Late Lessons, Jensen Huang and AI
Markdown version

LL1-11 — Ch11 MTBE in petrol as a substitute for lead#

Late lessons from early warnings: the precautionary principle 1896–2000 (EEA Environmental Issue Report No 22, 2001). Chapter 11, report pages 110–125 (PDF pages 110–125; in this chapter the printed and PDF page numbers coincide). Chapter text runs pp. 110–120. Table 11.1 is on p. 121. References run pp. 121–125.

Reading notes. I read the whole text extract in order, to the final page marker (p. 125). I checked pp. 110, 112–115 and 119–121 visually in the PDF. The extract is faithful. Two things are errors in the printed original, not in the extraction: - The micro sign is missing throughout. “180 g/l (micrograms per litre)” (p. 112), “5 g/l” (p. 112), “30 g/l” (p. 114), “5 to 30 g/l” (p. 114) and “below 5 g/l” (p. 119) should all read µg/l. - “Østergaard” is printed as “stergaard” (p. 113).

“NOx” is split across lines in the extract (pp. 111, 118). Box 11.1 (p. 113) and Table 11.1 (p. 121) were read from the page images. This 2001 chapter has no panels or commentaries. Panels are a feature of the 2013 volume.

Authors and standpoint#

Authors. Martin Krayer von Krauss and Poul Harremoës (p. 110). The chapter does not give their affiliations.

Peer reviewers (acknowledgements, p. 6). The report does not say which reviewer read which chapter. The reviewer list does include, in one run, Arturo Keller, Finn Bro Rasmussen, Sandrine Dixson-Decleve, Hilkka Vahervuori, Grete Østergaard, Erik Arvin and Eberhard Morgenroth. Several of them are also the chapter’s main sources: - Keller is lead author of the University of California reports, Keller et al. 1998a/b. - Dixson-Decleve is cited through personal communication for the European Chemicals Bureau (ECB) decision and for the Dekant expert review. - Østergaard is cited through personal communication for the threshold reasoning. - Arvin is cited through personal communication and as the early warner. - Morgenroth co-authored with Arvin.

My inference is that these were the chapter’s reviewers, since the order of the list appears to follow the chapters. The overlap between reviewers and sources means the chapter was checked mainly by people inside the debate it describes. No industry reviewer is identifiable. The chapter describes Dixson-Decleve as “Environmental Consultant, Brussels office” (p. 122) and does not name her clients.

Evident stance. The chapter is explicitly normative and pro-precaution. Its frame is sustainability and “futurity” (p. 110). Its stated goal leans illustrative: “to examine how foresight, embodied in the precautionary principle, can lead to robust and flexible decisions” (p. 110). The discussion’s aim is “to illustrate the potential of applying foresight as part of the precautionary principle” (p. 115). The chapter does also pose a checkable question: it “will examine whether, at the time of MTBE’s introduction, it would have been possible … to foresee” the problem (p. 110). But it answers that question in the course of illustrating a principle it already holds.

The chapter is unusual in this report because the case was still open. The editing was “finished on 10 May 2001”, and the authors warn that “aspects of the paper may be quickly outdated” (p. 110). The last row of Table 11.1 reads “2001 The debate goes on” (p. 121). This is a live case, not a closed historical one, and the “lesson” is partly prospective.

Boxes and tables. - Box 11.1, “Carcinogenicity — complex science, borderline results and conflicting interests” (p. 113), is a balanced survey of divergent institutional judgements on cancer. - Table 11.1, “MTBE: early warnings and actions” (p. 121), is credited “Source: EEA”. It was presumably compiled editorially, in the report’s standard format.

Cross-references outside the section. I located these by text search only and have not analysed them; they belong to another strand. The editors reuse MTBE in the synthesis chapter at pp. 170, 172, 173, 174, 176, 177 and 182. Examples: - novelty as a warning sign (p. 170) - persistence “apparent at the outset” (p. 173) - introduction “based on bodies of knowledge concerning engines, combustion and air pollution” (p. 174) - tank leakage “underestimated in the US regulatory appraisal” (p. 174) - MTBE “apparently promised a simple solution” to lead (p. 176) - “little formal scrutiny” of alternatives at adoption (p. 177)

Some of these editorial framings go beyond what Chapter 11 itself documents. For example, Chapter 11 documents underestimated tank leakage for Denmark (p. 114), not a US regulatory appraisal. Its US tank evidence (Couch and Young, p. 115) shows only that well-designed tanks can leak through improper installation, which the synthesis also paraphrases. Likewise, the synthesis’s “simple solution” point (p. 176) is hedged there (“perhaps more readily overlooked”). That gap needs checking when the synthesis is analysed.

Section-by-section notes#

11.1 Introduction (p. 110)#

11.2 Lead in petrol (p. 110)#

11.3 The MTBE case (pp. 110–111)#

11.4 Benefits (p. 111)#

11.5 Impacts (pp. 111–114)#

11.5.1 Persistency (pp. 111–112) - Mobility. MTBE moves “at about the same speed as groundwater” (Barker et al., 1990). - Detection. The 1996 US Geological Survey (USGS) report found it the second most frequently detected VOC in shallow urban groundwater from eight areas, 1993–94. - Solubility. About 50,000 mg/l. Petrol with 10% MTBE gives about 5,000 mg/l at equilibrium. - Degradation. Rapid in air. Underground: - no degradation in two aerobic field studies; - “very slow but measurable” degradation in the laboratory; - anaerobic degradation seen in field studies; - but “significant rates” observed only in the laboratory. - Conclusion: “a potential risk to groundwater” (p. 112). - Check: the field/laboratory labelling may not match the cited titles. Landmeyer et al., labelled a laboratory study, is titled as an aquifer study (p. 123). Conversely, two of the “field investigations” of anaerobic degradation carry titles suggesting soil or subsurface-sample work (“in soils”, Yeh and Novak, p. 125; “in the terrestrial subsurface”, Suflita and Mormile, pp. 124–125). The chapter also labels the 1990 Barker et al. finding “only at the laboratory scale” (p. 115), but that paper’s title (p. 121) does not say whether it was a laboratory or field study. The same group’s later work is a long-term study in the Borden aquifer (Schirmer and Barker, 1998, p. 124). This matters because the “regulators waited for full-scale results” argument (p. 115) rests on the laboratory/field distinction. Verify against the originals.

11.5.2 Aesthetics (p. 112) - Thresholds. A “potent terpene-like odour”. Odour thresholds of 180 µg/l (Denmark) and 5 µg/l (California): a 36-fold spread in a “threshold”. Some drinking water exceeds them. - Consequence. “Relatively small amounts of MTBE may thus render large reserves of groundwater useless” (p. 112). - Cause. The main cause is leaking underground storage tanks (USTs) and overfilling. Severity depends on reliance on groundwater and the condition of tanks. - EU evidence is thin: - “not routinely monitored in groundwater”; - “little public information” (Arthur D. Little, 2001). That report used unpublished data “as a significant source of information”, combined with data available from six Member States (Denmark, Finland, France, Germany, Sweden, UK); - “none of the findings indicated widespread or serious” contamination “on the same scale as the USA”. - EU conclusions nonetheless: the documented cases justify concern, and MTBE “is causing a risk for the aesthetic quality of drinking water” (p. 112). - Note: the core documented harm is loss of use triggered by human perception, not demonstrated toxicity.

11.5.3 Cancer (p. 112) - The EU risk assessment called MTBE “borderline”. The European Chemicals Bureau (ECB) decided not to classify it (source: Dixson-Decleve, personal communication). - “no comprehensive pre-marketing carcinogenicity testing was carried out” (p. 112).

11.5.4 Asthma (pp. 112–113) - The hypothesis. Joseph’s February 2000 editorial raised concern about epidemiological studies “indicating” increases in asthma and other respiratory or inflammatory conditions in the years after MTBE’s introduction in Philadelphia and New York. - In their own voice, the chapter’s authors add that “The past five years have seen reports of alarmingly high rates (above 20 %) of asthma among children in several large cities on the east coast of the United States where MTBE was in use” (Mangione; McBride; Hathaway; Leighton). The word “alarmingly” is theirs, not Joseph’s. - It “has been suggested” (Joseph, 1999; Leikauf et al., 1995) that an unknown exhaust product of MTBE combustion is responsible (p. 112). - UC Davis view: plausible, but no designed studies; “little evidence at present either to implicate or exonerate” (p. 113). - Evidence quality (references, pp. 122–124): - the sources are an editorial, conference papers, a newspaper article, a local health-department survey, prevalence studies that do not mention MTBE in their titles, and one journal article evaluating “a possible association of urban air toxics and asthma” (Leikauf et al., 1995, Environmental Health Perspectives), which also has no MTBE in its title (the chapter does not say whether it was peer reviewed); - the evidence is ecological and circumstantial; - “alarmingly” describes the reported prevalence rates, which may be accurate. What outruns the evidence is the implied link carried by the phrase “where MTBE was in use”. The section does close on the balanced UC view.

11.5.5 Other impacts (p. 113) - “A few investigations” suggest endocrine disruption, with unclear mechanisms. - The EU assigned no no-observed-adverse-effect level (NOAEL) for lack of data.

Box 11.1 — “Carcinogenicity — complex science, borderline results and conflicting interests” (p. 113) - Timing. Comprehensive tests began only “in the 1990s”. - Animal evidence: - inhalation: kidney and testicular tumours in male rats, liver tumours in mice; - oral: lymphomas and leukaemias in female rats, testicular tumours in male rats; - apparently tumours at “multiple sites, in multiple species, and via multiple routes”. - Critique (Dekant, 2000). Study “conduct and reporting … is inadequate”, and “rodent-specific mechanisms” make the results “inapplicable to man”. The reference list shows this was an “Expert review”, apparently unpublished, obtained via Dixson-Decleve as a personal communication (p. 122). Its title, “The mechanistic toxicology of MTBE does not support a classification as Category 3 carcinogen”, shows it was written to argue a position in the EU classification decision. - Did not classify: - IARC (1999): “not classifiable”. No analytical epidemiological studies had addressed human cancer, and the animal evidence was “limited”. - The National Toxicology Program’s board “voted 6 to 5” not to list it. - IPCS (1999) echoed IARC. - Found carcinogenic potential: - Mehlman (2000), cited for “Other institutions and individual scientists”: “probable human carcinogen”. The reference (p. 124) shows this is a two-page piece in Archives of Environmental Health (the same issue as Joseph’s editorial) titled “Misclassification of carcinogenic methyl tertiary butyl ether (MTBE) by the National Toxicology Program Board: Smokescreen in, science out?”. So the pro-classification pole, like the anti-classification Dekant review, is represented by an advocacy piece rather than a primary assessment; - University of California: “potential human carcinogen, although further studies would be needed to make a conclusive determination”; - the White House science council (NSTC, 1997): “carcinogenic hazard potential for humans”; - US EPA (1997): “carcinogenic potential to humans”. - EU rapporteur (draft, 2001): - tumours equivocal or of questionable relevance, mostly at “very high and systemically toxic doses”; - MTBE not genotoxic; - yet the human relevance of the testicular adenomas, seen in two rat strains, “cannot be neglected”, and “some uncertainty remains” about the lymphatic tumours, given study limitations and “somewhat inadequate” reporting; - verdict: “a borderline case between non-classification and Carcinogenicity Category 3”. - ECB decision (November 2000). Rejected Category 3 with risk phrase R40. - The authors’ “probable basis” for the decision. This is the authors’ surmise. The chapter does not report the ECB’s stated reasons. - non-mutagenic carcinogens are “generally accepted” to have a threshold; - positive in-vitro mutagenicity is attributed to extracellular exposure to the metabolite formaldehyde, so MTBE “is regarded as non-mutagenic” (Finnish EPA, 2001); - staying below the organ-damage NOAEL prevents tumours (Østergaard, personal communication); - hence “the risk of carcinogenicity posed by such substances is considered manageable” (unattributed closing sentence). - Reading: - the box is largely balanced and does not adjudicate. Its opening line leans one way, though: the three bioassays “demonstrated that chronic exposure to MTBE … causes cancers in animals”. The critiques follow only after that. - the same data produced divergent institutional verdicts. US EPA, NSTC and the UC review leaned to concern. IARC, IPCS and the EU declined to classify. The US NTP board split 6–5 against listing, so the divide is not simply US against international. - “conflicting interests” is never unpacked in the text. The references list the 1992 inhalation bioassays as Bushy Run Research Center reports tied to Union Carbide Chemicals and Plastics Co., submitted to US EPA under a TSCA Section 4 Testing Consent Order (p. 122).

11.5.6 MTBE in drinking water (p. 114) - Santa Monica. - Detection in late 1995; by June 1996 the city was “forced to shut down some of” its groundwater supply wells. - About 10 potential sources within 1 km. - The city lost “71 % of its local water supply” and now buys about half its total water “at a cost of USD 3.5 million a year”. - Standish, Maine. One road accident spread MTBE more than 0.7 km through fractured rock and contaminated more than 20 domestic wells. - US scale. “many documented cases”, and the number is increasing. - Costs. - Granular activated carbon treatment to the very low limits (5–30 µg/l) “could be enormous”. - Tank and pipeline leaks could cost “tens to hundreds of millions of dollars per year in California alone”. - Caveat: unpublished work suggests biological degradation in ordinary sand-filter waterworks (Arvin, personal communication).

11.6 Responses (pp. 114–115)#

11.8 Discussion (pp. 115–120)#

Framing (pp. 115–116) - Counterfactual date. The analysis asks what could have been asked “at the time of the introduction of MTBE in the early 1980s”. This sits awkwardly with 1973 production and Italian use from the late 1970s (p. 110). - Prior knowledge. Ethers’ low biodegradability was published in 1954 and was “textbook knowledge” by 1960. So “competent chemists and microbiologists could have anticipated” persistence. But “Documentation for this argument has however not been found” (p. 115). - Warnings and delay. - The first MTBE-specific warnings (1990) were “only at the laboratory scale”; “regulatory agencies did not react”. - Only the “full-scale results” of 1996 alerted the system. - The delay is said to be “documented in the United States and Denmark”. Only Denmark’s is shown. - Concession. There were “no indications of any obvious harmful effects” when MTBE was chosen. So the real issue is whether knowledge “was adequate at the time to foresee that persistency in itself might eventually be considered problematic” (p. 116). - Reference-list note (p. 121): Barker et al. (1990) was presented at a Houston conference co-organised by the American Petroleum Institute. The first warning was aired at an industry-co-organised venue in the year of the US oxygenate mandate. The chapter does not remark on this.

11.8.1 Historical perspective (pp. 116–117) - Concern about persistence as a class predates MTBE’s introduction: - Silent Spring (1962) and the persistent pesticides; - 1960s river foaming from non-biodegradable detergents, which led the detergent industry and the OECD to develop degradability tests; - an OECD working group (1978/79) generalising those methods, with governments, industry and academia discussing “the already accepted concept of persistency”; - standard tests from US EPA (1979), ECETOC (1982) and ISO (1984). - Law. Persistency “appeared early in the development of EU environmental legislation as an undesirable property”, citing the 1967 directive and 1984 test methods together (p. 116). The chapter does not say exactly when persistency entered the law, so “from 1967” would be too precise. The sixth amendment (September 1979) required pre-market notification of new substances, with data scaled to volume. Together with toxicity, bioaccumulation and suspected carcinogenicity, persistency could trigger classification as “particularly dangerous to the environment”. - Grandfathering. MTBE had been marketed before the 18 September 1981 cut-off, so it was an “existing substance” outside those rules. Rules on existing substances “were adopted in 1993”. The Finnish Environment Institute’s risk assessment was only reaching draft by end-2000, with the final version expected in 2001 (p. 116). - Citation check: the text cites “European Commission, 1993” for the existing-substances rules. The reference given is Commission Directive 93/67/EEC, which by its own title covers risk assessment of substances “notified in accordance with” Directive 67/548/EEC, i.e. new substances (pp. 122–123). The citation appears mismatched. From background knowledge, to verify: the existing-substances instrument was Council Regulation (EEC) 793/93, also from 1993. The date and the argument are unaffected. - Inference. Society “was indeed prepared to implement responses to address the impact of chemical persistency in combination with other negative characteristics” (pp. 116–117). The qualifier matters, because the chapter’s closing question is whether persistency alone suffices (p. 120). Projected high volume “could have been expected” to prompt study, “although there was no requirement at the time”, and “Such investigations would have revealed the persistency of MTBE to decision-makers” (p. 117). - Precautionary question. Does MTBE have “any other characteristic which, when combined with persistency, requires a response”? - Taste and odour were “known for a long time” but “overlooked during the screening of MTBE” (no screening is documented). The sentence opens “As has been mentioned previously”, but the chapter text never earlier says when taste and odour became known. Its only odour data are from 1997 and 1999 (p. 112). The sole support is Table 11.1’s unreferenced 1960 row, “Information about taste and odour in water … available in textbook” (p. 121). - “It seems that” a threat to groundwater “was never considered”. The claim is hedged in the original. - Cancer testing came only in the 1990s. - Verdict. - “There is no doubt” that the combination of persistence, taste and odour, “potential carcinogenicity and other adverse effects could have sparked concern in the early 1980s. Studies should therefore have been done at that time.” - The choice “was not grounded in precaution, as the possible threats … were not adequately characterised” (p. 117).

11.8.2 Modern-day perspective (p. 117) - Irreversibility. “persistency is unique in that it implies irreversibility”. It magnifies the cost of a “false diagnosis” and of effects science is “still unaware” of. - Open questions. Whether persistency alone should trigger a response, and whether it should trigger systematic investigation, are “still pending”. But MTBE’s volume and solubility “should definitely give rise to concern”. - Remediation. Once the chemical is spread, remediation of a newly discovered effect is “likely to be out of reach”. - The principle as interpreted. Always consider “the risk of being wrong”. Uncertainty “cannot be used as an excuse to postpone action”, and “scientific uncertainty and ignorance warrant precautionary measures”.

11.8.3 Alternatives (pp. 117–118) - Substitution principle. “There is no doubt that MTBE is a reasonably better option than lead”. But irreversibility requires investigating substitution “as an alternative to risk reduction” (p. 117). - Options (p. 118): - Other oxygenates (ethanol, ETBE, TAME, DIPE). - Ethanol needs farmland, fertiliser and pesticides and is energy-intensive. It “may not lower greenhouse gas emissions significantly” over the full cycle. - Other ethers “could have health effects similar to those of MTBE”. - Gasoline direct injection. It has lower octane needs at part load only, and oxygenates would still be needed as aromatics are cut. - Restructuring branched alkanes. This “may however be opposed by the petrochemicals industry” because of “major investment costs”. This is asserted without evidence. - Deeper issue. MTBE is “only the symptom”. As India and China industrialise, “future generations cannot afford to operate a mass transit system” (the chapter’s term for mass motorised transport) that emits VOCs, CO, CO2, NOx and other pollutants. Incremental alternatives will not be enough, so the principle “would now beckon us to invest seriously” in “alternatives entirely different”. The chapter describes two paths: - Demand reduction: “some European governments” question “the basic need for a mass transit system”. Denmark’s government holds that, “with advances in communication technologies”, private cars are increasingly a “want” rather than a “need”. It has enacted economic policies to discourage driving. - Radical engine change: hydrogen fuel cells, whose by-products are said to be “limited to oxygen and water”, with pilot fuel-cell buses in Chicago and Vancouver. My note: this is imprecise and ignores emissions from producing the hydrogen. - Readiness. None of these is ready for wide-scale use in the short term, so alternatives need active R&D (p. 118).

11.8.4 Cost-benefit analysis (pp. 118–119) - Principle. Under ignorance, a cost–benefit analysis (CBA) “is of only limited value” and “cannot, of itself, form the sole basis for a policy decision”. Systematic comparison of options remains useful (p. 118). - The California CBA (Keller et al., 1998a). It covered six cost categories and compared MTBE, ethanol and non-oxygenated petrol against pre-RFG petrol. - Air benefits were similar across the options and “relatively minor”: USD 14–78m/yr. - MTBE water treatment cost USD 340–1,480m/yr, assuming treatment to below 5 µg/l with granular activated carbon; “there may be cheaper methods”. The equivalent costs for ethanol and non-oxygenated petrol would be small. - Other costs: consumer fuel USD 435–1,055m/yr; recreational boating USD 160–200m/yr. - The report’s verdict: non-oxygenated petrol is cheapest, then ethanol (p. 119). - Provisional in both directions. An asthma link would raise MTBE’s costs; waterworks degradation would lower them. On what is “currently known with reasonable certainty”, MTBE is “a very expensive option”. A worst case would make correction “far outweighing the cost of prevention”. - Proportionality. Very high potential costs “justify a proportionately costly response” (p. 119). - Responsibility. “both industry and society (represented by regulators) were short-sighted”. “Who is the polluter, who should pay?” Most costs so far have been “borne by society”, including “human health costs”, which the chapter has not shown. - Prescription. Choose alternatives on “best environmental practices, even if these alternatives are more costly in the short term” (pp. 119–120).

11.9 Conclusion (p. 120)#

Table 11.1 — MTBE: early warnings and actions (p. 121; “Source: EEA”)#

Sixteen dated rows run from 1954 (low biodegradability of ethers) to “2001 The debate goes on”. The dates are integrated into the Case timeline below. - The table places the first lab warning and the US Clean Air Act scale-up in the same year (1990) but draws no connection. - It omits the regulatory architecture: 1979, 1981, 1985, 1993 and 1995. - It has minor inconsistencies with the text: - Santa Monica closures dated 1995, against June 1996 in the text; - field studies dated 1997, against 1996 “full-scale results”; - the 1960 textbook entry adds taste and odour, while the text cites the 1960 textbook only for biodegradation; - the ECB non-classification is dated 2001, but Box 11.1 places the Working Group decision in November 2000 (the 2001 date is that of the personal communication); - the asthma and endocrine “indications” are dated 2000, although the text cites asthma reports from 1994–99 and endocrine studies from 1998 (p. 113). The table appears to date them to when they were compiled (Joseph 2000; Williams et al. 2000).

Case timeline#

What was known when, and by whom (all from the chapter unless flagged)

Date Event / knowledge Strength of warning Page
Ancient–1800s Lead toxicity known; formalised by occupational health Strong (for lead) 110
Early 1920s Wisdom of lead in petrol questioned — (lead) 110
1954 Low biodegradability of ether family (Mills and Stack) Generic, not MTBE-specific 115, 121
1960 Ether resistance to biodegradation “textbook knowledge” (Sawyer) Generic 115, 121
1962 onward Silent Spring; persistent-pesticide debate; research on biodegradability and mobility Class-level concern about persistence plus toxicity 116
Early 1960s Detergent foaming in European rivers; industry and OECD develop degradability tests (1965, 1970) Class-level; shows persistence recognised as a problem property 116
1967 EC classification/labelling directive (cited with 1984 test methods as showing persistence “appeared early” as undesirable; exact date of entry not stated) Institutional 116
1973 Commercial MTBE production in Europe — 110
Late 1970s Italy adds MTBE to petrol; lead phase-out begins — 110
1978/79 OECD group on degradation/accumulation test methods Institutional 116
1979 US commercial production; EC sixth amendment (pre-market notification of new substances, scaled to volume) Institutional; did not apply to MTBE 110, 116
18 Sept 1981 Cut-off: MTBE classed as an “existing substance”, exempt from new-substance rules — 116
Early 1980s The authors’ counterfactual “introduction” date; ECETOC (1982) and ISO (1984) tests — 115–116
1985 Directive 85/536/EEC: Member States cannot prevent oxygenates below set limits (MTBE ≤15%) — 111
Mid-1980s Most other European countries adopt MTBE — 110
1990 First MTBE-specific warnings (lab scale): Barker et al. (moves at groundwater speed); Jensen and Arvin. Arvin informs Danish EPA, which dismisses it. Credible but lab-scale 111, 114, 115
1990 US Clean Air Act amendments require oxygenates; US MTBE use “greatly increased” — 110–111, 121
1990s Comprehensive carcinogenicity bioassays (1992 inhalation; 1995/98 oral; 1997) Mixed/borderline 113
1993 EU existing-substances rules adopted (chapter’s citation points to Directive 93/67/EEC on notified substances; see 11.8.1 note) Institutional 116, 122–123
1993–94 USGS sampling: MTBE second most frequent VOC in shallow urban groundwater Strong (field) 111
Jan 1995 RFG “containing more MTBE” introduced in severe-ozone US metropolitan areas (present-day average about 11%, per sources dated 1997 and 2000) — 111
1995 MTBE third most produced organic chemical in the US (about 8.0 Mt) — 110
Late 1995 to June 1996 Santa Monica detection; some supply wells shut; 71% of local supply lost; USD 3.5m/yr to buy in about half its water Harm realised 114
1996 USGS report (Squillace et al.): concern becomes widespread; “regulatory system was alerted”. NRC finds MTBE does not significantly affect CO/NOx/VOC emissions Strong 111, 115
1997 NSTC and US EPA: carcinogenic potential/hazard for humans; Danish odour threshold 180 µg/l Contested 112–113
1998 Danish ministry report admits the 1990 warning; tentative 30 µg/l limit; UC task force reports (recommend phase-out; CBA); EPA Blue Ribbon Panel set up; NTP board votes 6–5 against listing; EU fuel-quality directive 98/70 — 113–115
1999 California Executive Order D-5-99 (calls for a removal timetable by 1 July 1999, with removal no later than 31 Dec 2002); IARC “not classifiable”; IPCS; NRC: oxygenates negligible for ozone; Blue Ribbon Panel report; CAL-EPA odour threshold 5 µg/l — 111–114
Feb 2000 Joseph editorial on asthma Weak/hypothesis 112
Mar 2000 US EPA announces steps to “significantly reduce or eliminate” MTBE Response announced 114–115
Aug 2000 Danish EPA lists MTBE as undesirable; plans economic instruments Response 115
Nov 2000 ECB working group rejects Category 3/R40 classification Non-action 113
Jan–Mar 2001 Finnish EPA draft risk assessment and risk reduction strategy: risk reduction “justified” — 115–116, 123
19 Apr 2001 Arthur D. Little report on USTs: EU focus on tank risk reduction; full effect expected by 2005 Response (risk reduction, not substitution) 115
10 May 2001 Chapter editing finished; “The debate goes on” — 110, 121

Responses by actor - Industry. Chose MTBE on cost, blending and infrastructure criteria (p. 110). Detergent-industry precedent shows industry can engage in persistence testing (p. 116). No direct evidence is given of how petrochemical firms assessed MTBE or responded to warnings. A claim that industry would resist refinery changes is asserted (p. 118). - Regulators. - Denmark dismissed the 1990 warning (p. 114), set a tentative limit in 1998, and listed MTBE as undesirable in 2000 (pp. 114–115). - US federal law drove scale-up (1990). EPA then convened a panel (1998) and announced a reduction or elimination (2000) (pp. 110–115). - California acted first, through the Legislature, its university task force and an Executive Order (p. 114). - The EU harmonised permissive use (1985) and left evaluation to the existing-substances track (1993 to 2001). The ECB working group declined carcinogen classification (2000). As of 2001, EU risk-assessment and risk-reduction reports leaned toward tank-focused risk reduction, which the authors describe, hedging, as the position the EU “tends to” hold (pp. 111, 113, 115–116). - Scientists. Groundwater scientists raised the persistence and mobility warnings (Barker; Arvin; USGS). Toxicologists split on carcinogenicity (Box 11.1). State-commissioned academics (UC) produced the synthesis that recommended phase-out, which the chapter says led (“As a result”) to the Executive Order calling for a removal timetable, plus a separate CBA (pp. 114, 118–119). - Publics and local officials. Concern from “State and local officials and the public” prompted the EPA Blue Ribbon Panel (p. 114). Local utilities (Santa Monica) and private well owners (Standish) bore the first harms (p. 114). - Courts. Not discussed. Liability is raised only as the open question “Who is the polluter, who should pay?” (p. 119).

Lags (computed from the chapter’s dates) - Generic knowledge that ethers persist (1954/1960) predates European production (1973) by 13–19 years. No documentation shows it was applied (p. 115). - First MTBE-specific warning (1990) to regulatory alert (1996): about 6 years (p. 115). To the Danish tentative limit (1998): 8 years. To the California Executive Order (1999): 9 years. To the US EPA announcement and Danish “undesirable” listing (2000): about 10 years. To California’s phase-out deadline (end-2002): about 12 years. - European production start (1973) to the draft EU risk assessment (2000/01): about 28 years. - Production start (1973 Europe / 1979 US) to the first comprehensive carcinogenicity bioassay reports (1992): about 13–19 years. Classification was still unresolved in 2001. - Effective action, as of the chapter’s writing: in the US it was pending (an announced federal intent plus a state removal timetable). In the EU, the direction of the risk-assessment and risk-reduction reports was risk reduction on tanks, with national tank rules due to take full effect by 2005, rather than restriction of MTBE. - Scale-up and warning coincided (1990). In the US, the mandate that multiplied exposure arrived just as the first evidence of mobility was being presented.

Harms and costs documented in the chapter - Contamination: widespread in the US; the second most frequent VOC in urban shallow groundwater in 1993–94 (p. 111). - Santa Monica: 71% of local supply lost; USD 3.5m/yr for bought-in water (p. 114). - Standish, Maine: more than 20 domestic wells contaminated from one accident (p. 114). - California cost estimates: - water treatment USD 340–1,480m/yr; - consumer fuel costs USD 435–1,055m/yr (the chapter’s wording is “the direct price paid by consumers for oxygenated petrol”, so this may not be specific to MTBE); - recreational boating USD 160–200m/yr; - against air-quality benefits of USD 14–78m/yr (p. 119). - Separately, tank and pipeline leak costs of “tens to hundreds of millions” per year (p. 114). - EU: contamination not on the US scale, but cases are sufficient for concern, and monitoring is absent (p. 112). - Health: cancer is borderline or unresolved, and asthma and endocrine effects are in “near ignorance” (pp. 112–113, 120). No human health harm is demonstrated in the chapter.

The authors’ own lessons and conclusions#

Lessons the authors derive from their evidence 1. Foreseeability. By applying foresight at introduction, “it could have been possible to predict” that persistency “would eventually be deemed problematic by society” (p. 120; also p. 117). Basis: generic knowledge of ether persistence (1954/1960) and the established institutional concern with persistence (1960s–1980s) (pp. 115–116). Caveat the authors state themselves: “Documentation for this argument has however not been found” (p. 115). The argument in question is that “competent chemists and microbiologists could have anticipated that MTBE might be persistent in groundwater”; no contemporary record shows anyone did. 2. Triggers for investigation. Persistency, mobility and high production volume “should have prompted systematic, comprehensive investigations”, which could have found taste, odour and cancer suspicions earlier, “and could have acquitted the chemical or confirmed suspicions” (p. 120). 3. Warnings ignored. The groundwater risk “was ignored for a long time”. The 1990 warnings drew no regulatory reaction until the 1996 field results (pp. 115, 120). The Danish case is documented by the Danish EPA’s own admission (p. 114). 4. Testing after deployment. No comprehensive pre-marketing carcinogenicity testing was done, and uncertainty persists long after scale-up (pp. 112, 117, 120). 5. Irreversibility. Persistence makes MTBE’s risk “ongoing, everlasting”, including unknown effects (pp. 117, 120). 6. Costs. On current knowledge, “water treatment costs make MTBE a very expensive option” relative to modest air-quality benefits (p. 119). 7. Shared short-sightedness. “both industry and society (represented by regulators) were short-sighted” (p. 119). 8. Not grounded in precaution. The choice “was not grounded in precaution, as the possible threats … were not adequately characterised” (p. 117).

Recommendations and advocacy (normative; not derived from evidence alone) - R1. Before “any release of large volumes of a persistent chemical”, carry out systematic investigations into “all known possible adverse effects”, reopened as “new categories of adverse effects are discovered” (p. 120). - R2. Seek alternatives to persistent chemicals “whenever possible” (p. 120). - R3. Pursue best environmental practices such as risk reduction, and actively encourage R&D into alternatives, because risk reduction may be insufficient (p. 120). - R4. Precautionary interpretation: always consider “the risk of being wrong”. Uncertainty “cannot be used as an excuse to postpone action”, and “uncertainty and ignorance warrant precautionary measures” (p. 117). - R5. Proportionality cuts both ways: large potential costs “justify a proportionately costly response” (p. 119). - R6. Select alternatives on best environmental practice “even if these alternatives are more costly in the short term” (p. 119). - R7. Address the “deeper issue”: invest in “alternatives entirely different”, including demand reduction for private motoring or radical engine technology (p. 118). - R8. Substitution principle: investigate substitution “as an alternative to risk reduction” (p. 117).

Lessons reported from other bodies (attributed; not the authors’ own) - University of California: any additive at a “significant fraction” of petrol “may have unexpected environmental consequences”, so fully assess any alternative. Media-specific agencies produce “fragmented and incomplete” assessments (p. 114). - EPA Blue Ribbon Panel: “full, multi-media assessment … of any major new additive to gasoline prior to introduction” (p. 114). - EU (Finnish EPA; Arthur D. Little): risk reduction is justified. Tanks need “strong enforcement” and deterrent penalties (p. 115).

Mechanisms and dynamics#

1. Substitution chosen by the problem it solves, not by its own failure modes. - MTBE was selected on anti-knock performance, cost and compatibility with refinery and pipeline systems (p. 110). - Its environmental fate was apparently not part of the evaluation: “It seems that the possibility that MTBE could pose a threat to groundwater reservoirs was never considered” (p. 117; hedged in the original). - On a comparison with lead, MTBE wins. “There is no doubt that MTBE is a reasonably better option than lead” (p. 117) is the authors’ own 2001 judgement, not a record of how the original choice was framed. The chapter has no such record. A substitute judged relative to a notorious incumbent can pass without being judged on its own terms. The synthesis makes this point for MTBE, hedged (“perhaps”, p. 176, outside section). - The same dynamic recurs down the chain: - MTBE also displaces octane from benzene and aromatics (p. 111); - EU rules to cut benzene and aromatics could increase MTBE (p. 115); - the US exit from MTBE leads toward ethanol, which carries its own combustion-product, land, energy and greenhouse-gas questions (pp. 114, 118). - The chapter supplies each link of what amounts to a cascade of substitutions, each addressing the last hazard. The cascade framing is this note’s synthesis. The chapter endorses UC’s and the Blue Ribbon Panel’s call for full pre-introduction assessment of the next substitute (p. 114).

2. Regulation as a driver of scale, and fragmentation by medium. - The large US increase followed the 1990 Clean Air Act oxygenate requirement and the 1995 RFG programme (pp. 110–111). - The EU directive prevented Member States from blocking oxygenates below the set limits (p. 111). - The air-quality benefit specific to MTBE was small or negligible per the NRC (p. 111). In the California CBA the air benefits of reformulation were “relatively minor” (USD 14–78m/yr) and “essentially the same” whether MTBE, ethanol or non-oxygenated petrol was used, so the CBA credits MTBE with no air benefit over the alternatives (p. 119). - So a regulatory programme aimed at air produced harm in water. UC’s diagnosis of “fragmented and incomplete” media-specific assessment (p. 114) names the institutional mechanism. - The authors distribute blame to “industry and society (represented by regulators)” (p. 119). They do not analyse why the air programme specified an oxygen-content requirement rather than an emissions outcome. My inference, from the text’s own juxtaposition of the mandate (pp. 110–111) and the NRC findings (p. 111), is that the mandate locked in an ingredient class whose specific benefit was doubtful.

3. Grandfathering and legal timing. - The legal concept that persistence is undesirable, and the machinery of pre-market notification scaled by volume, existed from 1979. MTBE escaped both because it was marketed before the September 1981 cut-off (p. 116). - Evaluation of existing substances was legislated only in 1993 and reached MTBE around 2001 (p. 116). - This shows how knowledge institutionalised in law can fail to reach the very items already in widest use. Generalising from this one case (this note’s inference), the largest-volume, earliest-established products can face the least scrutiny.

4. Volume as a hazard multiplier. - MTBE was a very high production-volume chemical (pp. 110, 117). UC stresses quantities that “constitute a significant fraction of the total content of gasoline” (p. 114). - At that scale, routine small losses (overfilling, leaks) and rare accidents (Standish) add up to widespread contamination (pp. 112, 114). - The authors treat projected high volume as something that “could have been expected … to have given rise to concern” (p. 117). The 1979 EU rules already scaled data requirements to volume (p. 116).

5. Hazard emerging from combined properties meeting real infrastructure. - No single property is damning on its own. The combination is: high solubility (50,000 mg/l), mobility at groundwater speed, slow subsurface degradation, and a very low sensory threshold (5–180 µg/l) (pp. 111–112). - Together with leaky storage infrastructure (the chapter points to “the condition of petrol station underground storage tanks” and to improper installation) and reliance on groundwater for drinking, these properties produce loss of the resource (pp. 112, 115). - The authors’ precautionary question is exactly about combinations: “does MTBE exhibit any other characteristic which, when combined with persistency, requires a response” (p. 117). - The harm is systemic. It sits in the interaction between the product, the logistics chain and the water system, not in the product alone.

6. How warnings were discounted. - Scale of evidence. The 1990 warnings were “only at the laboratory scale”. Regulators waited for the “full-scale results” of 1996 (p. 115). An implicit standard of proof required field demonstration. - Beliefs about infrastructure. The Danish EPA judged tank leakage “a minor problem” (p. 114). US data later showed even double-walled tanks with detectors leak (p. 115). - Absence of evidence read as evidence of absence. Denmark dismissed the warning partly because “problems with gasoline components were rarely found in groundwater at the time” (p. 114). But MTBE was “not routinely monitored in groundwater” even in 2001 in the EU (p. 112). The chapter supplies both facts but does not join them: a problem nobody is looking for is “rarely found”. Caveat: this link is this note’s synthesis. The monitoring gap is documented for MTBE across the EU in 2001. The Danish remark concerns “gasoline components” generally in 1990, and the chapter says nothing about Danish monitoring practice then. - Venue. The first warning (Barker et al.) was presented at a conference co-organised by the American Petroleum Institute (reference, p. 121). This suggests the finding was aired in a forum with industry involvement. It does not show that industry decision-makers registered it, and the chapter does not examine what industry did with it.

7. Handling uncertainty: classification defaults, thresholds and split votes. - On carcinogenicity (Box 11.1, p. 113), the same studies produced “carcinogenic potential” (US EPA, NSTC), “potential human carcinogen” (UC), “probable human carcinogen” (Mehlman), “not classifiable” (IARC, IPCS), a 6–5 vote against listing (NTP board), and “borderline” followed by non-classification (EU). - Three mechanisms tip borderline cases toward non-action: - the absence of human epidemiology counts as “inadequate evidence”; - mode-of-action arguments (rodent-specific mechanisms; non-genotoxic with a threshold) discount animal findings; - the default when evidence is borderline is non-classification. - The regulators’ mental model, as the authors reconstruct it, is that a threshold exists, so the risk is “considered manageable” (p. 113). - The authors do not directly challenge the threshold reasoning in the box. Their challenge is structural: persistence, irreversibility and ignorance should shift the burden (pp. 117, 120).

8. Ignorance, not just uncertainty. - The chapter distinguishes uncertain effects (cancer) from effects in “near ignorance” (endocrine, asthma; p. 120). - It argues that persistence exposes present and future generations to effects “of which we are still ignorant” and that “the list of known negative features … is still expanding” (pp. 117, 120). - Ignorance is sustained by the fact that studies were neither required nor designed: “there have been no studies to date designed to address this issue” (asthma, p. 113), and “no requirement at the time” (p. 117).

9. Risk reduction versus substitution, and the mental models behind each. - In the authors’ hedged characterisation (“it seems that the position in the EU tends to be”), the EU sees tank leakage as “a technical problem that can be managed by risk reduction”. The EU reports expect risk to remain “low in the future” and rely on enforcement and penalties (p. 115). - The chapter notes, as the EU report itself does, that installation errors defeat sophisticated tanks and that non-compliant tanks remain in the interim (p. 115). The authors’ own conclusion is that, because of “the possibility of risk reduction being insufficient”, alternatives must be researched and substitution investigated (pp. 117, 120). - This is a contrast in confidence. Regulators trust engineered containment plus enforcement. The authors trust neither to be perfect when the failure is irreversible.

10. Incumbent infrastructure, sunk capital and lock-in. - MTBE’s selling points were that it fits existing refineries and pipelines (p. 110). - The alternative of restructuring branched alkanes “may however be opposed by the petrochemicals industry” because of “major investment costs” (p. 118). This is asserted without evidence. - Alternatives were underdeveloped when needed: “none of them has been investigated to the point where they could be implemented on a wide scale in the short term” (p. 118). - Lock-in here is less contractual than infrastructural and cognitive. The path of least disruption was taken, and no parallel investment was made in alternatives.

11. Asymmetric time profiles of benefits and harms. - The benefits were modest and declining as the fleet modernised (p. 111). - The harms are persistent, accumulating and potentially “everlasting” (pp. 117, 120), with remediation costs “likely to be out of reach” once the chemical is spread (p. 117). - The chapter’s CBA (p. 119) makes the asymmetry concrete in money terms for California.

12. Distribution of costs and diffusion of responsibility. - This note’s inference, which the chapter does not state: the benefits (cheap octane, blending convenience, compliance with the air-quality mandate) accrued mainly to fuel producers and, diffusely, to air quality (pp. 110–111, 119). - The costs fell on water utilities and municipalities (Santa Monica), private well owners (Standish), consumers (fuel price) and the public purse (risk reduction, monitoring) (pp. 114, 119). - Because regulators mandated or permitted the use, the polluter is ambiguous: “Who is the polluter, who should pay?” (p. 119). The chapter concludes the majority of costs have been “borne by society” (p. 119).

13. Divergence between jurisdictions from context and from monitoring. - The US used far more MTBE. The chapter’s volume figures are not like-for-like: US production was about 8.0 Mt/yr in 1995, against EU production of 3.3 Mt in 1999, of which 2.3 Mt was used in the EU (p. 110). No US consumption figure is given. By content, US RFG averages about 11% and US conventional petrol 2–3%, against an EU average across all petrol of 2.1% (p. 111). (Comparing the RFG figure with the EU all-petrol average would overstate the gap.) Europe therefore saw problems “later” (p. 115). Local severity depends on dependence on groundwater and the condition of tanks (p. 112). - EU confidence (“not on the same scale as the USA”) rests partly on unpublished and sparse monitoring (p. 112). Lower-use jurisdictions can learn from the first movers, but they can also mistake unmeasured for absent. - The same evidence led to a US move toward phase-out (announced, not yet completed, in 2001) and an EU risk-management path. The ENDS headline “EU backs MTBE despite US prohibition” (p. 122) captures the perceived split, though “prohibition” overstates the US position at the time.

14. Crisis-driven institutional learning. - Harm realised in Santa Monica (1995–96) and the USGS data (1996) produced: - a legislatively created scientific task force (UC); - a CBA; - an Executive Order; - a federal panel convened in response to public and local pressure; - a federal announcement (pp. 114–115). - The UC task force and the Blue Ribbon Panel each distilled a generalisable rule: full (multi-media) assessment before introducing any major new additive or alternative (p. 114). - The mechanism is that an acute, visible, local loss (a city’s wells) converts diffuse technical warnings into political action.

15. Framing and language. - The product. The title and opening frame MTBE as “a substitute for lead”, a remedial technology (pp. 110, 117). The US programme’s names (“reformulated gasoline”, Clean Air Act “oxygenates”) give the additive environmental credentials (pp. 110–111). This framing plausibly lowered scrutiny. The editors suggest as much later, hedged: problems were “perhaps more readily overlooked” (p. 176, outside section). Chapter 11 only implies it. - Regulators’ language, and the authors’ reconstruction of it. Only one of these phrases is quoted from a regulatory or consultancy source: risk “remains low in the future” (ADL/EU report, p. 115). “considered manageable” (p. 113) closes the authors’ own reconstruction of the ECB’s “probable basis”. “a technical problem that can be managed by risk reduction” (p. 115) is the authors’ hedged characterisation of the EU position. Together they present regulatory thinking in the language of containment and control, but that is partly the authors’ framing. - The authors’ language. - Moral and temporal vocabulary: “futurity”, “foresight”, “beckon(s)” (used four times, pp. 118–120), “everlasting risk”, “unwelcome ‘surprises’”, “guarantor of futurity” (p. 120). - Emotive in places: “alarmingly high rates” (p. 112). - Candid in others: “Documentation for this argument has however not been found” (p. 115); “could have acquitted the chemical” (p. 120).

Transferable insights (technology-neutral)#

  1. A substitute judged against the hazard it replaces can escape evaluation on its own terms. “Better than the incumbent” is not “adequately characterised”. - Evidence: selection criteria (p. 110, from a 1996 retrospective source); “It seems … never considered” (p. 117); “reasonably better option than lead” alongside “not grounded in precaution” (p. 117); UC’s call to fully assess any alternative (p. 114). - Strength: moderate–strong. The outcome is well documented, and official bodies drew the same lesson. The decision process itself is not documented (p. 115), so the mechanism is inferred rather than shown.
  2. Substitution chains: fixing one hazard creates pressure toward the next substitute, often less studied. - Evidence: lead → MTBE (p. 110); benzene/aromatics cuts → more MTBE (pp. 111, 115); MTBE → ethanol or other ethers with their own concerns (pp. 114, 118). - Strength: moderate. Each link is documented. Harm from later links is anticipated, not shown.
  3. Rules that reach only new entrants leave the most established, highest-volume items the least examined. - Evidence: the 1979/1981 new-substance regime did not cover MTBE; existing-substances regulation came in 1993; the MTBE risk assessment was only reaching draft by end-2000, with the final expected in 2001 (p. 116). - Strength: strong that MTBE fell outside the new-substance regime and was evaluated late. Moderate on the counterfactual that coverage would have changed the outcome: the chapter says persistency “could” have triggered classification, and it does not show that the notification data set would have exposed the groundwater problem. (The chapter’s 1993 citation is also mismatched; see 11.8.1.)
  4. Scale changes the nature of risk. Deployment volume should itself trigger scrutiny, because small per-unit loss rates add up to systemic exposure. - Evidence: high-production-volume status (p. 110); USGS detection frequency (p. 111); small spills causing plumes (p. 114); UC’s “significant fraction” warning (p. 114); volume-scaled data rules (p. 116). - Strength: strong for this case; moderate as a general rule.
  5. Hazards can emerge from combinations of individually modest properties interacting with the real-world infrastructure a product moves through. Assess the system, not the substance alone. - Evidence: solubility, mobility, persistence and a low odour threshold, combined with leaky tanks and reliance on groundwater (pp. 111–112, 115); the combination question (p. 117). - Strength: strong. Well evidenced within the chapter.
  6. Harm can be loss of use of a shared resource at trace levels, not toxicity. Such harm is real, costly and falls on third parties. - Evidence: “render large reserves of groundwater useless” (p. 112); Santa Monica 71% and USD 3.5m/yr (p. 114); California treatment costs (p. 119). - Strength: strong.
  7. Programmes built for one objective or medium can create harm in another. Fragmented, single-medium institutions produce incomplete assessments. - Evidence: the air mandate drove use (pp. 110–111); water harm (pp. 112, 114); UC on media-specific agencies (p. 114); the Blue Ribbon Panel’s multi-media recommendation (p. 114). - Strength: moderate–strong. Official bodies drew this conclusion. The chapter does not analyse the mandate’s design.
  8. Benefits of a package can be wrongly credited to one component. Test component-specific benefit before mandating or scaling the component. - Evidence: RFG compared with MTBE; NRC 1996 and 1999 findings (p. 111); similar air benefits across options in the CBA (p. 119). - Strength: moderate. Rests on cited NRC assessments, summarised briefly.
  9. Early small-scale warnings tend to be discounted until large-scale harm appears, and absence of monitoring gets read as absence of a problem. - Evidence: the 1990 lab warnings were ignored until 1996 (p. 115); the Danish EPA’s reasoning (p. 114); no routine EU monitoring (p. 112). - Strength: moderate. Strong for Denmark, where there is an official admission. US documentation is asserted rather than shown (p. 115). The “absence of evidence” link is my synthesis of two passages.
  10. Borderline evidence is decided by defaults. The same data yield opposite institutional conclusions, and mode-of-action or threshold reasoning can downgrade concern.
    • Evidence: Box 11.1 (p. 113): the 6–5 NTP vote, IARC/IPCS against EPA/NSTC, and the ECB rejecting classification despite a “borderline” rapporteur view.
    • Strength: strong as a description of divergence. The chapter does not show which judgement was right.
  11. When key testing is not required before deployment, it arrives decades later, and uncertainty outlives the decision to scale.
    • Evidence: “no comprehensive pre-marketing carcinogenicity testing” (p. 112); tests in the 1990s (p. 113); “near ignorance” on endocrine and asthma effects in 2001 (p. 120).
    • Strength: strong on the timeline; moderate on the counterfactual that earlier testing would have changed the decision.
  12. Irreversibility creates asymmetric costs of error. Where harms persist and accumulate while benefits are modest and eroding, a lower threshold for investigation and more investment in alternatives are warranted.
    • Evidence: benefits declining (p. 111); persistence “implies irreversibility” (p. 117); CBA asymmetry (p. 119); “everlasting” risk (p. 120).
    • Strength: moderate. A coherent normative argument, supported by the CBA. “Everlasting” is overstated given the chapter’s own evidence on degradation (pp. 111–112, 114).
  13. Engineered containment is only as good as its installation, maintenance and enforcement. Transition periods leave legacy exposure.
    • Evidence: the Couch and Young finding on double-walled tanks; interim non-compliant tanks; the need for enforcement and penalties (p. 115).
    • Strength: moderate. One cited US study plus the EU report’s own caveats.
  14. When an authority mandates or permits the activity, responsibility for harm diffuses and costs default to the public and to third parties.
    • Evidence: “Who is the polluter, who should pay?”; majority of costs “borne by society” (p. 119); local utilities and well owners (p. 114).
    • Strength: moderate. The examples are concrete. The “majority” claim and “human health costs” are asserted.
  15. Formal cost–benefit analysis under ignorance is provisional in both directions. Comparative assessment of options remains useful, and proportionality can justify costly responses when potential costs are very large.
    • Evidence: pp. 118–119, including both upward (asthma) and downward (waterworks biodegradation) revisions.
    • Strength: moderate. The argument is balanced. The numbers come from one state-commissioned analysis under Californian conditions: RFG with high MTBE content, contamination of both groundwater supplies and surface reservoirs, and treatment assumed to be granular activated carbon to below 5 µg/l.
  16. Alternatives must be developed before they are needed. When a problem surfaces, the absence of mature replacements lengthens dependence on the problematic option.
    • Evidence: no alternative ready for wide-scale short-term use (p. 118); recommendation to push R&D so alternatives become “feasible as soon as possible” (p. 120).
    • Strength: moderate.
  17. Contextual divergence: the same product carries different risk across jurisdictions depending on intensity of use, infrastructure condition, reliance on the vulnerable resource and monitoring. Late adopters can learn, or can mistake unmeasured for absent.
    • Evidence: pp. 111–112, 115.
    • Strength: moderate.
  18. Acute, visible local losses convert diffuse technical warnings into political action and institutional learning.
    • Evidence: the sequence from Santa Monica and USGS to the UC task force, Executive Order, Blue Ribbon Panel and EPA announcement (pp. 114–115).
    • Strength: suggestive–moderate. The sequence is documented. The chapter implies causation but does not trace it.
  19. The component problem may be a symptom of an underlying demand. Durable solutions may require questioning the need or investing in radically different systems.
    • Evidence: p. 118.
    • Strength: asserted. Normative and not evidenced in the chapter.
  20. Decision robustness and flexibility: build in reopening points as new categories of harm are recognised, and keep alternatives alive.
    • Evidence: pp. 110, 120.
    • Strength: asserted. The organising frame of the chapter, not tested within it. The US moves of 1999–2000 (a state removal timetable and a federal announcement, pp. 114–115) show that reversal was being attempted. As of 2001 it was not complete, and the costs were already high.

Limitations, contestation and bias check#

Where the chapter is advocacy rather than analysis - Stated purpose. The chapter poses a checkable question: could MTBE’s problems have been foreseen (p. 110)? But its declared aim is to “illustrate” foresight (pp. 110, 115). The case is mainly used to demonstrate a principle rather than to test it. - Section 11.8.3 on transport. The “deeper issue” of transport, India and China, want versus need, and hydrogen fuel cells (p. 118) goes well beyond the evidence. Neither the effectiveness nor the costs of these options are examined. - “Everlasting risk” (p. 120) sits uneasily with the chapter’s own evidence: - anaerobic field degradation (p. 111); - “very slow but measurable” aerobic degradation (p. 111); - possible degradation in sand filters (pp. 114, 119), though that bears on treatability at waterworks rather than persistence in aquifers. On the other side, the chapter’s field evidence on aerobic degradation is negative (“could not be detected”, p. 111), and “significant rates” were seen only in the laboratory (p. 112). “Long-lived” would be defensible; “everlasting” is rhetorical. - “alarmingly high rates” of asthma “where MTBE was in use” (p. 112) implies a link the cited sources cannot carry. That sentence rests on a journal item on two Philadelphia schools, a Stamford health-department survey, a Hartford newspaper report and a CDC conference paper (pp. 122–124), none of which mentions MTBE in its title. The wider asthma paragraph adds an editorial and conference papers by Joseph. - “human health costs” borne by society (p. 119) is asserted. The chapter demonstrates no human health harm.

Hindsight bias and the counterfactual - Undocumented anticipation. The central counterfactual rests on generic knowledge of ether persistence (1954/1960) and class-level concern about persistence. The authors concede “Documentation for this argument has however not been found” (p. 115). “Could have anticipated” is plausible, but it is not evidence that anyone did, or that the decision-makers of the time would have weighed it heavily. - Anachronistic list. The list of properties that “could have sparked concern in the early 1980s” (p. 117) includes “potential carcinogenicity and other adverse effects”. Those suspicions emerged only from the 1990s bioassays (p. 113) and later hypotheses (2000). Only persistence, and taste and odour, were plausibly knowable then; even taste and odour rests only on Table 11.1’s unreferenced 1960 row. A charitable reading is that the sentence is hypothetical: a chemical displaying these properties could have sparked concern, and studies “should therefore have been done” to find out (p. 117). On that reading the claim is about what investigation would have revealed. But the chapter shows only that carcinogenicity testing came late. It does not show what earlier testing would have found or how it would have been judged. - Moving standard. The standard is whether properties would “eventually be deemed” undesirable by society (pp. 110, 116, 120). That social-evaluative test becomes easier to meet in hindsight, because the social judgement it refers to happened later. - Unfixed date. The “introduction” date is not fixed. Production began in 1973 and Italian use in the late 1970s (p. 110), but the counterfactual is set in the “early 1980s” (p. 115). An earlier date slightly weakens the claim about how developed test methods and legal concepts were at the time. - Undocumented screening. “Overlooked during the screening of MTBE” (p. 117) presumes a screening that the chapter never documents.

Evidence gaps - No primary sources on the decision-makers. There are no industry documents, regulatory files or decision records on how MTBE was selected in the late 1970s and 1980s, or on the 1990 US oxygenate decision. “the petrochemicals industry chose MTBE” (p. 110) attributes agency without evidence. - The US mandate is under-examined. The decisive US scale-up was a legislative mandate for oxygenates (pp. 110–111), not simply industry’s choice of a lead substitute. The chapter does distinguish the two roles, anti-knock and oxygenate, in 11.4 (p. 111). But the title’s framing (“substitute for lead”) fits Europe’s octane use better than the US oxygenate programme. This note infers that the US programme drove most of the documented harm, since the harm documented is overwhelmingly American. The chapter’s analysis of “the choice … to replace lead” (p. 117) partly conflates the two roles. Who pushed the oxygenate requirement, and why, is not examined. - Reliance on grey literature. Heavy use of drafts and personal communications for load-bearing points: - the ECB decision (Dixson-Decleve); - the anti-classification review (Dekant, via Dixson-Decleve; by its title, a position paper against Category 3); - the threshold reasoning (Østergaard); - biodegradation in sand filters (Arvin); - draft Finnish EPA documents. - One CBA. The only CBA is Californian (high MTBE content in RFG; contaminated groundwater supplies and surface reservoirs; GAC treatment to below 5 µg/l). Transfer to EU conditions is not argued. The chapter does not claim it transfers, but it uses the CBA to call MTBE “a very expensive option” in general (p. 119). - Conflicts named but not substantiated. The box title flags “conflicting interests” (p. 113) but the text does not substantiate any conflict. Readers need the reference list to see who funded what. - Referencing slips. The existing-substances citation points to the new-substances risk-assessment directive (p. 116 against pp. 122–123). “As has been mentioned previously” on taste and odour (p. 117) has no antecedent in the text. There are small date mismatches between Table 11.1 and the text (see Table 11.1 notes). None of these overturns an argument, but they call for care when citing the chapter’s details.

Counter-arguments acknowledged - The chapter does acknowledge: - MTBE is better than lead (p. 117); - no obvious harms were known at introduction (pp. 115–116); - EU contamination is not at the US scale (p. 112); - cheaper treatment may exist (pp. 114, 119); - the ECB did not classify MTBE as a carcinogen (pp. 112–113); - investigation could have “acquitted” it (p. 120); - the CBA might move downwards (p. 119). - The carcinogenicity box presents critics fairly. - By the standards of the report, this is factually balanced even though its normative direction is clear.

Counter-arguments not addressed - Risk–risk trade-offs. Holding back MTBE in the 1980s or 1990s might have meant more aromatics or benzene for octane (p. 111 raises this), or a slower lead phase-out (this note’s own counter-argument; the chapter does not raise it). The costs of precaution (false positives, delay) are not weighed. - Cost of the prescriptions. “Investigations into all known possible adverse effects” before any large-volume release of a persistent chemical, reopened as new effect categories emerge (p. 120), is not assessed for feasibility or opportunity cost. - Ignored warnings or genuine ambiguity? The 1990 lab warning was one signal among many. Whether regulators could reasonably have prioritised it is not tested against what else was on their agenda.

Case selection - An unresolved case in 2001 (“The debate goes on”, p. 121) is used as a “late lesson”. Its status as a lesson depends on subsequent events, so hindsight checking is essential. - The US and EU regulatory responses diverged. The chapter tends to treat the US phase-out as the lesson learned and the EU’s risk-reduction path as the one needing correction. That preference is a normative judgement.

Positionality - One author chaired the report’s editorial team (title page). - The key early warner appears to share the authors’ institution (a background inference). - Several apparent reviewers are also the chapter’s main sources (p. 6). - None of this discredits the chapter, but it narrows the range of perspectives that tested it. No industry or pro-MTBE reviewer is visible.

Panels or dissent - There are none; the 2001 format has no panels. Internal dissent appears only through Box 11.1’s presentation of institutional disagreement.

Notable quotes#

  1. “no comprehensive pre-marketing carcinogenicity testing was carried out” (p. 112)
  2. “Relatively small amounts of MTBE may thus render large reserves of groundwater useless.” (p. 112)
  3. The Danish EPA’s reason for dismissing the 1990 warning: “because problems with gasoline components were rarely found in groundwater at the time” (quoted p. 114)
  4. UC: “The current structure of state agencies which focus on specific media (land, air, water), leads to fragmented and incomplete environmental impact assessments.” (quoted p. 114)
  5. “Documentation for this argument has however not been found.” (p. 115)
  6. “It seems that the possibility that MTBE could pose a threat to groundwater reservoirs was never considered.” (p. 117)
  7. “Of all the negative features of MTBE, persistency is unique in that it implies irreversibility.” (p. 117)
  8. “In the face of ignorance, a cost-benefit analysis is of only limited value to decision-makers and cannot, of itself, form the sole basis for a policy decision.” (p. 118)
  9. “Who is the polluter, who should pay?” (p. 119)
  10. “does persistency alone, without indication of other adverse effects, give reason to apply the precautionary principle?” (p. 120)

Open questions#

Within the text 1. How was MTBE actually evaluated when it was adopted, in Europe (1970s–80s) and for the 1990 US oxygenate programme? Is there any primary record of the environmental-fate question being raised or dismissed? The chapter found none (p. 115). 2. Who advocated the US oxygenate requirement, and on what evidence of air-quality benefit, given the NRC’s later findings (p. 111)? Why an ingredient requirement rather than a performance standard? 3. What did industry, including the American Petroleum Institute as co-organiser of the 1990 conference (p. 121), do with the early mobility findings? 4. Is the chapter’s split between field and laboratory studies consistent with the cited papers? Landmeyer et al. appears to be an aquifer study. Two anaerobic “field” citations look like soil or subsurface-sample studies (p. 111). Was Barker et al. (1990), the “laboratory scale” first warning (p. 115), actually laboratory work? 5. Did the EU’s Directive 85/536/EEC (p. 111) constrain Denmark into economic instruments (p. 115) rather than a ban? The chapter does not say. 6. Did the EU’s position as a net exporter of MTBE (1.1 Mt out, 0.2 Mt in, 1999; p. 110) shape its regulatory stance? The chapter provides the data but no analysis. 7. What was Sandrine Dixson-Decleve’s affiliation or client in 2000–01? She is the chapter’s sole source for the ECB decision and the Dekant review (p. 122).

For the hindsight strand

The pointers below come from my general background knowledge. They were not verified in this pass and must be checked against primary sources. - Persistence and clean-up. Did MTBE persistence in groundwater prove as “everlasting” as claimed (p. 120), or did natural attenuation and bioremediation (including biological treatment at waterworks, per Arvin, p. 114) turn out to be viable? What happened to the degradation product TBA? - Carcinogenicity. Did any body change its classification after 2001? My recollection is that IARC remains Group 3 and the EU never classified MTBE as a carcinogen. - Asthma and endocrine hypotheses (pp. 112–113). Were they substantiated or dropped? - US policy. Did the California deadline hold? My recollection is that it was postponed by a year, to end-2003. What happened federally? My recollection is that the 2005 Energy Policy Act removed the RFG oxygen requirement without a liability shield for MTBE producers, after which refiners switched largely to ethanol around 2006. Did many states ban MTBE? - Liability. What did US litigation by water suppliers and states establish? Recollection: multi-district MTBE product-liability litigation, a large settlement around 2008, and verdicts against an oil major in New York City (2009) and New Hampshire (2013). This bears directly on “Who is the polluter, who should pay?” (p. 119). - EU outcomes. Did the EU settle at 1–4% MTBE (p. 115)? Did the tank regulations achieve “full effect by 2005” and keep groundwater risk low? My recollection is that there was substantial substitution toward ETBE, another ether the chapter flagged as possibly similar (p. 118). What does later EU groundwater monitoring show? - Ethanol. Did the chapter’s caveats on land use, fertiliser, energy and greenhouse gases (p. 118) prove prescient? Did ethanol raise new groundwater issues, for example effects on benzene plume behaviour? - Regulatory learning. Were the recommendations for full multi-media pre-introduction assessment of new fuel additives (p. 114) and for treating persistence as a trigger (p. 120) ever institutionalised? My recollection is that the EU chemicals regime later added persistence, mobility and toxicity as hazard criteria (EU classification rules revised around 2023). If so, that would be a direct later echo of this chapter’s “persistency plus mobility” argument. Verify. - Santa Monica. How did the case end (settlements, treatment plant, restored supply), and who paid?

Audit log#

Independent audit against the text extract (pp. 110–125, all pages read) and the PDF. PDF pages 1, 6, 112, 113 and 121 were re-rendered and checked visually. The synthesis cross-references (pp. 170–182) were confirmed by text search of the PDF. Both printed-original errors the reading notes report were confirmed: the micro sign is missing and “stergaard” appears for Østergaard.

Second pass (resumed audit). I re-read the full extract (pp. 110–125) and verified the title page, the acknowledgements (p. 6) and the synthesis cross-reference quotes (pp. 170–177, 182) by extracting text from the PDF. The first-pass fixes above hold. Further changes: - 11.5.4: “showing” became the source’s “indicating”; added “other respiratory or inflammatory conditions”. - 11.5.4: Leikauf et al. is no longer called “peer-reviewed” (the chapter does not say); “alarmingly” is now separated from the real problem, the implied MTBE link carried by “where MTBE was in use”. - Box 11.1: IARC “no epidemiology” became “no analytical epidemiological studies”, as in the source. - Box 11.1: added the Mehlman (2000) reference detail (a two-page polemical commentary on the NTP vote). The pro-classification pole, like Dekant’s, rests on a position piece, which balances the earlier scrutiny of Dekant alone. - Box 11.1: restored UC’s caveat “although further studies would be needed to make a conclusive determination”. - Lesson 1: tied the “Documentation … not been found” caveat to the specific argument it qualifies (chemists could have anticipated persistence). - Responses by actor: “the CBA that underpinned California’s phase-out” became the UC synthesis leading to the Executive Order, plus a separate CBA. - Mechanism 2: the CBA’s “relatively minor” air benefits apply to all three formulations equally. They are not an MTBE-specific benefit. - Mechanism 13: flagged that the US figure (8.0 Mt, 1995) is production and the EU 2.3 Mt (1999) is use; added EU production of 3.3 Mt. - Insight 11: restored “comprehensive” in “no comprehensive pre-marketing carcinogenicity testing”. - Insight 12: “persistence equals irreversibility” became the source’s “implies irreversibility”. - Limitations: rewrote the “alarmingly” bullet to list the sources actually cited for that sentence and to locate the problem in the implied link. - Digest (second pass): restored “comprehensive” in pre-market testing; added the Mehlman/Dekant symmetry; added that CBA air benefits were the same across options and the GAC/”cheaper methods” caveat; tied the “Documentation” caveat to its specific argument; split Insight 10’s strength (strong on timeline, moderate on counterfactual).