LL2-04 — Ch4 Too much to swallow: PCE contamination of mains water#

Late lessons from early warnings: science, precaution, innovation (EEA Report No 1/2013), Part A “Lessons from health hazards”, Chapter 4. Report pages 76–91 (PDF pages 78–93). Main text pp. 76–88; Table 4.1 and references pp. 89–91. Panel 4.1 (pp. 84–85) and Panel 4.2 (p. 87) sit inside the chapter. Photos on pp. 86 and 87 are decorative (pipes; garments on a rail).

Reading note: I read the full text extract page by page to the final marker (PDF 93 / report p. 91). I checked the opening summary box (p. 76), both panels including their tables (pp. 84–85, 87) and Table 4.1 (p. 89) visually in the PDF. The extract is accurate for these; the only garbled element is Figure 4.1 (p. 78), a set of chemical structure diagrams with no analytical content beyond the text. Author details come from Annex 1 of the same report (pp. 695–697).

Convention: “[Background — verify]” marks points I add from general knowledge, not from the chapter. These are for the hindsight pass and must not be read as the chapter’s claims. No web sources were consulted when the notes were drafted; the audit consulted one PubMed search listing (author’s co-authored drinking-water studies), marked “outside source” where used.

Authors and standpoint#

Chapter author: David Ozonoff (sole author, p. 76). Annex 1 (p. 696) gives him as Professor of Environmental Health, Boston University School of Public Health. He was the founding Chair of its Department of Environmental Health (1977–2003). His research covers “community health effects of toxic exposures, especially from hazardous waste sites”, the mathematical foundations of epidemiology and “the use of scientific evidence in court”. He has been “principal or co-investigator of several major studies of waste sites” and directed a Superfund Basic Research Program funded by the National Institute of Environmental Health Sciences (NIEHS).

Relevant standpoint signals: - He has previously written critically about Johns-Manville on asbestos. The chapter cites his own “Failed Warnings: Asbestos Related Disease and Industrial Medicine” (Ozonoff, 1988) as showing that Johns-Manville “wilfully disregarded and concealed scientific evidence” on asbestos (p. 84). - The chapter is organised around legal questions of foreseeability (“Foreseeable harm?”, p. 77; “could this have been foreseen?”, pp. 77–78), state of knowledge (Johns-Manville’s defence that it “did not and could not have known”, p. 77) and duty of care (pp. 76, 88). This is the vocabulary of product liability. It fits his stated interest in scientific evidence in court. - The detailed account of the 1966–1968 Providence pipe tests (p. 77) carries no source citation. Neither do the absence-of-record statements (“There is no indication…”, “Nor is there any record…”, p. 77). It reads like an account built from documents such as litigation discovery records. The chapter does not say where the material came from, and it makes no disclosure of any role in the pipe litigation. This is an inference, not a finding. It is flagged under Open questions. - Undisclosed research on this exposure (outside source, confirmed). A PubMed listing (checked during the audit) shows Ozonoff as co-author of case-control studies of PCE-contaminated public drinking water in Massachusetts: Aschengrau, Ozonoff et al. 1993, “Cancer risk and tetrachloroethylene-contaminated drinking water in Massachusetts”, Arch Environ Health 48(5):284–92 (PMID 8215591); and Aschengrau, Rogers and Ozonoff 2003, breast cancer on Cape Cod, Environ Health Perspect 111(2):167–73 (PMID 12573900). The chapter mentions that “several environmental studies” of contaminated drinking water gave “mixed” results (p. 86) but cites none of them, including his own. Whether those studies concern exactly this pipe source was not re-checked in the audit (abstracts not accessible); treat that link as [Background — verify]. - Evident stance: firmly pro-precaution and critical of industry. The overall judgement is that the harm was foreseeable and that “nobody made them care” (p. 86). The chapter is candid, though, that there is no evidence of concealment by Johns-Manville over PCE (p. 84). It offers corporate indifference as the plausible explanation, which is more careful than a cover-up story.

Panel 4.1 “Differences between risk assessments drawn from the same basic data”: Christina Rudén (pp. 84–85). Annex 1 (p. 697): Professor, Department of Philosophy, Royal Institute of Technology (KTH), Stockholm. Her research is on regulatory aspects of toxicology, with the stated aims of improving “the scientific basis of risk assessment and management” and strengthening protection of health and the environment. She sits on the Eurotox Executive Committee and the Swedish Chemicals Agency’s Supervisory Council. The panel is analytical and measured. It treats divergence among assessments as partly legitimate (knowledge changes over time) and partly the result of data selection, interpretation and assessment policy. Its recommendation is about transparency and consistent terminology, not a charge of bad faith. Its tone is noticeably cooler than the main text’s closing claims about manufactured doubt (p. 88).

Panel 4.2 “Wet cleaning technology eliminates PCE use in dry cleaning”: Joy Onasch (p. 87). Annex 1 (pp. 695–696): she oversees the community programme at the Toxics Use Reduction Institute (TURI), UMass Lowell, “works with the dry cleaning sector to reduce their use of Perchloroethylene”, and is an engineer and registered Professional Engineer. The panel is openly policy-oriented. Its italic lead paragraph argues for “further policy measures” to phase out PCE and for utility partnerships (p. 87). Its main quantitative evidence is her own single-facility case study (Onasch, 2011).

No panel dissents from the chapter. There is no industry or regulator response panel in this chapter, so the industry position appears only as the chapter author describes it.

Section-by-section notes#

Opening summary box (p. 76)#

• The case. PCE was used to make plastic linings for drinking-water pipes “in the late 1960s and 1970s”. The pipe was a “new and relatively untested type” and went into “over 700 miles of New England’s water distribution systems”. Leaching was found only in 1976, and supplies “still today require continuous remediation” (p. 76).
• Prior knowledge. There was “a substantial amount of scientific information” on PCE hazards before production. It did not cover today’s main concerns (carcinogenicity, teratogenicity, low-level effects), “but many early warnings suggested the need for caution” (p. 76).
• Later evidence. Studies covered hookworm side effects, occupational groups (aircraft workers, small firms in countries requiring biological monitoring, dry cleaners) and drinking-water/cancer links. “Results were mixed and the chemical industry consistently denied that PCE was a human carcinogen” (p. 76).
• Scope. The pre-1970 toxicity record, and “the failure of one manufacturer, Johns-Manville Corporation, to recognise the warning signals”. It “examines why a new product was deployed without thought to the public health consequences and why evidence of the potential hazard was ignored” (p. 76).
• Thesis (repeated almost verbatim on p. 88). “The science has not been hidden. It has been ineffective in guiding and catalysing action.” Whether the problem is “a failed duty of care or a lack of clarity about what evidence will trigger action”, the interpretive argument “is irresolvable within science itself”. There are “no overarching criteria from the philosophy of science that can dictate a solution” (p. 76; p. 88 has “force a solution”).

Introduction and 4.1 PCE linings in water mains (p. 77)#

• General claim. In institutions “large and small”, foresight “based on existing information — is often absent”. The late 1960s and early 1970s were “a strategic and historical turning point” in awareness of environmental carcinogens and teratogens. The public health implications of low-level exposure via water mains “remain unresolved” (p. 77).
• The pipe. It was installed 1969–1979 and is “now known to have caused widespread PCE contamination of water supplies” in Massachusetts (Demond, 1982; MDEE, 1982; Larsen et al., 1983). The reference titles call it “vinyl-lined asbestos-cement pipe” (pp. 89–90).
• Origin. In the early 1960s the Providence (Rhode Island) Water Supply Board wanted to replace cast-iron mains in low-flow areas “troubled with colour and taste problems”. Johns-Manville, “a manufacturer of asbestos cement water mains”, tried clear plastic linings: the plastic was dissolved in PCE and the slurry painted inside the pipe (p. 77).
• 1966 trial. The water had a “slight chemical taste and odour, whose origin was not revealed by routine water quality tests like pH, alkalinity and hardness” (p. 77).
• Early 1968 test.
• Air trapped in the new clear-lined pipe smelled “similar to chloroform and strongly resembling a commercial dry-cleaning fluid used at a local cleaning plant”. Water tests found nothing.
• Under static conditions a slight odour persisted “even after substantial volumes of water had run through the pipe”.
• Company representatives were shown the odour at first hand.
• At a newly installed 277-foot, eight-inch pipe, neither they nor the water supply chemist smelled anything.
• The company said the odorous pipe had been kept in brown paper and so had “not ‘cured’ completely” (p. 77).
• Argument from the record. “There is no indication that the taste and odour incident prompted Johns-Manville to investigate the curing process to evaluate whether PCE remained in the liner.” There is also no record of any tests beyond routine measures, “which did not indicate the nature or amount of organic contaminants” (p. 77).
• Discovery. “Not until 1976, when over 700 miles of this pipe had been installed … was it accidentally discovered that PCE had been leaching” (p. 77). The chapter does not say how or by whom.
• My analytical note, grounded in p. 77. The product was designed for low-flow areas, and the company’s own tests showed the odour persisting under static conditions. The intended operating environment was therefore the one most favourable to residual solvent reaching the water. The chapter places these facts side by side without drawing the link.

4.2 Foreseeable harm? and 4.2.1 An insight from the early history of PCE (pp. 77–78)#

• Framing. “A classic case of deploying a new product without considering the public health consequences” (p. 77).
• In the “ensuing battle over who should pay for the damage”, Johns-Manville argued it “did not and could not have known” that PCE was a chemical of public health concern (p. 77).
• Post-1970 work revealed potential “various cancers, birth defects and autoimmune disease”. This sentence is uncited (p. 77).
• The framing questions: could this have been foreseen? If not, why not? If so, “what factors prevented adequate foresight?” (pp. 77–78).
• Faraday. Faraday made hexachloroethane (C2Cl6), “the first chlorinated hydrocarbon to be synthesised”, and got PCE by heating it (Williams, 1965).
• His interest in chlorine is traced to a “philosophical dispute” between Humphry Davy (“Humphrey” in text) and Lavoisier, over reconciling Newtonian mechanics with free will.
• Davy and Faraday were “Kantians and deeply religious”. The result was disputes with Berzelius and Dalton (Sharlin, 1966) (p. 78).
• Lesson. The key point from this period is “the way that non-scientific concerns can distort scientific disputes. Ideology can make it impossible for protagonists to reverse a course of action or alter a position. But other interests, such as money, market share and reputation, can have similar effects” (p. 78).
• Caveat the chapter states first. “Of course, disputes and doubts are normal in science, particularly when the consequences matter” (p. 78). The corollaries below are framed as consequences of that normal condition, not as a claim that all dispute is illegitimate.
• Two corollaries (p. 78). These are among the chapter’s most portable claims. 1. A finding may become disputed, “whether real or manufactured to prevent action”, “because the outcome matters to someone with the means to challenge the finding and delay action”. 2. “If nobody cares or nobody with means cares, there will be little pressure to challenge a scientific finding or explore an issue in greater detail”. “Results of potentially great significance in other contexts may fail to influence the public health landscape”.
• Note. The Faraday episode is analogy and scene-setting, not evidence about PCE. [Background — verify: Lavoisier died in 1794, so the “dispute” was with his chemical system, mainly over whether chlorine contains oxygen. Faraday’s C2Cl6/PCE work dates to about 1820–21.] Table 4.1 (p. 89) prints “1860 PCE synthesised”, which conflicts with the Faraday attribution. The conflict is also internal: p. 79 says that “for more than a hundred years after its discovery by Faraday” PCE had no significant commercial use and little literature “until the 1920s”, which implies discovery before about 1825, not 1860.

4.3 PCE and the chlorinated ethylenes (pp. 78–79)#

• The chemical family. Figure 4.1 shows the series formed by swapping hydrogen for chlorine on the ethylene backbone, one atom at a time:
• vinyl chloride (“a known human carcinogen”);
• the dichloroethylenes (DCEs);
• TCE;
• PCE.

All are feedstocks, and several are degreasing or dry-cleaning solvents. TCE and PCE were “among the highest production volume chlorinated solvents in the twentieth century”. Their medical uses (TCE as anaesthetic, PCE as anthelmintic) go back “almost a century” (p. 78). - Key interpretive claim. “The familiarity and benefits of these substances should have alerted us to the fact that exposure to these chemicals has biological effects that could also be harmful” (p. 78). - Water mains in context. Lining water mains was “just one of many applications”, but PCE’s use near drinking water came “at a time when problems could have been foreseen” (pp. 78–79).

4.4 Discovery of PCE’s toxic effects, and Phase I: 1925–1940 (pp. 79–80)#

• Why PCE entered medicine.
• For a century after Faraday there was little commercial use and little toxicity literature.
• Hookworm afflicted “millions of rural poor” and caused substantial economic losses (Rosenau, 1935).
• The Rockefeller Foundation campaign used “relatively toxic” thymol. Carbon tetrachloride (CCl4) followed, then related compounds including PCE (p. 79).
• Start of evaluation. Hall and Shillinger (1925) began “a process of toxicological evaluation of PCE that has continued to the present day”. New uses followed: dry cleaning in 1934 and degreasing in 1939.
• Two phases. Oral medicine (1925–40), then inhalation (1940–70) (p. 79).
• Hall and Shillinger’s evaluation (p. 79).
• Dog tests: 3 of 55 died at supposedly therapeutic doses, and none of those had the largest doses. They were “the first of many to comment on potentially significant differences in individual susceptibility”.
• Self-test: one researcher took 1 cc and reported muscle relaxation, slight cerebral discomfort and a levitation dream.
• They judged PCE’s safety comparable to CCl4 (a known liver toxicant) and recommended hospital trials, with contraindications for alcoholism, liver disease, infection and debility.
• Growing use despite harm. Use grew despite signs of harm “even in relatively small quantities” (Manson, 1934; AMA Council, 1936; Wright et al., 1937; Fernando et al., 1939). PCE was “a commonly used drug” but “not a completely safe one” (p. 79).
• Sandground (1941). He reported two cases of unconsciousness after a normal therapeutic dose. His own conclusion was that “for want of a better drug [these cases] should not discourage the use of tetrachlorethylene [PCE]”. He added Maurice Hall’s rule that no anthelmintic can be assumed “entirely safe for human use until there are reliable reports on at least a million treatments without any untoward effects” (p. 79). Note: the contemporaneous actor accepted the risk given the benefit; the chapter uses him for the “million treatments” caution, not as a critic of the drug’s use.
• Phase I summary (p. 80).
• Serious effects appeared in “a small percentage” and were “not necessarily related to the dosage”.
• Some human effects were “extremely serious or fatal at doses as small as half a teaspoon (2–3 cc)”.
• The number of other severe side effects “is difficult to estimate, although the question was being considered at the time that the PCE-lined pipes were installed (Bwibo, 1969)”.
• “There remained uncertainties about the degree of absorption of the drug in humans and the individual variation in susceptibility”.
• “PCE’s potentially lethal side-effects were tolerated because hookworm disease was a major public health problem”. That was “a balance not found in newer uses”.

4.4.2 Phase II: 1940–1970, degreasing and industrial uses (pp. 80–82)#

• Shift in exposure. Therapeutic data were “sufficient to arouse concern”. Exposure was now mainly by inhalation and skin, and chronic effects joined acute ones. Epidemiology “was still primitive and did not enter into most decisions” (p. 80).
• 1930s warnings (p. 80).
• Zernik (1933) set Lamson et al.’s (1929) optimistic assessment against Beyer and Gerbis (1932), who reported fatal stomach and liver disease after chronic inhalation of a PCE-based solution. Other ingredients could not be excluded.
• Alice Hamilton (1936, NEJM), “an industrial health pioneer”, made two points. Animal data may be hard to apply to humans (humans showed more liver damage and less kidney damage than animals). And low chronic exposures can do more damage than large acute ones, as with lead and benzene.
• A Massachusetts physician in the discussion of her paper complained that manufacturers marketed products “under trade names, with little information for physicians about the effects of the chemicals” (chapter’s paraphrase).
• The chapter then quotes, as “Hamilton (1936) had observed”, that this is “a problem for the general practitioner since the use of these solvents in industry is increasing by leaps and bounds each day. I have seen many individuals … suffering from conditions brought about by prolonged exposure” (p. 80). Attribution note: the quote follows the discussant’s point and reads like a practitioner’s first-person observation; the chapter attributes it to Hamilton (1936), which may mean the published paper including its discussion. Treat the speaker as uncertain.
• St George (1937) said chronic intoxications are “difficult to recognise” and that chemicals may be transformed or retained in the body. He called for worker instruction and container warnings, e.g. use only “in a room with at least one window wide open”.
• Carpenter (1937) (pp. 80–81).
• Rats: no pathology at 70 ppm over 10 weeks; kidney swelling and minimal liver changes at 230 ppm.
• Humans (himself and colleagues), 500–5,000 ppm: odour detectable at 50 ppm, irritation, nausea at 500 ppm for 2 hours, and CNS depression at higher levels.
• He concluded that a safe daily level “probably lay somewhere between 100 ppm and 500 ppm”, pending more human data.
• The 1940s (p. 81).
• “Little original work was done, perhaps because of the war effort and the pressures of industrial production”. General reviews (Lehmann and Flury, 1943; Sappington, 1943) recycled the drug-use data and the findings of Carpenter (1937) and Barrett et al. (1939).
• Morse and Goldberg (1943) said Carpenter “by no means clarified the toxicity”. Headache, nausea and dizziness were common among degreaser operators “even when concentrations were well within the generally accepted toxic limit”.
• Writers were “alarmed by the lack of hard data and the misperception that PCE was non-toxic based on its therapeutic use”.
• Fairhall’s 1949 textbook said PCE is not harmless, can be more toxic than CCl4 under some conditions, and affects individuals very differently.
• Dow, 1952 (Rowe et al.), “a major manufacturer of PCE for degreasing use” (p. 81).
• Focused on acute effects, in laboratory animals and human volunteers.
• Proposed an average limit of 100 ppm and a maximum of 200 ppm.
• Identified eye irritation and central nervous system depression as “the prime toxic effects”, and “considered serious organic injury to be unlikely”.
• Coler and Rossmiller (1953) (p. 81). (“Rosmiller” is used once in the text.)
• A pump factory degreasing with 99% PCE at 200–400 ppm.
• A 35-year-old worker had severe cirrhosis and ruptured oesophageal varices.
• Co-workers reported intoxication, passing out, eyes that “did not coordinate”, and slowed thinking and memory.
• 3 of 7 tested had abnormal liver function. The authors concluded that PCE liver toxicity “should be investigated more thoroughly rather than disregarded”.
• The chapter’s verdict: “contemporary clinical observations contradicted the Dow studies”.
• Audit note: the measured worksite levels (200–400 ppm) were at or above the 200 ppm ceiling Dow had proposed. What the clinical series contradicts is Dow’s judgement that serious organic injury was unlikely, not its exposure limits. It is a small clinical series (one cirrhosis case; 3 of 7 abnormal liver tests), and the chapter does not discuss other possible causes.
• Lob (1957), “Les dangers du perchlorethylene” (pp. 81–82). The chapter paraphrases his view that “a toxic industrial chemical is often considered harmless until experience shows the opposite”.
• Animal data were meagre and heterogeneous, and TCE showed how poorly animal results transfer.
• He added ten poisoning cases: 1 fatal, 2 severe chronic cases with autonomic damage, and 7 milder cases that resolved when exposure stopped.
• Plaa et al. (1958). “No correlation between the dose that caused liver damage and the lethal dose”. Newer animal methods were “beginning to show inconsistent effects” (p. 82).
• Dow’s 1960s programme (Stewart et al. 1961–1970; Irish 1962; Rowe et al. 1963; Gehring 1968) (p. 82).
• It used new gas chromatography with IR or electron-capture detection.
• It found slow excretion and accumulation: “acute exposure to PCE might, in fact, be a chronic exposure from the body’s standpoint”.
• Liver tests may not become abnormal for 2–3 weeks after exposure.
• Irish (1962): “the mistaken perception that PCE was relatively non-toxic encouraged careless use”.
• Stewart et al. (1970) found an “unexpected prevalence of light-headedness and abnormal neurological results … at the lowest exposure levels”, which “the Dow authors could not interpret”.
• Smyth et al. (1969). 27 industrial chemicals were given to rats in all possible pairs (the reference says “intubated”, i.e. by mouth; 351 pairs by my arithmetic). The endpoint was death. “Most combinations showed no tendency to produce lethal effects in excess of” additivity. Of the 9 combinations that deviated significantly, 4 contained PCE, “making it the chemical most often associated with causing a net effect greater than the sum”. The chapter says this “strongly suggested” that PCE “could have a potentiating effect on the toxicity of other chemicals (and vice versa)” (p. 82). Audit note: the signal is acute lethality in rats, from a handful of pairs; the step to low-level chronic human exposure is the chapter’s.
• 4.4.3 The view from 1970 (p. 82). The chapter’s synthesis of what was knowable when the pipes went in:
• continuing workplace poisonings (CNS and liver);
• uncertain thresholds, “with some writers considering PCE to be more dangerous than conventionally believed”;
• big advances in measurement;
• Dow’s in-house research “published in the open scientific literature”;
• wide variation in susceptibility;
• slow excretion causing “a chronic, low-level internal exposure”;
• suspected synergy “of unknown magnitude”.

4.5 Implications for PCE use in water supply infrastructure (pp. 82–83)#

• Aesthetic standard. The 1968 US Public Health Service Drinking Water Standards (as dated in the chapter) said water “should contain no impurity which would cause offence to the sense of sight, taste, or smell”. The purpose was to stop consumers turning to “alternative but less safe sources”. The chapter’s claim is hedged: public health experts “might have worried that the new pipes could cause a health problem on that basis alone, as a chemical odour was an initial concern” (p. 82). [Background — verify: the standards then in force are usually dated 1962.]
• Individual variation. Responses varied with inherent factors and with health, diet and other exposures (p. 83).
• Risk–benefit does not transfer. Side effects tolerated in hookworm treatment were “of little relevance to water managers who had no interest in purveying an anthelmintic drug through the water mains” (p. 83).
• “Bathtub” accumulation (p. 83).
• The author’s own estimate “using data from Stewart et al. (1965)” (the chapter says it “can be estimated”): elimination of about 25% a day via the lungs, a half-life of about 2.5 days, and steady state after about two weeks (“four to five half-lives”).
• Solubility caps PCE in water at 100–150 ppm, and “no measurements of PCE appear to have been made at the time the pipes were installed”.
• Worst case, 125 ppm × 2 L/day: a steady body burden of about 1 g, “close to the dose used to treat hookworm”.
• At lower exposures the internal burden is about four times the daily intake.
• My check: 250 mg/day ÷ 0.25/day = 1,000 mg, so the arithmetic is consistent. The chapter labels it a worst case.
• Audit note: the comparison sets a steady-state amount held in the body against a single oral therapeutic dose. These are different quantities (a constant internal burden versus a bolus), so “close to the dose used to treat hookworm” is an order-of-magnitude illustration, not a like-for-like dose comparison. The chapter does not present it as more than that, but it should not be quoted as one.
• Other grounds for concern. Inconsistencies between animal and human data, and possible synergy, “might” also have troubled experts (p. 83).
• Hindsight rebuttal. Experts “would have been unlikely to view the presence of any PCE in their water favourably unless it was unavoidable. This is not merely a statement based on hindsight” (p. 83).
• Colten (1991): by the early 1950s, agencies, professional bodies and trade associations “all publicly recognised the hazards posed by the surface disposal of liquid wastes”.
• Amter and Ross (2001) and other historians: that chlorinated solvents contaminate groundwater was “generally understood in the 1940s–1960s”.
• Entry via the pipe lining rather than land disposal “was irrelevant” (p. 83).

4.5.1 Why did Johns-Manville fail to foresee the potential harm? (pp. 83–86)#

• Could have known. Johns-Manville may have seen the harm only with “hindsight”, but “it is quite clear they could have done so”. Its options were to redesign the product, warn water managers about an “insufficiently ‘cured’ product”, or carry on as before (p. 83).
• Did not consider it. “The company did not even consider or worry about the possibility of insufficient curing, and once the problem was detected, it denied that there were any health hazards — a necessary position if it were to avoid paying damages for a faulty product” (pp. 83–84).
• Nuance: on p. 77 the company’s own representatives cited incomplete curing to explain the odour. The chapter’s point is that they did not follow that implication through to pipes in service.
• Indifference, not concealment. There is “abundant evidence” that Johns-Manville “wilfully disregarded and concealed” asbestos evidence (Ozonoff, 1988), but “there is no such evidence in relation to PCE”. The next step is explicitly conditional: “Assuming that knowledge was not hidden, the proposition that Johns-Manville was merely indifferent to these dangers is a plausible explanation for its action”, with “many similar examples involving other companies in this period” (uncited) (p. 84). The chapter does not rule concealment out; it says there is no evidence of it.
• Evidential norms. “The lack of epidemiological evidence would not have been a reason to delay action”. Reconfirmation by epidemiology “was not yet the norm” (p. 84).
• Restricted circulation.
• Evidence “mainly circulated in the restricted arena of medical specialist literature”.
• Physicians and workers did not know what they were exposed to, which kept solvent toxicity “off the agenda”.
• That also kept “workers, their unions and their advocates” out of the conversation, so water contamination was “unlikely to be well recognised” (p. 84).
• 1970 turning point. The US moved “from minimal federal engagement in workplace and environmental concerns to a prominent role”, with new concern about water (pp. 84, 86).
• Core diagnosis (p. 86).
• PCE science “never figured in the water mains product design. It was ignored or invisible”. Johns-Manville did not care enough because “nobody made them care”.
• Firms “have interests and intentions, which are rarely related to public health and environmental concerns. The job of a company is to make money for its owners”. The product’s nature was of “secondary importance”.
• Remedy (p. 86). “Criminal and civil liability, both of which rely on state enforcement of legal rules”. Moral pressure and voluntary standards “must all pass the acid test: are they sufficient to make the company care?”
• Power and uncertainty (p. 86). Whether a problem can be ignored depends on “the state of knowledge, power relations, economic considerations and political arrangements”. High stakes bring resources “more at the disposal of large corporations”. So “uncertainty favours the side of inaction”.

Panel 4.1 (Rudén): Differences between risk assessments drawn from the same basic data (pp. 84–85)#

These views are Rudén’s alone and should not be merged with Ozonoff’s.

• TCE review. Rudén (2002) reviewed 29 TCE carcinogenicity assessments from 1973–1997 to see how “selection, interpretation and weighting of primary data” drove different conclusions (p. 84). The results table (“Classification of TCE risk in risk assessment reports 1973 and 1997”, p. 84):

• Pattern by assessor type (p. 84).
• “Eight of the 10 evaluations that identified a risk for animals but not for humans were conducted by international organisations or industry”.
• “Eight of the nine assessments that concluded a human risk based on animal evidence were conducted by government or academic authors”.
• Rudén reads this as possibly reflecting “more risk-averse assessment policies” in government and academia, and “a tendency for industry to apply less precautionary criteria”.
• The variation persisted in 1995–96, “when the evaluating bodies were working from the same available body of knowledge”.
• PCE conclusions (table, p. 85).
• ECETOC (1999): animal evidence plausible, human risk not plausible.
• IARC (1995b): animal and epidemiological evidence plausible, human risk plausible.
• ACGIH (1993): animal evidence plausible, human risk not plausible (animal data not considered relevant).
• MAK (1992): animal evidence plausible, human risk plausible (from animal data).
• Legitimate and discretionary divergence. “Scientific knowledge increases over time … a time-dependent and natural part of the scientific and regulatory process. However, risk assessors may also select different sets of data … and may interpret key studies in different ways” (p. 85).
• Worked example: PCE and non-Hodgkin’s lymphoma (NHL) (Rudén, 2006, p. 85). IARC 1995 read the evidence as positive, citing raised NHL risk in Blair (1990), Spirtas (1991) and Anttila (1995). ECETOC 1999 read it as negative:
• it discounted Spirtas because the study was started over a priori concerns about lymphatic cancer;
• it called Anttila’s excess non-significant;
• it said Blair “did not provide results for NHL as a cause of death”;
• it cited no excess in Ruder (1994);
• it concluded the studies were “either negative or were not sufficient”.

() ECETOC called Anttila positive on its p. 143 and negative on its p. 136, an internal inconsistency. () ECETOC did not treat Spirtas as a key study. - Context. Such differences are “not uncommon” (EEA, 2001) and are “attracting increasing attention from regulators and policymakers” (EEA, 2008) (p. 85). - Recommendation. Evaluators “must communicate better about the approach they use to evaluate the strength of evidence and scientific uncertainties” and use “clear and consistent terminology” (p. 85). - Note.* The panel does not say that ECETOC is industry-funded. [Background — verify: it is a chemical-industry-funded scientific association.] The TCE statistic groups “international organisations or industry” together.

4.6 The view from 2013 (pp. 86–88)#

• Shift in concern. Pre-1970 work “had revealed the outline of acute reactions to PCE, and established concern about chronic and delayed effects”. Soon after installation, “an entirely new dimension” emerged: carcinogenicity, teratogenicity and other effects (p. 86).
• Trigger. The early-1970s finding that vinyl chloride monomer is a human carcinogen (cross-reference to Chapter 8) “immediately raised the question” for TCE and PCE. By then dry cleaning exposed workers and patrons, and “numerous small firms” produced groundwater contamination through improper disposal (p. 86).
• Animal bioassays (NCI 1977; Mennear et al. 1986). Both solvents proved animal carcinogens, “although by this time the methodology and validity of animal bioassays had become a matter of dispute and no finding went unchallenged” (p. 86).
• Epidemiology. Attempts at verification “were difficult because of the long latency for cancer, lack of exposure information and low statistical power of most studies” (p. 86). Results from occupational and drinking-water studies were mixed, and industry “consistently denied that PCE was a human carcinogen” (pp. 86–88).
• Symmetric scepticism. “In each of the individual studies it was possible to find limitations or alternative explanations for positive results (and for negative ones).” Arguments “were spun out” in turn: first epidemiological, then “sophisticated toxicological arguments as to why PCE could be a carcinogen in rodents but not a carcinogen in humans”. So scientists “could look at exactly the same set of data and come to opposing conclusions” (p. 88).
• Net trajectory. Still, “taken as a whole … the literature shows a clear and consistent progression towards increasing concern about the carcinogenic effects of PCE” (p. 88). Karstadt (1998) is quoted on IARC’s reviews of both solvents (three for PCE, four for TCE): until volume 63 (1995), animal evidence for both was “limited” and human evidence “inadequate”; volume 63 raised both to “sufficient in animals and limited in humans”, showing “the gradual accretion of human evidence” (p. 88).
• Visibility after 1970. “Throughout this period PCE has been on the radar screen of the occupational health and environmental scientific communities, unlike the period before 1970” (p. 88).
• Uneven standards. “Fairly strict community drinking water standards have been established, although occupational standards have lagged behind”. This “may partly result from the combination of a weak labour movement and some highly publicised environmental cases involving childhood cancer (e.g. Lagakos et al., 1986)”, the Woburn case (p. 88).
• Regime change. “The problem is no longer invisibility and neglect, but intense scrutiny” (p. 88).
• Manufactured doubt (asserted, uncited). Industry “has been active and aggressive in countering new information through the strategy of artificially and purposefully creating doubt and uncertainty in the minds of decision-makers”. With PCE “now on the cusp of being declared a confirmed human carcinogen in some major national markets”, industry is “essentially buying extra time (and creating continuing exposure and disease)” (p. 88).
• Verdict. “40 years after the hard lesson of the water mains in Massachusetts, a sound precautionary strategy for continued exposure to PCE has still not been initiated” (p. 88).
• Then and now (p. 88). “The means used to avoid or promote action today are different from those of 1970, employing many sophisticated means to create doubt and increase uncertainty about the true value of a regulatory action”. In 1970, evidence “was available and not acted on for reasons not complicated by complex regulations, the potential of lawsuits or the activities of environmental or activist organisations”. Information was “restricted or non-existent and available primarily to scientists”. Industry “felt no special need to consult it (although they had contributed to it) and apparently did not”.
• Stated lesson for both periods. “Mechanisms are needed to force the production, sharing and publication of information about exposure and effects; normative and legal requirements concerning the duty of care of employers and manufacturers are also required. Alternatives are available” (Panel 4.2) (p. 88).
• Close. Interpretation is “irresolvable within science itself … there are no overarching criteria from the philosophy of science that can force a solution”. “The history of PCE will continue in the future as it has in the past”. Then the thesis is repeated (p. 88).

Panel 4.2 (Onasch): Wet cleaning technology eliminates PCE use in dry cleaning (p. 87)#

These views are Onasch’s alone.

• Policy lead. Massachusetts lists PCE as a “higher hazard substance” under its Toxics Use Reduction Act. The panel calls for “further policy measures”, conversion assistance and partnerships with utilities, since conversion saves electricity and gas (p. 87).
• Status quo (p. 87).
• PCE is the most widely used US dry-cleaning solvent; the EPA estimates about 85% of cleaners use it.
• It is a major contributor to contamination at dry-cleaning sites, “mainly due to past unsafe handling practices”.
• It is “reported to be the chemical most widely found in groundwater contamination at Superfund sites (TURI, 2007)”.
• Technology. Wet cleaning has existed “for several decades”, but only “in the last 10 years or so” can “100 % of garments” be wet-cleaned. Keoleian et al. (1997) suggested mixed-mode shops (p. 87).
• Diffusion (p. 87).
• There are “over 150 dedicated wet cleaners” in California, where California Air Resources Board (CARB) amendments phase out PCE dry-cleaning machines “by 1 January 2023”.
• “The shift … has been slow, especially where regulations phasing out solvent use do not exist”.
• Sinsheimer et al. (2007): switchers kept service and customers, cut operating costs, found the transition not greatly difficult, and were “highly satisfied”.
• Single-shop economics (Onasch, 2011; Bellingham, MA) (p. 87).
• Electricity and gas use fell up to 20%, and water use also fell.
• Over 12 months: equipment −USD 500, claims −USD 1,000, operational costs “mainly due to costs of detergents” +USD 1,069, resource use −USD 2,318 (“calculated using normalised rates”). Net saving USD 2,749 (check: 500 + 1,000 − 1,069 + 2,318 = 2,749).
• To replace its solvent machine the facility spent about USD 12,000 “(in actual costs, but not factoring in discounts and grant monies received)”, giving payback in “just under 4.5 years” (check: about 4.4).
• Audit note: the parenthesis most naturally means the USD 12,000 is before deducting discounts and grants, so subsidies do not flatter the payback figure; the wording is ambiguous. The weaker points are elsewhere: one shop, one 12-month period, a study by the panel author, and the largest saving item (USD 2,318) computed at normalised rates rather than from actual bills.
• Time savings were “difficult to quantify” but said to be possible with training (less pre-spotting, separate washer and dryer).
• Staff reported “‘whiter’ whites”.
• Indirect benefits: air quality, reduced liability, “elimination of regulatory oversight” and green marketing.

Table 4.1 Early warnings and actions (p. 89)#

• As printed (p. 89):
• 1860: PCE synthesised.
• 1920s–1960s: used against hookworm.
• 1925: first toxicological evaluation.
• 1925–40: therapeutic problems found, and “responses of different subjects varied”.
• 1940–70: inhalation and chronic effects (CNS depression) studied; gas chromatography introduced.
• 1970–present: the vinyl chloride finding prompts “competing accounts” of PCE carcinogenicity. Standards generate “controversy couched in scientific terms”; “PCE figures in lawsuits”.
• Today: PCE is regulated, but “controversy continues over where to set standards and whether PCE has caused harm in many legal cases”.
• Note. The table leaves out the pipe-specific events (1966 and 1968 odour tests, 1969–79 installation, 1976 discovery). It is a timeline of PCE toxicology, not of the pipe case.

Case timeline#

Lags: - Product-specific warning to discovery: from the 1966 and 1968 odour tests (p. 77) to the 1976 discovery (p. 77) is about 8–10 years. Installation started only about a year after the 1968 tests (p. 77). - General toxicology to deployment: about 40 years of accumulating warnings (1925–1968) came before deployment (pp. 79–82). - Discovery to “effective action” on the pipes: the chapter does not say what remedial action was taken, when, or at what cost. It says only that some supplies “still today require continuous remediation” (pp. 76–77). - Carcinogenicity suspicion to precautionary strategy: from the early 1970s to 2013 is about 40 years with, in the chapter’s view, still no sound strategy (p. 88).

Harms and costs: the chapter gives no quantified measure of exposure, affected population, health outcomes or remediation cost for the pipe contamination. Harm is described qualitatively as “widespread contamination” and “continuous remediation” (pp. 76–77). Potential effects are listed without citation: “various cancers, birth defects and autoimmune disease” (p. 77). For PCE more broadly the chapter points to groundwater contamination from dry cleaning (p. 86), with Panel 4.2 adding Superfund prevalence (p. 87), and to “continuing exposure and disease” from delay (p. 88), again unquantified.

The authors’ own lessons and conclusions#

Ozonoff’s lessons derived from the evidence#

1. Harm was foreseeable from existing information. The early warnings “suggested the need for caution” (pp. 76–77). A late-1960s public health expert reading the literature would have been uneasy (pp. 82–83), and aversion to solvents in water was established before the pipes, so this “is not merely a statement based on hindsight” (p. 83).
2. Non-scientific interests shape scientific disputes. Ideology, money, market share and reputation can lock positions in place. Findings get disputed when someone with means cares, and neglected when no one with means does (p. 78).
3. Therapeutic tolerance does not transfer. Side effects accepted for a serious disease would “probably have been of little relevance” to a use that brings the exposed no matching benefit; the hookworm balance was “a balance not found in newer uses” (pp. 80, 83).
4. The exposure route is irrelevant to foreseeability. Solvent in water was already recognised as a threat, so entry via the pipe lining rather than land disposal did not matter (p. 83).
5. Indifference, not concealment, is a plausible explanation of Johns-Manville’s conduct on PCE, “assuming that knowledge was not hidden”; there is no evidence of concealment (p. 84).
6. Missing epidemiology would not, at the time, have been a reason to delay (p. 84).
7. Knowledge confined to specialists, together with ignorance among physicians and workers, kept the issue off the agenda and shut out unions and advocates (p. 84).
8. “Nobody made them care.” Firms’ interests are rarely aligned with public health, and “uncertainty favours the side of inaction” where large firms have high stakes and resources (p. 86).
9. The failure mode has changed from neglect to contested scrutiny. The same inaction now comes from “intense scrutiny” and doubt-creation (p. 88).
10. The science has not been hidden but has been ineffective in guiding and catalysing action. The interpretive dispute cannot be resolved within science (pp. 76, 88).

Ozonoff’s recommendations and advocacy#

• Mechanisms “to force the production, sharing and publication of information about exposure and effects” (p. 88).
• “Normative and legal requirements concerning the duty of care of employers and manufacturers” (p. 88).
• Criminal and civil liability backed by state enforcement as the main means to “make the company care”. Voluntary or moral measures must pass that “acid test” (p. 86).
• Implicitly, clarity “about what evidence will trigger action” (pp. 76, 88).
• Pointing to available alternatives, via Panel 4.2 (p. 88).
• Claims that are advocacy rather than demonstrated in the chapter: that industry is “artificially and purposefully creating doubt” and “buying extra time” (p. 88); and that “the history of PCE will continue in the future as it has in the past” (p. 88).

Rudén’s lessons (Panel 4.1)#

• Divergent conclusions from the same data are common. They reflect data selection, interpretation, weighting and assessment policy (more or less precautionary), as well as legitimate updating (pp. 84–85).
• Recommendation: evaluators should be transparent about how they judge strength of evidence and uncertainty, and use “clear and consistent terminology” (p. 85).

Onasch’s lessons (Panel 4.2)#

• A substitute for PCE in dry cleaning exists, works for all garments, can cut costs, and pays back within about 4.5 years in one documented case (p. 87).
• Diffusion is slow without regulation. Policy support, conversion help and utility partnerships are recommended (p. 87).

Mechanisms and dynamics#

1. How the product-level warning was neutralised. - The only direct, product-specific warnings were sensory: odour and taste in 1966 and 1968 (p. 77). - The formal tests (pH, alkalinity, hardness) could not see organic contamination (p. 77). - A later clean result, plus a local explanation (brown paper, incomplete curing), closed the anomaly; there is no record it was tested (p. 77). - That explanation implied residual solvent, which was the actual hazard pathway, but the chapter finds no record of follow-up tests (pp. 77, 83). - The utility’s own chemist also smelled nothing on the second visit (p. 77). My inference: the buyer’s quality control seems to have been geared to the old problem (colour and taste in cast-iron mains) and the old tests; the chapter does not examine the utility’s testing.

2. Knowledge and tools stayed upstream. - In the 1960s Dow’s researchers were measuring PCE in people with gas chromatography, and publishing openly (p. 82). There is no record that the pipe was tested for organic contaminants (p. 77). - “Industry” here means two actors. Dow made PCE and produced toxicology. Johns-Manville used PCE as a process solvent and never drew on that toxicology. The chapter says industry “had contributed to” the information but “felt no special need to consult it” (p. 88). - The knowledge lived in specialist journals and with the chemical’s maker. The product integrator and the water utilities worked with routine tests.

3. Familiarity and benefit bred a presumption of safety. - Therapeutic use created a “misperception that PCE was non-toxic” (p. 81), which “encouraged careless use” (Irish, 1962, p. 82). - Lob (1957) made it a general pattern: a chemical is assumed harmless until shown otherwise (p. 81). - The chapter reverses the inference: biological efficacy “should have alerted us” (p. 78).

4. Acceptable risk belongs to the use, not the substance. - Serious side effects were accepted for a major disease (p. 80). - Drinking-water consumers received no benefit from PCE itself (p. 83). - The chapter treats the risk–benefit balance as non-transferable between uses.

5. Toxicological complexity defeated threshold thinking. - The complexities: - individual susceptibility (pp. 79–83); - effects not tied to dose (p. 80); - chronic low exposures worse than acute ones (p. 80); - accumulation, and delayed liver markers (p. 82); - unexplained effects at the lowest doses (p. 82); - animal–human inconsistency (pp. 80, 82); - synergy with other chemicals (p. 82). - Standard-setting still sought a “safe concentration” band: Carpenter’s 100–500 ppm and Dow’s 100/200 ppm (p. 81). - Workers had symptoms “well within the generally accepted toxic limit” (Morse and Goldberg, 1943, p. 81), and clinical findings contradicted the manufacturer’s view that serious organic injury was unlikely (p. 81), though those findings came from exposures at or above the manufacturer’s proposed ceiling.

6. Attention follows stakes and means. - The chapter’s two corollaries (p. 78) amount to a political economy of attention. Evidence is contested when someone with resources is threatened by it, and neglected when nobody with resources cares. - Before 1970 PCE was neglected: information was specialist and “of no interest” to industry (pp. 84, 88). After 1970 it was contested: “intense scrutiny” (p. 88). - The same outcome, inaction, came from opposite conditions of attention.

7. Opacity excluded the people affected. - Trade names hid chemical identity from physicians (p. 80). - Specialist-only circulation, and workers’ and physicians’ ignorance of what they were exposed to, kept unions and advocates out and the issue off the agenda (p. 84).

8. Burden and standard of proof rose over time. - In 1968, missing epidemiology “would not have been a reason to delay”. Epidemiological reconfirmation became the norm later (p. 84). - After 1970 the bar rose in steps: - bioassays were disputed; - epidemiology was structurally weak (long latency, poor exposure data, low statistical power); - arguments moved on to rodent-to-human relevance (pp. 86–88). - Panel 4.1 shows how assessment policy decides the verdict: whether animal data are treated as relevant, and how particular studies are weighted (pp. 84–85). - The chapter concludes that the choice of trigger is normative, since no internal scientific criterion can settle it (pp. 76, 88).

9. Incentives, liability and legal positioning. - The firm’s goal is profit, and the product’s nature is secondary (p. 86). - After detection, denial was “a necessary position” to avoid damages (p. 84). Litigation made historical foreseeability the battleground (p. 77). - The chapter relies on liability as the corrective (p. 86). In its own account, though, liability appears only after the fact, as a “battle over who should pay” (p. 77). It did not shape the 1968 design decision. - The chapter’s power claim: high stakes plus superior resources mean “uncertainty favours the side of inaction” (p. 86).

10. Institutional timing and uneven protection. - Deployment came just before the 1970 US shift to federal engagement (pp. 84, 86). - In 1970, inaction was not “complicated by complex regulations, the potential of lawsuits or the activities of environmental or activist organisations” (p. 88). - Later, drinking water standards tightened while occupational standards lagged. The chapter attributes this partly to a “weak labour movement” and to publicised childhood cancer cases (p. 88). Protection tracked political salience and the organised power of the exposed.

11. Durable infrastructure. - Some 700 miles of buried pipe turned one design choice into a long-lived exposure source needing “continuous remediation” (pp. 76–77). - The chapter does not use the terms “lock-in” or “irreversibility”. Its account nonetheless shows the cost of a mistake in durable infrastructure persisting for decades.

12. Innovation and substitution. - The pipe was a well-meant innovation to fix a real problem (p. 77). It was deployed at scale while “relatively untested” (pp. 76–77). - For PCE’s wider uses, Panel 4.2 describes a substitute that is technically complete and claimed to save money, yet spreads slowly “especially where regulations phasing out solvent use do not exist” (p. 87). The panel associates the larger number of dedicated wet cleaners (over 150) with California’s regulatory phase-out; it does not show that regulation caused the uptake.

13. Mental models and blind spots, as far as the text supports. - Johns-Manville. Solving the customer’s colour and taste problem, judging adequacy by routine tests and smell, explaining the anomaly away, and later taking a defensive legal stance (pp. 77, 83–86). The chapter calls this indifference, not malice (p. 84). - Providence utility. Noticed the odour, raised it, and accepted a later clean result (p. 77). The chapter does not examine utilities’ purchasing decisions. - Early pharmacologists. Weighed risk against benefit explicitly, and were humble about rare harms (Hall’s “million treatments”, p. 79). They still judged safety by analogy with a known toxicant, CCl4 (p. 79). - Industrial hygienists and Dow in 1952. Focused on acute effects and threshold bands; “serious organic injury… unlikely” (p. 81). By the 1960s, Dow’s own researchers were flagging accumulation and a “mistaken perception” of low toxicity (p. 82). - Post-1970 assessors. Divided systematically by institutional type and precautionary policy (pp. 84–85). - Chemical industry after 1970, as Ozonoff characterises it. Denial, a string of iterative counter-arguments, and doubt as strategy (pp. 76, 88).

14. Framing and language. - “Cured” / “not ‘cured’ completely” (p. 77). This turns a chemical-exposure signal into a manufacturing-quality glitch. - “Relatively non-toxic” and “harmless” (pp. 81–82). A reputation borrowed from medical use. - “Safe concentration” (p. 81). Threshold language applied to a chemical with variable susceptibility and accumulation. - “Hindsight” (p. 83). The defence frame, which the chapter rebuts. - “Could not have known”, “foreseeable” and “duty of care” (pp. 77, 88). Legal framing. - “Make the company care” (p. 86). Corporate responsibility framed in terms of incentives. - “Invisibility and neglect” versus “intense scrutiny” (p. 88). The two failure regimes. - The title, “Too much to swallow”. My reading, not stated in the chapter: a pun on ingestion and on implausible defences.

Transferable insights (technology-neutral)#

1. Test regimes built for a familiar problem can be blind to the new hazard a novel product introduces. Routine water tests (pH, alkalinity, hardness) could not detect organic contamination. The only signal was sensory, and it was explained away (p. 77). Strength: moderate. The case facts are specific and plausible. The narrative source is uncited, and this is one case, though the mechanism is general.

2. When an anomaly is explained away, the explanation’s implications must themselves be tested. “Not cured completely” implied residual solvent, the actual hazard, yet there is no record of follow-up tests (pp. 77, 83). Strength: moderate. A clear episode, but it rests on absence-of-record claims.

3. Hazard knowledge often exists but sits in the wrong place. It was in specialist literature and with upstream producers, while downstream integrators and buyers did not consult it. Dow generated and published it; Johns-Manville’s design never drew on it (pp. 82, 84, 86, 88). Strength: moderate–strong. The existence of the knowledge is well documented (pp. 79–82). Non-consultation is inferred from the lack of records.

4. Risk acceptance belongs to the use, not the substance. Harms tolerated where the exposed get a large direct benefit cannot justify exposure in a new use where they get none (pp. 80, 83). Strength: strong. This is a sound analytical and normative point clearly grounded in the text.

5. A record of beneficial use can create a false presumption of safety. Biological efficacy is a reason for more scrutiny, not less (pp. 78, 81, 82; Lob 1957, Irish 1962). Strength: moderate–strong. Several contemporaneous sources in the chapter make the same observation.

6. Rare, idiosyncratic, dose-independent, delayed, cumulative and interactive effects escape small or short tests. Safety claims need scale-appropriate evidence. Evidence: Hall’s “million treatments” rule (p. 79), individual susceptibility (pp. 79–83), accumulation and delayed markers (p. 82), synergy (p. 82), and symptoms within “accepted” limits (p. 81). Strength: strong for the toxicological record as reported; strong as a general principle.

7. Attention to evidence follows stakes and means. Findings are contested when a party with resources is threatened, and ignored when nobody with resources cares (p. 78; illustrated pp. 84, 86, 88). Strength: suggestive–moderate. It is a stated theoretical corollary, consistent with the case’s two eras, but not tested systematically.

8. Inaction can come from two opposite conditions: neglect, where evidence is invisible, and saturation, where evidence is intensely contested. Remedies differ for each (pp. 84, 86, 88). Strength: moderate. The framing is well constructed. The “saturation” side leans on uncited claims of manufactured doubt.

9. The same evidence yields different verdicts depending on data selection, weighting and assessment policy, and verdicts correlate with the assessor’s institutional type (Panel 4.1, pp. 84–85). Strength: strong for the documented divergence (systematic review of 29 TCE assessments; the PCE NHL example, including ECETOC’s internal inconsistency). Moderate for the causal attribution to institutional policy (small numbers, correlational, and “international organisations” grouped with industry).

10. Disputes over interpretation cannot be settled by science alone. Decision triggers (what evidence, at what strength, prompts what action) need to be set explicitly and transparently (pp. 76, 88; Panel 4.1 recommendation p. 85). Strength: moderate. The philosophical claim is asserted, but Panel 4.1’s evidence supports it.

11. Evidential standards can ratchet upward over time, and demands for confirmatory evidence that is structurally slow to arrive (long latency, poor exposure data, low power) favour delay (pp. 84, 86). Strength: moderate. A historical claim about norms that is plausible but only briefly argued.

12. Uncertainty favours inaction when the party facing the cost of action has more resources to contest than those bearing the harm (p. 86). Strength: suggestive. The mechanism is plausible and widely argued, but asserted here rather than shown with case data.

13. Harm does not need concealment. Organisational indifference is enough, and non-binding pressure is effective only if it actually changes what the organisation cares about (pp. 84, 86). Strength: moderate. The chapter’s own careful inference (no evidence of concealment for PCE), explicitly conditional (“assuming that knowledge was not hidden”, p. 84), and the “many similar examples” are uncited.

14. After a problem is detected, legal exposure shapes public statements about hazard, and the argument moves to what could have been known when (pp. 77, 83–84). Strength: moderate. Directly described, but based on one firm; any cross-chapter pattern is for the synthesis phase to establish.

15. Opacity about what people are exposed to keeps them and their advocates out of the conversation and keeps hazards off the agenda (pp. 80, 84). Strength: moderate. Supported by contemporaneous testimony (Hamilton’s discussant, St George).

16. Failures embedded in durable, distributed infrastructure produce exposure and remediation costs that last for decades (pp. 76–77). Strength: suggestive–moderate. Stated but not quantified in the chapter.

17. Protection follows political salience and the organised power of affected groups, not only risk. Drinking water standards tightened while occupational standards lagged (p. 88). Strength: suggestive. The chapter’s own hedge is “may partly result”.

18. A novel, untested product can be deployed at scale before hazard-specific testing, and discovery may then be accidental (pp. 76–77). Strength: strong as case fact (cited to state and technical reports), moderate as a generalisation.

19. Substitutes can be technically adequate and even cost-saving yet spread slowly without regulatory pressure. Being cheaper does not by itself overcome incumbency (Panel 4.2, p. 87). Strength: moderate. The slow diffusion is conceded honestly; the cost evidence is thin (one self-authored shop study plus one California study), and the link between regulation and faster uptake is an association, not a demonstrated cause.

20. A new pathway for a known class of hazard is not a new hazard. Foreseeability can come from analogy with established knowledge about the same class of contaminant (p. 83, citing Colten 1991; Amter & Ross 2001). Strength: moderate. The historical sources are cited, but the analogy is contestable, as a defendant would argue.

Limitations, contestation and bias check#

Where the chapter is strong - The pre-1970 literature review (pp. 79–82) is detailed, cites primary sources, and is balanced in places. It records optimistic assessments (Lamson 1929; Dow 1952) alongside warnings, and credits Dow for publishing its research openly (p. 82). - It declines to allege concealment where there is no evidence (p. 84), which is more restrained than a simple cover-up story. - It acknowledges that sceptical arguments applied to both positive and negative studies (“(and for negative ones)”, p. 88). - It confronts the hindsight objection directly with contemporaneous groundwater history (p. 83).

Where it is advocacy or thinly evidenced - Manufactured doubt (p. 88): the claim that industry has been “artificially and purposefully creating doubt” and “buying extra time (and creating continuing exposure and disease)” has no specific evidence or citation in this chapter. Panel 4.1, the chapter’s own analytical support, is more measured. It attributes divergence to legitimate updating and to different assessment policies, not to bad faith. The strong claim may well be defensible from other literature, but in this section it is asserted. - Industry scientific arguments are not engaged on their merits. Arguments about rodent-versus-human relevance are called “sophisticated” and said to be “spun out” (p. 88), but their content is never described or assessed. [Background — verify: some mode-of-action arguments in chlorinated-solvent assessments, such as rodent-specific kidney or liver mechanisms and high background tumour rates in certain rat strains, have been debated in good faith and sometimes accepted by regulators for other chemicals.] A fair analysis would separate legitimate scientific disagreement from strategic doubt. The chapter does not. - Harm is not quantified. There are no exposure measurements, population figures, health outcome data or remediation costs for the pipe contamination (pp. 76–77, 83). The adverse effects listed (“various cancers, birth defects and autoimmune disease”, p. 77) are uncited. [Outside source, checked in audit: Ozonoff co-authored at least two of these Massachusetts studies of PCE-contaminated public drinking water (Aschengrau, Ozonoff et al. 1993, Arch Environ Health; Aschengrau, Rogers and Ozonoff 2003, Environ Health Perspect; PubMed listing). Background — verify: the wider Boston University Cape Cod series on PCE from vinyl-lined asbestos-cement pipes also covered reproductive outcomes, birth defects and neurobehavioural outcomes. None is cited here, although p. 86 refers to “several environmental studies” with “mixed” results.] The omission is significant. The chapter’s lesson is about process failure and foreseeability, not about demonstrated magnitude of harm. - The worst-case exposure calculation (p. 83) assumes water saturated with PCE (125 ppm) and yields a body burden of about 1 g, “close to the dose used to treat hookworm”. The chapter labels it a worst case and notes no measurements were made at installation. It is vivid but an extreme upper bound. [Background — verify: concentrations later reported in Massachusetts pipe systems were mostly in the µg/L range, i.e. about three orders of magnitude below saturation (125 mg/L), with peaks at low-flow dead ends reported in the thousands of µg/L, still more than an order of magnitude below saturation.] The chapter’s secondary point, that steady-state internal burden is about four times daily intake at any exposure level, holds regardless. The comparison with a hookworm dose also sets a steady-state body burden against a single oral dose, which is illustrative rather than like-for-like (see 4.5 note). - The foreseeability argument could rest on simpler ground. The toxicological warnings came almost entirely from high-dose inhalation (hundreds to thousands of ppm in air) and gram-level therapeutic doses. The leap to low-level chronic ingestion is made through the accumulation argument, the aesthetic standard and the general norm against solvents in water. That last route (pp. 82–83) is actually the strongest and needs no toxicology: a known solvent in a product touching drinking water, an observed odour, and an established norm against solvent contamination. The long toxicology review does more work for the chapter’s broader point (knowledge existed but was ineffective) than for the specific foreseeability claim. - Sourcing of the core narrative. The 1966–1968 Providence account and the absence-of-record statements (p. 77) are uncited. Readers cannot judge what records were examined. Possible litigation involvement is not discussed (see Open questions). - Scope of blame. The chapter concentrates on Johns-Manville. Its own counterfactual, that public health experts “would have been unlikely to view the presence of any PCE in their water favourably” (p. 83), sits awkwardly with the fact that water utilities and their public health overseers bought and installed some 700 miles of the pipe (p. 77). Their decision-making, testing duties and any regulatory approval process go unexamined. Purchaser and regulator failures are part of the story but receive no analysis. - Generalisation from one firm. The chapter opens with “Institutions, large and small” (p. 77) and says “there are many similar examples involving other companies in this period” (p. 84) without citing them. - Theory of the firm. “The job of a company is to make money for its owners” (p. 86) is a model, not a finding. It motivates a liability-centred remedy, but the chapter does not test whether liability actually changed behaviour here. Liability appears only after the fact (“battle over who should pay”, p. 77). [Background — verify: Johns-Manville filed for bankruptcy in 1982 because of asbestos litigation. That bears on whether liability deters or merely compensates late, and on how the PCE claims were resolved.] - The alternatives lesson fits the case loosely. Panel 4.2’s substitute concerns dry cleaning, not pipe linings (p. 88). The chapter does not discuss what alternatives existed for lining or replacing cast-iron mains in 1968. Panel 4.2 rests heavily on a single-shop, 12-month study by the panel author (Onasch, 2011), whose largest saving is computed at “normalised rates” (p. 87). (Its equipment cost is stated “not factoring in discounts and grant monies received”, which on the natural reading makes the payback conservative rather than flattered.) Its title (“eliminates PCE use”) is stronger than its evidence of real-world diffusion, which it admits is slow (p. 87). - Historical and internal inconsistencies: - Table 4.1 dates PCE synthesis to 1860 (p. 89), while the text credits Faraday (p. 78) and implies discovery more than a hundred years before the 1920s (p. 79). [Background: Faraday’s work was about 1820–21.] - The chapter attributes a first-person clinical quote to “Hamilton (1936)” immediately after reporting a discussant’s remark; the speaker is uncertain (p. 80). - Installation dates vary: “late 1960s and 1970s” (p. 76), “1969–1979” and “late 1960s and early 1970s” (p. 77). - “Rosmiller/Rossmiller” (p. 81). - The Davy–Lavoisier “dispute” is compressed (Lavoisier died in 1794). - The ACGIH is misnamed “Government” in the references (p. 89). - The PHS standard is dated 1968 (p. 82). [Background: the 1962 standards may be meant.] None of these changes the argument, but they signal a lightly edited historical treatment.

Hindsight bias check. The chapter tackles hindsight explicitly and partly succeeds. The groundwater-solvent literature and the aesthetic standard were contemporaneous (pp. 82–83), and the odour signal was direct (p. 77). The claim that carcinogenicity could have been foreseen is not made: the chapter concedes that current concerns were not part of the 1960s understanding (pp. 76–77). The risk of hindsight is higher in the claim that toxicological concern was obvious for low-level ingestion, since that literature concerned different routes and magnitudes.

Case selection and the pro-precaution frame. This is a case chosen to show a failure. Nothing in the chapter considers cases where similar warnings were heeded, or where caution about a solvent-applied product proved unnecessary. The precautionary frame leads the chapter to treat post-1970 scientific disagreement mainly as strategic delay. Panel 4.1 offers an alternative reading, that divergence is policy-laden but not necessarily bad faith, and the chapter cites it without resolving the tension.

Dissent within panels. Neither panel disagrees with the main text. Panel 4.1’s tone and diagnosis (communication and terminology) are more modest than the main text’s (duty of care, liability, manufactured doubt). That difference should be kept visible rather than smoothed over.

Fair to industry? The chapter is fair to Dow (p. 82) and careful about Johns-Manville’s intent (p. 84). It is less fair to “the chemical industry” after 1970 (p. 88), which is characterised without evidence and without its arguments being set out.

Notable quotes#

1. “The science has not been hidden. It has been ineffective in guiding and catalysing action.” (pp. 76, 88)
2. “Institutions, large and small, make decisions every day where a conscious application of foresight could prevent a later hazard. Yet such foresight — based on existing information — is often absent.” (p. 77)
3. “…if nobody cares or nobody with means cares, there will be little pressure to challenge a scientific finding or explore an issue in greater detail.” (p. 78)
4. “The familiarity and benefits of these substances should have alerted us to the fact that exposure to these chemicals has biological effects that could also be harmful.” (p. 78)
5. “…one cannot assume that any anthelmintic is entirely safe for human use until there are reliable reports on at least a million treatments without any untoward effects.” (Maurice Hall, via Sandground 1941, p. 79)
6. “The Dow researchers noted that acute exposure to PCE might, in fact, be a chronic exposure from the body’s standpoint because of the slow excretion rate.” (p. 82)
7. “…the principal reason that Johns-Manville did not care enough to examine thoroughly the risks of using PCE was that nobody made them care.” (p. 86)
8. “In such circumstances, uncertainty favours the side of inaction.” (p. 86)
9. “The problem is no longer invisibility and neglect, but intense scrutiny.” (p. 88)
10. “…irresolvable within science itself because the same evidence can be interpreted differently and there are no overarching criteria from the philosophy of science that can force a solution.” (p. 88)

Open questions#

1. Sources and standpoint. What records underlie the Providence 1966–1968 narrative and the absence-of-record claims (p. 77)? Did Ozonoff serve as an expert in the pipe litigation? If so, how does that bear on the chapter, both for access to documents and for adversarial framing?
2. Harm magnitude. How many people were exposed, at what concentrations and for how long? What did the Cape Cod epidemiology (Boston University School of Public Health, including Aschengrau, Ozonoff et al. 1993 and Aschengrau, Rogers and Ozonoff 2003) find, and how robust is it? Why does the chapter not cite it, given the author’s co-authorship?
3. Remediation and cost. What was done after 1976 (flushing, bleeders, pipe replacement)? At what cost and borne by whom? How were liability claims against Johns-Manville resolved, given its 1982 bankruptcy?
4. Purchaser and regulator roles. Why did utilities adopt a “relatively untested” pipe without organic-contaminant testing? What approval or specification regimes applied? Were alternatives, such as other lining methods, available in 1968?
5. The 2013 predictions. Was PCE “declared a confirmed human carcinogen in some major national markets” after 2013 (p. 88)? [Background — verify: IARC re-evaluated PCE in 2014 and, as I understand it, kept it in Group 2A (probably carcinogenic). The US EPA’s 2012 IRIS assessment called it “likely to be carcinogenic to humans”. The EU classification is Carc. 2. If correct, the “cusp” prediction did not materialise in the form stated.]
6. “No sound precautionary strategy” (p. 88). [Background — verify: California’s phase-out of PCE dry-cleaning machines by 2023 (as the panel anticipated, p. 87); a US EPA TSCA risk evaluation and a final rule in late 2024 restricting most PCE uses and phasing out dry cleaning over about a decade (reportedly revisited in 2025); national phase-outs in some EU states, e.g. France for dry cleaning in residential buildings.] Was action finally driven by cancer evidence, neurotoxicity, or substitution economics?
7. Wet-cleaning diffusion. Did diffusion speed up where regulations required it, and stall where they did not? Did the cost savings claimed in Panel 4.2 hold up in larger samples?
8. Risk assessment practice. Did the divergence Panel 4.1 documents narrow after 2013 as structured, transparent evidence-evaluation frameworks spread? Rudén’s recommendation (p. 85) invites this check.
9. Generalisability of the “two regimes” model (neglect, then contested scrutiny) across other Late Lessons chapters. Is it a common trajectory?
10. Mode-of-action arguments. Which rodent-to-human arguments about PCE were scientifically reasonable and which were strategic? The chapter does not separate them (p. 88).

Audit log#

Independent audit against the full text extract (PDF pp. 78–93), with Panels 4.1 and 4.2, Table 4.1, the p. 86 photo and Annex 1 bios checked against the PDF. No strand contamination and no contemporary-technology references were found. Changes:

1. Authors: added Ozonoff’s Annex 1 line “principal or co-investigator of several major studies of waste sites”.
2. Authors: added outside-source (PubMed) confirmation that Ozonoff co-authored Massachusetts PCE drinking-water case-control studies (Aschengrau et al. 1993; 2003), which the chapter does not cite or disclose; updated Limitations, Open question 2 and the Convention line to match.
3. Opening box: added the scope sentence “why evidence of the potential hazard was ignored” and noted “dictate a solution” (p. 76) versus “force a solution” (p. 88).
4. 4.2.1: made the lesson quote verbatim (“the way that non-scientific concerns can distort scientific disputes”).
5. 4.2.1: added the chapter’s own caveat that “disputes and doubts are normal in science”.
6. 4.2.1: made corollary 2 verbatim (“Results of potentially great significance in other contexts…”).
7. Faraday note, timeline and inconsistencies list: showed that the “1860” date in Table 4.1 conflicts with the chapter’s own p. 79 wording, not only with background knowledge.
8. Phase I: added Sandground’s own conclusion that the cases “should not discourage the use” of PCE.
9. Phase I: replaced the paraphrase “still open in 1969” with the verbatim “being considered at the time…”, and added the uncertainties about absorption and susceptibility.
10. Hamilton (1936): split her points from the discussant’s, and flagged that the chapter’s attribution of the first-person “leaps and bounds” quote to Hamilton is uncertain; added the clinical-observation part of the quote.
11. 1940s: added the general reviews (Lehmann and Flury; Sappington) and Barrett et al. (1939).
12. Dow 1952: added eye irritation and CNS depression as “the prime toxic effects”.
13. Coler and Rossmiller: noted that exposures (200–400 ppm) were at or above Dow’s 200 ppm ceiling, so the “contradiction” is with Dow’s injury judgement, not its limits; lowered the timeline rating from “strong” to “moderate”; carried the same nuance into Mechanism 5 and the digest.
14. Smyth et al. (1969): added the acute-lethality endpoint, oral route in rats, and that most pairs were additive; lowered the timeline rating to “suggestive–moderate”.
15. View from 1970: added “some writers considering PCE to be more dangerous than conventionally believed”.
16. 4.5: softened “the odour alone was therefore a public health concern” to the chapter’s hedged “might have worried”.
17. Bathtub model: marked the 25%/day and 2.5-day half-life as the author’s own estimate from Stewart et al. data; added that comparing a steady body burden with a single hookworm dose is illustrative, not like-for-like (notes and digest).
18. 4.5.1, Lesson 5, Insight 13 and digest: added the explicit condition “assuming that knowledge was not hidden” to the indifference explanation.
19. 4.6: “raised concern” corrected to “established concern”; Karstadt quote now covers both TCE and PCE; added the “on the radar screen” sentence and the verbatim “avoid or promote action” wording.
20. Panel 4.2: added “normalised rates” for the largest saving; corrected the reading of “not factoring in discounts and grant monies” (on the natural reading it makes the payback conservative); added the time-savings point; revised the Limitations bullet.
21. Timeline: toned down the 1966 and 1968 odour-test ratings (“strong” to a faint or moderate direct sensory signal); noted that the 1966 location is implied, not stated; replaced the unsourced “2007” CARB date with a background-verify flag.
22. Lesson 3: “have no bearing” corrected to the chapter’s “of little relevance”.
23. Mechanisms 1 and 2 and Insight 2: replaced “no follow-up tests were done” and “never tested” with absence-of-record wording; labelled the claim about the buyer’s quality control as my inference.
24. Mechanism 12 and Insight 19: softened “regulation in California did [drive adoption]” to an association the panel reports.
25. Mechanism 14: labelled the reading of the title as a pun as my interpretation.
26. Insight 14: removed the unsupported claim that it is “consistent with the pattern in other chapters”.
27. Limitations: refined the background note on measured concentrations relative to saturation.
28. Digest: fixed the conflation of the 1966 and 1968 tests; replaced “banned offensive odours” with the standard’s verbatim wording; tied “not merely hindsight” to the right claim; added the bathtub estimate.
29. Digest: corrected “read the same lymphoma studies oppositely” (the two bodies agreed on one study) and added the four-body PCE conclusions table.
30. Digest: added the stakes-and-means corollaries and the missing-epidemiology point to the lessons; rewrote the garbled recommendations line; labelled “liability acting only after the fact” as the auditor’s observation.
31. Digest caveats: added the uncited co-authored epidemiology, the worst-case comparison, internal evidence for the date error, and thin evidence on substitution.