LL2-A2 — Annex 2 An overview of Late lessons from early warnings vol. 1 (2001)#

Report: Late lessons from early warnings: science, precaution, innovation (EEA Report No 1/2013) Section: Annex 2, “An overview of Late lessons from early warnings: the precautionary principle 1896–2000 vol. 1” Report pages: pp. 701–716 (PDF pages 703–718)

How this was read: - I read the full text extract in order, up to the final page marker. That is p. 716, which ends with the annex’s single reference, Gee 2009. - I rendered all sixteen PDF pages and checked them by eye. The extract matches the printed pages throughout. The only defects are broken SO2 subscripts and repeated running heads. - There are no panels in this section. - I did not open vol. 1 itself. It was outside the files named for this task, although Strand A would permit it. So none of these summaries has been checked against the original 2001 chapters.

Authors and standpoint#

What the text is#

An unsigned, 16-page recap of the fourteen case studies in vol. 1. It has four parts:

1. An introduction (p. 701).
2. Table A2.1 (p. 702). One line per case, giving the date of the first early warning, the date of effective risk reduction, and “Years of substantial inaction”. Its source is given as “Gee, 2009”, a Pathophysiology paper titled “Late Lessons from Early Warnings: Towards realism and precaution with EMF?” (reference on p. 716).
3. Fourteen short case summaries (pp. 703–716), each under the title and author names of the matching vol. 1 chapter.
4. Fourteen chronology tables, each headed “[case] — early warnings and actions” (Tables A2.2–A2.15).

The parts were written at different times, and the text doesn’t say so#

• The summaries were at least edited for the 2013 volume. They may have been adapted from older vol. 1 material, which I can’t check.
• They are editorial paraphrases in the third person (“The chapter analyses…”, p. 703; “This case study analyses…”, pp. 704, 716), not the chapter authors’ own words.
• They cite the 2013 volume: “as noted in Chapter 3 on lead in petrol” (p. 712). Vol. 1 had no chapter on lead.
• The radiation summary’s “now” about CT scans (p. 704) probably speaks from 2013.
• The chronologies date from 2001.
• They end in 2001 and keep present-tense entries from that year:
  • “The debate goes on” (p. 712)
  • “judgement expected end 2001” (p. 710)
  • “EU still suffers from sanction” (p. 715)
  • convention text “to be finalised” (p. 714)
  • “are they happening fast enough…?” (p. 703)
• They look like vol. 1’s chapter chronologies reprinted with a single later, undated update: a note that the “EU won the case against Pfizer Animal Health” (p. 710).
• So the evaluative phrases in them (“At last”, “Obvious lack of precaution”, “Continuous inaction”) are probably the chapter authors’. I can’t confirm this without vol. 1.
• Table A2.1 dates from 2009. It includes dates after 2001 (2008, 2010).

How I attribute views in these notes. “The annex” means the unsigned editorial voice: the introduction, the summaries and Table A2.1. “The chronology” means the per-case tables. Neither should be read as the full argument of the named authors.

Named vol. 1 chapter authors#

Who the authors were#

• What the annex says. Only that they were “all experts in their particular field of environmental, occupational and consumer hazards” (p. 701). No affiliations are given.
• What I know from general knowledge (verify):
• Gee was the EEA’s lead editor of both volumes.
• Harremoës chaired vol. 1’s editorial team.
• Farman co-discovered the ozone hole while at the British Antarctic Survey.
• Infante was a US federal occupational epidemiologist.
• Santillo and Johnston worked at the Greenpeace Research Laboratories.
• van Zwanenberg and Millstone were science-policy researchers at Sussex and critics of UK BSE policy.
• Bridges chaired EU scientific committees.
• Edqvist was a senior Swedish veterinary official.
• Some authors were actors in their own stories.
• The annex itself shows “Infante et al.” publishing the pivotal 1977 benzene cohort study (p. 705), and “Farman, Gardiner and Shanklin” publishing the 1985 ozone-hole paper (p. 708).
• Gee co-wrote the asbestos case and is also the source of Table A2.1.
• Insider authorship brings depth. It also risks accounts that vindicate the authors.

The annex’s stance#

• Openly pro-precaution. This starts with vol. 1’s title, “the precautionary principle 1896–2000”.
• Case titles carry verdicts:
• “from ‘magic’ to malevolent mineral” (p. 706)
• “resistance to common sense” (p. 710)
• “how reassurances undermined precaution” (p. 716)
• “Years of substantial inaction” frames the whole set (p. 702).
• Only one title questions precaution. The hormones case asks: “the precautionary principle or a political risk assessment?” (p. 715).

Panels and opposing views#

There are no panels. Opposing positions appear mostly as entries in the chronologies. One summary also records one: the hormones summary says “expert scientific committees concluded that growth hormones were safe to use” (p. 715). Their reasoning is rarely given. The exceptions are the JECFA’s “standard risk assessments” (p. 715) and the Phillips Inquiry’s stated conclusion (p. 716). The positions are:

• the UK Factory Department’s “insufficient evidence” (p. 706)
• the Swedish Agriculture Board’s “negligible” (p. 710)
• the Lamming Committee and WHO/FAO JECFA finding some growth promoters safe (p. 715)
• the European Chemicals Bureau (ECB) declining to classify MTBE as a carcinogen (p. 712)
• Hooker Chemical’s denial (p. 713)
• the American Petroleum Institute’s (API) and the pharmaceutical industry’s court challenges (pp. 705, 710)
• the Phillips Inquiry (p. 716)

Section-by-section notes#

Introduction (p. 701)#

• Rationale. Science’s “growing powers of innovation seem to be outstripping its ability to predict the consequences of its applications”. Meanwhile, the growing scale of human intervention makes hazards potentially “serious and global” (p. 701). This is asserted, not argued.
• Aim. To draw lessons “particularly in terms of providing information and identifying early warnings” (p. 701).
• This focus on information fits the EEA’s role as an information agency.
• But in many of the cases the information existed and was not acted on (Insight 10).
• Premise. “The lessons of history have rarely been used in trying to manage current and future risks” (p. 701). Asserted.
• Case selection. Fourteen cases were chosen “where sufficient is now known about their impacts” to draw conclusions about “how well they were addressed by governments and civil society” (p. 701).
• This is selection on outcome.
• So the set contains no cases where the warnings proved unfounded.
• Method claim. The conclusions were “based on ‘the spirit of the times’ and not on the luxury of hindsight” (p. 701). Assessed under Limitations below.
• Author brief. Authors were asked to date the early warnings, analyse how information “was used or not used”, and describe “costs, benefits and lessons” (p. 701).
• The annex carries almost nothing on costs, benefits or lessons that cut across the cases.

Table A2.1 (p. 702)#

Title: “Case studies in Late lessons from early warnings: the precautionary principle 1896–2000”. Source: Gee, 2009.

My arithmetic check of the table#

• Figures that match simple subtraction: radiation, benzene, asbestos, antimicrobials, MTBE, hormones, BSE. DES matches after rounding: the dates give 33–47, and the table says 30–50.
• Figures that are only approximate:
• Halocarbons: the dates give 13–36 years; the table says 10–30.
• SO2: the dates give 11–49; the table says 25–55.
• TBT: the dates give 1–32; the table says 5–30.
• PCBs. The 1970s–80s restrictions give 71–90 years, and the “phase out by 2010” gives 111. “c. 100” sits between the two, so it’s ambiguous which action is being counted.
• A figure that can’t be derived from the action the table lists:
• Great Lakes. The DDT bans give 6–10 years using the chronology’s 1969 and 1972 dates, or at most 17 across “the 1970s”. “45+” implies counting to about 2008. In effect the table treats the DDT bans as not the effective action, while listing them as such.
• “Effective action” means different things in different rows:
• Benzene. The table uses a voluntary withdrawal from consumer products, though the case is about workplace exposure. The 1 ppm workplace standard came in 1987, which would make the gap 90 years.
• Asbestos. It uses the 1999 ban decision rather than its 2005 effect.
• Fisheries. The action is conceded to be “not very effective”.
• MTBE. The substance was still “permitted elsewhere”.
• Radiation. It counts to the first UK regulations on radioactive substances (1961) and ignores voluntary protection rules from 1913 and the international committee of 1928 (p. 704).
• “First early warning” also means different things:
• Direct observations of the same harm: asbestos 1898; radiation 1896; the TBT oyster collapse.
• Generic signals from another endpoint or class of substance:
  • PCBs 1899: chloracne in “chlorinated organic industry” workers, thirty years before PCBs were mass-produced (p. 707).
  • DES 1938: cancer in animals (p. 709).
  • Benzene 1897: aplastic anaemia. Leukaemia wasn’t first reported until 1928 (p. 705).
• A theory: ozone 1974.
• Data that could have been known: MTBE, a 1960 textbook.
• A pre-modern call for action: fisheries 1376 (“a call for precautionary actions in fishing by controlling the mesh of the nets”, p. 703).
• A different pathway from the one action was based on: hormones. The warning concerns wildlife (1972/73), but the ban rested on uncertainty about human health (p. 715).
• The dates are not systematically the earliest available. The table ignores Fleming’s 1945 warning (p. 710), the 1954 MTBE paper (p. 712), and the 1970–71 imposex reports (p. 714).
• The lags shrink with the era of the first warning:

• Using range midpoints and leaving out fisheries, the median is about 40 years.
• The decline could reflect better institutions after 1970.
• It could also be truncation. A warning from the 1970s can’t show a 60-year lag in a 2001 set of cases chosen because their outcomes were “now known”.
• The annex doesn’t discuss either possibility.

Upshot. The table gives a powerful headline: decades pass between warning and action. But it is not a consistent measure. It mixes delays in knowing with delays in responding, and each row follows its own rules. Treat the figures as orders of magnitude only.

Fisheries: taking stock — MacGarvin (p. 703; Table A2.2)#

Summary. - Measures for sustainable fishing “have been known for centuries” (p. 703). - Industrial fishing intensified in the second half of the 20th century. - “Necessary safety margins were underestimated”, bringing North Sea herring “to the brink of collapse by the 1970s” (p. 703). - The chapter compared Canada, the US and the EU, and weighed “the implications of international documents introducing the term ‘precautionary approach’” (p. 703).

Key dates. - 1376–77: the English parliament answers a call to control net mesh by setting up a committee. - 1866–93: “Following uncertainty about the consequences of fisheries growth”, official enquiries, but “no action is taken”. - California sardines: scientists call for precaution in the mid-1920s. “Continuous inaction” leads to collapse in 1942. Recovery only begins in the mid-1980s. - Northern cod: - From the late 1970s Canada manages the fishery, “claiming it is doing so precautiously”. - 1986: the Keats Report, “commissioned by inshore fishers”, finds fishing pressure severely underestimated. - 1988: the Alverson Report finds “consistent overestimation of the stock size leads to overfishing”. - 1989: a new government assessment recommends halving the offshore catch. - 1990: the Harris Report, “a new independent assessment”, confirms overfishing. - 1992: collapse and moratorium. - 1999: fishing restarts, with “dissidents” saying the level is still too high. - 1990s: “The ecosystem approach slowly settles in fisheries management procedures”. - 1995: FAO Code of Conduct and UN Straddling Fish Stocks Agreement. - 2001: European Commission Green Paper. The chronology closes: “Positive changes are occurring, but are they happening fast enough to avoid further collapses?” (p. 703).

Observations. 1. The basic remedy was known for centuries, so this was largely an institutional failure. But it was not purely one. The chronology also records uncertainty (1866–93) and assessment errors (1986–88), and the summary says “Necessary safety margins were underestimated” (p. 703). So gaps in knowledge played a part too. 2. All three recorded assessment errors ran in the direction that favoured continued fishing: underestimated safety margins, underestimated fishing pressure, overestimated stock. 3. The first correction came from a report commissioned by inshore fishers. They were a different stakeholder group from the offshore fleet that would bear the 1989 catch cut. My reading is that this combines local knowledge with a conflict inside the industry; the annex doesn’t say so. 4. “Precaution” was claimed as rhetoric. 5. Committees stood in for action. 6. Action came only when the stock collapsed. 7. This is one of the clearest cases where those causing the harm also bore much of it. (Radiologists harmed by their own practice, p. 704, are another.) Long-run self-interest still didn’t produce precaution (my inference).

Radiation: early warnings; late effects — Lambert (p. 704; Table A2.3)#

Summary. - “Injuries and fatalities” were acknowledged “as early as 1896”, but “the serious consequences of widespread use… were not recognised” (p. 704). - “Absence of an agreed unit to measure radiation made the introduction of effective standards of protection difficult” (p. 704). - “Ill-conceived” uses went on through the 1930s–50s: fitting children’s shoes, treating ringworm, removing hair. - Stewart’s evidence linking X-rays in pregnancy to childhood leukaemia was “controversial at first”. Her work, “along with that of other researchers”, eventually led to the principles of “justification, optimisation and dose limitations” for patients as well as radiologists. - “Similar concerns are now being expressed at the frequent over-exposure to radiation from CT scans” (p. 704).

Key dates. - 1896: injuries noted by Edison, Tesla and Grubbe, the technology’s pioneers. - 1899: a journalist, John Dennis, campaigns for licensing of radiologists. - 1904: Edison’s assistant dies. Rollins publishes warnings, “including pregnant women”. - 1913: the German Radiological Society publishes the first voluntary protection rules. - 1924: Blum identifies “radium jaw” in dial painters but “wrongly attributes this to phosphorous”. Martland identifies radium as the cause (1925–29). - 1928: an international protection committee is formed (later the ICRP). - 1934: more than 200 radiologists’ deaths from radiation-induced cancer are reported. - 1949: the ICRP concludes there is no dose threshold for cancer. - 1958: Stewart’s finding, “Not generally accepted until the 1970s”. - 1961: UK regulations. - 1977: the ICRP updates its recommendations and “links dose limits to risk”. - 1988: UK regulations on doses to patients. - 1990–97: the UK National Radiological Protection Board (NRPB) finds 20 % of medical X-rays “probably clinically unhelpful” and 50 % of the collective dose avoidable. Doses for the same X-ray vary a hundredfold between hospitals. - 1990: ICRP Publication 60 puts cancer risk at 4–5 times the 1977 estimate and cuts the occupational limit to 20 mSv/yr. - 1996: EU Directive (p. 704).

Observations. 1. The pioneers saw the harm first, yet trivial uses spread for decades. 2. Blaming an already-known hazard delayed recognition. 3. Measurement had to come before standards. 4. Radiologists were protected before patients, although Rollins had already warned in 1904 about patients, “including pregnant women”. Harm to the unborn (Stewart, 1958) was not accepted until the 1970s, more than a decade later. 5. Risk estimates were revised sharply upward decades into regulated use. 6. Much exposure was unjustified: 20 % of X-rays “probably clinically unhelpful” and half the collective dose avoidable. That suggests it could be cut at little cost to patients, but the annex gives no costs, so this is my inference. 7. Voluntary rules (1913, 1928) came long before law. Table A2.1’s “65” ignores them.

Benzene — Infante (p. 705; Table A2.4)#

Summary. - Benzene was known early as a bone-marrow poison. It was still used as a rubber solvent from 1910, and then widely. Leukaemia and other diseases followed. - “Exposure recommendations were set slowly and remained at high levels for many years” (p. 705). - “From the 1970s the benzene industry began to attack the science demonstrating its harm”, which “contributed to delays in government action” (p. 705).

Key dates. - 1897: aplastic anaemia reported (Santesson; printed “Santessen”). - 1926: Greenburg and colleagues observe abnormally low white blood cell counts in benzene workers. - 1928: first leukaemia case reported (printed “Dolore and Borgomano”). - 1939: substitution with other solvents recommended; “not implemented”. - 1946: the ACGIH sets 100 ppm, “even though some cases of benzene poisoning were associated with levels of 25 ppm and 10 ppm”. The limit falls to 50 ppm (1947), 35 ppm (1948) and 25 ppm (1957). - 1948: the API concludes “the only absolutely safe level is zero, but recommends 50 ppm or less”. - 1950s–60s: “Obvious lack of precaution… with fatal consequences”. - 1977: the Infante et al. cohort study links benzene directly to leukaemia. The Department of Labor seeks 1 ppm and is “challenged in the courts by API”. - 1978: voluntary withdrawal from consumer products. - 1980: the US Supreme Court’s “Benzene Decision severely limiting regulatory actions”. - 1987: a 1 ppm standard is set. “This 10-year delay caused more than 200 deaths in the United States.” - 1996: studies show disease at 1 ppm. - 2001: benzene in petrol remains a public exposure (p. 705).

Observations. 1. Limits were set above levels already linked to poisoning. 2. An industry body said no level was absolutely safe, recommended 50 ppm anyway (higher than the ACGIH’s 35 ppm that same year), and later litigated. 3. A court decision limited the regulator. The annex says only “severely limiting regulatory actions”. Reading this as a higher burden of proof is my gloss from general knowledge. The method behind the deaths estimate isn’t given here. 4. A substitute was recommended and ignored. 5. The “safe” level kept moving. 6. This is a participant’s history.

Asbestos: from ‘magic’ to malevolent mineral — Gee and Greenberg (p. 706; Table A2.5)#

Summary. - Asbestos was valued for sound absorption and resistance to fire, heat, electricity and chemicals. - A latency of “20–40 years” slowed recognition, “although inspectors had warned about the risks as early as 1898” (p. 706). - Mesothelioma, identified from 1959, is “now causing the bulk of current and future harm… including some 400 000 expected deaths in Europe over the next three decades” (p. 706).

Key dates. - 1898: factory inspector Lucy Deane warns of “harmful and ‘evil’ effects”. - 1906: a French report of 50 deaths among female textile workers, with a “recommendation of controls”. - 1911: rat experiments give “‘Reasonable grounds’ for suspicion”. - 1911 and 1917: the Factory Department finds “insufficient evidence”. - 1918: US insurers refuse cover to asbestos workers “due to assumptions about injurious conditions”. - 1930: the Merewether Report finds asbestosis in 66 % of long-term Rochdale workers. - 1931: regulations cover manufacturing only and are “poorly implemented”. - 1935–49: lung cancer cases reported in asbestos manufacturing workers. - 1955: Doll shows high lung cancer risk. - 1959–64: mesothelioma found in South Africa, then in UK and US workers, “bystanders” and relatives. - 1969: regulations “improve controls, but ignore users and cancers”. - 1982–89: media and union pressure tightens controls and “stimulates substitutes”. - 1998–99: EU and French bans. - 2000–01: the WTO upholds the bans against Canada’s challenge (p. 706).

Observations. 1. Regulators wanted proof while insurers acted on “assumptions”. 2. The harm spread across diseases and groups. The disease found last became the largest. 3. Rules were narrow and poorly enforced. 4. The chronology credits the 1980s tightening to “media, trade union and other pressure”. It doesn’t mention new science at that point, though it doesn’t rule it out. The pressure also “stimulates substitutes”. 5. Latency locks in deaths long after the ban. 6. A trade tribunal upheld the bans against a trade challenge. 7. The 400,000 figure has no base year. It may be carried over from vol. 1 (unverified), so it’s unclear which three decades it covers. 8. Minor: Table A2.1 says “malevolent material” while the case heading says “malevolent mineral”.

PCBs and the precautionary principle — Koppe and Keys (p. 707; Table A2.6)#

Summary. - PCBs made electrical equipment “smaller, lighter, and what were thought to be safer” (p. 707). - Chloracne and liver damage in workers were “well known in the 1930s”. - From the 1960s came infertility in wildlife, then contamination of breast milk and endocrine effects. - “Prohibition only happened gradually with each government action taken on the basis of a high level of scientific proof” (p. 707).

Key dates. - 1899: chloracne in workers handling chlorinated organics. - 1929: mass production begins. - 1936–37: liver damage in workers and rats. The results “did not gain attention from policymakers but both labour regulators and manufacturers were made aware”. - 1966–69: Jensen finds unknown molecules in sea eagles and identifies them as PCBs. - 1968: Yusho in Japan: 1,800 people poisoned by contaminated rice oil. This was “the first well-publicised warning”. - 1970s: high levels of PCBs found in infertile seals of three species. - 1972: Sweden bans “open” uses. - 1976: the US Toxic Substances Control Act (TSCA) allows only “totally enclosed” use. - 1979: 2,000 poisoned in Taiwan. “25 % of children born of poisoned mothers died before the age of four years.” - 1990s: IQ effects from exposure in the womb. “Fetotoxicity represents a new paradigm for toxicology.” - 1996: EU phase-out set for 2010. - 1999: contaminated animal feed found in Belgium (p. 707).

Observations. 1. A safety claim was part of the benefit case. 2. Regulators and manufacturers knew from 1937. 3. Detection limited the environmental signal. 4. A mass poisoning made the hazard public, and the same kind of poisoning happened again 11 years later. 5. Control came in stages, by type of use. 6. Each step required a high level of proof. 7. Persistence outlasts bans. 8. The key target shifted to the foetus.

Halocarbons, the ozone layer and the precautionary principle — Farman (p. 708; Table A2.7)#

Summary. - CFCs were used from the 1930s. Concern “did not arise before the 1970s”. - Five WMO reports (1985–99) found halocarbons responsible “beyond reasonable doubt”. - They were replaced by HCFCs, still ozone-depleting (printed “less harmful that CFCs”), and HFCs, “powerful greenhouse gases”. - “The political process failed to stimulate more radical changes towards halocarbon-free and energy-efficient technology” (p. 708).

Key dates. - 1907 and 1934: laboratory chemistry. - 1973: Lovelock’s global survey. - 1974: Molina and Rowland. - 1977: the US bans CFC aerosols “based on ‘reasonable expectation’ of damage”, followed by Canada, Norway and Sweden. UNEP research plan. - 1980: a European aerosol restriction, but “rising use in refrigerators, etc. marginalises this restriction”. - 1985: the Vienna Convention (“restrictions if and when justified”). Farman, Gardiner and Shanklin report the Antarctic hole (printed “Antartica”). - 1987: the Montreal Protocol, with different timetables for developed and developing countries. - 1990s: finance for developing countries. - 1997: amendments aimed at restoring chlorine levels by 2050–60. - 1999: the Beijing Declaration on illegal trade (p. 708).

Observations. 1. Early action by one country on a low standard of evidence. 2. A partial measure was swamped by growth in other uses. 3. A framework came first, then a protocol. Montreal followed the measured hole by two years, but the annex just lists the dates and doesn’t claim cause. 4. Differentiated obligations plus finance. Illegal trade became the enforcement gap. 5. Substitution traded one harm for another. The annex’s (unsigned) summary of Farman’s chapter criticises the political process for not pushing innovation further. The criticism is presumably his, but this can’t be confirmed without vol. 1. 6. Harm is committed for decades. 7. The author co-discovered the hole (with Gardiner and Shanklin). 8. This is usually told as the success story. Here the success is qualified.

The DES story — Ibarreta and Swan (p. 709; Table A2.8)#

Summary. - In 1970 seven young US women were diagnosed with a rare vaginal cancer, later linked to their mothers taking DES. - DES was prescribed from 1947 “in the belief that it prevented miscarriages”. - “Prescription continued even though trials in the 1950s established the ineffectiveness of DES” (p. 709). - Science had to abandon the idea of “the foetal environment as a safe place, protected by the placental ‘barrier’”, in favour of “the extreme vulnerability of the developing foetus” (p. 709).

Key dates. - 1938: DES synthesised; first report of cancer in animals. - 1939–42: first use in patients (1939). Approval by the American Council of Pharmacy and Chemistry, and first report of use to prevent abortion (1942). - 1947: FDA approval for “threatened or habitual abortion”. - 1948: use rises after a large-scale US study is published. - 1953: the first large placebo-controlled randomised trial shows no benefit. - 1970: the seven cases. - April 1971: the link to DES. - November 1971: FDA withdraws approval for use in pregnancy. - 1972: a registry of clear-cell adenocarcinoma in young women is established. - 1978: reanalysis of the 1953 (Dieckmann) data shows DES “actually increased the risk of miscarriage”. - 1985: last reported use in pregnancy worldwide (p. 709).

Observations. 1. A cancer signal in animals came before clinical use. 2. Prescription outlasted a negative randomised trial by 18 years in the US and 32 years worldwide. The 1948 study that boosted use shows where the belief in benefit came from. The 1971 withdrawal was prompted by harm, not by the trial’s finding of no benefit. 3. A harm signal sat in existing trial data for 25 years. 4. Once a specific harm could be attributed, US action took about seven months. 5. A blind spot built into the prevailing paradigm. 6. The harm crossed generations.

Antimicrobials as growth promoters: resistance to common sense — Edqvist and Pedersen (p. 710; Table A2.9)#

Summary. - The growth effects were found in the late 1940s. Resistance has “possible effects on human health”. - “The number of independent scientific studies… was relatively small compared to the number of studies supported by the pharmaceutical industry” (p. 710). - Independent studies only increased once resistance linked to animal feed “was made clear”.

Key dates. - 1945: Fleming (printed “Flemming”) warns that “microbes are educated to resist”. - 1950s–60s: vertical and then horizontal transfer of resistance recognised. - 1969: the Swann Committee recommends severe restrictions, mostly implemented in the 1970s. - 1975: the Swann restrictions are relaxed. The entry is garbled (“tolysin… vancomycin comes into use”); it probably means tylosin, and possibly avoparcin. Check. - 1977: the Swedish Agriculture Board judges the risk “negligible”. - 1984: Swedish farmers ask for a ban “because of health and consumer concerns”. - 1985: Sweden bans them “on grounds of antibiotic resistance in animals and ‘uncertain’ long-term effects”. - 1997: a Swedish report finds the risk “‘far from negligible’”. WHO calls it “‘essential to replace’” these uses. - 1998: the EU bans four as a “‘precautionary’ measure”. - 1999: the EU Scientific Steering Committee recommends phase-out, and industry sues. A later, undated note says the EU won against Pfizer Animal Health. - 2000: WHO recommendation (p. 710).

Observations. 1. The discoverer warned about the mechanism decades before regulation. 2. Restrictions were implemented, then relaxed. 3. The official risk judgement reversed. 4. A user group, the farmers, moved first, citing “health and consumer concerns”. The consumer part suggests a market motive, but health is cited too. 5. The evidence base followed the funding. 6. The ban survived legal challenge. 7. Sweden acted 14 years before the EU. 8. The title assumes the answer, but the chronology records a genuine official risk judgement in 1977.

Sulphur dioxide — Semb (p. 711; Table A2.10)#

Summary. - The London smogs of 1952 and 1962 drew attention to health effects. Measures were “slowly taken”. - Energy growth in the 1960s “resulted in large emission increases”. - Acid rain and its long-range effects needed cooperation across borders. This was “extremely difficult during the Cold War period”, but progress came in the 1980s and 1990s (p. 711).

Key dates. - 1952: smog “kills more than 2 000”. - 1952: the Beaver Report, “prepared by a parliamentary commission”, proposes “modest remedies”. Check this date and description. From general knowledge, it was a government committee reporting in 1953–54. The 1956 Clean Air Act is not listed. - 1962: a second smog, 800 deaths. - 1968: acidification in Sweden linked to other countries’ emissions. - Late 1960s: an OECD Air Management Sector Group is set up. - 1972: further Swedish evidence goes to the Stockholm conference; an OECD acid rain study and an OECD programme on long-range transport are launched. - 1977: OECD report on the relationship between emission and deposition. - 1979: the LRTAP Convention on long-range air pollution; WHO air quality guidelines. - 1980s: forest death in Germany, Poland, Czechoslovakia and North America. - 1985: a protocol for a 30 % cut. - 1988: the large combustion plants directive (printed “published, and amended in 1988”, which is garbled). It was “very efficiently used by Germany against the British opposition”. - 1994: a protocol based on critical loads. - Table A2.1: emissions c. 90 % below 1975 levels by 2010 (pp. 702, 711).

Observations. 1. A disaster that drew public attention prompted only modest remedies. 2. Growth outpaced controls. 3. The harm crossed borders. Sweden, a receiving country, raised the alarm, and Germany used the 1988 directive “against the British opposition”. Calling Germany a victim state and the UK a net exporter is my reading; the annex says only that the UK opposed emission controls. 4. Shared monitoring science (the OECD programmes and 1977 report) and critical-load maps underpinned the agreements. The 1994 protocol is explicitly “based on the critical load concept”. That they made agreement possible is my inference. 5. Geopolitics limited cooperation. 6. Action worked.

MTBE in petrol as a substitute for lead — Krayer von Krauss and Harremoës (p. 712; Table A2.11)#

Summary. - After debate about lead in petrol “in the 1960s”, industry chose MTBE as its substitute. It’s unclear whether the 1960s date refers to the debate or the choice. - From about 1990, groundwater contamination “rendered large water reserves useless”. Links to asthma, cancer and endocrine effects were suspected. - The chapter asks “the extent to which it would have been possible to foresee undesirable properties of MTBE at the time of its introduction” (p. 712).

Key dates. - 1954: a paper shows ethers break down poorly in water. - 1960: the same information, plus taste and odour, appears in a textbook. - 1990: laboratory indications of groundwater risk. A “Significant increase in use… due to amendments to the Clean Air Act”. “Comprehensive investigations of carcinogenicity” begin in the 1990s. - 1995: Santa Monica “lost 71 % of its local water supply”. - 1996: a US Geological Survey (USGS) report spreads concern. - 1997: field studies show MTBE is soluble, mobile and persistent. - 1998: the Danish EPA acknowledges the government “was informed in 1990”, and presents an action plan for remediation and risk reduction. - 1999: California recommends removal “as soon as possible, but not later than end 2002”. - 2000: the US EPA moves to reduce or eliminate it. Denmark lists it as undesirable. There are “indications” of asthma and endocrine effects. - 2001: the ECB decides MTBE “should not be classified as a carcinogen”. “The debate goes on” (p. 712).

Observations. 1. A clear case of a regrettable substitute. The chronology shows the problem properties (low biodegradability, taste and odour) were in the literature by 1954–60. Whether they were weighed when MTBE was introduced is the chapter’s question. The annex doesn’t answer it. 2. A rule made for air quality drove up use, and with it harm to water. 3. The Danish government’s knowledge went unused for eight years (1990–98). 4. The documented harm is loss of a water resource. Health effects are only suspected, and the chronology reports both sides. 5. Action varied by jurisdiction (“permitted elsewhere”, p. 702).

The precautionary principle and early warnings of chemical contamination of the Great Lakes — Gilbertson (p. 713; Table A2.12)#

Summary. - In 1972 the US and Canadian governments responded to organochlorines in wildlife with the Great Lakes Water Quality Agreement (GLWQA) and chemical bans. - “In the 1980s, however, the political climate changed.” Claims of harm “were met with scepticism and demands for proof of a causal relationship before appropriating funds for remedial works” (p. 713). - “Decades later the efforts to clean up the area continue” (p. 713).

Key dates. - 1962: Silent Spring, “the significant early warning”. - 1963: “changes in eggshell quality” observed on Pigeon Island, Lake Ontario. - 1966: first analytical data (Hickey). - 1969 and 1972: DDT banned in Canada and the US (dieldrin 1973), followed by “gradual improvements in wildlife”. - 1974: concerns about possible effects on human health. - 1978: health effects at Love Canal “denied by Hooker Chemical Company”. The GLWQA is renegotiated with “a precautionary policy, but it is not properly implemented”. - 1980: 900 families relocated from Love Canal. - 1984: infants exposed to PCBs are born smaller. - 1996: the same behavioural effects are found in Lake Ontario children. - 2000: “the specific relationship between behavioural anomalies and prenatal exposures to the highly chlorinated biphenyls is determined”. “Reluctance to undertake costly remedial actions even after causal relationship is proven” (p. 713).

Observations. 1. A popular book served as the early warning. 2. The bans worked for wildlife. 3. A company denied the harm. 4. A precautionary policy was adopted in 1978 but was “not properly implemented”. 5. A shift in political climate raised the burden of proof. 6. Proof wasn’t enough to overcome the cost of cleanup. 7. Exposure passed down the generations through diet. 8. The Table A2.1 figure implicitly rejects the DDT bans as the effective action.

Tributyltin (TBT) antifoulants — Santillo, Johnston and Langston (p. 714; Table A2.13)#

Summary. - Preventing fouling of hulls “has a huge effect on the fuel consumption of boats” (p. 714). - TBT paint arrived in the late 1960s. The first imposex (male sex organs developing in female snails) was “discovered in France in 1970”. - Only in the early 1980s were “analytical techniques… sensitive enough” to identify TBT as the cause of “sex reversals in oysters and snails” (p. 714). - The case follows prohibition for vessels under 25 m, then for all vessels. - Check: the table cites Blaber (1970) and Smith (1971), which general knowledge places in the UK and US, not France. And TBT’s effects on oysters were larval failure and shell deformity rather than sex reversal.

Key dates. - Early 1970s: “Rapid increase in the use of TBT antifouling paints on vessels of all sizes” and first reports of imposex (Blaber, 1970; Smith, 1971). - 1976–81: the Arcachon Bay oyster fishery nearly collapses. - 1982: France bans TBT on small vessels. - 1985: first UK controls. - 1986: Bryan et al. report widespread imposex in dogwhelks. - 1987: further UK restrictions on TBT content (January), then a UK retail ban for vessels under 25 m and fish cages (May). The Paris Commission (PARCOM) recommends the same ban across the NE Atlantic (Recommendation 87/1). - 1988: US restrictions. Waldock et al. show shipyards are a significant source. - 1989: Canada, Australia and New Zealand act. - 1991: harmonised EU ban on retail sale of TBT paint. - 1994: imposex offshore, linked to shipping. - 1995: the Esbjerg declaration commits to a global phase-out through the International Maritime Organization (IMO). - 1997–99: the IMO’s Marine Environment Protection Committee (MEPC) agrees a phase-out in principle; draft rules follow. The OSPAR Convention aims to end releases by 2020. IMO Resolution A.895(21) sets deadlines. - 2001: convention text to be finalised. New use to stop in 2003 and existing coatings to be removed by 2008 (p. 714).

Observations. 1. Analytical limits delayed attribution by about a decade. 2. An identifiable, organised economic loss (oysters) preceded fast national action. This is my reading of the sequence. 3. The scope of controls widened as monitoring found large ships and shipyards were significant sources. 4. The product had a real benefit (fuel). The costs of control presumably fell on shipping, but the annex doesn’t discuss them. 5. One of the shortest lags in the set.

Hormones as growth promoters: the precautionary principle or a political risk assessment? — Bridges and Bridges (p. 715; Table A2.14)#

Summary. - Hormones have been used to promote growth since the 1940s, with DES favoured. - Concern grew in the early 1970s as DES was “confirmed to be a human carcinogen”. - “Although expert scientific committees concluded that growth hormones were safe to use, the EU banned several hormones” in 1988, and later extended the ban to imports (p. 715). - “This can be seen as an early example of application of the precautionary principle.” The chapter covers the WTO implications.

Key dates. - 1972: Peakall (printed “Peakal”) on effects of DES on wildlife, “ignored until the late 1980s”. - In the US: - DES banned (1972), then reinstated (1974). - The FDA sets “the minimum detectable level” (1976). - Banned again (1979) because of “the impossibility of identifying levels below which it would not be carcinogenic”. - 1982: the Lamming Committee finds some growth promoters safe. - 1985: first EU ban, “ignoring results from the Lamming Committee because the scope of their assessments had not been broad enough”. - 1987: “Lamming Committee disbanded by EU and their results were not published”. - 1988: EU ban “based on uncertainty of their effects on humans”. JECFA, “using standard risk assessments”, reaches the same conclusion as Lamming. - 1989: ban extended to imports “from third world countries” (probably meaning “third countries”, i.e. non-EU). The Pimenta Report finds illegal use in member states. - 1989–96: US trade retaliation. - 1995: uncertainties still acknowledged. - 1999: the EU’s Scientific Committee on Veterinary Measures relating to Public Health (SCVPH) finds no safe threshold can be defined for six promoters. - 2000: effects on wildlife confirmed. - 2001: “EU still suffers from sanction to its exports of around EUR 160 million per year” (p. 715).

Observations. 1. This is the counter-case in the set. The EU overrode its own expert committee, disbanded it and left its findings unpublished. Expert advice was set aside, and kept from view, in the precautionary direction. Two things temper this. The chronology gives the EU’s stated reason: the committee’s assessments were not broad enough in scope (1985). And later EU bodies moved toward the EU’s position: the 1995 conference found uncertainties remained, and in 1999 the SCVPH found no safe threshold could be defined for six promoters. The dispute is as much about when, and by whom, the science was judged as about precaution versus science. 2. At bottom this is a dispute about how widely a risk assessment should be drawn and framed. 3. Actual misuse (Pimenta) differed from the conditions that were assessed. 4. US policy swung back and forth, and a detection limit served as the de facto standard. 5. Precaution’s cost is explicit here, and exporters bore it. 6. Table A2.1’s “warning” (wildlife) concerns a different pathway from the one the ban was based on. 7. The summary’s hedged claim that the ban “can be seen as an early example” of precaution sits a little awkwardly with the chapter title’s question mark. The outcome at the WTO isn’t recorded.

Mad cow disease 1980s–2000: how reassurances undermined precaution — van Zwanenberg and Millstone (p. 716; Table A2.15)#

Summary. - BSE was acknowledged in 1986. The suspected source was slaughterhouse waste infected with scrapie and recycled into animal feed. Transmission to humans was confirmed “a decade later”. - The chapter analyses decisions made “when weighing human health and financial consequences”. - “The lack of precautionary measures and information to the public about the risks contributed to prolonging the crisis and the collapse of public trust in scientists and governments” (p. 716).

Key dates. - Mid-1970s: the US bans scrapie-infected sheep and goat meat from cattle feed. - 1979: the UK Royal Commission on Environmental Pollution (RCEP) recommends processing standards for rendering. - 1986: first cases. - 1988: first official acknowledgement that BSE may transmit to humans. The Southwood Committee is set up and recommends that “clinically affected cattle should not go into human and animal food”. - 1989: bans on ruminant feed and on specified bovine offal (SBO). Check this: general knowledge dates the feed ban to 1988. - 1995: “Almost 50 % of the abattoirs checked” are failing to comply with the SBO ban. Evidence of a link to Creutzfeldt-Jakob disease (CJD) emerges. - 1996: “At last, experiments start…”. The crisis over variant CJD (vCJD) breaks. - 1998–2000: the Phillips Inquiry. The chronology, probably in the case authors’ voice, says its conclusions “do not seem sufficiently rigorous on judging government actions over time”. It found decisions “appropriate… although not always timely, or adequately implemented or enforced” (p. 716).

Observations. 1. A precedent existed abroad. 2. The domestic warning was generic. 3. The suspected source was slaughterhouse waste recycled into feed. That this created a loop amplifying the infectious agent is my gloss; the annex says only “suspected”. 4. Partial measures came quickly but were poorly enforced. 5. The key experiment was delayed by about ten years. 6. The official inquiry was judged too lenient. 7. The mechanism in the case’s own title, reassurance, doesn’t appear in the chronology. Within this annex it is an asserted claim.

Case timeline (cross-case)#

How to read the table. - The “credible, specific warning” column is my reading of the chronologies, not the annex’s. The strength ratings in the second column are also mine. - The last column is my arithmetic. Counting from specific rather than generic warnings shortens many of the lags. - This matters for separating delays in knowing from delays in responding.

Reading across the table. - Fast responses. Harms that were vivid, specific, attributable and fell on identifiable groups got a response within months or a few years: the DES cancers, US ozone action, French TBT action, and the first BSE measures. - Slow responses. Chronic, latent, diffuse harms took decades, even counting from specific warnings: asbestos, benzene, PCBs in workers, DES’s lack of efficacy, and antimicrobials. - So the long figures in Table A2.1 are partly, but not only, an artefact of choosing early, generic warnings as the starting point.

The authors’ own lessons and conclusions#

There is no lessons section. Vol. 1’s synthesis, including its “twelve late lessons” (general knowledge), is not reproduced.

Lessons the annex draws from evidence, as presented#

• Overall: long periods of “substantial inaction” (p. 702).
• Fisheries: safety margins were underestimated (p. 703).
• Radiation: without an agreed unit of measurement, standards were weak. Stewart’s work led to the principles of justification, optimisation and dose limitation (p. 704).
• Benzene:
• Exposure limits were set slowly and stayed high.
• Industry attacks on the science delayed action.
• More than 200 deaths are attributed to the 1977–87 delay (p. 705).
• Asbestos: latency delayed recognition. Mesothelioma causes most of the harm, with about 400,000 deaths expected (p. 706).
• PCBs: prohibition was gradual because each step needed “a high level of scientific proof” (p. 707).
• Ozone: substitution happened, but politics failed to push radical alternatives (p. 708).
• DES: prescribing continued despite trial evidence. The idea of the placental barrier had to be abandoned (p. 709).
• Antimicrobials: independent research lagged behind industry research (p. 710).
• SO2: cooperation across borders was needed and was hampered by the Cold War (p. 711).
• MTBE: the chapter asks whether MTBE’s properties were foreseeable. The chronology implies they partly were (p. 712).
• Great Lakes: the 1980s political shift meant proof was demanded before funding (p. 713).
• TBT: analysis was not sensitive enough to identify the cause (p. 714).
• BSE: the lack of precautionary measures and of public information “contributed to” prolonging the crisis and to “the collapse of public trust in scientists and governments” (p. 716).

Framing and verdicts closer to advocacy#

• The introduction’s premises (p. 701).
• The “spirit of the times” claim (p. 701).
• The “substantial inaction” label (p. 702).
• The evaluative case titles.
• Verdicts inside the chronologies:
• “At last” (p. 716)
• “Obvious lack of precaution” (p. 705)
• “Continuous inaction” (p. 703)
• the Phillips Inquiry “not… sufficiently rigorous” (p. 716)
• Calling the hormone ban “an early example” of precaution (p. 715).

Forward-looking statements that can be checked#

• CT over-exposure (p. 704).
• The pace of fisheries reform (p. 703).
• MTBE: “The debate goes on” (p. 712).
• Benzene in petrol (p. 705).
• Chlorine levels restored by 2050–2060 (p. 708).
• TBT deadlines of 2003 and 2008, and OSPAR’s 2020 target (p. 714).
• PCB phase-out by 2010 (p. 707).

Mechanisms and dynamics#

Case-level detail sits under each case above. This section pulls out the patterns that recur across cases.

1. How warnings arose#

• People working with the technology and its pioneers (pp. 704, 707).
• Front-line officials (p. 706).
• Clinicians who noticed clusters of illness (pp. 704, 709).
• Outsiders to the formal expert system:
• a journalist (p. 704)
• inshore fishers (p. 703)
• farmers (p. 710)
• insurers (p. 706)
• a popular book (p. 713)
• media and unions (p. 706)
• a water utility’s loss (p. 712)
• Dissenting scientists (pp. 703, 704, 707, 708, 710).
• Focusing events, i.e. single dramatic harms:
• Yusho (p. 707)
• the London smogs (p. 711)
• the ozone hole (p. 708)
• fishery collapses (pp. 703, 714)
• Love Canal (p. 713)
• vCJD (p. 716)

Action often follows these more closely than it follows slowly accumulating evidence of chronic harm. But not always: the 1952 London smog, for example, drew only “modest remedies” (p. 711). - Measurement is a precondition for any warning (pp. 704, 707, 708, 712, 713, 714).

2. How warnings were contested, ignored or suppressed#

• Findings of “insufficient evidence” (p. 706).
• An official “negligible” risk judgement, later reversed (p. 710).
• Harm blamed on the wrong cause (p. 704).
• Defence of the prevailing paradigm (pp. 704, 707, 709).
• Industry attacks, litigation and denial (pp. 705, 710, 713).
• Knowledge held but not used (pp. 705, 707, 709, 712).
• A regulator disbanding an expert committee and not publishing its finding that some growth promoters were safe (p. 715).
• An official inquiry largely endorsing the government’s decisions, with qualifications on timing and enforcement (p. 716).

3. Burden and standard of proof#

The standard chosen set the timing. - Low standards, fast action: - “reasonable expectation” of damage (p. 708) - “‘uncertain’ long-term effects” (p. 710) - “uncertainty of their effects on humans”, and no identifiable safe level (p. 715) - High standards, delay: - “insufficient evidence” (p. 706) - “high level of scientific proof” (p. 707) - “proof of a causal relationship before appropriating funds” (p. 713) - (The WMO’s 1985–99 finding that halocarbons were responsible “beyond reasonable doubt”, p. 708, is a scientific conclusion, not a standard demanded before action. The US had already acted in 1977 and Montreal was signed in 1987. So it does not belong in this list.) - Courts cut both ways. - The US Benzene Decision was “severely limiting regulatory actions” (p. 705). I read this as raising the burden on the regulator. - The Pfizer case upheld an explicitly “‘precautionary’” EU ban (p. 710). The WTO upheld the EU/French asbestos bans (p. 706), which came after strong evidence of harm rather than on a low standard of proof. - Proof may still not produce a remedy (p. 713).

4. Interests, incentives and funding#

• Funding was lopsided: industry studies outnumbered independent ones (p. 710).
• An industry body recommended its own limit and litigated against a tighter one (p. 705).
• Economic actors weighed risk according to their own exposure:
• insurers (p. 706)
• farmers (p. 710)
• oyster growers (p. 714)
• a water utility (p. 712)
• The industry was split internally (p. 703).
• Trade retaliation (p. 715).
• National interest in cross-border pollution (p. 711).
• Health was explicitly weighed against finance (p. 716).

5. Institutions and implementation#

• Committees and enquiries in place of action (pp. 703, 711).
• In the ozone case, though, research-first frameworks (1977, 1985) were followed by the Montreal Protocol (p. 708). My reading is that they were productive; the annex just lists the sequence.
• Voluntary and professional rules came before law (pp. 704, 705).
• Rules adopted but not implemented (pp. 706, 708, 713, 715, 716).
• Backsliding and swings in policy (pp. 710, 715).
• Political climate (pp. 711, 713).
• Groups with an organised voice were protected first:
• radiologists before patients (p. 704)
• manufacturing workers before users (p. 706)

6. Scope, growth, lock-in and substitution#

• Narrow rules were bypassed (pp. 706, 708, 714).
• Growth outran controls (p. 711).
• Substitutes brought new harms (pp. 708, 712).
• Policy itself pushed one substitute into wide use (p. 712).
• Substitution was sometimes stimulated, as with asbestos under 1980s pressure (p. 706), although the annex doesn’t evaluate those substitutes. Sometimes it was recommended but never implemented, as with benzene in 1939 (p. 705).
• Economic lock-in is barely addressed. The nearest the text comes is “dependence on ozone depleting substances” (p. 708).

7. Time#

• Latency: harm appears long after exposure (pp. 706, 707, 709).
• Persistence: substances stay in the environment (pp. 707, 712, 713, 714).
• Harm continues after action is taken (pp. 706, 708, 713).
• Recovery does happen (pp. 702, 703, 713).

8. Distribution of harms and benefits#

• Harm fell largely on people with little voice:
• women textile workers and radium-dial painters (pp. 704, 706)
• workers generally (pp. 705–707)
• patients, the unborn and daughters (pp. 704, 709)
• neighbours and relatives (p. 706)
• wildlife (pp. 707, 713–715)
• lakes and forests in other countries (p. 711)
• poisoned consumers (p. 707)
• Love Canal families (p. 713)
• Benefits went to:
• producers and users (pp. 706, 707)
• farmers and suppliers (p. 710)
• shipping (p. 714)
• the petrochemical industry (p. 712)
• Costs are counted unevenly.
• The only explicit cost of precaution is EUR 160 million a year (p. 715).
• Costs of inaction appear as scattered figures (pp. 705, 706, 707, 711, 712, 713).
• There is no systematic cost–benefit accounting, despite the brief (p. 701).

9. How key actors thought: mental models and blind spots#

• Overstated benefits: a “magic” material (p. 706); equipment “thought to be safer” (p. 707).
• Beliefs that resisted evidence: belief in DES’s efficacy outlasting a trial, and the placental barrier (p. 709).
• Threshold thinking and “standard risk assessments” (pp. 705, 715), against findings of no threshold (pp. 704, 715).
• Optimistic stock models (p. 703).
• A detection limit treated as the safety standard (p. 715).
• Non-compliance and illegal use found once checked (pp. 715, 716). My inference is that compliance had been assumed.
• Early risk estimates too low (p. 704).

10. Framing and language#

• Precaution claimed without being practised (pp. 702, 703, 713). By contrast, the EU’s 1998 “‘precautionary’ measure” (p. 710) was a real ban.
• Scare quotes: “‘responsible’” management (p. 702).
• “Reassurances” cast as the enemy of precaution (p. 716).
• “Resistance to common sense” assumes the answer was obvious (p. 710).
• A question mark in the hormones title (p. 715).
• “Magic” turning “malevolent” (p. 706).

11. System dynamics#

• Build-up through food webs (pp. 707, 713).
• Long-range transport of pollutants, and critical loads (p. 711).
• Resistance passing between bacteria (p. 710).
• Feed recycling suspected as the source of the epidemic (p. 716).
• Fisheries as whole ecosystems (p. 703).
• A global commons, needing universal participation, finance and control of illegal trade (p. 708).

Transferable insights (technology-neutral)#

1. Delays of decades between early warning and effective action are common. How long they look depends heavily on how “warning” and “action” are defined. - Evidence: Table A2.1 figures run from 5–30 years to over a century, with a median of about 40 (p. 702). Counted from specific warnings, the lags are shorter, but still decades for chronic, diffuse harms (pp. 703–716). - Strength: moderate. The pattern is consistent across 14 diverse cases. But the measure is computed inconsistently and the cases were selected because harm occurred.

2. Separate delays in knowing from delays in responding. They need different remedies. - Evidence:

   • Delays in knowing (harm not yet detectable or attributable): TBT (p. 714); PCBs in wildlife (p. 707); DES crossing the placenta (p. 709).
   • Delays in responding (after the knowledge existed): DES efficacy (1953–71), benzene (1977–87), Danish MTBE (1990–98), Great Lakes cleanup (pp. 705, 709, 712, 713).
   • Strength: strong as a distinction the chronologies support. The annex itself blurs the two.

3. The first harm recognised is often not the biggest. Harms spread to new illnesses, people, generations, species and places. - Evidence:

   • asbestosis, then lung cancer, then mesothelioma; from workers to bystanders (p. 706)
   • from workers to wildlife to the foetus (p. 707)
   • from mothers to daughters (p. 709)
   • from lungs to forests abroad (p. 711)
   • from small boats to offshore shipping lanes (p. 714)
   • from radiologists to patients (p. 704)
   • Strength: strong within the annex, though the sample was chosen for serious outcomes.

4. As evidence accumulates, risk estimates for known hazards tend to be revised toward more protection, and supposedly safe thresholds recede. - Evidence:

   • ICRP 60 put the risk at 4–5 times the earlier estimate (p. 704).
   • Benzene limits fell from 100 to 1 ppm, then disease was found at 1 ppm (p. 705).
   • Findings of no threshold (pp. 704, 715).
   • The API: “only absolutely safe level is zero” (p. 705).
   • Strength: moderate. This is consistent across cases, but the set leaves out hazards whose risks were later revised down.

5. The standard of proof chosen, not the science alone, sets the pace of action. Courts and trade bodies can raise that standard or endorse a low one. - Evidence:

   • Low standards led to fast action (pp. 708, 710, 715).
   • High standards led to delay (pp. 706, 707, 713).
   • A court limited the regulator (p. 705).
   • A court upheld an explicitly precautionary ban (p. 710), and the WTO upheld the asbestos bans, which were not low-bar (p. 706).
   • Strength: moderate–strong. There are many examples, but no counterfactuals. The low/high sorting is mine, not the annex’s. And one low-bar action (hormones) stayed contested.

6. Establishing cause is necessary but not sufficient. The cost of a remedy and the political climate can still block action after proof. - Evidence: Great Lakes (p. 713); BSE, where health was weighed against “financial consequences” (p. 716). - Strength: suggestive–moderate. Stated outright mainly in one case.

7. Adopting a measure is not the same as reducing risk. Measures fail through narrow scope, poor implementation, backsliding, or growth elsewhere. - Evidence: pp. 706, 708, 710, 711, 713, 715, 716. - Strength: strong. Stated explicitly in seven or more cases.

8. Substitutes chosen under pressure to fix one problem can carry hazards that were knowable in advance. A rule in one area can cause harm in another. - Evidence:

   • MTBE: data on persistence existed in 1954 and 1960, and an air-quality law increased its use (p. 712).
   • HCFCs and HFCs replaced CFCs, and radical alternatives were not pursued (p. 708).
   • Strength: moderate. Two well-documented cases. The annex also records pressure that “stimulates substitutes” for asbestos (p. 706), without evaluating them.

9. Claimed benefits deserve the same scrutiny as risks. A large share of hazardous exposure may deliver little or no benefit, and cutting it is cheap. - Evidence:

   • DES was ineffective (1953) and harmful (1978) but was prescribed for decades (p. 709).
   • 20 % of X-rays were unhelpful and 50 % of the collective dose avoidable (p. 704).
   • Trivial uses of radiation (p. 704).
   • Strength: moderate. Well documented in two cases.

10. Knowledge of a hazard often exists inside organisations long before the public or policymakers act on it. Having the information is not the same as using it.

    • Evidence: pp. 705, 707, 712, 716. Unpublished expert findings show the same thing in the opposite direction (p. 715).
    • Strength: moderate.

11. Who funds and does the research shapes what is known and when. Independent research tends to follow recognition of harm rather than lead it.

    • Evidence: pp. 705, 710.
    • Strength: suggestive. Asserted without numbers here.

12. Early warnings often come from outside the core expert and regulatory system, or from dissenting insiders. Acceptance can take a decade or more.

    • Evidence: pp. 703, 704, 706, 710, 713.
    • Strength: moderate. Many instances, but there is no systematic comparison with warnings from inside the system.

13. The ability to monitor and measure is a precondition for warnings and for standards.

    • Evidence: pp. 704, 707, 708, 712, 713, 714.
    • Strength: moderate–strong.

14. When those who profit from an activity are involved in assessing it, errors can lean consistently one way.

    • Evidence: fish stocks were overestimated and fishing pressure underestimated. The correction came from a report commissioned by another stakeholder group (p. 703).
    • Strength: suggestive. One case; the causal reading is my inference.

15. Precautionary language can be adopted without precautionary practice.

    • Evidence: pp. 702, 703, 713.
    • Strength: moderate.

16. Vivid harms that can be traced to a cause and that hit identifiable, organised groups bring action within months or a few years. Diffuse, delayed harms to workers, wildlife or future generations move far more slowly.

    • Evidence: DES, TBT and BSE (pp. 709, 714, 716), and the Montreal Protocol two years after the ozone hole (p. 708), compared with asbestos and benzene (pp. 705–706).
    • Counter-evidence within the set: the 1977 US aerosol ban came before any observed harm, on a theory (p. 708). The 1952 London smog was vivid but drew only “modest remedies” (p. 711).
    • Strength: suggestive–moderate. The pattern is inferred from the timelines. The annex doesn’t argue it.

17. Harms that cross borders need shared monitoring science and cooperative institutions. Jurisdictions that suffer the harm push for action and those that cause it resist. Differentiated obligations and finance widen participation.

    • Evidence: pp. 708, 711, 714.
    • Strength: moderate.

18. Action works, but recovery is slow. Latency and persistence lock in harm long after controls arrive.

    • Evidence: pp. 702, 703, 706, 707, 708, 713.
    • Strength: strong.

19. Precautionary action taken against current expert advice can carry real costs, and possibly costs to legitimacy. Being open about expert advice and about the scope of an assessment matters in both directions.

    • Evidence: the hormones case (p. 715). The ban overrode the Lamming Committee and JECFA and cost about EUR 160 million a year in sanctions. Later EU bodies (1995, 1999) supported the concerns behind it.
    • Strength: moderate for the economic cost; suggestive for legitimacy, which the annex doesn’t discuss. One case.

20. Reassurance and withholding information under uncertainty can prolong a crisis and damage trust in science and government.

    • Evidence: BSE (p. 716).
    • Strength: asserted within this annex. The chronology contains no instances of reassurance.

21. Assessments that assume ideal conditions of use can miss real-world practice: misuse, illegal use, non-compliance.

    • Evidence: pp. 708, 715, 716.
    • Strength: suggestive–moderate. The annex records the misuse and non-compliance. It doesn’t say what the assessments assumed.

22. A technology’s pioneers may see its harms first, yet low-value uses can keep spreading for decades.

    • Evidence: radiation (p. 704).
    • Strength: suggestive. One case.

Limitations, contestation and bias check#

A summary, not evidence#

The annex compresses fourteen chapters into sixteen pages. Many of its strongest claims arrive without data: - industry “attack[s]” on the science (p. 705) - the imbalance in research funding (p. 710) - reassurance undermining trust (p. 716) - “more than 200 deaths” (p. 705) - “400 000 expected deaths” (p. 706)

Treat these as pointers to the full chapters.

Cases chosen because harm occurred#

The cases were chosen “where sufficient is now known about their impacts” (p. 701). So every case involves real harm. The set cannot show: - how often warnings turn out to be false alarms - what precaution would have cost in those cases - what benefits earlier action would have forgone

This is the largest bias in any lesson drawn from the annex. The 2013 volume elsewhere has a chapter on false alarms (Hansen and Tickner, as I recall; cross-check).

Hindsight, despite the “spirit of the times” claim#

• Some first-warning dates only look like warnings in retrospect: the 1376 petition, chloracne in 1899, animal cancer in 1938.
• The chronologies record warnings but rarely the reasons for inaction. For example, they don’t say what the Factory Department weighed in 1911/1917 (p. 706).
• Evidence against the warnings at the time mostly goes unrecorded. Contrary official judgements are listed (pp. 706, 710, 712, 713, 715, 716), but their grounds are given only on pp. 715 and 716.

So the claim is only partly met. The chronologies do document some knowledge that genuinely existed at the time: the API in 1948, the 1953 trial, the 1960 textbook, PCBs in 1937, Denmark in 1990. But the overall design favours the warners.

Problems with the Table A2.1 metric#

(Details under Table A2.1 above.) - The rules for start and end dates are inconsistent. - Some “effective” actions are conceded to be ineffective, or applied only in some places. - The Great Lakes figure can’t be derived from the table’s own entries, and it’s ambiguous which action the PCB figure counts to. - Recent cases are truncated. - The label “substantial inaction” discounts voluntary and partial measures, some of them significant (pp. 704, 706, 708, 710). The qualifier “substantial” concedes that some action happened, but the figures still count those years as inaction.

Dates mixed without warning#

The chronologies stop in 2001 and the annex doesn’t say so. A 2013 reader could take “The debate goes on” (MTBE) or “EU still suffers from sanction” (hormones) as current. Most developments after 2001 are missing.

Authors inside their own stories, and framing#

Several authors were participants in the events they describe. The case titles and chronology entries carry verdicts. Only the hormones case is framed as an open question.

Benefits and costs under-represented#

• Benefits appear only in passing: fire resistance, lighter equipment, fuel savings, productivity, refrigeration, diagnostics (pp. 704, 706–708, 710, 714). None are quantified.
• Costs of control are almost absent, apart from EUR 160 million a year (p. 715).
• MTBE’s air-quality benefits are never weighed against its water costs (p. 712).

This limits any conclusion about the net costs of precaution versus inaction.

Where the evidence is thin or contested#

• Hormones. Expert committees found some growth promoters safe (1982, 1988). But the 1999 EU scientific committee found no safe threshold could be defined for six of them. The table’s “warning” concerns a different pathway from the ban’s basis. The WTO outcome isn’t recorded (p. 715). The case supports lessons about framing, procedure and the costs of precaution, rather than about neglected warnings.
• MTBE. The health effects are only “suspicions” and “indications”. The documented harm is loss of a water resource (p. 712).
• Antimicrobials. The size of the human-health effect was still debated in 2001. The case title presumes the answer (p. 710).
• BSE. Reassurance, the case’s central mechanism, is asserted rather than documented (p. 716).

Possible factual slips and garbles (verify)#

• Beaver Report dated 1952 and called a “parliamentary commission” (p. 711).
• Large combustion plants directive “published, and amended in 1988” (p. 711).
• First imposex “discovered in France in 1970”, although the table cites Blaber (1970) and Smith (1971). Also “sex reversals in oysters” (p. 714).
• “Tolysin” and “vancomycin comes into use” (p. 710).
• “Third world countries” (p. 715).
• The 1952 smog “more than 2 000” deaths is low against common estimates (p. 711).
• Ruminant feed ban dated 1989 (p. 716).
• Misspellings and typos: Flemming, antimcirobials, Peakal, Antartica, Dolore, Santessen, phosphorous, “harmful that”, Creutzfeld, “promotors” (p. 702), “cases studies” (p. 701), “lead the way” for “led” (p. 704).
• “Malevolent material” versus “malevolent mineral” (pp. 702, 706).

Fair in both directions#

The annex does include material that cuts against a simple “regulators are always too slow” story: - Quick action: DES in 1971, the US aerosol ban, French TBT, the first BSE measures. - Successes: SO2, ozone, and wildlife recovery in the Great Lakes. - Precaution adopted against the advice of expert committees: the hormones case. - An official body declining a precautionary classification: the ECB on MTBE. - Courts upholding bans.

It also shows precautionary rhetoric that turned out to be empty (fisheries; the Great Lakes agreement).

Notable quotes#

1. “science’s growing powers of innovation seem to be outstripping its ability to predict the consequences” (p. 701)
2. “based on ‘the spirit of the times’ and not on the luxury of hindsight” (p. 701)
3. “‘responsible’ management (which is not very effective)” (p. 702)
4. “claiming it is doing so precautiously” (p. 703)
5. “Absence of an agreed unit to measure radiation made the introduction of effective standards of protection difficult.” (p. 704)
6. “the only absolutely safe level is zero, but recommends 50 ppm or less” (p. 705)
7. “Prohibition only happened gradually with each government action taken on the basis of a high level of scientific proof.” (p. 707)
8. “The political process failed to stimulate more radical changes towards halocarbon-free and energy-efficient technology.” (p. 708)
9. “Reluctance to undertake costly remedial actions even after causal relationship is proven” (p. 713)
10. “Lamming Committee disbanded by EU and their results were not published” (p. 715)

Open questions#

1. How faithful are these summaries and Table A2.1 to vol. 1 and to its “twelve late lessons”? Checking this needs vol. 1, which is outside this strand.
2. What rule, if any, did Gee (2009) use to define “first early warning” and “effective action”? Would a consistent rule change the picture? For example, counting from the first peer-reviewed evidence of the specific harm to the first binding measure.
3. Do the shorter lags after 1970 reflect real institutional improvement, or just truncation? The later cases in the 2013 volume could test this.
4. Hormones: - What did the Lamming Committee actually conclude, and why was it left unpublished? - What did the WTO decide? - Does this case belong in a set of “late lessons” at all?
5. How far do the 2013 volume’s chapters and panels update or revise these vol. 1 cases? Examples include antifoulant substitutes after TBT, DDT, and lead.
6. The annex frames the problem as one of information and early warning (p. 701). How far does the evidence point instead to failures of response, incentives and institutions? Does the 2013 volume shift its emphasis?
7. What would a matched set of cases show? Specifically, cases where warnings proved false, or where precaution caused net harm.

Pointers for the hindsight strand#

These come from general knowledge and are unverified. Check them against primary sources.

• Ozone: the Kigali Amendment (2016) phasing down HFCs. The WMO/UNEP 2022 assessment projects ozone recovery around mid-century, later over Antarctica. Compare “2050–2060” (p. 708).
• TBT: the Anti-Fouling Systems (AFS) Convention, adopted 2001 and in force from 2008. Also data on recovery from imposex, and the OSPAR 2020 target (p. 714).
• Antimicrobials: EU ban on all growth promoters from 2006; the Pfizer judgment in 2002. What happened to resistance afterwards (p. 710)?
• Hormones: the 1998 WTO Appellate Body ruling (EC–Hormones) and later rulings; the 2009 US–EU arrangement. Compare against EUR 160 million (p. 715).
• Asbestos: the 2001 WTO Appellate Body ruling upholding the French ban. Compare observed mesothelioma deaths in Europe with the 400,000 projection. Also the EU’s lower exposure limit adopted in 2023 (p. 706).
• Benzene: later evidence of effects below 1 ppm; the IARC re-evaluation; EU limits now below 1 ppm (p. 705).
• MTBE: US use largely ended in the mid-2000s, with litigation. Check its status in the EU and its IARC classification (p. 712).
• Fisheries: the 2013 reform of the EU Common Fisheries Policy; the status of northern cod (moratorium reportedly lifted in 2024) (p. 703).
• Radiation: paediatric CT cohort studies from 2012 onward (p. 704).
• PCBs: Stockholm Convention deadlines of 2025 and 2028; whether the EU’s 2010 target was met (p. 707).
• SO2: EEA emissions trends against the “c. 90 %” claim (p. 702).
• Great Lakes: the 2012 GLWQA; progress on remediation (p. 713).
• BSE: the final vCJD toll; evaluations of the feed and offal bans (p. 716).

Audit log#

Independent audit against the text extract (pp. 701–716) and spot-checks of PDF pp. 704, 712 and 713, which match the extract. All quotes in “Notable quotes” were checked and are verbatim. The author and page attributions are correct.

• How this was read: vol. 1 was not opened because of the task’s file restrictions, not the strand rules (Strand A permits it).
• Summaries: “date from about 2013” softened to “at least edited for the 2013 volume”. The Pfizer note is now described as later and undated, not “2013”.
• Panels and opposing views: this said opposing positions appear “only” in chronologies with reasoning “never” given. Corrected: the hormones summary also records the committees’ finding of safety, and some reasoning is given (pp. 715, 716).
• Lamming/JECFA findings corrected from “growth hormones safe” to “some growth promoters safe” (1982 entry wording).
• Introduction: added the selection criterion’s purpose (“how well they were addressed by governments and civil society”).
• Table A2.1: DES moved to “matches after rounding” (the dates give 33–47). PCBs reclassified from “can’t be derived” to “ambiguous”, because the listed 2010 phase-out gives 111 and c. 100 lies between. Great Lakes range corrected to 6–10 (or at most 17). Radiation “statute” changed to “regulations”. SO2 action text restored to “EU and other”. Fisheries 1376 “petition” changed to the source’s “call”.
• Fisheries: added the 1866–93 “uncertainty” context, the 1989 “government assessment”, the 1990s ecosystem approach and the 2001 “Positive changes are occurring”. Observation 1 softened: knowledge gaps and assessment error also played a part. Observation 2 now counts the three recorded errors. Observation 3 flagged as inference. Observation 7 “only case” corrected (radiologists are another).
• Radiation: “injuries and fatalities” restored; Stewart’s work credited “along with that of other researchers”; added the 1977 ICRP entry. Observation 4 notes Rollins’ 1904 warning about patients. Observation 6’s “cheap to reduce” flagged as inference.
• Benzene: added the 1926 Greenburg entry. Observation 2 notes the API’s 50 ppm was above the ACGIH’s 35 ppm that year. Observation 3’s “raised burden of proof” flagged as a gloss on “severely limiting regulatory actions”.
• Asbestos: added the 1906 recommendation of controls, 1935–49 lung cancer cases and the 1969 “improve controls”. Observation 4 (“rather than new science”) and observation 6 (“upheld precaution”) softened. Observation 7’s claim that the 400,000 figure is “repeated from 2001” is now marked unverified.
• PCBs: added the 1970s seal infertility entry.
• Halocarbons: the political-process criticism is now attributed to the unsigned summary of Farman’s chapter. “Discovered” changed to “co-discovered”.
• DES: added the 1939–48 entries (including the 1948 study that increased use) and the 1972 registry. Observation 2 reworded from “belief” to “prescription”, noting that the 1971 withdrawal was for harm, not inefficacy.
• Antimicrobials: restored the 1985 ban’s first ground (“antibiotic resistance in animals”). Observation 4’s “for market reasons” corrected to the source’s “health and consumer concerns”.
• SO2: added the late-1960s OECD group and the 1972 OECD programmes. Observations 3 (victim and exporter states) and 4 (science “made agreement possible”) flagged as inference.
• MTBE: added the 1990 carcinogenicity investigations and the 1998 Danish action plan. California 1999 corrected from “commits” to “recommends”. Observation 1 (“weren’t evaluated”) softened, because foreseeability is the chapter’s open question. Observation 3 specified as Danish.
• Great Lakes: “thinning eggshells” corrected to “changes in eggshell quality”. Added the 1974 health concerns and dieldrin 1973. The 2000 entry now quoted precisely. Observation 4’s “on paper only” softened to “not properly implemented”.
• TBT: the fuel quote’s subject corrected (preventing fouling, not the paint). Added the early-1970s growth in use, January 1987, and the 1991 ban as a retail-sale ban. Observation 4 flagged as inference.
• Hormones: “suppression of expert advice” softened. Added the EU’s stated reason (narrow scope) and the later 1995 and 1999 EU findings that support the ban’s concerns. Observation 7 notes the “can be seen as” hedge.
• BSE: added Southwood’s recommendation. The Phillips critique is now attributed to the chronology, “probably” the case authors. The feed-loop amplification is flagged as a gloss on “suspected”.
• Lessons: BSE lesson restated with the source’s “contributed to” and “collapse of public trust”.
• Mechanisms: added the London smog as a counter-example to focusing events. Hormones “suppressing” and Phillips “clearing” softened. Removed the WMO “beyond reasonable doubt” from the high-standard/delay list, because it is a scientific conclusion, not a demanded standard. Asbestos WTO ruling no longer called a “precautionary”/low-bar ban. “Industry shaped the standards” made specific. Ozone “research-first productive”, asbestos “substitution worked” and “compliance assumed” flagged or softened. The page citation for precaution claimed without being practised corrected (p. 710 changed to p. 702).
• Insights: 5 (WTO asbestos not low-bar; the sorting is mine), 8 (“successful substitution” changed to “stimulates substitutes”), 16 (ozone 1977 and the 1952 smog added as counter-evidence), 19 (legitimacy now suggestive; later EU findings added) and 21 (downgraded to suggestive–moderate).
• Limitations: hindsight bullet corrected on contrary judgements. PCB metric claim corrected. “Substantial inaction” qualifier nuance added. SCVPH 1999 added under hormones. Typos list extended. “Precaution overriding science” changed to “against the advice of expert committees”.
• Digest: matching corrections made (summary dating, the Table A2.1 excerpt flagged as selected rows with the median, the 400,000 base year, DES US versus worldwide, MTBE wording, hormones balance, BSE lesson wording, insights 14 and 16, the PCB caveat, and the introduction’s aim).
• No mentions of contemporary technologies or companies outside the report were found. None were added.