LL1-05 — Ch5 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, report pages 52–63 (PDF pages 52–63; in this report the PDF and printed page numbers coincide). Text body pp. 52–61; Table 5.1 p. 61; references pp. 61–63.

Read in full from the text extract. The PDF was checked visually for p. 52 (Figure 5.1), p. 55, p. 57 (Box 5.1), p. 58 and p. 61 (Table 5.1). The extract is faithful: no garbled passages were found. Author biographies (p. 196) and the report’s editorial note on author involvement (p. 12) were also checked in the PDF for the standpoint section.

Authors and standpoint#

Authors: David Gee and Morris Greenberg (p. 52).

• David Gee (bio, p. 196) trained in economics and politics. Since 1974 he had worked “at the interface of science, economics, production and policy-making” in occupational and environmental health, for trade unions and NGOs. He is a former Director of Friends of the Earth (England, Wales and Northern Ireland) and had been at the EEA since 1995 on emerging issues and scientific liaison. He was also one of the report’s two EEA editors and one of its two project managers (both roles shared with Sofia Guedes Vaz, p. 1), and the Executive Director’s preface thanks him as the person “who initiated this report” (p. 5). The chapter itself discloses that Gee visited UK asbestos plants in the 1980s “as the union health and safety adviser” (p. 60). That first-hand experience is the source of the “pensioners’ party fallacy”.
• Morris Greenberg (bio, p. 196) is a former Medical Inspector of Factories at the UK Health and Safety Executive and a former member of the Environmental Toxicology Unit at the UK Department of Health. He “helped in setting up the first asbestos mesothelioma register in the United Kingdom” and is described as an expert in the medical aspects of inhaled fibres. The bio does not say so, but his cited papers (e.g., Greenberg 1994 in Social History of Medicine) show that he also writes on the history of the asbestos hazard. His own papers are cited five times in the text (Greenberg 1993 twice, 1994 once, 2000 twice) and underpin key claims: Tage, Legge’s dismissal of Tage, knowledge before 1930, the Knox quotation, and the “ignorant expert”.

Report-level disclosure. The report’s introduction notes that case-study authors “are not without strong views, being for the most part active participants” in the histories they recount. It names Greenberg’s role in the mesothelioma register as an example, says authors worked pro bono and were asked to keep chapters brief, and says they were expected to be “as objective as possible” in answering four standard questions (p. 12). The chapter’s structure appears to follow those questions: early warnings (5.2–5.4), actions and inactions (5.5), costs and benefits (5.6), lessons (5.7). That is an inference from the headings.

Evident stance. This is a strongly pro-precaution, pro-worker-protection narrative, written by a former regulator-physician and an environmental-policy advocate with a trade-union background. The tone is prosecutorial toward industry (Turner Brothers, cartels, company doctors) and critical of UK regulators (non-enforcement). It is sympathetic to lay observers, victims, unions, investigative media and independent epidemiologists. The title’s moral register (“malevolent”) mirrors and inverts the industry-era “magic mineral” framing. It is itself a framing choice, and it echoes the title of Tweedale’s history Magic mineral to killer dust (2000), one of the chapter’s two most-cited secondary sources (Tweedale 2000 and Castleman 1996 each have five in-text citations).

Panels/commentaries. None in this chapter, and no dissenting or industry-authored commentary. Industry and dissenting voices appear only as actors the authors quote or report (for example the Asbestos Textile Institute representative, Dr Knox, the Lancet, Sells), usually in order to rebut them. The one dissenting historian in the reference list (Bartrip 1998) is cited as if he supported the authors (see References). The non-narrative elements are: - Figure 5.1 (p. 52): UK asbestos imports and predicted mesothelioma deaths. Source: Peto 1999. - Box 5.1 (p. 57): “WTO upholds French and EU ban on asbestos”, apparently written by the chapter authors or the EEA. - Table 5.1 (p. 61): “Asbestos: early warnings and actions”. Source: EEA.

Main sources relied on. Tweedale 2000 (company history of Turner & Newall; that it draws heavily on litigation-discovered company documents is general knowledge, not stated in the chapter); Castleman 1996 (medico-legal history); Greenberg’s own papers; Peto (1978 Lancet, 1998 BBC radio interview, 1999 BJC projections); primary Factory Inspectorate reports (Deane 1898, Collis 1911, Merewether & Price 1930); primary epidemiology (Doll 1955, Wagner et al. 1960, Selikoff et al. 1964, Newhouse & Thompson 1965, Knox et al. 1965, Hammond et al. 1979); Sells 1994 (Harvard Business Review, by a former Johns Manville executive); Heerings 1999 (a study for Greenpeace Netherlands); and the WTO panel and Appellate Body reports (2000, 2001).

Section-by-section notes#

Epigraph (p. 52)#

The chapter opens with Thomas Legge, ex-Chief Medical Inspector of Factories, writing in Industrial maladies (1934): opportunities “for discovery and prevention of asbestos disease were badly missed.” The choice matters. It shows a senior insider lamenting missed opportunities as early as 1934, so the “late lesson” was already visible to contemporaries and is not purely hindsight. Legge reappears (p. 55) as someone who, in 1932, advised the TUC it could be “quit of” a troublesome investigator. The chapter reports both facts but does not remark on the tension between them (see Mechanisms).

5.1 Introduction (pp. 52–53)#

• Opening case (p. 52). On 20 May 2000 the family of a senior UK hospital surgeon was awarded GBP 1.15 million after his death at 47 from mesothelioma. He had been exposed to “blue” asbestos (crocidolite) from damaged pipe insulation in the tunnels under Middlesex Hospital, London, where he worked for four years as a student and trainee during 1966–73 (BMJ 2000). The authors call this an “environmental exposure” (p. 52). Rhetorically, the case establishes that exposure reached far beyond asbestos workers, to a professional in a hospital, from building materials, decades after the hazard was known.
• Scale of burden (p. 52). “The main cause of mesothelioma is asbestos.” Some 250,000 mesothelioma cases, “normally fatal within one year”, are estimated for the EU over the next 35 years (Peto 1999). Adding lung cancer and asbestosis, the total “could be around 250 000–400 000 deaths”. Asbestosis was the first disease associated with asbestos. (Arithmetic oddity: the lower bound of the total equals the mesothelioma estimate alone, so the range as written implies near-zero additional lung cancer and asbestosis deaths at its low end. On p. 58 the upper figure is reused as “400 000 European asbestos cancer deaths … over the next few decades”.)
• Figure 5.1 (p. 52; Peto 1999). Bars show annual UK asbestos imports in thousands of tons, 1900–1990, split into chrysotile, amosite and crocidolite. Bars are at five-year intervals. Imports rise from under 5 thousand tons around 1900 to about 25 in the 1920s, about 60 by 1935 and about 95 around 1940, dip to about 62 around 1945, then run at 120–170 through the 1950s–1970s, peaking at about 170 in the early-to-mid 1960s. They fall to about 89 around 1980, about 35 around 1985 and about 15 around 1990. Chrysotile dominates. The amosite segment reaches about 28 on the axis around 1960; the bars appear stacked with crocidolite at the base, so amosite alone is roughly 24 thousand tons. Crocidolite is small (at most about 6). Two red curves project annual male mesothelioma deaths on the right axis:
• “Men born before 1953” rises from about 200 around 1970 to a peak of about 2,700 around 2015, then falls steeply to about 350 by 2035.
• “All men, assuming risk to men born after 1953 is 50% of 1943–48 birth cohort” peaks at about 3,300 around 2020, then declines to about 2,300 by 2045.
• The text summarises that the import peak is followed “some 50–60 years later by the estimated peak of mesotheliomas” (p. 52).
• Values are read off the chart and approximate. The figure covers males only and is a projection.
• Origins (p. 52). “A new global public health hazard was born” when chrysotile (“white”) mining began at Thetford, Canada, in 1879. Crocidolite (“blue”) and amosite (“brown”) were later mined in Australia, Russia, South Africa and elsewhere. World production of all types “grew to 2 million tonnes in 1998”. EU imports peaked in the mid-1970s, stayed above 800,000 tonnes a year until 1980, and fell to 100,000 tonnes by 1993. (Caution: “grew to” is misleading. USGS Circular 1298 (Virta 2006, Worldwide asbestos supply and consumption trends from 1900 through 2003) puts peak world production at “almost 4.8 Mt” a year around 1977, and notes that annual world production “by 1965 had already exceeded 2” Mt. So by 1998 output had more than halved from its peak and was back near its mid-1960s level. See the Limitations section.)
• Legacy (p. 52). “A substantial legacy of health and contamination costs has been left for both mining and user countries, and asbestos use is continuing, now largely in developing countries.”
• Scope (p. 53). The focus is “primarily on the United Kingdom”. The authors say similar histories occurred in France, Germany, Italy, Scandinavia and the US (Castleman 1996) and in the mining countries (Australia, Canada, Russia, South Africa). They add that the histories “are now being repeated, albeit with some differences, in Asia, Africa and South America” (p. 53). No evidence is offered for this last claim.

5.2 The first “early warnings” of asbestosis and some responses (pp. 53–54)#

• Speed of diffusion (p. 53). “Within 20 years of the start-up of asbestos mining, over 100 products made from the ‘magic mineral’ had been developed”, while “reports of serious disease had also begun to appear.”
• Lucy Deane, 1898 (p. 53). Deane was one of the first UK Women Inspectors of Factories. Her report included asbestos among four dusty occupations observed that year “on account of their easily demonstrated danger to the health of workers” and of “ascertained cases of injury to bronchial tubes and lungs medically attributed to the employment”. She noted that the “evil effects” of the dust had prompted a microscopic examination by HM Medical Inspector, which revealed “the sharp glass-like jagged nature of the particles” (Deane 1898). The warning therefore rested on case observation plus a plausible physical mechanism, not a quantified study. The chapter says Deane made “proposals” about “the kinds of mortality statistics that would be helpful” (p. 53). But her own words, quoted later (p. 60), call “reliable comparative statistics of mortality” the only convincing proof and say they were “practically unattainable in the case of any given factory, or at any rate with the time and opportunity at present at our disposal”. So she identified the method but judged it beyond the inspectorate’s then resources.
• Further inspectorate reports (p. 53). Similar observations by Women Inspectors followed in 1909 and 1910, published in the widely circulated annual reports of HM Chief Inspector of Factories. The authors’ judgement: the observations of “these laywomen might not have been categorised as ‘expert opinion’”, but they were “competent observers” whose discussion “would have done credit to a medical scientist”, and “Their reports were not refuted but simply ignored” (p. 53). Note the scope: the “simply ignored” verdict is about the Women Inspectors’ reports; the French report is separately said to have been “largely ignored”.
• Dr Montague Murray, 1899/1906 (p. 53). Murray, of Charing Cross Hospital, saw the “first reported case of lung disease attributed to inhaled asbestos dust” in a 33-year-old man who had worked 14 years, 10 of them in the carding room. The patient said he was the only survivor of ten people who had worked in that room, the others dying “somewhere about thirty years of age”. Murray added: “I have no evidence except his word for that” (Murray 1906). The observation “was brought to the attention of” the 1906 UK government inquiry into compensation for industrial diseases, to which Murray gave evidence (the chapter uses the passive voice; the reference is to his minutes of evidence).
• Auribault, France, 1906 (p. 53). A French Factory Inspector reported “some 50 deaths amongst female asbestos textile workers”. The report covered processing, hazards in spinning and weaving, and designs for dust-capture apparatus. It was “largely ignored”. The authors link it forward: the French ban about 90 years later led to the 1999 WTO case.
• The 1906 inquiry’s decision (p. 53). The inquiry did not include asbestos as a cause of industrial disease. Murray had stated that “one hears, generally speaking, that considerable trouble is now taken to prevent the inhalation of the dust, so that the disease is not so likely to occur as heretofore.” The authors say this “may have influenced the committee”. They add that no one checked the patient’s claim about nine co-worker deaths or investigated the surviving workers, “despite the proposals from Lucy Deane” about useful mortality statistics.
• First named fallacy (p. 53). “Dr Murray’s view that ‘no evidence of harm’ is the same as, ‘there is evidence of no harm’, is an early example of a common fallacy”. The authors say this fallacy has inhibited identification of many substances first considered harmless (“false negatives”).
• Collis 1910, rats 1911 (pp. 53–54). Evidence was noted in workers in 1910 (Collis 1911) and in “pioneering dust experiments with rats in 1911”. This evidence “was later considered to have been ‘reasonable grounds for suspicion that the inhalation of much asbestos dust was to some extent harmful’, such that the Factory Department pressed for the installation of exhaust ventilation in the dusty processes” (Merewether 1933). The sentence is ambiguous about timing: it can be read as a retrospective 1933 judgement, but Table 5.1 dates the “reasonable grounds” to 1911, and the ventilation pressure appears to have followed the 1910–11 evidence. Either way, the chapter’s own account shows the Factory Department responded to the early evidence to some degree.
• Inaction (p. 54). “Subsequent Factory Department inquiries in 1912 and 1917 found insufficient evidence to justify further action.” Table 5.1 gives “1911 and 1917”, an internal inconsistency.
• Insurers, 1918 (p. 54). In the US and Canada, insurers had “seen enough proof” by 1918 to decline cover for asbestos workers “due to the assumed injurious conditions in the industry” (Hoffman 1918). The authors call this “early precautionary action” that “was later forgotten”, so that asbestos costs became “hugely damaging” to US insurers in the 1990s. This is a striking point: actors with money at risk acted decisively on the early evidence at a time when UK regulators judged it insufficient (1912, 1917). But (this is the note-taker’s observation, not the chapter’s) the precaution protected the insurers, not the workers: declining cover limits an insurer’s exposure and does nothing to reduce the hazard itself. The chapter itself presents it simply as “early precautionary action”. Table 5.1 mentions only “US insurers”; the text says the United States and Canada.
• Rochdale and Leeds, 1924–1930 (p. 54). In 1924 at Rochdale, home of the Turner Brothers factory since 1880, came the first inquest and pathological examination of an asbestos worker. Nellie Kershaw was diagnosed with “asbestos poisoning” by her local doctor, Dr Joss, who said he saw 10–12 such cases a year. Pathologist W. Cooke corroborated this and published it (Cooke 1924, 1927). In Leeds, at another Turner factory, a local doctor found enough cases for an MD thesis (Grieve 1927). By 1930 there were “at least 12 deaths” at the two factories with asbestosis cited as a cause or partial cause (Tweedale 2000). “In some cases” diagnosis was complicated by tuberculosis, heart failure and pneumonia, “as they did for the next few decades” (p. 54). This is an important acknowledgement of genuine diagnostic uncertainty.
• Merewether & Price, 1930 (p. 54). Two 1928 papers (Simpson; Seiler), including four South African cases, together with the above, prompted a “major government inquiry” by Dr Merewether (Medical Inspector of Factories) and C.W. Price (Factory Inspector and “pioneer of dust monitoring and control”). It included “the first health study of asbestos workers”:
• of 363 workers, 25% had asbestosis on average;
• 66% of those employed 20 years or more were affected, against none of those employed under four years.
• The authors add that this was “probably an underestimate, as only current workers were examined, excluding those who had left employment through ill health”, an early instance of the healthy-worker bias they later name.
• The 1931 regulations (p. 54). The results led in 1931 to “the first asbestos dust control regulations, medical surveillance and compensation arrangements in the world”. These “remained largely unaltered (but also unenforced) until 1969”.

5.3 Early warnings on asbestos cancers (pp. 54–55)#

• Tage, 1932 (p. 54). In a report to the Trades Union Congress, the freelance investigator “Ronald Tage” (the reference list gives “Reginald Tage”, p. 62, and the cited paper’s published title, Greenberg 1993, Am. J. Ind. Med. 24: 521–524, is “Reginald Tage: A UK asbestos prophet — A postscript”, so “Ronald” in the text is an error) drew attention to three asbestosis cases from the Cape Asbestos Company in Barking, London, that were complicated by cancer (Greenberg 1993).
• Lung cancer literature (p. 54). Lung cancer associated with asbestos was reported in the US, German and UK literature in the 1930s and 1940s (Lynch & Smith 1935; Gloyne 1935; “Heuper” [Hueper] 1942; Wedler 1943), including the 1938 Report of the Chief Inspector of Factories.
• Germany (p. 54). In 1938, “when lung cancer was generally much less prevalent”, German authorities “were persuaded that the association was causal”, and asbestos lung cancer became a compensatable industrial disease in 1943. The authors note in parentheses that smoking-induced lung cancer later made the link “that much harder to prove”. The political context of this recognition is not discussed.
• 1949 and the industry’s unpublished studies (p. 54). The Chief Inspector’s 1949 Annual Report found a high rate of lung cancer at autopsy in asbestosis cases. Industry “had two unpublished US reports of an excess of respiratory cancers in mice” (Scheper 1995), meaning hazard evidence was held by industry and not published.
• In-house studies (p. 54). “Three in-house investigations of cancer mortality in the Rochdale district had failed to find evidence of lung cancer in asbestos workers” (cited as Knox 1952 and 1964). Yet “the company doctor admitted that his knowledge of statistics was ‘nil’” (Tweedale 2000, p. 148). Caution: the Knox 1952 and 1964 references are titled as reports of visits to Canadian mines (p. 62), so the citation may be mismatched.
• Doll, 1953–55 (p. 54). In 1953 Turner Brothers asked Richard Doll, “an independent epidemiologist”, to study Rochdale workers. He found a lung cancer risk ten times that expected among those exposed for 20 years or more. “Despite attempts by the Turner directors to suppress these findings they were published” (Doll 1955). The mechanism of suppression is not described.
• Recognition lag (pp. 54–55). It was “another 30 years before the government accepted lung cancer from asbestos as a compensatable industrial disease, and then only if it was accompanied by asbestosis”. This was partly because studies were “increasingly complicated by the rising trends of lung cancer caused by smoking”, which Doll had also discovered “in a study of British doctors in 1955” (p. 55). (Minor dating issue: Doll and Hill’s smoking work dates from 1950 and 1954.)
• Synergy (p. 55). Later studies showed that smoking and asbestos multiplied lung cancer risk. Asbestos alone gave about 5-fold, smoking alone about 10-fold, and both together “not 15 times the risk (an additive effect) but over 50 times the risk, a multiplicative or synergistic effect” (Hammond 1979). A similar synergy is noted for smoking and radiation “from uranium and other mining” in radiation-exposed workers (Archer 1973).
• The “latency lacuna” (p. 55). Few workers had worked under the “new conditions” of improved dust control for the 20–25 years needed before lung cancer appeared. So in 1955 it was not possible to say what the current risks were “until many more years had passed”, by which time dust control had improved again and it was once more “impossible to know what the risks could be”. The authors name this the “‘latency lacuna’, characteristic of all long-latent-period hazards under conditions of technological change”, and call it “a major reason why preventative action is often too late” (p. 55). This is one of two terms the chapter explicitly coins (“which might be called”); the other is the “pensioners’ party fallacy” (“could be called”, p. 60).

5.4 Early, devastating warnings about mesothelioma cancer (pp. 55–56)#

• Mesothelioma recognised (p. 55). Mesothelioma is “a normally very rare cancer of the lining of the chest or abdomen”. Cases had been seen alongside asbestos exposure in the 1940s–50s, but in 1955 Dr Sleggs, a local South African doctor, noticed a cluster “at the centre of the asbestos mining areas” and sent samples to the pathologist Dr Wagner. They toured the mining areas reconstructing exposure histories “by talking to colleagues and families”. Of 47 cases, all but two had earlier asbestos exposure. “Many of them were environmental cases including children exposed when playing on waste dumps.” The work was published in 1960 (Wagner et al. 1960). The method matters: it relied on local clinical observation plus lay testimony (families, colleagues).
• Why it was “devastating” (p. 55). Exposure needed to cause mesothelioma “seemed to be a matter of months only”. Lung cancer and asbestosis needed 10 or more years. Mean latency for mesothelioma was about 40 years, against 20–25 for lung cancer.
• Causal acceptance by 1964 (p. 55). Wagner provided a very strong association, but “by 1964 most experts accepted that the relationship was causal”, based mainly on Selikoff (US) and Newhouse (UK). “Both worked independently of the industry, using case data from unions and hospital records respectively.” Independence from industry is explicitly flagged as significant.
• Selikoff (p. 55). He observed 15 of 17 patients from one asbestos plant with asbestos disease. “As he was refused access to company records” he used trade union records. These showed that users such as insulation workers were at greater risk than manufacturing workers: of 392 with 20 or more years’ exposure, 339 had asbestosis, lung cancer was seven times normal, and several had mesothelioma (Selikoff et al. 1964).
• Low power of “negative” studies (p. 55). The lung cancer excess “only became statistically clear after 25 years of follow-up”. This illustrates “one of the serious limitations of so called ‘negative’ cancer studies”: their power to detect long-latency cancers “can be very low unless some 20–30 years of follow-up has been possible.”
• Attacks on messengers (p. 55). An Asbestos Textile Institute representative called Selikoff a “disturbing sore thumb” (Tweedale 2000, p. 183 fn 17). The authors liken this to Legge, as TUC Medical Adviser in 1932, suggesting the TUC could be “quit of” Tage by paying him a small fee (Greenberg 1993).
• Ibsen (pp. 55–56). “The practice of attacking the purveyors of news about hazards” is illustrated by Ibsen’s An enemy of the people (1882). In the play a doctor who notices a hazard threatening the town’s economy “descends from public hero to public enemy” as the mayor, media and citizens grasp the economic implications. This is a literary analogy, not evidence.
• Newhouse (p. 56). Using London Hospital pathology records for 1917–1964, Newhouse found that of 76 mesothelioma cases, over 50% had occupational or domestic exposure (living in the house of an asbestos worker). Of the others, one third lived within half a mile of the Cape asbestos factory (Newhouse & Thompson 1965). About 30 years later, children exposed near factories who developed mesothelioma became “the first successful environmental exposure cases against Turner Brothers in the United Kingdom” (Tweedale 2000, p. 272).
• New York Academy of Sciences conference, October 1964 (p. 56). Newhouse and Selikoff presented there. At the same meeting, a study of Doll’s Rochdale cohort from regulated areas (Knox et al. 1965) was presented as supporting the view that “it is possible that the specific occupational hazards to life have been completely eliminated”. The authors call this “perhaps another example of the ‘latency lacuna’”. Neither Selikoff nor the UK Factory Department found falling disease rates, “mainly because they were including severe cases of dust exposure amongst the users, not just the manufacturers”, whose conditions were better, “at least in the regulated factory production areas”.
• Worst-case blindness (p. 56). “This failure to appreciate the ‘worst case’ asbestos exposure scenarios was part of the reason for the delayed and inadequate responses.” Julian Peto called the focus of cancer studies on factories rather than users a “stupid mistake” (Peto 1998, a BBC Radio 4 interview).
• Industry insider hindsight (p. 56). A former director of Johns Manville (“the world’s biggest asbestos company”) shared this view in reviewing why the firm, “although still profitable”, filed for bankruptcy in 1982 “as a means of dealing with asbestos pollution claims”. He argued that “medical research, assiduous communication, insistent warnings and a rigorous dust reduction programme” “could have saved lives and would probably have saved the stockholders, the industry and, for that matter, the product” (Sells 1994). This is the chapter’s main evidence that precaution could have served industry’s own long-run interest.

5.5 Actions and inactions by regulatory authorities and others (pp. 56–58)#

• Non-enforcement (p. 56). The 1931 regulations were only partially enforced, with “only two prosecutions between 1931 and 1968” (Dalton 1979). Their focus on parts of the manufacturing process meant “riskier user activities were neglected”. The authors add: “However, the issue of dangerous asbestos was not neglected”. This introduces the role of media and politics.
• Media and political pressure, 1964–75 (p. 56). US and UK media kept asbestos “high on the political agenda” (Sunday Times 1965, “Urgent probe into ‘new’ killer dust disease”). Two documentaries about Cape’s Acre Mill plant in Yorkshire, ITV’s World in Action (1971) and BBC’s Horizon (1975), “helped to initiate action by authorities”. This included a Parliamentary Ombudsman report prompted by a complaint from the local MP, Max Madden, about non-enforcement of the 1931 regulations. The report “was very critical of the Factory Inspectors”, and the government responded by appointing the Simpson Committee in 1976.
• 1969 regulations (p. 56). A factory dust limit of 2 million fibres/m³ (2 f/ml) was to be introduced gradually. This “hygiene standard” “did not include consideration of the lung or mesothelioma cancer hazards”. Julian Peto, in evidence to Simpson, later criticised it strongly and linked it to high asbestosis levels (“one worker in 10 would get the disease”) (Peto 1978). Table 5.1 summarises 1969 as “improve controls, but ignore users and cancers” (p. 61).
• Simpson Report, 1979 (p. 56). It recommended:
• banning blue asbestos, “which had already been withdrawn by the industry”;
• banning insulation spraying, “also largely defunct by then”;
• licensing contractors for asbestos removal;
• a limit of 1 f/ml for white asbestos by 1980, with a target of 0.5 f/ml for brown, which was considered more hazardous. Two of the four recommendations (the blue asbestos and spraying bans) ratified what industry had already done; licensing and the lower limits were new. The chapter does not say why industry had withdrawn blue asbestos and spraying.
• Fibre and fibrils (pp. 56–57). A visible fibre is about the diameter of a human hair (40 microns) but is “composed of a bundle of about 2 million fibrils” that abrasion or processes in the body can release (Selikoff & Lee 1978). “Electron microscopy is needed to accurately monitor” such fibrils in air or tissue. Read with p. 57 (optical microscopy was “the prevailing dust monitoring method”), this implies the standard monitoring method could not accurately track the finest fibrils.
• Fibre-type controversy (p. 57). “There was, and still is, scientific controversy about the relative cancer and asbestosis potencies of the three types”, with white “often being regarded as less hazardous”. By 1986 IARC had concluded that all three were carcinogenic and that, “as with other carcinogens, there was no known safe level of exposure”. The chapter records the controversy but does not set out the case for lower chrysotile potency.
• Public exposure standard (p. 57). There was no hygiene standard for the public until the late 1980s, when the HSE recommended the lowest limit detectable by optical microscopy (0.1 f/ml). The recommended public limit was pegged to the detection limit of the prevailing method; the chapter gives no health-based rationale for it, and it describes a recommendation rather than a statutory limit.
• Alice, a Fight for Life, 1982 (p. 57). Yorkshire TV screened a two-hour prime-time documentary about Alice Jefferson, aged 47, who contracted mesothelioma after “working for a few months” at Acre Mill. It “had an immediate impact, even though some, like Sir Richard Doll, criticised the programme for being unscientific and emotional.” (“Unscientific and emotional” is the authors’ summary of Doll’s criticism, not a quotation from Doll; no source is cited for it.) The government “responded to the programme” by implementing Simpson and, in 1984, introducing licensing regulations and lower limits: 0.5 f/ml white and 0.2 f/ml brown. A voluntary labelling scheme was introduced for some uses. The causal attribution (“responded to the programme”) is asserted rather than documented.
• Continuing pressure (p. 57). Pressure came from local MPs, some unions and victims’ representatives such as the asbestos widow Nancy Tait. She helped reveal that Turner Brothers’ compensation arrangements, “such as GBP 1 a week for the widows of workers with asbestosis”, had been “largely unchanged since the 1930s”, and her work helped force improvements.
• Tightening to ban (p. 57). New regulations came in 1987, tightened in 1989. In 1998 the UK government adopted a ban on all forms, implemented in 1999, alongside an EU ban to be implemented by Member States by 2005. Canada filed a trade complaint at the WTO against the French and EU bans. The Disputes Panel rejected it, Canada appealed, and the Appellate Body “found in favour of France and the EU”.
• Box 5.1: WTO upholds French and EU ban (p. 57).
• Background. France banned all forms in 1997 “to protect the health of workers and consumers”. Existing white asbestos products could be exempted “on an exceptional, temporary and annually reviewed basis” if no effective lower-risk substitute was available. (This exemption is relevant to the “irreplaceability” debate, p. 58.)
• Proceedings. The Panel found for France in September 2000. (Precision from the WTO record, not the chapter: the dispute, EC – Asbestos, concerned the French Decree, with the European Communities as formal respondent; the EU-wide ban was not itself challenged. And the Panel’s ruling was mixed. It found chrysotile and substitute fibres to be “like products” and the Decree to violate GATT Article III:4, but held the Decree “justified” under Article XX(b) (Panel conclusions as summarised in AB report para. 4). The Appellate Body later reversed the likeness finding.) Canada appealed, and the EU cross-appealed to uphold the main findings and correct “errors”. The Box also says “The US cross-appealed against the panel’s judgement that glass fibres were as carcinogenic as asbestos”. This is wrong on the WTO record. The Appellate Body report (WT/DS135/AB/R, 12 March 2001, pp. 1, 4) lists Canada and the European Communities as appellants/appellees and Brazil and the United States as third participants; the US filed a third participant’s submission, not a cross-appeal. Nor did the Panel find glass fibres “as carcinogenic as asbestos”: it declined to consider health risk in its “likeness” analysis at all (Panel paras 8.130, 8.132), and the US argued, on the EC’s side, that this was an error (AB report, para. 48). The Appellate Body reported on 12 March 2001.
• Main points listed by the authors, “which also have implications for other hazardous agents”:
  • all forms are carcinogenic, with no known threshold of safety;
  • the risk from white asbestos in products rests on evidence that “tends to show” a risk rather than not (this is the Panel’s phrase, Panel para. 8.193, which the AB quoted and upheld at para. 157);
  • workers handling products (building, brake lining) are at risk;
  • “there is no WTO requirement for countries to provide quantitative risk assessment data: qualitative evidence is sufficient”;
  • countries may rely on “qualified and respected scientific opinions held by only a minority of scientists”, quoting the AB: “a Member is not obliged, in setting health policy, automatically to follow what, at a given time, may constitute a majority scientific opinion” (AB p. 64, para. 178; verified). The Box adds that a Panel therefore need not decide on “a ‘preponderant weight of the evidence’, but on a lower level of proof”. The first half closely tracks the AB’s own next sentence (“a panel need not, necessarily, reach a decision under Article XX(b) of the GATT 1994 on the basis of the ‘preponderant’ weight of the evidence”, para. 178, AB p. 65). Only “but on a lower level of proof” is the Box’s gloss; the AB did not name an alternative standard;
  • the efficiency of “controlled use” was not demonstrated, residual risk to workers would remain significant, and so it “was not a reasonable ‘alternative’ measure to the asbestos ban”;
  • on “like products”, WTO criteria (properties, end uses, consumers’ tastes and habits) led the AB to find that the panel erred in treating glass fibre products as “like” asbestos products: they were not, “principally because they were not as carcinogenic”.
• Check against the AB report. The qualitative-evidence point (para. 167: “A risk may be evaluated either in quantitative or qualitative terms”) and the controlled-use point (para. 174: efficacy “remains to be demonstrated”, with a “significant residual risk”, and “particularly doubtful for the building industry and for DIY enthusiasts”) are accurately summarised. The likeness point is simplified. The AB reversed the Panel’s likeness findings and held that carcinogenicity is “a defining aspect of the physical properties” of chrysotile (para. 114). But its formal finding was that Canada had “not satisfied its burden of proving” likeness (paras. 141, 147), and for cement-based products the decisive gap was the absence of evidence on consumers’ tastes and habits. The AB also found (para. 168) that the risk from the substitute fibres is “less” than chrysotile’s but not shown to be “non-existent”, and that it is “perfectly legitimate” to halt a highly risky product while allowing a less risky one. That point bears on Lesson 10 and is not in the Box.
• WTO procedures criticised (p. 58). One of the scientific advisers in the case criticised WTO procedures for handling scientific complexity (Castleman 2001, a draft conference paper). Details are not given.
• Current burden (p. 58). The Health and Safety Commission estimated UK deaths from asbestos-related mesothelioma and lung cancer at “around 3 000 deaths per year and rising” (HSC statistics for 1994–95). The text says “see Figure 5.1”, but the figure shows male mesothelioma only, so it does not directly display this total. “Despite huge amounts of research, many issues of biological mechanisms and dose-response relationships remain unclear, illustrating the limited relevance of more research to disease.” The phrasing is compressed, probably meaning “to disease prevention”. The claim is that more research does not by itself deliver prevention.

5.6 The costs and benefits of actions and inactions (pp. 58–59)#

• Scope disclaimer (p. 58). A full cost-benefit evaluation is “beyond the scope” of the chapter (it points to Castleman 1996, pp. 8–9). “A few illustrative figures” are given instead.
• Company and insurer costs (p. 58). Turner Brothers arranged in 1994 to pay up to GBP 1 billion in claims. Lloyd’s of London “faced near collapse in the early 1990s” from US pollution claims, many of them asbestos compensation and clean-up. Neither figure carries a citation in the chapter.
• Monetised deaths (p. 58). At EUR 1 million per life, “common in transport studies”, the estimated 400,000 European asbestos cancer deaths cost EUR 400 billion. “The human costs in terms of suffering are not calculable.” Safe removal from buildings “will cost further billions”. “Earlier actions to reduce asbestos exposure could have saved many of these costs.” The calculation is undiscounted and uses the upper end of the 250,000–400,000 range.
• Dutch counterfactual (p. 58). A ban in 1965, after the mesothelioma evidence was widely accepted, instead of 1993, would have saved the Netherlands “some 34 000 victims and NLG 41 billion in building and compensation costs”. The comparison is with 52,600 victims and NLG 67 billion expected over 1969–2030, as estimated by the Dutch Ministry of Health and Social Security (via Heerings 1999, a study for Greenpeace Netherlands).
• US (p. 58). “Asbestos compensation settlements reached USD 2 billion, with Lloyds syndicates paying around half of that.” This figure is ambiguous and probably understated relative to total US asbestos litigation costs (see Limitations).
• Benefits (p. 58).
• Employment.
• A 1919 estimate that theatre fires in the 1870s–80s caused 2,216 deaths, 95% of which asbestos fire insulation could have prevented, “it was claimed” (Summers 1919). The reference list shows Summers is cited second-hand (“Cited in Tweedale, P5, fn 10”, p. 63).
• Boiler insulation saved energy.
• Brake linings saved lives, “though the extra vehicle speeds it allowed complicates the picture”. This is a risk-compensation point.
• The Lancet, 1967 (p. 58). “It would be ludicrous to outlaw this valuable and often irreplaceable material in all circumstances (as) asbestos can save more lives than it can possibly endanger.” The authors respond that, beyond a “gross underestimate of the health impacts” (on which, they concede, the doctors’ “expertise was at least relevant”), “the replaceability of asbestos is a technological and economic question which doctors are not well qualified to judge”. They add that little evidence was offered for “irreplaceable”. This anticipates Lesson 9 (“ignorant expert”).
• Substitutes (p. 58). “Substitutes for most uses of asbestos were available by the 1970s and, in some cases, much earlier”. Many US oil refineries were insulated with mineral wool in the 1940s–50s (Castleman 1996, pp. 456–457). Substitutes spread slowly for two reasons:
• “asbestos industry cartels worked to inhibit their spread” (Castleman 1996, pp. 34–38);
• the market price was “very low compared to its full production, health and environmental costs”. The authors generalise: “This failure of market prices to reflect full environmental and health costs is the common cause of delay in replacing hazardous materials.” This is a strong generalisation asserted from one case.
• Profits and externalisation (p. 58). “Many jobs, much profit and high dividends”. Turner Brothers’ profits rose strongly after 1947 and peaked at almost GBP 9 million a year in 1965 (Tweedale 2000, p. 9). These profits “suffered little” from ill-health and contamination costs, which were “‘externalised’ onto workers with disease, their families, the health service, insurance carriers and building owners.”
• Litigation as a way of producing knowledge (pp. 58–59). An “often ignored but significant non-financial benefit” of compensation trials is that they “frequently uncover many of the contradictions between company words on asbestos and their actions to reduce hazards” (Castleman 1996). Some of the chapter’s own evidence visibly derives from documents discovered in litigation: the Knox references cite “the Chase Manhattan microfilms” and “Discovered documents” (p. 62). How much of the Tweedale and Castleman material rests on such documents is not stated in the chapter.

5.7 What are the lessons of the asbestos story? (pp. 59–61)#

Ten numbered lessons, said to be relevant to “numerous other agents or activities that have long-term hazardous impacts” (p. 59). Recorded in full under “The authors’ own lessons” below.

Table 5.1: Asbestos: early warnings and actions (p. 61; Source: EEA)#

Note the small discrepancies with the text: 1911 versus 1912; 1959–60 versus 1955/1960; 1962/64 versus 1964/65; and France’s ban dated 1998–99 in the table but 1997 in Box 5.1. The table’s 1998–99 row also omits the UK ban, which the text dates to 1998 (adopted) and 1999 (implemented) (p. 57). Note also that the table’s claim that pressure “stimulates substitutes” is not substantiated in the text.

References (pp. 61–63): notable points#

• The sources mix primary government and epidemiological records with secondary histories, an advocacy-commissioned study (Heerings 1999, for Greenpeace NL), a radio interview (Peto 1998), a business-magazine retrospective (Sells 1994), a draft conference paper (Castleman 2001) and a letter to the editor by a co-author (Greenberg 2000, “Trust me, I’m a doctor”).
• Citation errors:
• “Bartrip, P., 1931” is given as Medical History Vol. 42. The article is Bartrip 1998 (Medical History 42(4): 421–438; confirmed from the journal record); “1931” is the year of the regulations in its title. It is cited on p. 59 as “(Greenberg, 1994; Bartrip, 1931)”. The citation is also substantively misleading. Lesson 4 cites Bartrip among “others” who “have noted” that better controls should have come “before 1930”. Bartrip’s article argues the opposite. It sets out “to question the prevailing consensus” that UK regulation arrived “late”, contends that “there was no compelling medical or scientific evidence against asbestos and no sustained call for regulation until the late twenties”, and says the contrary view “rests heavily … upon the use of hindsight” (Bartrip 1998, pp. 421–422). Greenberg (this chapter’s co-author) and Nick Wikeley published a reply to Bartrip in Medical History 43(4), 1999, pp. 508–510. The chapter therefore cites, as support, a historian with whom one of its authors had publicly disagreed on this very point, and it does not tell the reader so.
• “Heuper” should be Hueper.
• “Ronald” (text, p. 54) should be “Reginald” Tage (reference list and the published title of Greenberg 1993).
• “WT/DST35/R” should be WT/DS135/R.
• The Hammond, Selikoff and Seidman reference lacks year and volume.
• The Knox 1952 and 1964 reports are Canadian mine visit reports, yet they are cited on p. 54 for Rochdale in-house cancer mortality studies. (Knox 1952 does fit the p. 60 quotation from Knox’s 1952 Canadian visit, which the text instead cites to Greenberg 2000.)
• Acheson and Gardner 1983 (HSE, The control limit for asbestos) is in the reference list (p. 61) but is not cited in the text.
• Summers 1919 is cited second-hand via Tweedale (p. 63).

Case timeline#

Key: W = warning; R = response; A = action; H = harm or cost. Strength assessments are mine, based on what the chapter reports.

Lags between credible warning and effective action (derived from the chapter’s dates): - Deane’s 1898 warning to the first regulation (1931): about 33 years. The regulation was narrow and unenforced until 1969. - The first quantitative study (Merewether & Price 1930) to regulation (1931): 1 year. This was fast, but the rules covered manufacturing processes only, not users, and there were two prosecutions in 37 years. - Doll’s 1955 lung cancer finding to UK compensation for lung cancer (about 1985): about 30 years, and only if asbestosis was also present (pp. 54–55). - Mesothelioma causal consensus (1964) to the first UK limits that addressed cancers (1984 at the earliest; the 1969 regulations “ignore … cancers”): about 20 years. This is an inference: the chapter does not say whether the 1979 recommendations or the 1984 limits were set with cancer in mind. To the UK ban (1999): about 35 years. To the EU-wide ban (2005): about 40 years. - First warning to ban: about 100 years in the UK (1898 to 1999) and about 90 years in France (1906 to 1997) (p. 53). - The Dutch counterfactual measures delay from “widely accepted” evidence (1965) to ban (1993): 28 years (p. 58). - The harm lag is separate from the policy lag: peak mesothelioma deaths are projected 50–60 years after peak use (p. 52). Even a ban at peak use would have been followed by decades of rising deaths.

What was known when (in the chapter’s account): - By 1911: “reasonable grounds for suspicion” that much dust was “to some extent harmful” (case reports, microscopy, animal studies). The chapter’s own wording is suspicion, not established knowledge. - By 1930: a quantified dose-duration relationship for asbestosis. - By the late 1930s to 1940s: suspicion of lung cancer, accepted as causal in Germany. - By 1955: a strong cohort finding for lung cancer. - By 1960–64: mesothelioma after brief and non-occupational exposure, with causal consensus by 1964. - By 1986: IARC’s view that all types are carcinogenic with no known safe level.

The authors’ own lessons and conclusions#

The ten lessons (pp. 59–61) are paraphrased below, with key phrases quoted, and each is classified as derived (it follows from evidence presented in the chapter), normative/recommendation (a policy prescription), or mixed.

1. Take lay and “competent observer” knowledge seriously (p. 59). The experiences of victims, lay people and “competent observers”, such as factory inspectors and family doctors, “should be taken seriously” and followed up by investigation; “They can anticipate the views of scientific experts, sometimes by many years.” - Classification: mixed. The observation is derived from Deane, the Women Inspectors, Joss, Sleggs and Tage (pp. 53–55); the prescription is normative.

2. Do the long-term surveys; build institutions for long-term monitoring (p. 59). The 1898–1906 warnings were not followed by the long-term medical and dust-exposure surveys “that would have been possible at the time”. The authors quote Peto (1999): the mesothelioma epidemic, “which far exceeds the combined effects of all other known occupational industrial carcinogens, cannot be adequately monitored.” And: “Long-term environmental and health monitoring rarely meets the short-term needs of anyone, thus requiring particular institutional arrangements”. - Classification: mixed. The failure to survey is derived. That such surveys “would have been possible at the time” is asserted. Deane’s proposals (p. 53) give it some support, but the chapter’s own quotation of Deane (p. 60) cuts the other way: she called reliable comparative mortality statistics “practically unattainable in the case of any given factory, or at any rate with the time and opportunity at present at our disposal”. Read together, the evidence supports “possible with dedicated resources that were not provided”, which fits the lesson’s institutional prescription better than its feasibility claim. The institutional prescription is normative.

3. Implementation and sanctions were weak (p. 59). The 1931–32 prevention and compensation laws “were not well implemented, and the sanctions were trivial, a pattern that was repeated down the long history”. - Classification: derived for poor implementation (two prosecutions, p. 56; Ombudsman report, p. 56; GBP 1 a week for widows, p. 57, which bears on compensation rather than sanctions). That “the sanctions were trivial” is asserted: the chapter gives no figures on penalties.

4. Earlier action would have avoided much loss, and tighter regulation would have driven innovation (p. 59). Heeding warnings “before 1930” (as Legge and others noted), or in the 1950s–60s “when new cancer hazards emerged and economic circumstances were good”, would have avoided “much tragic loss”. Controlling asbestosis before the cancers were discovered would have “minimised the impact of these later ‘surprises’”. More strategically, tighter regulation “would have raised its market price to capture more of its costs”, “stimulating the innovation that belatedly led to better and often cheaper substitutes”, and improved engine and building designs that produce less waste heat. - Classification: the first half is a counterfactual that the evidence strongly supports for the 1950s–60s window (dose-response, latency, Dutch estimate). The “before 1930” window is historically contested. Of the two works cited for it, Greenberg 1994 is by a co-author, and Bartrip (1998, misdated “1931”) argues the opposite: that there was “no compelling medical or scientific evidence … until the late twenties” and that earlier readings rely on hindsight (see References). The chapter presents a disputed historical judgement as an agreed one. The innovation claim is a hypothesis asserted with little evidence in the chapter; the only support is “substitutes … available by the 1970s” and cartels (p. 58). The claim that substitutes are “often cheaper” is unsupported.

5. Economic interests and externalities (p. 59). Economic factors “played a key role”. Employers’ need for profits and workers’ need for jobs “can together produce an alliance which may not be in the long-term interests of workers or society”. The larger the external cost, the more diverging private and social costs “will inhibit preventative action”. Remedy: “polluter pays”, liability, regulation and taxes to align private and social costs, “allowing the market place to operate more efficiently”, with penalties “commensurate with the costs they inflict on others”. But “It is very difficult for governments to overcome powerful economic interests that usually operate on the same short timescales as most politicians”. Institutional arrangements are needed for society’s long-term interests, a “governance” issue deferred to the report’s final chapters. - Classification: mixed. The externalisation is derived (profits, costs borne by others, p. 58). The jobs-profits alliance is not directly evidenced; the nearest material is the pensioners’ party anecdote (p. 60), in which workers themselves cited survivors as proof of safety. The policy prescriptions are normative and draw on standard environmental economics.

6. The “latency lacuna” and precaution for long-latency hazards (pp. 59–60). A main reason for failing to implement controls was the view that “current exposures … are so much lower than past exposures and should therefore be safe”. It is presented in quotation marks and attributed to Murray in 1906 “and repeated by many others ever since”, but it is a paraphrase, not the Murray words quoted on p. 53. With a 10–40-year latency, by the time evidence on “today’s” exposures arrives, exposures have fallen again, so it is “once again possible to say that risks in the new ‘today’ are much less than in the past, or non-existent”. “The point cannot be proved decisively one way or the other until another 20–40 years have passed.” This is the error of treating “absence of evidence of harm” as “evidence of absence of harm”. “It does not.” - Prescription: “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 where there is “no known threshold”. The measures should follow the proportionality principle: expected benefits of prevention, including “secondary benefits”, should be “significant in relation to the costs”. - Error balance: this approach would mean “switching the current bias within normal scientific methods away from avoiding ‘false positives’” towards a “better balance” between false positives and false negatives. The authors concede this “would increase the chances of generating the costs of restricting a substance or activity that might later turn out to be safe”. They nonetheless conclude that “the asbestos case strongly suggests that society would gain overall from a more ethically acceptable and economically efficient balance”. - Classification: the latency lacuna is a derived analytical point (p. 55, p. 56 Knox et al.). The precautionary default and the error-rebalancing are normative. The rebalancing is argued from a single, extreme false-negative case, which the authors partly acknowledge with “strongly suggests”.

7. The healthy survivors fallacy (p. 60). This fallacy “gives rise to a general but false reassurance of safety, as it does with the general hazard of smoking”. It was first described for asbestos by Deane in 1898: there is “always a certain proportion of ‘old workers’ — the survivors of their mates” who “appear to thrive on their unhealthy calling”. Deane added that the only convincing proof, “reliable comparative statistics of mortality”, is “practically unattainable” for any given factory. Examples: - Knox (Turner company doctor) in 1952 at the Canadian mines: “I am assured that many workers over 70 years of age are still employed and are active and vigorous” (Greenberg 2000). Note that Knox was relaying what he had been told. - Gee in the 1980s as union adviser: workers pointed to retirees at the annual pensioners’ party as proof of low risk. Hence the “‘pensioners’ party fallacy’, as it was the workers who did not make it to the party who provided the proof of harm”. Healthy survivors “needed to be related to non-survivors via appropriately analysed mortality statistics”. - Classification: derived (a documented recurring argument) plus a communication recommendation (“needs to be widely communicated”).

8. Speedy, affordable, transparent compensation, anticipatory where possible (pp. 60–61). The lesson is hedged (“It seems necessary”): establish these arrangements “based on agreed liabilities, as soon as any harmful effects become known”. The aims are to increase incentives for prevention and “improve the chances of recording accurate exposure histories”. The authors cite anticipatory elements in the early nuclear industry: state assumption of accident liabilities up to limits (UK Nuclear Installations Act 1965), and what “seems to be” a “unique example”: a radiation-induced cancer compensation scheme for British Nuclear Fuels workers (cross-referencing the Radiation chapter). - Classification: normative. Its rationale draws on the compensation failures (GBP 1 a week, p. 57) and exposure-history reconstruction problems (Wagner’s reconstruction, p. 55). The chapter offers no evidence on how well the nuclear schemes worked.

9. Draw on a wide range of disciplines; curb the “ignorant expert” (p. 61). Specialists in one discipline, “for example medicine”, gave “expert” opinions on others, such as dust monitoring and control (occupational hygiene, ventilation engineering) or the availability of substitutes. “These opinions were often mistaken but went largely unchallenged, and this contributed to misplaced complacency” (Greenberg 2000). - Classification: mixed. The one worked example is the Lancet 1967 editorial (p. 58). The general claim rests on a co-author’s letter.

10. Anticipate “surprises” and take care with substitutes (p. 61). If substitutes reproduce asbestos’s physical form, “long, respirable (< 3 microns in diameter) and durable fibres”, they are “likely” to be carcinogenic too (Roller & Pott 1998). The UK HSE predicted this in 1979, and IARC later confirmed it “for some forms of synthetic mineral fibres”. But mineral wool and glass fibre “appear to be much less hazardous”, and they can be made “good enough for insulation but not as thin, or durable enough in human tissue, to be carcinogenic”. Therefore “‘Clean’ production and user techniques that minimise exposures … via ‘closed loop’ and eco-efficient systems, are therefore essential with whatever materials are being used”. This “minimises the size of any future ‘surprise’ impacts from substitutes, which is an important benefit of applying the precautionary principle.” - Classification: mixed. The fibre-paradigm prediction is evidence-based. The “closed loop” prescription is normative and characteristic of the EEA and eco-efficiency discourse of the time. The WTO Appellate Body took a compatible line (AB report para. 168, not cited in the chapter): substitute fibres carry a “less” but not demonstrably “non-existent” risk, and it is “perfectly legitimate” to halt a highly risky product while allowing a less risky one in its place.

Other conclusions embedded in the narrative (not in the numbered list): - “Latency lacuna” as “a major reason why preventative action is often too late” (p. 55). - The limited statistical power of “negative” studies of long-latency cancers (p. 55). - Failure to consider “worst case” exposure scenarios (users rather than producers) as part of the reason for delay (p. 56). - “No evidence of harm” is not “evidence of no harm” (p. 53). - Market-price failure as “the common cause” of delay in replacing hazardous materials (p. 58). - The “limited relevance of more research” to prevention (p. 58). - Compensation trials as a source of truth about company conduct (pp. 58–59). - The WTO asbestos ruling’s “implications for other hazardous agents” (p. 57).

Mechanisms and dynamics#

1. How warnings arose#

• From people close to the harm. Warnings came from people who were close to it and outside the core expert establishment: women factory inspectors (p. 53), a family doctor seeing 10–12 cases a year (p. 54), a local South African doctor noticing an unusual cancer cluster (p. 55), a freelance union investigator (p. 54), patients’ own testimony (p. 53), and families and colleagues whose accounts let Wagner reconstruct exposure (p. 55). Formal science often lagged, confirmed, or was pulled in by these observations.
• Through independent access routes. The decisive epidemiology came from researchers “independently of the industry, using case data from unions and hospital records” (p. 55). When Selikoff was refused company records, union records supplied the denominator (p. 55). So the institutional location of data, and who controlled access to it, shaped which evidence could exist.
• From actors with financial exposure. Insurers declined cover in 1918 (p. 54). This is a market signal that functioned as a warning, though it protected the insurers rather than the workers. The chapter says it was “later forgotten”, without saying by whom or how; a loss of institutional memory among insurers is the natural reading but is not spelled out.

2. How warnings were contested, ignored or suppressed#

• Ignoring rather than refuting. The authors say this of the Women Inspectors’ reports (p. 53), and call the French report “largely ignored”. The wider official picture is more nuanced than “ignored”, however. The Factory Department pressed for exhaust ventilation after 1911 and then held inquiries (1912, 1917) that judged the evidence insufficient (p. 54). The more precise mechanism is that an evidentiary threshold for further action was set above what was then available, and that nobody commissioned the studies that could have met it (p. 53: no follow-up of Murray’s patient’s co-workers; Deane’s proposed statistics not collected).
• Reassurance from assumed improvement. Murray in 1906 said “considerable trouble is now taken” (p. 53). Knox et al. in 1965 said hazards were possibly “completely eliminated” (p. 56). Both rested on the belief that controls in the monitored part of the system had solved the problem. The 1969 hygiene standard (p. 56), which addressed asbestosis only, can arguably be read in the same light, though the chapter criticises it for its narrow endpoint rather than for complacency about improvement.
• Weak in-house science. Three in-house investigations failed to find lung cancer, and the company doctor admitted his knowledge of statistics was “nil” (p. 54). The chapter implies, but does not state, that he conducted them. Industry held unpublished animal cancer data (p. 54).
• Suppression attempts. Turner directors tried to suppress Doll’s findings (p. 54). The mechanism is not described.
• Denial of data access. Selikoff was refused company records (p. 55).
• Marginalising messengers. Selikoff was a “disturbing sore thumb” (p. 55). Legge suggested the TUC pay Tage a small fee to be “quit of” him (p. 55). The Ibsen analogy frames this as a general social pattern driven by the local economy’s dependence on the hazard (pp. 55–56). The Legge example matters because Legge was not an industry actor. He was a former chief regulator, working for the unions, and the author of the epigraph lamenting missed opportunities. Gatekeeping of outsiders by reform-minded professionals is a distinct mechanism from industry suppression. The chapter records it but does not analyse it.
• Dismissing affective or public evidence. Doll, the independent epidemiologist who had established the lung cancer risk, criticised the 1982 documentary, in the authors’ paraphrase, “for being unscientific and emotional” (p. 57; not a direct quotation of Doll, and uncited), yet the chapter credits that documentary with triggering action. This shows a tension between scientific norms of evidence and the political effectiveness of individual stories. The same scientist can be a key source of warning in one period and a critic of how the warning is communicated in another.

3. Uncertainty, ignorance and the burden and standard of proof#

• Genuine uncertainties. The chapter acknowledges several: diagnostic confounding with TB, heart failure and pneumonia (p. 54); confounding and synergy with smoking (pp. 54–55); fibre-type potency differences (p. 57); unresolved mechanisms and dose-response (p. 58); detection limits (p. 57).
• Structural sources of false negatives:
• latency (the latency lacuna, pp. 55, 59–60);
• low power of studies with short follow-up (p. 55);
• healthy-survivor selection (pp. 54, 60);
• sampling from the wrong population: producers in regulated areas rather than users (p. 56);
• measurement not seeing the relevant agent: fibrils need electron microscopy (p. 57);
• standards covering only the first-recognised endpoint: the 1969 hygiene standard covered asbestosis only (p. 56). Each of these biases pushes toward “no evidence of harm”. Together they make a system in which reassurance is structurally more likely than alarm.
• Standard of proof. The authors argue that conventional science’s bias against false positives produced false negatives. They advocate rebalancing (p. 60). The WTO Box shows a legal forum accepting qualitative evidence, reliance on respected minority scientific opinion, and the inadequacy of “controlled use” (p. 57). The authors read this as a “lower level of proof” than preponderance, which is their gloss.
• Burden of proof. This is implicit throughout. Harm had to be demonstrated before controls were imposed (1906, 1912, 1917), while safety was assumed from improvement (1906, 1965). The authors’ prescription reverses the default for carcinogens without a threshold: “assume they are unsafe” absent good evidence of safety (p. 60), tempered by proportionality.
• Divergent weighting across jurisdictions. Germany accepted causation in 1938 and compensated from 1943 (p. 54). The UK accepted lung cancer compensation about 30 years after 1955 and only with asbestosis (pp. 54–55). The UK held at least as much evidence yet acted far later, which suggests evidentiary thresholds are partly institutional choices. The comparison is not clean, though: the chapter itself notes that the German decision came “when lung cancer was generally much less prevalent”, and that rising smoking-related lung cancer later made the asbestos link “that much harder to prove” (pp. 54–55). So part of the difference lies in the evidence situation, not only in the threshold.

4. Industry, incentives and conflicts of interest#

• Control of knowledge production. Industry commissioned studies (in-house; Doll), controlled records (Selikoff refused, p. 55), held unpublished animal data, and attempted suppression (p. 54).
• Cartels inhibiting substitutes (p. 58).
• Profits. Profits rose through 1947–1965 while health costs were externalised (p. 58).
• Self-defeating defensiveness. Sells’s retrospective (p. 56) says that “medical research, assiduous communication, insistent warnings and a rigorous dust reduction programme” “would probably have saved the stockholders, the industry and … the product”. The mental model of the industry leadership is portrayed, through this insider’s hindsight, as short-termist and defensive in a way that proved very costly to the firms and their insurers: Manville’s bankruptcy filing (while still profitable), Turner’s arrangements to pay up to GBP 1bn, Lloyd’s near collapse. This is an important nuance. The chapter frames precaution as a potential private good as well as a social one.
• Industry withdrawals ahead of regulation. Industry had already withdrawn blue asbestos, and insulation spraying was “largely defunct”, before the 1979 Simpson recommendations (p. 56). The chapter does not say why. On these two points regulation codified moves already made; on licensing and exposure limits it went further.

5. Regulators and institutional behaviour#

• First mover, weak enforcer. The UK was the world’s first to regulate (1931, p. 54) but enforced weakly: two prosecutions in 37 years (p. 56), with trivial sanctions (p. 59). Early regulation is not effective regulation, and a regulatory “first” can coexist with decades of harm.
• Scope set by where regulators looked. Rules covered parts of manufacturing and neglected users (pp. 54, 56). Regulators and researchers attended to the visible, bounded workplaces (factories) rather than the dispersed and harder-to-monitor downstream users (insulators, builders, brake workers), who had the worst exposures.
• Accountability came from outside. It came through an MP’s complaint and the Parliamentary Ombudsman (p. 56), media (pp. 56–57) and victims’ advocates (p. 57), not from internal regulatory review.
• Standards shaped by measurement capability. The HSE’s recommended public limit was pegged to optical microscopy’s detection limit (p. 57), and the 1969 standard reflected one endpoint (p. 56). What the regulator could measure shaped what it recommended or regulated.

6. Scientists and experts: mental models and blind spots#

• Wrong reference population. Researchers studied producers in regulated areas instead of users: the “stupid mistake” (p. 56).
• Overreach outside expertise. Doctors pronounced on engineering and on substitutability (Lancet 1967, p. 58; Lesson 9, p. 61).
• Aggregate benefit reasoning. The Lancet said asbestos “can save more lives than it can possibly endanger” (p. 58). This is a confident risk-benefit framing made without data on either side, and it shows how an apparently rational benefit argument can entrench a hazard.
• Faith in improvement. Improvement in current conditions was taken as evidence of safety (latency lacuna, pp. 55–56, 59–60).
• Research as a substitute for action. “Huge amounts of research” left mechanisms and dose-response unclear (p. 58). The authors imply that waiting for mechanistic clarity is itself a choice with consequences.

7. Workers, lay publics, media and courts#

• Workers’ own reassurance. Workers used the pensioners’ party to reassure themselves (p. 60), and the jobs-profits alliance could align worker and employer interests against prevention (p. 59). The chapter treats worker knowledge as both a source of warning (Murray’s patient, p. 53) and of false reassurance (pensioners, p. 60).
• Media as agenda-setter and trigger. The media acted from 1964 to 1982 (pp. 56–57), and the Alice documentary is presented as the proximate cause of the 1984 regulations.
• Victims’ advocates. Nancy Tait exposed static compensation (p. 57).
• Courts and litigation. Inquests produced the first pathology (p. 54). Compensation trials surfaced company documents (pp. 58–59). Environmental exposure cases succeeded about 30 years after Newhouse (p. 56). The 2000 surgeon award (p. 52) shows litigation extending to non-occupational victims. Liability eventually led Johns Manville, “although still profitable”, to file for bankruptcy as a way of handling claims, and brought Lloyd’s near collapse (pp. 56, 58).
• Trade law as a counter-pressure. A producer country (Canada) used trade law to challenge importers’ bans (p. 57). The WTO outcome is presented as legitimising precautionary health measures under trade rules.

8. Economics, externalities, lock-in and substitutes#

• Low prices from externalised costs slowed substitution (p. 58).
• Cartels actively inhibited alternatives (p. 58).
• The “irreplaceable” framing (Lancet 1967) versus the availability of substitutes “by the 1970s and, in some cases, much earlier” (p. 58). The chapter does not engage with the transition costs or technical performance gaps that France’s temporary exemptions imply (Box 5.1, p. 57).
• Lock-in through the built environment. Asbestos embedded in buildings becomes a long-term hazard and cost: pipe insulation in hospital tunnels (p. 52), and “Removing asbestos from buildings safely at the end of their life will cost further billions” (p. 58).
• Innovation. The authors argue that tighter regulation would have spurred substitutes and better designs (p. 59). Table 5.1 claims that 1980s pressure “stimulates substitutes” (p. 61). Both are asserted rather than demonstrated.

9. Time lags, irreversibility and distribution#

• Lags compound. There were three layers:
• a policy lag of decades between warning and effective action;
• a biological lag between exposure and disease of 20–25 years for lung cancer and about 40 for mesothelioma (p. 55), given as “10–40” years in Lesson 6 (p. 60);
• a use-to-harm lag of 50–60 years between peak imports and peak mesothelioma (p. 52). Harm is therefore locked in long before it is visible. That is the chapter’s implicit case for acting under uncertainty.
• Distribution of benefits and harms:
• Benefits went to firms, shareholders and dividends (p. 58); to workers through jobs (pp. 58, 59); and to consumers and the public through fire safety, energy saving and brakes (p. 58).
• Harms fell on workers (especially downstream users), families through domestic exposure (p. 56), neighbourhoods near factories (p. 56), children on mine dumps (p. 55), building occupants such as the surgeon (p. 52), the health service, insurers and building owners (p. 58), future generations (the mesothelioma peak decades later, p. 52) and developing countries where use continues (pp. 52–53).
• The dose needed for mesothelioma, “a matter of months” (p. 55), means that low-exposure bystanders can bear serious harm. This widens the circle of risk-bearers well beyond those who chose or were paid for the exposure.

10. Complexity and interaction#

• Synergy with smoking. Multiplicative interaction with smoking (p. 55) produced both a larger combined risk and an attribution problem that delayed compensation (pp. 54–55).
• Multiple pathways and populations. Occupational, para-occupational (domestic), environmental and building-related pathways (pp. 52, 55, 56) meant that any study of one pathway understated the total.
• Global systems. Use shifted geographically: asbestos use “is continuing, now largely in developing countries”, and the histories “are now being repeated … in Asia, Africa and South America” (pp. 52–53). Insurance markets transmitted liabilities internationally (Lloyd’s and US claims, p. 58).

11. Framing and language#

• “Magic mineral” (p. 53): the language of wonder and versatility attached to the material early.
• “Evil effects” (Deane, p. 53): a contemporary moral framing of harm.
• “Hygiene standard” (p. 56): a label implying safety while covering only one disease endpoint.
• “Controlled use” (Box 5.1, p. 57): the producer-country framing that risk can be managed without prohibition, rejected by the WTO Appellate Body as not demonstrated.
• “Irreplaceable” (Lancet, p. 58): a framing that closes off alternatives.
• “Unscientific and emotional” (the authors’ paraphrase of Doll’s criticism of Alice, p. 57, not Doll’s own words): boundary-work separating legitimate from illegitimate evidence.
• “Disturbing sore thumb” and “quit of” (p. 55): delegitimising messengers.
• Terms the authors explicitly coin: “latency lacuna” (“which might be called”, p. 55) and “pensioners’ party fallacy” (“could be called”, p. 60). Other terms they put in scare quotes: “ignorant expert” (p. 61), “surprises” (pp. 59, 61), “false negatives” (pp. 53, 60). The authors’ own title term “malevolent” is also a framing device.

Transferable insights (technology-neutral)#

Ratings: strong means well documented in the section and consistent with established methodology; moderate means supported by examples but with a thin or secondary evidence base, or a generalisation beyond the case; suggestive means one or two anecdotes or an analogy; asserted means claimed without supporting evidence in the section.

1. Proximate observers detect harm before formal experts do, and their reports tend to be set aside rather than tested. Inspectors, clinicians, affected workers and families recorded harm years or decades before expert consensus. The failure was not refutation but a lack of follow-up investigation. (pp. 53–55, 59 L1) - Rating: Strong for this case from the 1920s onwards, with multiple independent instances over 60 years. For the 1898–1910 warnings the rating should be tempered: whether they amounted to actionable evidence is disputed by at least one historian the chapter itself cites (Bartrip 1998; see References), and Deane herself judged the decisive statistics “practically unattainable” with the resources then available (p. 60). Only moderate as a general law, because the chapter cannot show how many lay warnings elsewhere proved wrong.

2. Latency plus improving conditions creates a self-renewing “all clear”. When harm appears decades after exposure and exposure conditions keep changing, evidence about current conditions always arrives too late. This allows repeated claims that today’s exposures are safe. Absence of evidence of harm is systematically mistaken for evidence of absence. (pp. 53, 55, 56, 59–60 L6) - Rating: Strong. The logic is structural, and the chapter documents repeated use of the argument (1906, 1965).

3. “Negative” studies of slow-emerging harms are often underpowered by design. Without follow-up as long as the latency period, null results are uninformative, but they function as reassurance. (p. 55) - Rating: Strong. This is a standard methodological point, illustrated by the 25-year follow-up needed in Selikoff’s data.

4. Survivor visibility produces false reassurance. Those still present (current workers, retirees at the party) are the survivors. The harmed have left and become invisible. Surveys of current populations understate harm. (pp. 54, 60 L7) - Rating: Strong. This is a well-established selection bias, recurring in 1898, 1930, 1952 and the 1980s.

5. Studying and regulating the best-controlled part of a system misses the worst exposures. Evidence and rules focused on bounded, monitored producers, while dispersed downstream users had the highest exposures. Apparent improvement in the monitored segment masked continued harm elsewhere. (pp. 54, 55, 56, 61) - Rating: Strong within the case: Selikoff’s data, Peto’s judgement, and the scope of the 1931 and 1969 rules.

6. Whoever controls data and funds studies shapes what evidence exists. Firms controlled records, commissioned weak in-house studies, held unpublished adverse findings, and tried to suppress an independent study. Decisive evidence came from researchers independent of the industry, using alternative data (union and hospital records). (pp. 54, 55, 58–59) - Rating: Moderate. The instances are documented but described briefly, and they rest on secondary historical sources (Tweedale, Castleman, Scheper).

7. Bearers of unwelcome news are marginalised, sometimes by reformers as well as incumbents. (p. 55 “sore thumb” and “quit of”; pp. 55–56 Ibsen) - Rating: Suggestive. There are two anecdotes plus a literary analogy.

8. Rules without enforcement, and penalties far below the harm caused, give the appearance of control without the substance. (pp. 54, 56, 59 L3, L5) - Rating: Strong for non-enforcement in the UK case: two prosecutions in 37 years, and an Ombudsman report critical of the inspectors. Asserted for the claim that penalties were far below the harm: the chapter calls sanctions “trivial” (L3) but gives no penalty figures.

9. When harm costs are borne by others, the incentive to prevent harm and to adopt safer alternatives is weak. Low prices that exclude health costs slow substitution. Private and social costs diverge most where external costs are largest. (pp. 58, 59 L5) - Rating: Moderate. The theory is robust and the case shows externalisation (profits versus costs borne by workers, the health service, insurers and building owners). But the claim that this is “the common cause” of delay is asserted, and the chapter does not quantify how much internalisation would have changed behaviour.

10. Incumbent coordination and claims of “irreplaceability” can delay available alternatives. (p. 58)

    • Rating: Moderate. The evidence is the Castleman citation on cartels and the mineral wool refinery example. The chapter does not engage with genuine performance gaps; France’s own temporary exemptions (p. 57) show some remained.

11. Experts who speak outside their competence can entrench complacency. Clinicians pronounced on engineering, exposure control and substitutability, largely unchallenged. (pp. 58, 61 L9)

    • Rating: Moderate. There is one clear quoted example (Lancet 1967), and the general claim is cited to a co-author’s own letter.

12. What can be measured, and which harm was recognised first, shape the standard. Limits were pegged to the available detection technology (the recommended public limit) or to the first-recognised endpoint (the 1969 standard). Later-recognised, more serious endpoints went unregulated. (pp. 56–57)

    • Rating: Moderate. Two concrete instances: the 1969 standard ignored cancers, and the HSE’s recommended public limit was pegged to the optical detection limit.

13. Interacting risk factors can both multiply harm and obscure attribution. A co-exposure that multiplies risk also provides an alternative explanation, delaying recognition and compensation. (pp. 54–55)

    • Rating: Strong for the synergy data. Moderate for the claim that it caused the delay: the chapter says “partly because”.

14. Public stories, political complaint and victim advocacy can move regulation where accumulated evidence did not. (pp. 56–57, 61)

    • Rating: Moderate. The temporal sequence is documented (Ombudsman leading to Simpson; documentary leading to the 1984 rules), but the causal attribution is asserted. The chapter also records a leading scientist dismissing the key programme.

15. Early precautionary signals from actors with money at stake can be lost through institutional forgetting. (p. 54)

    • Rating: Suggestive. It rests on a single source (Hoffman 1918), and “later forgotten” is asserted without mechanism. Such signals can also be purely self-protective: declining cover shielded the insurers without reducing anyone’s exposure.

16. Harm can peak decades after use peaks, so the decisions that matter are made long before harm is visible, and the legacy persists in infrastructure. (pp. 52, 55, 58)

    • Rating: Strong for the latency and legacy mechanism (Figure 5.1, latency data). The specific projected magnitudes are model outputs to be checked later.

17. Delay has large costs, and those costs can fall on the delaying industry itself. Earlier action would have saved lives and money. An industry insider concluded that research, communication, warnings and rigorous dust reduction would probably have saved the stockholders, the industry and the product. (pp. 56, 58, 59 L4)

    • Rating: Moderate. The direction is well supported by bankruptcies, liabilities and Lloyd’s. The magnitudes are crude or advocacy-sourced (EUR 1m per life, undiscounted; a Greenpeace-commissioned Dutch estimate), and the counterfactual is untestable.

18. Legal forums can accept qualitative evidence, respected minority scientific opinion, and doubt about “controlled use” as grounds for restriction. (p. 57, Box 5.1)

    • Rating: Strong that the WTO Appellate Body reasoned along these lines. Checked against WT/DS135/AB/R: qualitative risk evaluation (para. 167), reliance on “divergent, but qualified and respected” opinion and no obligation to follow “majority scientific opinion” (para. 178), and “controlled use” not a reasonably available alternative (para. 174). The AB itself said a panel “need not, necessarily” decide on the “preponderant” weight of the evidence (para. 178). Asserted only as to the Box’s further gloss that this means “a lower level of proof”, and as to any general precedent beyond this dispute.

19. Rebalancing the tolerance for false positives against false negatives would yield net social gain. (p. 60 L6)

    • Rating: Asserted (normative). It is derived from one extreme false-negative case. The authors acknowledge the cost of false positives but present no false-positive case.

20. More research does not automatically resolve decision-relevant uncertainty. Mechanisms and dose-response can remain unclear after decades of study, so waiting for research is itself a decision. (p. 58)

    • Rating: Suggestive. It is a one-sentence claim, but consistent with the fibre-potency controversy “still” unresolved (p. 57).

21. Substitutes can reproduce the hazard if they reproduce the hazardous property. Selecting and designing alternatives by the property that causes harm, and minimising exposure whatever the material, limits future “surprises”. (p. 61 L10)

    • Rating: Moderate. It is supported by an anticipatory HSE prediction (1979) and partial IARC confirmation, while noting that some substitutes appear much less hazardous.

22. Long-term monitoring serves no one’s short-term interest and needs dedicated institutions. Without it, even a major epidemic “cannot be adequately monitored”. (p. 59 L2)

    • Rating: Moderate. It is supported by the Peto quotation and the failure to follow up in 1906, but no institutional comparison is offered.

23. Pre-agreed, fast compensation can create prevention incentives and preserve exposure records. (pp. 60–61 L8)

    • Rating: Asserted. It is a recommendation by analogy with nuclear liability schemes, without evidence on outcomes.

24. Employer and worker interests in continued activity can align against prevention. Short political and economic time horizons compound this. (p. 59 L5; p. 60)

    • Rating: Suggestive. There is one first-hand anecdote (pensioners’ party), otherwise asserted.

25. Hazards restricted in one jurisdiction can migrate to others. Use shifts to places with weaker controls, and producer jurisdictions may resist restrictions through trade mechanisms. (pp. 52–53, 57)

    • Rating: Asserted for the migration claim within the chapter (no data are given). The trade challenge is documented (Box 5.1).

26. Similar evidence can be weighed differently across jurisdictions. The threshold for action is partly an institutional choice. (p. 54: Germany 1938/1943 versus the UK in the 1980s)

    • Rating: Suggestive to moderate. It is one clear comparison, but the chapter does not explain the difference, and it itself points to a confounder: German recognition came when background lung cancer was “much less prevalent”, before smoking made attribution harder (pp. 54–55).

Limitations, contestation and bias check#

Advocacy versus analysis. - The chapter is explicitly a lesson-drawing narrative by participants with strong views, disclosed at report level (p. 12). One author was, at the time of writing, EEA staff and one of the report’s own editors, a former environmental NGO director with a union background; the other is a former factory medical inspector who also writes on asbestos history. - Industry, regulators and the medical establishment appear almost entirely as obstacles. No one who defended asbestos use, the “controlled use” position or the chrysotile-potency argument is given a sympathetic hearing. Their positions are reported mostly in order to be rebutted: Murray, Knox, the Lancet, Doll on Alice. - The lessons mix derived findings with normative programmes: polluter pays, closed-loop production, error rebalancing, anticipatory compensation.

Hindsight bias. - Calling the 1898–1910 observations “competent” warnings that “would have done credit to a medical scientist” (p. 53) partly reflects knowledge of the outcome. At the time many dusty trades caused lung disease, TB confounded diagnosis (acknowledged on p. 54), and Murray’s case rested on hearsay he himself flagged (p. 53). - In fairness to the authors, they cite near-contemporary judgements that point the same way: the “reasonable grounds for suspicion” drawn from the 1910–11 evidence (Merewether 1933; Table 5.1 dates the judgement to 1911) (p. 54), and Legge’s 1934 “badly missed” (p. 52). So the missed-opportunity claim is not purely retrospective. Note, though, that the chapter’s own phrasing for that period is “suspicion”, not proof. - The chapter’s own quotation of Deane (p. 60) shows that she judged “reliable comparative statistics of mortality” to be “practically unattainable … with the time and opportunity at present at our disposal”. That tempers the claim (Lesson 2) that the needed surveys “would have been possible at the time”. - The early-warning reading is contested in the historical literature, and the chapter does not say so. Bartrip (1998), whom the chapter cites in support of pre-1930 action (Lesson 4), argues that “there was no compelling medical or scientific evidence against asbestos and no sustained call for regulation until the late twenties”, and that “cobbling together various references” to suggest early official knowledge “is misleading and rests heavily … upon the use of hindsight” (pp. 421–422). Greenberg, the chapter’s co-author, co-wrote a published reply (1999). The chapter takes one side of a live scholarly dispute without flagging it. - The Factory Department did something after the 1910–11 evidence: it pressed for exhaust ventilation (p. 54). Strictly, “not refuted but simply ignored” (p. 53) refers to the Women Inspectors’ reports, but the chapter’s overall framing of early official inaction sits in some tension with its own account of this response.

Case selection and generalisation. - Asbestos is arguably the paradigm false negative, with very large harm, a strong dose-response, a distinctive signature disease (mesothelioma) and clear substitutes. Generalising its lessons to “all long-latent-period hazards” (pp. 55, 60) risks overreach. - The report’s case studies are all false negatives (p. 12). In fairness, the editors say they tried to include false positives, inviting industry representatives to submit examples, but “no suitable examples emerged” that were robust enough (p. 12). Even so, the chapter’s argument for rebalancing false positives and false negatives (p. 60) is made without any counterexample. The authors acknowledge the cost of false positives but do not weigh any. - The UK focus (p. 53) means other national trajectories are asserted to be “similar” without evidence. This includes US regulatory history. The US Environmental Protection Agency’s 1989 asbestos phase-out rule and its 1991 court reversal (Corrosion Proof Fittings v. EPA) are not mentioned, even though they are directly relevant to burden-of-proof and cost-benefit themes (noted from general knowledge).

Thin or secondary evidence. - Key claims rest on secondary histories: suppression of Doll, cartels, company doctors, “disturbing sore thumb”. These are Tweedale and Castleman; Castleman was also an adviser in the WTO case (p. 58). - Some claims rest on co-author self-citation (Greenberg 1993, 1994, 2000). - The cost figures come from an NGO-commissioned study (Heerings 1999, for Greenpeace NL) and a crude, undiscounted value-of-life multiplication. The Turner GBP 1bn and Lloyd’s near-collapse statements carry no citation, and the one quantified benefit (theatre fires) is cited second-hand via Tweedale and hedged “it was claimed”. - The US figure (“USD 2 billion”, p. 58) is ambiguous and probably a large understatement of total US asbestos compensation. From general knowledge, later estimates of cumulative US asbestos litigation costs run to tens of billions of dollars by the early 2000s. To verify. - Causal attributions of regulatory change to media events (“The government responded to the programme”, p. 57) are plausible but not documented. - Lesson 4’s innovation claim is asserted (substitutes “often cheaper”).

Benefits and substitutes under-examined. - Benefits get one paragraph, with “it was claimed” hedges (p. 58). - The Lancet argument is dismissed on competence grounds rather than engaged on substance. - France’s 1997 ban allowed temporary exemptions where no effective lower-risk substitutes existed (Box 5.1, p. 57). That implies “irreplaceability” was not entirely empty even in the late 1990s, a tension the chapter does not address. - Risks of substitutes are acknowledged (Lesson 10), which is to the authors’ credit.

Fibre-type controversy. - The chapter notes that controversy over the relative potency of chrysotile, amosite and crocidolite “was, and still is” live (p. 57), but relies on IARC’s “no known safe level” and the WTO findings. - It does not set out the evidence for large potency differences, especially for mesothelioma. That evidence was the scientific basis of the “controlled use” position. From general knowledge, a widely cited 2000 potency analysis by Hodgson and Darnton, not cited in the chapter, estimated much lower mesothelioma potency for chrysotile than for the amphiboles. - A fair reading: the chapter’s policy conclusion (ban all types) does not depend on equal potency. It depends on chrysotile being carcinogenic without a demonstrated threshold, and on “controlled use” being ineffective in practice. The chapter asserts the latter via the WTO.

Internal inconsistencies and factual flags (to check before relying on specifics): - Dates. 1912 versus 1911 (p. 54 versus Table 5.1). Mesothelioma dates in Table 5.1 (1959–60, 1962/64) versus the text (1955/1960, 1964/65). France’s ban 1997 (Box) versus “1998–99” (Table). “Doll … British doctors in 1955” (p. 55); Doll and Hill’s findings date from 1950/1954. IARC “by 1986” (p. 57) is not wrong but understates how early IARC acted: the WTO record cites IARC recognition of chrysotile as carcinogenic “Since 1977” (AB report, footnote citing the Panel), and IARC’s Group 1 re-evaluation was in 1987 (general knowledge). - Production figure. “Grew to 2 million tonnes in 1998” (p. 52). USGS Circular 1298 (Virta 2006) reports peak world production of “almost 4.8 Mt per year” around 1977, and that annual world production “by 1965 had already exceeded 2” Mt. So “grew” misdescribes the trend: by 1998 production had fallen to well under half its peak, roughly back to its mid-1960s level. - Names. “Ronald” (p. 54) should be “Reginald” Tage (p. 62; confirmed by the published title of Greenberg 1993). “Heuper” for Hueper. - References. Bartrip dated 1931 but in Medical History Vol. 42 (p. 62); confirmed as 1998. More importantly, Bartrip is cited (p. 59) in support of a view his article disputes (see References above). Knox 1952 and 1964 references do not match the Rochdale in-house studies they are cited for (pp. 54, 62). “WT/DST35/R” is a typo. - Box 5.1 (verified against WT/DS135/AB/R, 12 March 2001). The statement that “The US cross-appealed against the panel’s judgement that glass fibres were as carcinogenic as asbestos” is wrong. The US was a third participant (AB report p. 1), not a cross-appellant. The Panel made no finding that glass fibres were equally carcinogenic: it declined to consider health risk in the likeness analysis (Panel paras 8.130, 8.132). The US argued that this was an error (AB para. 48), and the AB agreed and reversed (paras 114, 141, 147). The Box’s “they were not [like], principally because they were not as carcinogenic” simplifies an AB finding that Canada had “not satisfied its burden of proving” likeness. The “preponderant weight of the evidence” wording tracks AB para. 178; only “a lower level of proof” is the Box’s gloss. The other Box points check out (paras 157, 167, 174, 178). - Lesson 6 quotation. Lesson 6 puts in quotation marks a sentence (“current exposures … should therefore be safe”) attributed to Murray in 1906 that is a paraphrase, not the words quoted on p. 53.

Fairness in the other direction. - The chapter is candid about genuine scientific complications (TB, smoking, fibre types, measurement) and about asbestos’s benefits. - It notes that the UK regulated first in the world (p. 54). - It records industry moving ahead of regulators in withdrawing blue asbestos and spraying (p. 56). - It includes an industry insider’s view that precaution was in industry’s own interest (p. 56). - It acknowledges the proportionality principle and the costs of false positives (p. 60). - Its central analytical contributions (latency lacuna, survivor bias, worst-case neglect, underpowered negatives) are methodologically sound and not merely rhetorical.

Dissent within the text. - There are no panels. Differences of view surface through reported actors: Doll’s criticism of Alice (p. 57), the Lancet’s 1967 position (p. 58), Knox et al.’s 1965 optimism (p. 56), and the Canadian “controlled use” case (p. 57). All are presented and then rebutted by the authors. - One scholarly dissent is present only in the reference list, and inverted: Bartrip (1998), a revisionist account of the 1931 regulations that argues against the “too late” reading of the pre-1930 period, is cited as if it supported that reading (p. 59).

Notable quotes#

1. “Looking back in the light of present knowledge, it is impossible not to feel that opportunities for discovery and prevention of asbestos disease were badly missed.” (Legge, 1934, epigraph; p. 52)
2. “Their reports were not refuted but simply ignored.” (on the Women Factory Inspectors; p. 53)
3. “Dr Murray’s view that ‘no evidence of harm’ is the same as, ‘there is evidence of no harm’, is an early example of a common fallacy” (p. 53)
4. “This problem, which might be called the ‘latency lacuna’, characteristic of all long-latent-period hazards under conditions of technological change, is a major reason why preventative action is often too late.” (p. 55)
5. “could have saved lives and would probably have saved the stockholders, the industry and, for that matter, the product” (Sells, 1994, former Johns Manville executive; p. 56)
6. “it would be ludicrous to outlaw this valuable and often irreplaceable material in all circumstances (as) asbestos can save more lives than it can possibly endanger” (Lancet, 1967; p. 58)
7. “This failure of market prices to reflect full environmental and health costs is the common cause of delay in replacing hazardous materials.” (p. 58)
8. “Long-term environmental and health monitoring rarely meets the short-term needs of anyone, thus requiring particular institutional arrangements” (p. 59)
9. “it was the workers who did not make it to the party who provided the proof of harm” (on the “pensioners’ party fallacy”; p. 60)
10. “a Member is not obliged, in setting health policy, automatically to follow what, at a given time, may constitute a majority scientific opinion” (WTO Appellate Body, quoted in Box 5.1; p. 57)

Open questions#

1. Strength of early warnings in context. How distinctive were the 1898–1910 asbestos observations against the background of other dusty-trade diseases? What would a reasonable regulator of the 1900s–1910s have been expected to do, and what would the proposed mortality surveys have cost and shown? Read Bartrip (1998, Medical History 42: 421–438) and the Greenberg and Wikeley reply (1999, Medical History 43: 508–510) for the two sides of this debate before relying on the chapter’s pre-1930 claims.
2. Why did 1918 insurer precaution “get forgotten”? Was it competition in underwriting, changes in the product mix (from manufacturing to users), lack of data sharing, or something else? This is a potentially important mechanism of institutional memory loss that the chapter leaves unexplained.
3. Provenance of the 1969 hygiene standard. Who set the 2 f/ml “hygiene standard”, and on what data? Later historical work (e.g. Tweedale) links it to industry-generated data. Check how far industry shaped the standard that regulators adopted.
4. How was Doll’s 1955 paper nearly suppressed, and how was suppression overcome? And how should Doll’s later criticism of Alice be read? Later literature has examined Doll’s relationships with industry; check whether that bears on this episode, without presuming it does.
5. The Legge paradox. The same person lamented “badly missed” opportunities (1934) and advised being “quit of” a critic (1932). What does this say about professional gatekeeping by reform-minded insiders?
6. Germany’s 1938/1943 recognition. What evidence and institutional context produced earlier German recognition of asbestos lung cancer? The chapter omits the political context of that period, and it is worth understanding before using the comparison.
7. Did media pressure cause the 1984 regulations, or accelerate what was already coming? What does the documentary record of UK government decision-making show?
8. Fibre potency. How strong is the evidence for lower chrysotile potency, especially for mesothelioma, and does it matter for the ban-versus-“controlled use” choice? What happened to the “controlled use” position internationally after 2001?
9. WTO precedent. Did the EC–Asbestos Appellate Body reasoning (qualitative evidence, minority scientific opinion, controlled use not a reasonable alternative) become influential in later trade and health disputes? The AB itself said a panel “need not, necessarily” decide on the “preponderant” weight of the evidence (para. 178); whether that amounts to the Box’s “lower level of proof” is interpretive.
10. Projections versus outcomes. Did UK male mesothelioma deaths follow Figure 5.1, peaking at about 2,700–3,300 around 2015–2020, or peak lower or earlier? Did EU mesothelioma deaths approach 250,000 over 35 years? Did UK asbestos cancer deaths keep “rising” from about 3,000 a year?
11. Compensation. Were “speedy, affordable and transparent” compensation arrangements (Lesson 8) established in the UK, EU or US after 2001, and did they improve prevention incentives or exposure records?
12. Substitutes. Were the substitute predictions borne out (IARC re-evaluations of synthetic vitreous fibres)? Did “closed loop” approaches spread?
13. Global displacement. Did asbestos use continue to grow in Asia and elsewhere after 2001, as the chapter implies (pp. 52–53)? Did producer countries, including Canada, change position?
14. The US regulatory trajectory. The chapter omits the US regulatory story (1989 EPA rule; 1991 court reversal). What does it add to the burden-of-proof and cost-benefit lessons?
15. The cost counterfactuals. How robust is the Dutch 1965-versus-1993 estimate? What would a discounted, benefit-inclusive evaluation look like?
16. The innovation claim. Is there evidence that tighter asbestos regulation actually stimulated substitute innovation (Table 5.1, 1982–9; Lesson 4), or did substitutes spread for other reasons?

Audit log#

Independent audit against the text extract (all pages 52–63), the PDF (pp. 1, 5, 12, 52, 57, 61, 196), the WTO Appellate Body report WT/DS135/AB/R (wto.org) and USGS Circular 1298 (Virta 2006).

1. Authors: Gee’s role corrected to one of two EEA editors and two project managers (p. 1), with the initiation credit sourced to the foreword (p. 5).
2. Authors: Greenberg’s self-citations corrected from “four” to five in-text citations (1993 ×2, 1994, 2000 ×2); “historian” reframed as an inference from his cited papers, since the bio does not say so.
3. Standpoint: “most-used secondary source” for Tweedale changed to one of two most-cited (tied with Castleman 1996, five citations each).
4. Panels: “no dissenting or industry voices” corrected to no industry-authored commentary; industry voices appear as quoted actors.
5. Main sources: the litigation-document basis of Tweedale 2000 is now marked as general knowledge.
6. 5.1: added “four years” of exposure and the authors’ “environmental exposure” label; added the arithmetic oddity in the 250,000–400,000 range.
7. Figure 5.1: added the 1920s level, the c.1945 wartime dip and the five-year bar interval (checked visually).
8. World production caution: now sourced to USGS Circular 1298 (peak almost 4.8 Mt around 1977), replacing unsourced general knowledge.
9. 5.2 Women Inspectors: added the authors’ “laywomen”/”not … ‘expert opinion’” framing; the “simply ignored” verdict is scoped to their reports, not all early warnings.
10. 5.2 Murray: “Murray brought the case” corrected to the chapter’s passive “was brought to the attention of”, noting Murray gave evidence.
11. 5.2 Collis and rats: the verbatim wording is restored; flagged the ambiguity over whether “reasonable grounds for suspicion” was a 1911 or a 1933 judgement (Table 5.1 says 1911).
12. 5.2 Insurers: “before regulators did” softened; added that declining cover protected insurers, not workers, and that the Table says “US” only.
13. 5.2 Rochdale: restored “In some cases” to the diagnostic-confounding sentence.
14. 5.3: “two original conceptual contributions” changed to the two terms the chapter explicitly coins.
15. 5.4 Johns Manville: restored “asbestos pollution claims” and Sells’s full list of measures (removing the paraphrase “candour” here and in Mechanisms, Insight 17 and the digest).
16. 5.5 Simpson: “largely ratified what industry had already done” corrected to two of four recommendations (also in the timeline and Mechanisms 4); noted the chapter gives no reason for industry’s withdrawals.
17. 5.5 Fibrils and public limit: softened “could not see the relevant particles” and “not by health criteria” to what the text supports.
18. 5.5 Alice: flagged “unscientific and emotional” as the authors’ uncited paraphrase of Doll, not his words (also in Mechanisms 2 and Framing).
19. Box 5.1 proceedings: verified against WT/DS135/AB/R. The US was a third participant, not a cross-appellant, and the Panel never found glass fibres “as carcinogenic”; the notes now state that the Box is wrong on this.
20. Box 5.1 points: “tends to show” attributed to the Panel (para 8.193, upheld at AB para 157); the “preponderant weight” wording shown to be the AB’s own (para 178), with only “lower level of proof” the Box’s gloss.
21. Box 5.1: added the AB’s likeness reasoning (paras 114, 141, 147) and para 168 on less-risky substitutes, which the Box omits.
22. 5.5 Current burden: added the HSC 1994–95 source and noted that Figure 5.1 (male mesothelioma only) does not show the 3,000 total.
23. 5.6: noted that the Turner GBP 1bn and Lloyd’s claims are uncited and that Summers 1919 is cited second-hand via Tweedale; added the authors’ concession that doctors’ expertise was “at least relevant” to health impacts.
24. 5.6 Litigation: “Much of the chapter’s own evidence” derives from discovery softened to what the Knox references show.
25. References: added the uncited Acheson and Gardner 1983 entry, the fit between Knox 1952 and the p. 60 quotation, and Summers’ second-hand citation.
26. Timeline: Collis/rats row re-dated; insurer row notes self-protection; Knox row restores “I am assured”; Simpson row corrected; 2000 award actor changed from “Courts” to compensation award.
27. Lags and “What was known when”: marked the 1984 cancer-limit lag as an inference; changed “known … harmful” by 1910 to the chapter’s “reasonable grounds for suspicion” by 1911.
28. Lesson 3: classification split, since “sanctions were trivial” is asserted without penalty data.
29. Lesson 5: the jobs-profits alliance is now marked as not directly evidenced, rather than “supported” by the pensioners’ anecdote.
30. Lesson 6: added the omitted sentence that the point “cannot be proved decisively … until another 20–40 years have passed”.
31. Lesson 7: restored “I am assured that” in the Knox quotation, since he was relaying hearsay.
32. Lesson 8: noted the hedge “It seems necessary”.
33. Lesson 10: added the WTO AB para 168 as a compatible external view.
34. Mechanisms 2: in-house studies no longer attributed outright to the company doctor; the 1969 standard is no longer presented as the same “assumed improvement” pattern.
35. Mechanisms 3: the Germany–UK comparison now notes the chapter’s own confounder (lower background lung cancer in 1938), so it is not just “same evidence, different threshold”; Insight 26 re-rated from moderate to suggestive–moderate.
36. Mechanisms 4 and 7: “ultimately destroyed the firms” and “bankrupted firms” corrected (Manville filed while profitable; Turner provisioned; Lloyd’s is an insurer); added the missing page reference for Selikoff’s refusal.
37. Mechanisms 9 and 11: the latency range is reconciled (20–25 and about 40 years on p. 55; 10–40 on p. 60); coinages are separated from scare-quoted terms.
38. Insight 8: split the rating (strong for non-enforcement; asserted for trivial penalties). Insight 15: added the self-protection caveat. Insight 18: updated with AB paragraph verification.
39. Limitations: corrected “former EEA staffer” (Gee was EEA staff and a report editor in 2001); added the p. 12 statement that the editors sought false-positive cases and found none robust.
40. Limitations: hindsight and Factory Department bullets nuanced; IARC “by 1986” reframed as understated rather than wrong (the WTO record cites IARC since 1977); Box 5.1 flag replaced with verified findings.
41. Open question 9: updated with the AB’s own “preponderant” wording.
42. Digest: aligned with all of the above (Gee’s role, scope of “ignored”, Factory Department response, insurers, “seemed” months, WTO verification and the US third-participant error, HSC source, Lesson 3, Sells wording, two added insights, the p. 12 false-positive search, and errors confirmed against primary sources). The media’s role in tightening is now presented as the chapter’s claim, and “fibrils invisible” is softened to “not accurately monitored”.

Second audit (independent re-check)#

Re-checked against the full text extract (pp. 52–63); PDF pp. 1, 5, 12, 52 (Figure 5.1, rendered) and 61 (Table 5.1, layout extraction), and p. 196; WTO Appellate Body report WT/DS135/AB/R (paras 4, 157, 167, 168, 174, 178 and the title page); USGS Circular 1298; Bartrip 1998 (Medical History 42: 421–438, full text); and the journal records for Greenberg 1993 and the Greenberg and Wikeley reply (1999). The first audit’s WTO, USGS, authorship and quotation corrections were confirmed.

43. Authors: “foreword” corrected to the Executive Director’s preface (p. 5), quoting “who initiated this report”.
44. Figure 5.1: amosite reading clarified (segment top about 28 on the axis; about 24 of amosite if, as appears, stacked above crocidolite).
45. 5.1 production caution: added USGS’s statement that world output “by 1965 had already exceeded 2” Mt (also in Limitations and the digest).
46. 5.2 Deane: “proposed the kind of mortality statistics … (see also p. 60)” corrected. On p. 60 Deane calls such statistics “practically unattainable … with the time and opportunity at present at our disposal”.
47. 5.2 Insurers: “protected the insurers, not the workers” labelled as the note-taker’s observation; added the chapter’s own framing (“early precautionary action”).
48. 5.3 Synergy: Archer’s scope restored (“from uranium and other mining”).
49. 5.3 Tage: “Ronald” confirmed as an error. The cited paper’s published title is “Reginald Tage: A UK asbestos prophet — A postscript” (also in References, Limitations and the digest).
50. 5.5 Public exposure standard: “set” corrected to “recommended” (the HSE recommended 0.1 f/ml). Also changed in the timeline, Mechanisms 5 and Insight 12.
51. Box 5.1 proceedings: added from the WTO record that the dispute concerned the French Decree, with the EC as respondent, and that the Panel’s ruling was mixed (a GATT III:4 violation, justified under XX(b)). Also in the timeline and the digest.
52. Table 5.1: restored the full 1918 entry and “amongst others” in the 1962/64 entry; noted that the 1998–99 row omits the UK ban.
53. References, Bartrip: date confirmed as 1998. New finding: Lesson 4 cites Bartrip as noting that controls should have come before 1930, but his article argues the opposite (“no compelling medical or scientific evidence … until the late twenties”; the early reading “rests heavily … upon the use of hindsight”). Greenberg and Wikeley published a reply (1999). The chapter does not disclose the dispute.
54. Standpoint (Panels/commentaries), Limitations (hindsight; dissent within the text) and Open question 1: added the Bartrip dispute and the Deane “practically unattainable” caveat.
55. Lesson 2 classification: the feasibility claim is now weighed against Deane’s p. 60 words, which support “possible with resources not provided”.
56. Lesson 4 classification: “strongly supports” narrowed to the 1950s–60s window; the pre-1930 window is flagged as contested.
57. Lesson 8: restored the hedge “seems to be” on the “unique example” of the British Nuclear Fuels scheme.
58. Timeline: “Fast diffusion before any warning” corrected. The chapter says disease reports “had also begun to appear” within the same 20 years.
59. Mechanisms 10: “Production shifted geographically” corrected to use shifting (p. 52: use continues “largely in developing countries”).
60. Insight 1: the rating is qualified for the 1898–1910 warnings (disputed historiography; Deane’s resource caveat).
61. Digest: insurer self-protection marked as the notes’ reading; Deane and Bartrip caveats added to the core story, to Lessons 2 and 4 and to the caveats; the Sells passage now separates the chapter’s paraphrase from Sells’s quoted words; Lesson 6 restored to “today’s exposures to carcinogens”; “for most uses” and “arrangements to pay up to GBP 1bn” restored; the Doll 10-fold risk scoped to 20+ years’ exposure; p. 61 added for “ignoring users”; the Germany rating aligned with the notes; confirmed Tage and Bartrip errors moved from “to check” to “confirmed”; WTO Panel nuance added.