 166–323 million. It chose 0.01 f/cm³ as “balanced”. The European Parliament had pressed for a lower value; its exact figure was not re-checked in this session (COM(2022) 489). That is Lesson 6’s proportionality clause working as intended: further tightening yields diminishing returns.

The asbestos story has produced a risk–risk dispute: - UK. The House of Commons Work and Pensions Committee recommended “a deadline … for the removal of asbestos from non-domestic buildings within 40 years” (HC 560, 21 April 2022). The Government declined. It argued that a fixed deadline “would increase the opportunity for exposure” and “stimulate poor removal and disposal practices” (Government response, HC 633, 21 July 2022). - EU. The Commission promised a legislative proposal on mandatory screening and registration of asbestos in buildings for 2023 (COM(2022) 488). A search of the EU Publications Office catalogue for acts with “asbestos” in the title (June 2022 to September 2026) found no such proposal. The recast buildings energy directive instead requires Member States to “address … the removal of hazardous substances including asbestos” during major renovations (Directive (EU) 2024/1275, art. 8(3)). The European Parliament’s 2013 call for removal from public buildings by 2028 was not enacted (EEA 2013, Annex 3, p. 725).

So even with no uncertainty about the hazard, the precautionary response to installed legacy material is contested on risk–risk grounds.

The general claim about error balance remains contested: - In favour. The Late Lessons network’s own follow-up reviewed 88 alleged regulatory false positives and found “only four cases” that qualified (Hansen, Krayer von Krauss & Tickner, Risk Anal 2007). - Limits of that study. It was written by proponents of precaution. False positives are structurally hard to observe, because a restricted agent is seldom studied afterwards. That is itself an “absence of evidence” problem. - Critiques of the general frame. Critics argue that precaution can be incoherent when risks lie on all sides (Sunstein, Laws of Fear, 2005), or that it is applied selectively instead of consistently (Wiener & Rogers, J Risk Res 2002). - Scientific standards. No systematic shift in scientific standards of proof has occurred. IARC’s evidence categories are unchanged in structure. What changed is regulatory practice for non-threshold carcinogens.

A contrasting historiography. The chapter’s history of regulatory failure relies heavily on Tweedale (2000) and on the authors’ own work. The historian Peter Bartrip argues that the 1931 regulations “appeared to have addressed the problem effectively” until the 1960s discoveries (Bartrip, Postgrad Med J 2004; Bartrip, Med Hist 1998). Greenberg, a co-author of this chapter, disputed that view (Greenberg & Wikeley, Med Hist 1999). The chapter cites “Bartrip, 1931” (pp. 59, 62), which should read 1998, but does not engage his interpretation. For balance, the “missed opportunity” reading, though widely held, has had scholarly challengers.

Verdict: partly held up. - As applied to asbestos, the presumption of no safe level and the claim of net gain are strongly vindicated by hindsight, and regulators have adopted the approach. - As a general prescription to rebalance scientific error preferences, it remains a normative argument. It rests mainly on this case, has not been systematically tested, and is contested in the risk-regulation literature. - The asbestos legacy now shows the proportionality and risk–risk problems the lesson acknowledges but does not resolve.

Implication for weight. The “latency lacuna” mechanism (pp. 55, 59–60) is among the chapter’s strongest insights and should carry high weight. The prescription built on it should carry moderate weight and be presented as the authors’ position.

In technology-neutral terms: - Where harm appears decades after exposure and conditions keep changing, each new generation of exposure looks “safe” until too late. Evidence-of-absence should not be inferred from absence-of-evidence. - Once the hazard is established, the hard questions move to proportionality: how far to push exposure down, and whether removing installed hazards creates more exposure than leaving them in place. The chapter’s own lesson list does not settle those questions.

Cross-cutting observations for using this section as a lens#

1. The mechanism lessons outlast the numbers. Every structural mechanism in the chapter has independent later corroboration: - latency (pp. 55, 59–60); - short-follow-up “negative” studies (p. 55); - the survivor fallacy (p. 60); - studying the best-controlled segment (pp. 55–56); - producer control of evidence (pp. 54–55); - externalised costs (p. 58).

The numbers need correcting in both directions. 2. Industry and interest-group shaping of evidence is better documented now than in 2001. - UK: industry control of the Asbestosis Research Council (Tweedale 2000). - Quebec: sponsored “anything but chrysotile” science (Egilman et al. 2003). - France: the French Senate found that the Comité permanent amiante (1982–95), an industry-led body that included scientists, unions and ministry officials, functioned as “un ‘modèle’ de lobbying, de communication et de manipulation”, exploiting “pseudo incertitudes scientifiques” and promoting the “mythe de ‘l’usage contrôlé’” (Sénat 2005). - The same report records that the Conseil d’État found the French State liable (four decisions of 3 March 2004) for failing to regulate before 1977 and for late, inadequate regulation afterwards.

This strengthens the chapter’s moderate-confidence claim about data control (pp. 54–55, 58–59). 3. Legacy stock is now the main exposure route in banning countries. The EU (4.1–7.3 million exposed workers, 97% in construction), Britain (building workers most at risk), and WHO all say so. The chapter anticipated this cost (p. 58) but not the governance difficulty of removing it. 4. International learning is slow and domestic. The best predictor of a national ban was the country’s own identifiable mesothelioma burden (Chimed-Ochir et al. 2022). This qualifies the implicit premise of the whole Late Lessons project, that early warnings elsewhere will be heeded. 5. Slips identified in the digest, now resolved: - world production “grew to” 2 Mt: in fact it had fallen from a peak of about 4.8 Mt (USGS 2006); - “USD 2 billion” in US settlements: RAND counted USD 70 billion by 2002; - “The US cross-appealed”: the US was a third participant (AB report); - “Bartrip, 1931” should be Bartrip 1998; - the text names “Ronald Tage” (p. 54), while the cited source’s title reads “Reginald Tage” (Greenberg 1993, p. 62).

Sources#

All accessed 25 September 2026. Publication dates are given where known.

Official statistics and agencies - Health and Safety Executive (1 July 2026). Mesothelioma statistics for Great Britain, 2026. https://www.hse.gov.uk/statistics/assets/docs/mesothelioma.pdf - Health and Safety Executive (1 July 2026). Table MESO01: Death certificates mentioning mesothelioma, 1968–2024. https://www.hse.gov.uk/statistics/assets/docs/meso01.xlsx - Health and Safety Executive (July 2026). Asbestos-related disease statistics, Great Britain 2026. https://www.hse.gov.uk/statistics/assets/docs/asbestos-related-disease.pdf - US Geological Survey (February 2026). Mineral Commodity Summaries 2026: Asbestos. https://pubs.usgs.gov/periodicals/mcs2026/mcs2026-asbestos.pdf - Virta, R.L. (2006). Worldwide asbestos supply and consumption trends from 1900 through 2003. USGS Circular 1298. https://pubs.usgs.gov/circ/2006/1298/c1298.pdf - World Health Organization (11 August 2026). Asbestos fact sheet. https://www.who.int/news-room/fact-sheets/detail/asbestos - WHO/IPCS (1998). Environmental Health Criteria 203: Chrysotile asbestos. https://www.inchem.org/documents/ehc/ehc/ehc203.htm - IARC (2012). Monographs Vol. 100C: Arsenic, Metals, Fibres, and Dusts. https://publications.iarc.who.int/120 - IARC (2026). List of classifications by cancer sites with sufficient or limited evidence in humans, Vols 1–141. https://monographs.iarc.who.int/wp-content/uploads/2019/07/Classifications_by_cancer_site.pdf - US EPA (page updated 5 May 2026). Risk Management for Asbestos, Part 1: Chrysotile Asbestos. https://www.epa.gov/assessing-and-managing-chemicals-under-tsca/risk-management-asbestos-part-1-chrysotile-asbestos - US EPA (28 March 2024). Final rule, 89 FR 21970. https://www.govinfo.gov/content/pkg/FR-2024-03-28/pdf/2024-05972.pdf - Federal Register API search (run 25 Sept 2026; EPA documents mentioning asbestos, June 2024 to September 2026). https://www.federalregister.gov/api/v1/documents.json

EU law and policy - Commission Directive 1999/77/EC (26 July 1999). https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:31999L0077 - Regulation (EC) No 1272/2008 (CLP), Annex VI (Note Q; entries 650-016-00-2, 650-017-00-8). https://eur-lex.europa.eu/eli/reg/2008/1272/oj - Commission Regulation (EU) 2016/1005 (22 June 2016). https://eur-lex.europa.eu/eli/reg/2016/1005/oj - European Commission (28 Sept 2022). COM(2022) 488 final, on working towards an asbestos-free future. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52022DC0488 - European Commission (28 Sept 2022). COM(2022) 489 final, proposal amending Directive 2009/148/EC. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52022PC0489 - Commission Recommendation (EU) 2022/2337 (28 Nov 2022), European schedule of occupational diseases. https://eur-lex.europa.eu/eli/reco/2022/2337/oj - Directive (EU) 2023/2668 (22 Nov 2023). https://eur-lex.europa.eu/eli/dir/2023/2668/oj - Directive (EU) 2024/1275 (24 April 2024), energy performance of buildings (recast). https://eur-lex.europa.eu/eli/dir/2024/1275/oj - EU Publications Office SPARQL endpoint (catalogue search for “asbestos” titles, June 2022 to September 2026, run 25 Sept 2026). https://publications.europa.eu/webapi/rdf/sparql

Trade law and treaties - WTO Appellate Body (circulated 12 March 2001; adopted 5 April 2001). EC – Asbestos, WT/DS135/AB/R. https://docs.wto.org/dol2fe/Pages/SS/directdoc.aspx?filename=Q:/WT/DS/135ABR.pdf - WTO. DS135 dispute summary. https://www.wto.org/english/tratop_e/dispu_e/cases_e/ds135_e.htm - WTO Appellate Body (2007). Brazil – Retreaded Tyres, WT/DS332/AB/R. https://docs.wto.org/dol2fe/Pages/SS/directdoc.aspx?filename=Q:/WT/DS/332ABR.pdf - WTO. Appellate Body status page. https://www.wto.org/english/tratop_e/dispu_e/appellate_body_e.htm - Rotterdam Convention Secretariat. Chemicals recommended for listing. https://www.pic.int/TheConvention/Chemicals/RecommendedforListing/tabid/1185/language/en-US/Default.aspx - Rotterdam Convention Secretariat. COP reports and decisions (including RC-12). https://www.pic.int/TheConvention/ConferenceoftheParties/ReportsandDecisions/tabid/1728/language/en-US/Default.aspx

National law, courts and parliaments - Canada (in force 30 Dec 2018). Prohibition of Asbestos and Products Containing Asbestos Regulations, SOR/2018-196. https://laws-lois.justice.gc.ca/eng/regulations/SOR-2018-196/FullText.html - UK. Compensation Act 2006, s. 3. https://www.legislation.gov.uk/ukpga/2006/29/section/3 - UK. Mesothelioma Act 2014 (30 Jan 2014). https://www.legislation.gov.uk/ukpga/2014/1/introduction - House of Commons Work and Pensions Committee (21 April 2022). The Health and Safety Executive’s approach to asbestos management, HC 560. https://committees.parliament.uk/publications/21884/documents/162937/default/ - UK Government (21 July 2022). Government Response, HC 633. https://committees.parliament.uk/publications/23196/documents/169506/default/ - European Court of Human Rights (11 March 2014). Howald Moor and Others v. Switzerland, nos. 52067/10 and 41072/11. https://hudoc.echr.coe.int/eng?i=001-141952 - Committee of Ministers (25 Sept 2019). Resolution CM/ResDH(2019)232. https://hudoc.echr.coe.int/eng?i=001-196617 - Sénat (France) (26 Oct 2005). Le drame de l’amiante en France, Rapport d’information n° 37 (2005–2006). https://www.senat.fr/rap/r05-037-1/r05-037-1.html - RAND Institute for Civil Justice (9 May 2005). Carroll, S.J. et al., Asbestos Litigation, MG-162. https://www.rand.org/pubs/monographs/MG162.html - US GAO (2011). Asbestos Injury Compensation: The Role and Administration of Asbestos Trusts, GAO-11-819. https://www.gao.gov/products/gao-11-819

EEA follow-up - EEA (22 Jan 2013). Late lessons from early warnings: science, precaution, innovation, EEA Report 1/2013. Annex 3, “Asbestos – the campaign for a global ban” (Castleman & Gee), pp. 724–726; Panel 24.2 (McIntyre), pp. 591–592; Summary p. 34. Landing page: https://www.eea.europa.eu/en/analysis/publications/late-lessons-2 . Full report: https://op.europa.eu/en/publication-detail/-/publication/b0ad3e2e-01ba-4f1f-ac9d-0e382abcb4fd/language-en

Peer-reviewed literature - Peto J et al. (1995). Continuing increase in mesothelioma mortality in Britain. Lancet 345:535–539. https://doi.org/10.1016/s0140-6736(95)90462-x - Peto J et al. (1999). The European mesothelioma epidemic. Br J Cancer 79:666–672. https://doi.org/10.1038/sj.bjc.6690105 - Hodgson JT, Darnton A (2000). Ann Occup Hyg 44:565–601. https://doi.org/10.1093/annhyg/44.8.565 - Segura O, Burdorf A, Looman C (2003). Occup Environ Med 60:50–55. https://doi.org/10.1136/oem.60.1.50 - Egilman D, Fehnel C, Bohme SR (2003). Am J Ind Med 44:540–557. https://doi.org/10.1002/ajim.10300 - Pelucchi C et al. (2004). Br J Cancer 90:1022–1024. https://doi.org/10.1038/sj.bjc.6601638 - Baan RA, Grosse Y (2004). Mutat Res 553:43–58. https://doi.org/10.1016/j.mrfmmm.2004.06.019 - Hodgson JT et al. (2005). Br J Cancer 92:587–593. https://doi.org/10.1038/sj.bjc.6602307 - Baas P et al. (2006). Ann Oncol 17:848–852. https://doi.org/10.1093/annonc/mdl021 - Hansen SF, Krayer von Krauss MP, Tickner JA (2007). Risk Anal 27:255–269. https://doi.org/10.1111/j.1539-6924.2006.00874.x - Pierce JS et al. (2008). Crit Rev Toxicol 38:191–214. https://doi.org/10.1080/10408440701845609 - Straif K et al. (2009). Lancet Oncol 10:453–454. https://doi.org/10.1016/s1470-2045(09)70134-2 - Kameda T et al. (2014). Bull World Health Organ 92:790–797. https://doi.org/10.2471/blt.13.132118 - Grosse Y et al. (2014). Lancet Oncol 15:1427–1428. https://doi.org/10.1016/s1470-2045(14)71109-x - Järvholm B, Burdorf A (2015). Scand J Public Health 43:875–881. https://doi.org/10.1177/1403494815596500 - Ruff K (2017). IJERPH 14:1135. https://doi.org/10.3390/ijerph14101135 - Gilham C et al. (2018). Int J Epidemiol 47:1745–1756. https://doi.org/10.1093/ije/dyx276 - Garabrant DH, Pastula ST (2018). Toxicol Appl Pharmacol 361:127–136. https://doi.org/10.1016/j.taap.2018.07.003 - Jiang Z et al. (2018). Int J Cancer 142:514–523. https://doi.org/10.1002/ijc.31077 - Finkelstein MM (2019). Toxicol Appl Pharmacol 371:1–2. https://doi.org/10.1016/j.taap.2019.03.023 - Ferrante D et al. (2020). Am J Ind Med 63:135–145. https://doi.org/10.1002/ajim.23071 - Wong JYY et al. (2021). Occup Environ Med 78:199–202. https://doi.org/10.1136/oemed-2020-106665 - Hemminki K et al. (2021). BMC Cancer 21:1189. https://doi.org/10.1186/s12885-021-08913-2 - Kwon SC et al. (2021). IJERPH 18:10007. https://doi.org/10.3390/ijerph181910007 - Chen J et al. (2022). BMC Public Health 22:2012. https://doi.org/10.1186/s12889-022-14437-6 - Chimed-Ochir O et al. (2022). BMJ Glob Health 7:e010553. https://doi.org/10.1136/bmjgh-2022-010553 - Darnton L (2023). Environ Res 230:114753. https://doi.org/10.1016/j.envres.2022.114753 - Le HT, Dinh HT, Ngo TT (2023). Int J Occup Saf Ergon 29:263–267. https://doi.org/10.1080/10803548.2022.2035510 - Singh R, Frank AL (2024). Public Health Action 14:30–33. https://doi.org/10.5588/pha.24.0003 - Kurzhunbaeva Z et al. (2024). Med Lav 115:e2024042. https://doi.org/10.23749/mdl.v115i6.15453 - Li WZ et al. (2025). BMJ Public Health 3:e001717. https://doi.org/10.1136/bmjph-2024-001717 - Jin W et al. (2025). Lung Cancer 203:108534. https://doi.org/10.1016/j.lungcan.2025.108534 - Marinaccio A et al. (2026). J Epidemiol Community Health 80:719–726. https://doi.org/10.1136/jech-2025-224991

Historiography and critiques - Tweedale G (2000). Am J Ind Med 38:723–734. https://doi.org/10.1002/1097-0274(200012)38:6%3C723::aid-ajim11%3E3.0.co;2-g - Bartrip P (1998). Med Hist 42:421–438. https://doi.org/10.1017/s0025727300064334 - Greenberg M, Wikeley N (1999). Med Hist 43:508–510. https://doi.org/10.1017/s002572730006573x - Bartrip PW (2004). Postgrad Med J 80:72–76. https://doi.org/10.1136/pmj.2003.012526 - Wiener JB, Rogers MD (2002). Comparing precaution in the United States and Europe. J Risk Res 5(4). https://doi.org/10.1080/13669870210153684 - Sunstein CR (2005). Laws of Fear: Beyond the Precautionary Principle. Cambridge University Press. https://doi.org/10.1017/cbo9780511790850 - OECD (2012). Mortality Risk Valuation in Environment, Health and Transport Policies. https://doi.org/10.1787/9789264130807-en

Secondary (advocacy NGO; used for the list of national bans only) - International Ban Asbestos Secretariat (revised 23 June 2026). Current Asbestos Bans. http://www.ibasecretariat.org/alpha_ban_list.php

Access notes. The following could not be retrieved directly, and were used only as stated in the text: - the full text of the Rotterdam COP-12 report; - the IARC 100C chapter PDF (the IARC site-list PDF was used instead); - the OECD 2012 full text (only the headline VSL is cited); - the Japanese relief statute (flagged in Claim 9).