LL2-09 — Ch9 The pesticide DBCP and male infertility#

Late lessons from early warnings: science, precaution, innovation (EEA Report No 1/2013), in the part with the running header “Lessons from health hazards”. Report pages 203–214 (PDF pages 205–216). The chapter text runs pp. 203–211. Table 9.1 and the references are on pp. 212–214. Box 9.1 is on p. 205 and Box 9.2 on p. 208. A large photograph of a worker handling bagged banana bunches (credited to istockphoto/Francisco Orellana) fills the lower half of p. 209.

Read in full from the text extract. Every page was also checked visually against the PDF. The extraction is faithful, and no text was lost from boxes, the table or the two-column layout.

Authors and standpoint#

Chapter authors: Eula Bingham and Celeste Monforton (p. 203).

• The chapter says of Bingham only that she was “then US Assistant Secretary of Labor for Occupational Safety and Health” in July 1977 (p. 204). That post was head of OSHA (background; the chapter does not say so in these words). The US National Peach Council letter quoted on p. 206 was addressed to her, “co-author of this chapter”. So one author was the regulator who received the union’s alarm and, by inference from her post, oversaw the emergency rulemaking the chapter narrates. The chapter discloses her role but does not reflect on it. The account is partly an insider’s first-hand record.
• The chapter gives no affiliation for Monforton. Background knowledge, not verified in this pass: she is an occupational health policy researcher, formerly at OSHA and MSHA and later at George Washington University’s School of Public Health. Bingham was a toxicologist at the University of Cincinnati.
• Evident stance. Strongly pro-worker-protection and explicitly precautionary. The summary box ends by saying the lessons “may help in ensuring timely protection from harm, based on precautionary approaches to scientific evidence” (p. 203), and the conclusion repeats this (p. 211). The chapter’s heroes are clear: the workers and their wives, the union (OCAW), NIOSH and OSHA. Its critique is aimed at the manufacturers (Shell, Dow, and Amvac on exports), the fruit companies (Standard Fruit/Dole) and the pre-1970 regulators (FDA, USDA), who accepted industry reassurances. OSHA’s own record is presented favourably (“a flurry of activity”, p. 206; rules “not effectively challenged by any interested party”, p. 206).
• Discipline. Occupational health and toxicology. The chapter reads more as regulatory history and advocacy than as a systematic review. It relies heavily on a small set of sources: Thrupp (1991, a secondary analysis), the NIOSH health hazard evaluation (1977, primary) and criteria document (1978), Goldsmith (1997), and Misko et al. (1993), DBCP: The Legacy (Dallas, Texas). The 1958 Shell document (Lykken 1958) and the 1993 Panamanian medical evaluations (Navaro 1993) are cited only via Misko et al. The text calls the Lykken document “an internal memorandum prepared by Shell” (p. 204), but the reference list describes it as a “Letter to M.R. Zavon, Kettering Laboratory, University of Cincinnati, June 4, 1958” (p. 213), that is, correspondence with an outside laboratory. Several passages carry no citation at all, including the FDA–Shell–Dow discussions (p. 205), the NCI and Dow/Hazleton cancer studies (p. 207), the Dow–Standard Fruit contract (p. 209), all of the settlement figures in 9.8 (pp. 209–210) and the 2010 well-inventory figures (p. 210). The chapter does not say what kind of document Misko et al. is. Its Dallas imprint and its role as a repository for company correspondence suggest a compilation prepared in connection with litigation, but I have not verified this. It matters for source weighting (see Limitations).

Panels and commentaries. This chapter has no panels and no industry or regulator response, unlike several other 2013 chapters. It has one signed box:

• Box 9.2, “The Swedish film documentary, the Dole lawsuit and freedom of speech”, by Maria Albin (p. 208). The box gives no affiliation. Background, not verified: Albin is a Swedish occupational and environmental medicine researcher at Lund. She also co-authored Albin and Giwercman (2009), “Evidence of toxic effect of a chemical on reproduction ignored” (in Swedish, Läkartidningen), which is in the reference list (p. 212) but not cited in the chapter text. The box is sympathetic to the filmmakers and frames Dole’s lawsuit as a SLAPP. That framing is the filmmakers’ own characterisation, but it was later supported by a 2010 Los Angeles court ruling, which the box reports. Dole’s position gets one clause: the company withdrew the suit citing Swedish free-speech concerns “although it continued to believe in the merits of its case” (p. 208). Albin’s view should not be merged with Bingham and Monforton’s, though they are fully aligned.

Box 9.1 (animal toxicity data, p. 205) and Table 9.1 (timeline, p. 212) are unsigned and presumably by the chapter authors.

Section-by-section notes#

Summary box (p. 203)#

• DBCP is a soil nematicide used mainly on tropical fruit. It was “introduced into US agriculture in 1955 and approved for use as a fumigant in 1964” (p. 203). Apparent inconsistency: the body dates FDA registration “as a pesticide” to 1961 (p. 205), as does Table 9.1 (p. 212). The 1964 date could refer to a separate fumigant approval, but the chapter never mentions one or explains the difference. Table 9.1 puts the start of manufacturing in 1956, against 1955 on p. 204.
• The storyline in brief: rodent testicular effects known by 1961. The chemical was “widely marketed and became a commercial success” anyway. Production workers discovered their own infertility in 1977. US controls followed, but marketing and spraying continued in Latin America, the Philippines and Africa (p. 203).
• “By the 1990s, tens of thousands of plantation workers … had allegedly suffered adverse reproductive effects” (p. 203). This is the only explicit hedge on the headline figure. Elsewhere the chapter attributes the Philippine accounts to “reports collected by lawyers” (p. 209) and notes that “many of the relevant facts are still unclear and contested by the growers” (p. 210). The story “continues today” with compensation claims, water contamination, and “industry attempts to prevent a Swedish documentary on the issue from being screened” (p. 203).
• Significance claim: “the first clear example of reproductive damage to workers who manufactured and used a synthetic chemical”. It is also presented as one of “many examples” behind concern about rising reproductive disease and endocrine-disrupting chemicals (p. 203). Prescription: “closer integration of scientific disciplines, as well as government action” and “precautionary approaches to scientific evidence” (p. 203).

Epigraph and 9.1 The discovery 1977 (p. 204)#

• Epigraph: Charles Hine, “Shell Contractor, c. 1960”, remembered by Goldsmith (1997) as saying they had “identified” a “male birth control drug … not very pleasant to use” (p. 204). Hine was at UCSF and was also “working as an expert consultant for Dow and Shell” (pp. 204–205): a dual academic and industry role.
• In July 1977 Tony Mazzocchi (Vice President of OCAW) contacted Bingham, then Assistant Secretary of Labor for OSHA. Workers at a California plant had confided “during a lunch hour” that they were worried about not having children. “Their wives had been examined and now they thought it might be that they themselves had a problem” (p. 204). The union arranged sperm counts for seven men, which were “either zero, or so low” that they showed sterility. The union then sought a NIOSH Health Hazard Evaluation (p. 204).
• The heading’s quoted phrase “our union members are sterile” does not appear in the quoted text.
• Gendered sequence: the wives were examined first. My inference: this may reflect a default assumption that the problem lay with the women. The chapter does not comment, and it gives no evidence on whether this delayed the men’s testing.
• Chemistry and market: 1,2-dibromo-3-chloropropane, trade names Fumazone, Nemagon, Nemaset, Nematox. Applied at 10–125 kg/ha by soil injection or in irrigation water, mostly on tropical crops. Made by Dow, Shell and Occidental in the US, and also in the UK (International Chemical Company) and Japan. Peak use of “an estimated 14.7 million kg” a year before 1977. No specific source is given, and it is unclear whether this is a US or global figure (p. 204).

9.2 Early warnings: 1961–1975 (p. 204) and Box 9.1 (p. 205)#

• The heading says 1961–1975, but the text, Box 9.1 (“1958–1975”) and Table 9.1 all start in 1958.
• “The earliest research on DBCP toxicity was carried out by two chemical companies producing the compound” (p. 204). By 1958 Shell and Dow had rat data showing absorption “through the skin and by inhalation” and effects on “liver, lung, kidney and testes” (p. 204). Dermal uptake was known from the start.
• Hine’s Shell-contract data: at 5 ppm the testes shrank, at 10 ppm most were half normal size, and at 20 ppm “all the male rats were sterile” (p. 204). The 1958 Shell document (Lykken, cited via Misko et al. 1993; “an internal memorandum” in the text, a letter to M.R. Zavon at the Kettering Laboratory in the references) reads: Dow “are very upset by the effects noted on the testes” (p. 204). The authors infer company scientists were “clearly worried” (p. 204). My note: the memo supports that inference better than the party anecdote does, which could be read as casual familiarity.
• Published data in 1961 showed “two outstanding toxic effects”: an antispermatogenic effect in males and kidney damage in both sexes (p. 204).
• Box 9.1 (p. 205), from Torkelson et al. 1961 (rats; 7 h/day, 5 days/week, 10 weeks):
• 5 ppm: an 18.6% fall in testis weight, “not statistically significant”.
• 10 ppm: a significant 49% fall, and kidney weight up 31.7%.
• 12 ppm: degeneration of sperm-forming tissue, fewer and more abnormal sperm, and kidney and liver changes.
• Guinea pigs and rabbits at 12 ppm: significant testis-weight falls.
• “European laboratories” (Rakhmatulayve 1971; Reznik and Sprinchan 1975) found ~50% reductions in testis weight, sperm count and motility, and a disrupted reproductive cycle in female rats. My note: both are Soviet-journal papers; one is flagged “in Russian” in the references (p. 213). Their accessibility to US decision-makers is not discussed.
• The 5 ppm result is plausibly relevant to the 1958–61 dispute, though the chapter does not connect the two. The FDA saw adverse effects at 5 ppm “after repeated exposures”, while Shell told the USDA that “at 5 ppm no adverse effects had been reported” (p. 205). Shell’s claim also sits awkwardly with the chapter’s own account of Hine’s Shell-contract data, in which “at an exposure to 5 ppm … the testes in male rats shrank” (p. 204). The chapter does not say which dataset (Hine’s pre-1961 work or the published Torkelson et al. analysis) was on the table in the regulatory discussions.

9.3 Pesticide registration and inadequate “hazard control” 1961–1977 (pp. 204–205)#

• “The years 1958–1961 were a critical period” (p. 204). Hine, as consultant, supported registration but called for workplace levels below 1 ppm and “impermeable protective clothing … if skin contact was likely” (p. 205).
• The FDA noted adverse effects at the lowest level studied (5 ppm), so precautions “appeared inadequate. However, the Shell representative considered the Hine recommendations to be impractical” (p. 205). The company’s own consultant was more protective than the company. This paragraph carries no citation. It describes discussions with the regulator, not an internal company decision, and it does not say what role the Shell representative held.
• Torkelson, Hine et al. (1961) recommended <1 ppm and “stressed that suitable analytical methods rather than sensory perception should be depended upon for control”. Men briefly exposed to 1.7 ppm reported “a definite, not unpleasant odour” (p. 205).
• 1961: the FDA “approved and registered DBCP as a pesticide and recommended the exposure limit of 1 ppm” (p. 205). Note “recommended”: the chapter does not say the 1 ppm figure was enforceable. To check: before 1970, FIFRA registration was a USDA function and FDA set residue tolerances (background knowledge).
• The USDA-approved label said only “Do not breathe vapours, use only in a well-ventilated area and avoid prolonged breathing”. It had “no reference to testicular damage” (p. 205). The USDA had “reservations”, but accepted Shell’s argument that the 1.7 ppm odour threshold was an adequate “warning” (Thrupp 1991): “USDA accepted these reassurances even though no studies had been performed to indicate that this approach was safe” (p. 205).
• The authors themselves say the “‘odour threshold’ for DBCP exposure was too high to ensure reliable protection against the toxicity reported in the animal studies” (p. 205). Analytical note: the odour threshold (1.7 ppm) sat above the 1 ppm recommended limit, so it could only signal a breach after it had happened. The published study had also advised against sensory control.
• “Virtually no attempts were made to determine if the measures adopted were indeed safe for manufacturing workers or pesticide sprayers. Neither group was subjected to medical surveillance” (p. 205). Table 9.1 says that a 1961 plant medical examination did take place, but “testicular function is not examined” (p. 212). This is a minor tension with the body’s “no medical surveillance”, presumably resolved by reading the 1961 examination as a one-off rather than surveillance. DBCP fumes are only mildly irritating, so workers “could therefore be exposed to dangerous amounts of the chemical without being aware of it” (p. 205).
• Authors’ verdict: precautions “were therefore inadequate from the 1950s until 1977”. The 1961 data “was sufficient to have required specific health warnings, personal protective equipment and medical surveillance. None was provided” (p. 205). This is a normative counterfactual. It is plausible, since the industry’s own consultant recommended part of it (impermeable protective clothing and a sub-1 ppm limit), but the chapter does not report that anyone in 1958–61 recommended specific health warnings or medical surveillance, and the verdict is not benchmarked against 1961 practice or law.

9.4 Actions to reduce exposure in DBCP manufacturing: 1977 and 1978 (pp. 206–207)#

• NIOSH was created in 1970 (misnamed on p. 206). Its Health Hazard Evaluation at Occidental’s Lathrop plant found 0.29–0.43 ppm (8-h TWA). Of 13 production workers, 9 were azoospermic and 4 oligospermic, with a “clear increase in the prevalence of oligospermia with increasing exposure” (Whorton et al. 1977, Lancet). These levels were “far below” the animal doses and below the “‘safe’” 1 ppm (p. 206).
• Caveat not raised: 1977 air levels are a snapshot, and past exposures may have been higher.
• Sequence: OCAW’s President petitioned OSHA on 23 August 1977 for a 1 ppb limit and medical testing “to identify cases of sterility and cancer among exposed workers”. “a flurry of activity” followed. Emergency Temporary Standard on 9 September 1977; proposal in November (the references give 1 November 1977); hearings in December; final standard 11 March 1978 (p. 206). The references give 17 March 1978, which is inconsistent (p. 213).
• OSHA also used Dow’s Magnolia, Arkansas data: 0.04–0.4 ppm, with “50 % of the 106 workers examined” oligospermic or azoospermic (p. 206). The chapter says these data “suggested that exposures below 1 ppm were associated with adverse reproductive effects”, while noting that dermal uptake “may have contributed an unknown but potentially significant amount” (p. 206).
• The final rule set 1 ppb TWA and 10 ppb over 15 minutes, with initial and annual medical examinations, respiratory protection and training. It was based on human and animal reproductive effects and animal carcinogenicity. It took effect in April 1978 and was “not effectively challenged by any interested party” (p. 206).
• The US National Peach Council letter to Bingham (September 1977) sought delay: “While involuntary sterility caused by a manufactured chemical may be bad, it is not necessarily so” (p. 206). Its arguments: people pay to be sterilised; the question of age and whether workers still wanted children; older workers could “accept such positions voluntarily”; some might volunteer as an alternative to vasectomy or tubal ligation, or “as a means of getting around religious bans on birth control”. Its close: “there can be good as well as bad sides to a situation” (p. 206). The authors respond only that “This argument found little favour with OSHA” (p. 206).
• Israel (Potashnik et al. 1978): six production workers exposed for 2–10 years were sterile, with raised FSH (“which increases with testicular damage”) and damage to sperm-producing tissue. In a related study (uncited), 18 of 23 (78%) had abnormal counts, 12 of them azoospermic. After several years without exposure “some of the workers’ testicular function improved”, but men exposed for >120 hours “experienced no improvement” (“Potashnik, 1984”). Recovery at 17 years was “mixed” (Potashnik and Porath 1995) (pp. 206–207).
• Citation ambiguity: the only 1984 Potashnik reference is on sex-ratio reduction (p. 213). The sex-ratio paper and Goldsmith et al. (1984) on family reproductive outcomes are in the references but not discussed. Lesson 6 (on effects in sons of exposed women) has no support in the body.

9.5 DBCP and cancer? (p. 207)#

• 1975: the NCI found DBCP carcinogenic in rats and mice. Industry criticised the high doses at the 1977 OSHA hearing. A 1977 lower-dose Dow/Hazleton study found carcinomas of the stomach, liver and renal tubules at the top dose, and a “statistically non-significant increase” at the two lower doses (p. 207).
• The authors’ general argument: with ~20 animals per group, high doses are needed to detect hazards that many workers face at lower levels. “High doses have been shown, in very many cases, to be reliable predictors of the hazards humans face at much lower doses” (p. 207). No citation is given. The claim is asserted and contested in toxicology.
• EPA has classified DBCP as a “probable human carcinogen” since 1992. IARC classified it 2B (“possible”) in 1999, on sufficient animal evidence (p. 207). No human cancer evidence is mentioned. The heading’s question mark signals uncertainty.

9.6 DBCP risks: from manufacturing to pesticide spraying (p. 207)#

• In September 1977 EPA Administrator Douglas Costle suspended distribution under FIFRA emergency authority, which applies where a pesticide “appears likely to pose an unreasonable risk to man during the period necessary to conduct and complete a more lengthy administrative proceeding” (p. 207).
• EPA acted “even though early studies of DBCP sprayers in California and Israel had produced mixed results (Glass et al., 1979; Karraazi [Kharrazi] et al., 1980)” (p. 207). Chronology problem: those papers were published after the September 1977 decision. The underlying data may have been available earlier, or the sentence may refer to the 1979 decision, but the chapter does not say.
• The prohibition became permanent in November 1979 for all crops except Hawaiian pineapples, which were covered in 1985 (p. 207). The chapter’s own reference for the 1979 action is titled “Dibromochloropropane: Suspension order and notice of intent to cancel” (US EPA 1979, 9 November 1979; p. 213), which suggests the November 1979 step was a suspension with cancellation to follow rather than a final prohibition. The sequence of EPA actions is probably simplified; see Open questions.

9.7 DBCP exports from 1969 to the 1980s: spreading sterility? (pp. 207–209)#

• Standard Fruit tested DBCP on Central American bananas in the mid-1960s, with full-scale use in Costa Rica and Honduras by 1969 (p. 207). Containers carried brand names but “no warnings to workers about the risk of sterility. Moreover, the labels on the pesticides exported were in English”, and many sprayers were illiterate (p. 207).
• The US use ban “did not ban Shell and Dow from manufacturing it”, and growers kept using it (pp. 207–209).
• Contract episode: when Dow told Standard Fruit it was halting shipments, Standard Fruit threatened a breach-of-contract claim. “To settle the matter”, it agreed to indemnify Dow for injuries and to implement “‘applicable work standards in respect of protective clothing, training etc.’” as outlined in the OSHA standard (p. 209). The chapter implies, but does not state, that shipments then resumed. It gives no date or source.
• Litigation records show Shell selling to the Ivory Coast in 1977–80 and Amvac selling to the Philippines, Honduras and Nicaragua in 1979. Use in Central America continued “until at least 1985” (NYT 2003), and in the Philippines (Standard Fruit subcontractors) until about 1986 (p. 209). “According to reports collected by lawyers”, Philippine workers became sterile and had not been told of the risk or given PPE. Protection was also inadequate in Costa Rica (Thrupp 1991) (p. 209).
• In Panama (August 1993), 25 of 28 banana workers were diagnosed with damaged sperm (Navaro 1993, via Misko et al.) (p. 209). No comparison group or selection method is described.

Box 9.2 The Swedish film documentary, the Dole lawsuit and freedom of speech (Maria Albin, p. 208)#

• Bananas! (2009, Fredrik Gertten) follows 12 Nicaraguan workers suing Dole (“previously named Standard Fruit”) and “the controversial personal injury Californian lawyer, Juan Dominguez”. The box does not explain why he was controversial (p. 208).
• Dole’s actions (p. 208):
• It sought to have the film withdrawn from the June 2009 Los Angeles film festival, which moved it to a “special case study screening”.
• It wrote to the Swedish ambassador asking for “appropriate steps to limit its damaging impact”.
• In July 2009 it sued the director and producer for defamation, seeking an injunction against screenings, the website and interviews.
• Responses (p. 208):
• A free-speech petition by filmmakers, a letter from the German Documentary Film Association CEO, and condemnation by the International Federation of Journalists.
• The filmmakers called the suit a SLAPP, “a deliberate attempt by a wealthy party to silence its critics by outspending them in launching a legal action”, and filed an anti-SLAPP motion and a cross-complaint.
• The film was screened in the Swedish parliament. Two MPs from opposing parties wrote to Dole, “you are misled by your PR-firm”. A cross-party petition followed, joined by food-retail CEOs.
• Outcome (p. 208):
• Dole withdrew in October 2009, citing free-speech concerns while maintaining “the merits of its case”.
• The threat of reinstatement “hampered the distribution of the film”.
• In 2010 a Los Angeles court found the suit was a SLAPP and awarded ~USD 200,000 in fees and costs. The sequel, Big Boys Gone Bananas!, premiered in October 2011.
• The box’s sources are mainly documents hosted on the film’s own website (the 2010 SLAPP ruling, Dole’s July 2009 defamation complaint and Dole’s letter to the ambassador, so partly primary legal documents), plus the IFJ and an Inter-American Institute of Human Rights page on Nicaragua (p. 208). Dole’s side is given one clause. See Limitations for context the box omits.

9.8 Banana workers bring compensation cases: 1990–2010 (pp. 209–210)#

• Early 1990s: a Texas class action by more than 16,000 Central American and Philippine workers (p. 209).
• 1992: a Costa Rica settlement of USD 20 million for 1,000 workers (p. 209), which works out at about USD 20,000 each (my arithmetic).
• 1997: USD 41 million for 26,000 workers, an average of USD 1,500 each (pp. 209–210).
• 2002: a Nicaraguan tribunal awarded USD 489 million to 450 workers. Enforcement is not discussed (p. 210).
• 2007: Amvac paid USD 300,000 to 13 Nicaraguan workers (p. 210).
• “These settlements came 20 years after each of these firms knew about the potential reproductive health risks to DBCP pesticide spray workers” (p. 210). The firms nonetheless “marketed and sold DBCP abroad without ensuring that worker health would be adequately protected”. “Tens of thousands of banana workers still have suits pending”, but “many of the relevant facts are still unclear and contested by the growers” (p. 210).
• No figure in this section carries an in-text citation. The reference list contains likely sources (Beasley Allen 2007 and Miller 2007 on the Amvac settlement; WSJ 1997 on a Filipino settlement fund), but they are not linked to the text.

9.9 Environmental pollution of soils and water by DBCP (p. 210)#

• DBCP is “persistent and mobile”. Torkelson et al. (1961) had already noted “a long residence in the soil” (p. 210). The persistence warning is in the same 1961 paper as the testicular data.
• DBCP contaminates Sacramento Valley aquifers and “can persist for over a century” depending on temperature and pH. It was the most frequently detected contaminant in California wells in the early 1990s (p. 210).
• 2010 California Well Inventory: 254 of 1,312 wells with detections, at 0.01–1.7 ppb, against an MCL of 0.2 ppb (uncited; the reference list has only the 1995 update, Bartkowiak et al.). The 1986–2009 trend sentence is garbled in the original PDF as well as the extract. It probably means the share of wells above the MCL fell from 49% to 25% (p. 210).
• EWG (labelled a “US-based interest group”) analysed 20 million tap-water tests (2004–09). It found 191 systems in 18 states above health guidelines, of which 48 were above the enforceable MCL. The WHO guideline is 1 ppb (p. 210). “Over 4 million Californians” have tap water exceeding “health limits” (p. 210). Ambiguous: this probably means non-enforceable goals, not the MCL. In 38 communities levels were above the “‘negligible’ risk” level for carcinogens, and in 31 they were 20–200 times it, with particular concern for infant formula (p. 210). Table 9.1 turns this into “a contaminant of drinking water in 38 cities” (p. 212).

9.10 Some late lessons and 9.11 Conclusion (pp. 210–211)#

The lessons are reproduced and classified in the section below. The conclusion (p. 211) links DBCP to “widespread concern about male infertility, and related reproductive problems, such as testicular cancer and developmental defects, in wildlife, workers and consumers” (citing WHO in press; EEA 2012; BCPT 2008). It also links it to endocrine-disrupting substances “which seem to be playing a role” in reproductive ill health (cross-referring to Chapters 10 and 13). It repeats the call for disciplinary integration, government action and “precautionary approaches to the evidence from science” (p. 211).

Table 9.1 Key early warnings about and recognition of DBCP toxicity (p. 212)#

The table has 15 dated entries from 1956 to “2007–today”. They are folded into the Case timeline below. Entries that appear only in the table: a 1961 worker medical examination that did not examine testicular function; 1961 publication of water-solubility data (the 1961 soil-persistence point is also in the body, p. 210, via Torkelson et al.); 1999 California drinking-water regulation (the body cites a 1999 Cal EPA public health goal document only as a source on persistence, p. 210). Discrepancies with the body: - manufacturing starts in 1956 rather than 1955; - the 1977 row folds the EPA actions of 1977–79 into one year (the body dates the permanent prohibition, with its Hawaii pineapple exception, to November 1979); - compensation cases are placed in “South America” rather than Central America and the Philippines; - EWG’s “38 communities” above negligible cancer risk becomes “38 cities” where DBCP “remains a contaminant”.

References (pp. 212–214): noteworthy features#

• Listed in the references but not cited in the text: Albin and Giwercman 2009; Beasley Allen Law Firm 2007 (a plaintiff firm’s press release); Goldsmith et al. 1984; Kloos 1996 (DBCP and EDB in Fresno/Clovis wells); Miller 2007 (LA Times); OSHA 1978 (the final rule, 43 FR 11514); Potashnik et al. 1979; Reed et al. 1987 (UC Davis risk assessment); WSJ 1997.
• Cited only in passing: Clark and Snedeker 2005, a cancer-risk evaluation, is cited once on p. 204, only in the string of sources for trade names and uses. Its cancer findings are not used in 9.5.
• Cited only via Misko et al. (1993), DBCP: The Legacy (Dallas): the Lykken 1958 document (a letter to M.R. Zavon at the Kettering Laboratory, per the reference, though the text calls it an internal Shell memorandum) and Navaro 1993. The provenance of the Misko volume is not described.
• Typos: “DPCP” (pp. 209, 211); “Karraazi”/”Kharrazi”; “Navaro”; “Fumazon”/”Fumazone”; “National Institute for Safety and Health” (p. 206); “confidentially asserted” (p. 211, probably “confidently”).

Case timeline#

Lags. - First internal industry warning (1958) to effective US worker protection (September 1977): ~19 years. From the first peer-reviewed warning (1961): ~16 years. - Registration with a 1 ppm limit set below the lowest dose tested (1961) to the first human data (1977): 16 years without surveillance to test the assumption. - Lay detection (July 1977) to the Emergency Temporary Standard (9 September 1977): ~2 months. To the final standard (March 1978): ~8 months. This is a notably fast response (comparison with other chapters not checked here). - US suspension (1977) to the last documented use on Central American and Philippine plantations (“until at least 1985”; “until about 1986”): at least ~8–9 years. First warning to the last documented use abroad: at least ~27–28 years. - Firms’ knowledge of sprayer risk (1977) to settlements (1992–2007): 15–30 years. The chapter says “20 years” (p. 210). - First persistence data (1961) to California drinking-water regulation (1999 per Table 9.1): ~38 years. Caveat: the 1999 source is a public health goal document, and the body mentions an existing US EPA and Californian MCL of 0.2 ppb without dating it, so the lag to first drinking-water regulation may be shorter. Contamination is still detectable in 2010, and potential persistence exceeds a century (p. 210).

What was known when (as the chapter presents it). - By 1958 (internal) and 1961 (published), firms and regulators knew DBCP caused testicular atrophy and reduced sperm in three species. They knew it was absorbed through the skin, that the lowest tested dose (5 ppm) produced testicular shrinkage in Hine’s data (p. 204) and a non-significant 18.6% testis-weight fall in Torkelson et al. (p. 205), and that no safe level had been demonstrated. They knew odour was an unreliable control (the published authors themselves said so) and that DBCP persisted in soil. - Nobody knew the human dose–response, because no one looked. The first human data came from the workers.

The authors’ own lessons and conclusions#

The chapter (pp. 210–211) gives three lists. I reproduce them closely and mark whether each is derived from evidence the chapter presents, asserted (stated without supporting discussion in the chapter), or recommendation or advocacy.

Lessons for science (pp. 210–211) 1. “DBCP exposures below the lowest dose tested in animal studies were mistakenly assumed to be safe.” Derived: pp. 205–206; the human effects at 0.04–0.43 ppm versus the 5 ppm lowest animal dose. 2. Effects on crude testicular morphology and sperm counts were documented, but no further studies sought the exposure levels at which “more subtle indicators of early stage infertility” would show. Derived by absence (p. 205). 3. The early studies predated modern protocols, but continued use in developing countries “did not lead to the use of updated protocols to assess the toxicity in further detail”. Asserted: the body says nothing about protocols. 4. “Evidence of harm in animals was not seen by many scientists as relevant to humans.” Partly derived. Industry disputed the high-dose cancer studies (p. 207), but the body gives no evidence about “many scientists’” views on the reproductive data. 5. “Human reproduction may be sensitive to subtle derangements of physiological processes, thereby causing sub-fertility or infertility in the absence of obvious pathology.” General biological claim; asserted. 6. “No attention was paid to the possible effects on sons of exposed women.” Asserted: female exposure is not discussed in the body at all. 7. “Routine medical records and health statistics can be of limited use in regard to adverse effects on reproduction.” Implied by the lay discovery; not developed. 8. Skin exposure can contribute significant doses, so “air monitoring alone … underestimated total doses received.” Derived: pp. 204, 206. 9. “Independent expert assessments e.g. by the US governmental body, NIOSH, were needed to identify harm and to better protect employees.” Derived: p. 206.

Communication and use of research evidence (p. 211) 1. “The original evidence for DBCP effects on human male sterility came from the lay and local knowledge of the workers and their wives.” Derived: p. 204. 2. “It was confidentially [sic] asserted that DBCP was safe to use without there being any studies of workers or relevant animal studies to confirm this assumption: an example of the ‘authoritative assertion but without evidence’ which appears in other chapters.” The phrase is bold in the original. Derived: p. 205. This is an explicit cross-report theme. 3. Toxicity information was “translated only into very general warnings on labels: no translations were provided for products exported.” Derived: pp. 205, 207. 4. Animal findings “did not lead to any surveillance studies of men exposed to DBCP at the production plants until after evidence was observed by the workers.” Derived: pp. 205, 212. 5. “No action was taken to avoid the earlier biological ‘effects’ in animals until they had become ‘adverse effects’ in people.” Derived and interpretive. 6. DBCP “was considered essential by growers, including multinational companies, and they considered the toxicity concerns were too small to be significant.” Partly derived: the Peach Council letter (p. 206) and the Standard Fruit contract threat (p. 209). The claim about growers’ beliefs is otherwise asserted. 7. “The early scientific warnings were not widely reported but confined to specialist scientific journals or internal company communications.” Derived (pp. 204–205). The 1961 Torkelson paper was published openly, but in a specialist toxicology journal (Toxicology and Applied Pharmacology), which fits the lesson’s wording. The point is that the warnings did not reach the exposed or the wider public; the lesson does not claim the findings were secret. 8. “Knowledge from the manufacturing risks did not get taken up by the companies responsible for user risks.” Derived: pp. 207–209. 9. “National standards to control the risks of DBCP were not transferred into international standards to protect workers from globalised exposures to hazardous chemicals.” Derived: pp. 207–209. 10. “Early warnings about the persistence of DBCP in soils and water did not get acted upon until many years later.” Derived: pp. 210, 212.

Compensation for victims (p. 211) 1. “Much information about the responses of DPCP [sic] producers and user companies only emerged via legal procedures in the compensation cases.” Derived: p. 209. 2. Compensation cases “can be difficult, expensive and very time consuming to pursue”. Cross-reference to Chapter 24. Derived: pp. 209–210.

Conclusion and recommendations (p. 211; mirrored on p. 203) - DBCP lessons are “very relevant” to current concerns about male infertility and endocrine-disrupting substances. This is advocacy or analogy: it extends the case to a different and more contested causal domain. - Protection “needs the closer integration of scientific disciplines, and government actions” using “precautionary approaches to the evidence from science”. This is a recommendation and is not specified further. The chapter makes no concrete institutional proposals: no specific surveillance requirements, export rules, or burden-of-proof changes.

Mechanisms and dynamics#

1. Who produced the knowledge, and what was never produced#

• The early hazard data were generated by the producers. (Later corroboration came from “European laboratories”, apparently Soviet, in 1971–75 and from the NCI in 1975; pp. 205, 207.) “The earliest research on DBCP toxicity was carried out by two chemical companies producing the compound” (p. 204). The firms knew first and knew a great deal: absorption routes, target organs, dose-dependence, persistence (pp. 204–205, 210). This was not a case of ignorance. The problem was how existing knowledge was turned, or not turned, into protection.
• What was not studied is as important as what was. No studies below 5 ppm, so no demonstrated no-effect level (p. 210, lesson 1). No studies of subtler early indicators (lesson 2). No medical surveillance of manufacturers or sprayers (p. 205). A 1961 plant medical examination did not examine testicular function (p. 212). No quantification of dermal dose (p. 206). No attention to female exposure or offspring (lesson 6). Ignorance was produced by omission. The absence of human data was then taken as consistent with safety.
• The endpoint was crude in animals but legible in humans. Animal studies measured testis weight and histology. The human endpoint that mattered, the ability to have children, was noticed by the workers themselves through a social channel (lunchtime conversation, the wives’ examinations) (p. 204). Once a cheap, direct test was applied (sperm counts), the signal was unmistakable: 9 of 13 azoospermic (p. 206).
• Lay knowledge was decisive. The chapter’s first communication lesson says the human evidence “came from the lay and local knowledge of the workers and their wives” (p. 211). My inference: that the wives were examined first (p. 204) may show a background assumption about where fertility problems lie. The chapter does not comment, and it gives no evidence that this slowed recognition.
• Knowledge was fragmented across languages and borders. Corroborating Soviet studies (1971, 1975) are labelled “European” (p. 205). Whether they reached Western decision-makers is unexamined. Knowledge from manufacturing did not pass to the companies controlling use (p. 211, lesson 8).

2. Translation failures between evidence and protection#

The chapter supports at least five distinct points where knowledge failed to become protection: 1. Evidence to exposure limit. The recommended 1 ppm limit was set 5 times below the lowest tested dose, with no data below that dose to support it (pp. 205, 210). 2. Limit to monitoring method. Odour (threshold 1.7 ppm) was accepted as the “warning”, although it was above the 1 ppm limit and the study authors warned against sensory control. The chapter itself says the odour threshold “was too high to ensure reliable protection” (p. 205). 3. Knowledge of dermal absorption (1958) to controls focused on inhalation. The label said “Do not breathe vapours” (p. 205), and even the 1977 NIOSH evaluation measured air (p. 206). 4. Hazard to label. No reproductive warning on US labels. Exported labels were in English only, for workers many of whom were illiterate (pp. 205, 207). 5. Manufacturing knowledge to user protection. In the US, the gap closed quickly once human evidence arrived: OSHA protections for makers in 1977–78, and for users an EPA suspension of distribution in September 1977 and a use prohibition in November 1979 (Hawaiian pineapples excepted until 1985), taken despite “mixed” sprayer evidence (pp. 206–207). Abroad, the accounts the chapter cites describe little or no protection for plantation workers (pp. 207–209). The chapter’s lesson that manufacturing knowledge “did not get taken up by the companies responsible for user risks” (p. 211) applies mainly to the pre-1977 period and to the export trade.

3. Burden and standard of proof; how uncertainty was framed#

• Statistical non-significance possibly treated as absence of effect (my inference). Shell’s argument that “at 5 ppm no adverse effects had been reported” (p. 205) may reflect the 18.6% testis-weight decrease at 5 ppm being “not statistically significant” in Torkelson et al. (Box 9.1, p. 205). The FDA saw adverse effects at 5 ppm “after repeated exposures” (p. 205), and Hine’s data as reported showed testicular shrinkage at 5 ppm (p. 204). The chapter does not connect Box 9.1 to the dispute or frame it as a statistical-inference problem, and it does not say which data the parties were looking at. A parallel appears in the 1977 Dow/Hazleton cancer study, where the increase at the two lower doses was “statistically non-significant” (p. 207).
• Burden on the regulator, not the proponent. The USDA “accepted these reassurances even though no studies had been performed to indicate that this approach was safe” (p. 205). The regulators voiced reservations (FDA: precautions “appeared inadequate”; USDA: “reservations regarding the warnings”) but did not require the firm to demonstrate safety. The chapter names this “authoritative assertion but without evidence” (p. 211).
• “Effects” versus “adverse effects”. Lesson 5 (p. 211) captures a threshold problem: animal changes were treated as “biological effects”, not yet “adverse”, until they appeared as harm in people.
• Discounting animal evidence. “Evidence of harm in animals was not seen by many scientists as relevant to humans” (p. 210). Industry attacked the NCI cancer bioassay for its high doses (p. 207). The chapter’s counter-argument (p. 207) is that small animal groups require high doses, and that high doses have “in very many cases” predicted low-dose human hazards. For DBCP’s reproductive effects specifically, workers were affected in plants where measured air concentrations were 0.04–0.43 ppm, roughly 12–125 times below the lowest animal dose of 5 ppm (pp. 205–206). The chapter itself says only that these data “suggested” exposures below 1 ppm were associated with harm (p. 206); the range is plant-wide air sampling, not the dose at which any individual was affected. That is still strong case-internal support for not treating the lowest tested dose as safe. It is caveated by dermal co-exposure and by possible higher historical exposure.
• Action under uncertainty once institutions existed. EPA’s 1977 suspension used a legal standard that allowed interim action where a pesticide “appears likely to pose an unreasonable risk to man during the period necessary to conduct and complete a more lengthy administrative proceeding” (p. 207). OSHA used an emergency temporary standard (p. 206). These interim instruments let regulators act before full proceedings, and in EPA’s case despite “mixed results” on sprayers (p. 207).

4. Mental models, confidence and blind spots of key actors#

• Company scientists and consultant (Hine, Torkelson). They understood the hazard and made protective recommendations: <1 ppm, impermeable clothing, analytical rather than sensory control (p. 205). Hine’s remark about a “male birth control drug” (p. 204) suggests the effect was familiar, even joked about. The problem was not the scientists’ blindness. Their recommendations could be set aside by the company’s representative in dealings with the regulator (p. 205). Note too the dual academic and consultant role (UCSF and Shell/Dow). Here the consultant was more protective than the client, so conflict of interest did not run the obvious way.
• Company representatives in dealings with regulators (Shell). “The Shell representative” (the chapter does not give the person’s role) called the consultant’s recommendations “impractical” in discussions with the FDA (p. 205). Shell offered the odour threshold as reassurance to the USDA (p. 205), and after 1977 kept exporting (pp. 207–209). My inference of the implicit mental model: protection is a cost to be minimised, and absence of reported harm means absence of harm.
• Regulators before 1970 (FDA, USDA). Reservations were noted but deferred. There were no demands for surveillance or proof of safety (p. 205). The chapter does not explain why they deferred (mandate, capacity, statutory tools). That is an analytical gap.
• Downstream users (growers). DBCP was “considered essential by growers, including multinational companies, and they considered the toxicity concerns were too small to be significant” (p. 211). The Peach Council letter (p. 206) is the most revealing artefact in the chapter. It reframes involuntary sterility as possibly beneficial (“not necessarily so” bad) and discounts harm by workers’ age. It proposes voluntary assumption of risk, even selecting workers who want sterilisation, and ends with “good as well as bad sides”. Dependence on a product appears to shape how its harms are perceived and weighed. Caveat: this is one letter from one trade association. The chapter presents it without evidence that it was typical.
• Standard Fruit. Rather than accept Dow’s decision to stop supplying, it threatened legal action and, “to settle the matter”, accepted liability for injuries and promised OSHA-type work standards (p. 209). The chapter implies that supply then continued. My inference of its model: the hazard is a legal and contractual risk to be allocated, not a harm to be eliminated.
• OSHA and EPA in 1977. The chapter presents a regulator ready to act quickly once human evidence arrived. Bingham, the chapter’s co-author, was in post (pp. 204, 206).

5. Institutional behaviour, capacity and contingency#

• New institutions changed the outcome. NIOSH was created in 1970 (p. 206). The chapter’s lesson 9 credits independent public expertise (NIOSH) with identifying the harm (p. 211). In 1958–61 neither NIOSH nor OSHA existed (background: both date from the 1970 OSH Act; a smaller Public Health Service occupational health unit did exist earlier, without OSHA-type standard-setting powers). The chapter does not draw out how much the 1977 speed depended on institutional infrastructure that did not exist in 1961.
• Contingent actors. The speed of July to September 1977 coincided with an occupational-health-oriented union leader (Mazzocchi) contacting a sympathetic OSHA head (Bingham). The chapter does not reflect on this contingency, or on whether a less receptive agency would have acted as fast.
• Absence of dissent to the final rule. The rules “were not effectively challenged by any interested party” (p. 206). Once the human evidence was in, the case for regulation appears to have been politically overwhelming. The only pushback the chapter records is the Peach Council letter and industry criticism of the NCI high-dose cancer study at the 1977 OSHA hearing (p. 207).

6. Globalisation, jurisdiction and the export of hazard#

• Regulatory scope gap. The US use ban “did not ban Shell and Dow from manufacturing it” (pp. 207–209). Sales continued to the Ivory Coast (1977–80) and to the Philippines, Honduras and Nicaragua (1979). Use is documented until at least 1985 in Central America and about 1986 in the Philippines (p. 209).
• Weakest protection where the hazard went. English labels in Spanish-speaking countries and illiteracy (p. 207); subcontracted labour and, “according to reports collected by lawyers”, no risk information or PPE in the Philippines; “inadequate protection” in Costa Rica (Thrupp 1991) (p. 209).
• Risk transfer through contract. The Dow and Standard Fruit indemnity (p. 209) moved legal liability without removing the hazard. My inference: the promise to implement OSHA-type standards appears, from the Philippine and Costa Rican accounts, not to have been realised on the ground (p. 209). The chapter does not make this link explicitly.
• No international carry-over. National standards did not become international standards (p. 211, lesson 9).
• Distribution. My inference (the chapter does not discuss benefits): benefits went to fruit multinationals and chemical makers, and perhaps to consumers. Harms fell on US production workers (whom the system eventually protected once the harm was detected in 1977), on plantation workers in poorer countries from 1969 into the mid-1980s (whom it did not), and later on drinking-water users in agricultural California (pp. 206–210). The chapter does not quantify benefits at all, so it offers no real cost–benefit accounting.

7. Law: as discovery tool, as compensation channel, and as a weapon#

• Litigation as the route to disclosure. Export sales records came to light through litigation (p. 209). “Much information … only emerged via legal procedures” (p. 211).
• Compensation is slow, uneven and contested. Settlements came about 20 years after the firms knew (p. 210). Amounts varied widely: the 1997 settlement averaged USD 1,500 per worker (pp. 209–210), while the 1992 Costa Rica settlement works out at about USD 20,000 each (p. 209; my arithmetic). The Nicaraguan judgment of USD 489 million for 450 workers (p. 210) is reported without any note on enforcement. The chapter itself does not characterise the sums as small. “Many of the relevant facts are still unclear and contested” (p. 210).
• Litigation to suppress. Box 9.2 describes Dole’s defamation suit, the pressure on a film festival, and the letter to a foreign ambassador as efforts to limit dissemination (p. 208). Even after withdrawal, the threat of reinstatement “hampered the distribution of the film” (p. 208). A court later called the suit a SLAPP (p. 208).
• Countervailing public pressure. Swedish MPs across parties, retail CEOs, journalists’ federations and documentary associations were “sufficient to make Dole withdraw its legal action” (p. 208). My observation: reputational pressure in a consumer country secured the withdrawal of a lawsuit against a film. The chapter does not suggest that it did anything for worker protection or compensation.

8. Persistence, irreversibility and time#

• Partial biological irreversibility. Some Israeli workers’ testicular function improved after several years without exposure, but men exposed for more than 120 hours showed no improvement. Recovery 17 years later was “mixed” (p. 207).
• Environmental persistence. DBCP can persist “for over a century” depending on temperature and pH (p. 210). It was still in 254 of 1,312 sampled California wells in 2010 (p. 210). A decision made in the 1950s–70s created a liability lasting generations. The persistence warning was in the 1961 literature (pp. 210, 212).

9. Lock-in, dependence and alternatives#

• The chapter shows dependence (growers’ view that DBCP was “essential”, p. 211; contract enforcement, p. 209). It says nothing about substitutes: what growers used after the ban, whether substitutes were safer, or what the ban cost in yields. The references include Kloos (1996) on DBCP and ethylene dibromide in Fresno wells (p. 213), which hints at a related fumigant problem, but the text does not discuss it. The chapter’s “innovation” dimension is essentially absent.

10. Framing and language#

• The chapter’s scare quotes do analytical work: “hazard control” (p. 204 heading), “safe” levels (p. 206), the odour “warning” (p. 205), “effects” versus “adverse effects” (p. 211), “authoritative assertion but without evidence” (p. 211).
• Industry framing words: “impractical” (p. 205), “reassurances” (the chapter’s word for Shell’s arguments, p. 205), “not necessarily so” bad and “good as well as bad sides” (Peach Council, p. 206).
• The chapter’s own rhetorical choices: the epigraph (Hine’s joke); the heading “spreading sterility?”; the summary’s framing of Dole’s litigation as “industry attempts to prevent a Swedish documentary … from being screened” (p. 203); the full-page photo of a banana worker (p. 209). Hedges are relatively few: “allegedly” (p. 203), the question marks in two headings (9.5 “DBCP and cancer?” and 9.7 “spreading sterility?”, both p. 207), “suggested” and “may have contributed” on the Magnolia data (p. 206), “many of the relevant facts are still unclear and contested” (p. 210), and “seem to be playing a role” for EDCs (pp. 203, 211).

11. Complexity#

Modest. Multiple exposure routes (inhalation and dermal), species differences in sensitivity, and hydrogeological persistence and mobility make the system harder to predict than a simple dose–response. The chapter does not frame this as systemic complexity. DBCP’s causal story is comparatively simple (one agent, a specific target organ, a large effect).

Transferable insights (technology-neutral)#

1. The earliest and best hazard knowledge often sits with those who make the product. Whether it becomes protection depends on how the firm acts on it when dealing with regulators. The consultant’s registration report recommended <1 ppm and impermeable clothing where skin contact was likely. In discussions with the FDA, “the Shell representative” called these recommendations “impractical” (pp. 204–205). (The advice to rely on instruments rather than smell came in the 1961 published paper, p. 205.) Strength: strong for “producers knew first” (a published 1961 paper and a 1958 Shell document). Suggestive for the override dynamic, which rests on one uncited paragraph (p. 205) and does not describe internal company decision-making.

2. A limit set below the lowest level ever tested is an assumption, not a finding. Without follow-up, it can stand unchallenged for decades. DBCP’s recommended 1 ppm limit (1961) had no data beneath 5 ppm. Human harm then appeared in plants with measured air levels of 0.04–0.43 ppm (pp. 205–206, 210). Strength: strong. The numbers are in the record and not disputed. The magnitude is caveated by dermal dose, by possibly higher past exposures, and by the air levels being plant-wide ranges rather than individual doses.

3. When groups are small, “no statistically significant effect” at the lowest dose can be read as “no effect”. Shell told the USDA that “at 5 ppm no adverse effects had been reported” (p. 205); the published data showed a non-significant 18.6% testis-weight fall at that level (Box 9.1, p. 205), and Hine’s data as reported showed shrinkage (p. 204). Strength: suggestive. The data and the competing readings are in the chapter, but the link between them is my inference: the authors neither connect Box 9.1 to Shell’s claim nor analyse it as a statistical-inference problem.

4. Control systems built around what is easy to measure, not how exposure actually happens, underestimate harm. A “warning signal” can be set above the threshold it is meant to protect. Air-only monitoring missed skin absorption, which had been known since 1958. The odour threshold of 1.7 ppm was offered as a warning for a 1 ppm limit (pp. 204–206, 211). Strength: strong. This is internal logic plus documented facts.

5. Without surveillance of exposed people, harm is found by the harmed, late and by accident. No surveillance was done from 1961 to 1977, and the one 1961 examination ignored the relevant organ. Detection came from workers’ lunchtime conversation and union-arranged tests (pp. 204–205, 211–212). Strength: strong.

6. Legible, cheap, direct endpoints, together with organised affected parties and independent public expertise, can turn suspicion into action very quickly. It took about two months from the union’s alarm to an emergency standard (pp. 204, 206). Strength: strong for this sequence of events. Suggestive as a general rule, because the speed also depended on contingent factors: a sympathetic regulator in post (a chapter author) and a very large effect size.

7. Formally “available” knowledge is not the same as knowledge reaching the people at risk. Warnings stayed in specialist journals and internal memos and became generic labels, in a foreign language for exported product (pp. 205, 207, 211). Strength: moderate to strong. The label facts are documented, and the 1961 study appeared in a specialist toxicology journal, consistent with the authors’ lesson. What the chapter does not show directly is how far the warnings failed to reach workers (beyond the labels and the workers’ own surprise in 1977).

8. Regulatory boundaries let hazards persist elsewhere rather than disappear. Domestic prohibition of use did not stop manufacture and sale for use abroad. Hazardous practice continued in jurisdictions with weaker oversight, lower literacy and less legal recourse for at least 8–9 years after US action (pp. 207–209, 211). (Use there had begun in 1969, before the US discovery; the US action did not move it there but failed to stop it.) Strength: strong. Sales and use are documented through litigation records and press reports.

9. Contracts and indemnities can move responsibility without reducing the hazard, and so keep a hazardous practice going. Standard Fruit indemnified Dow to settle a threatened breach-of-contract claim after Dow moved to halt shipments; continued supply is implied rather than stated (p. 209). Strength: moderate to suggestive. One episode, undated and without a specific source.

10. Actors who depend on a product tend to shrink, reframe or even recast its harms, including by discounting whose harm counts. The Peach Council letter treats sterility as possibly beneficial and discounts it by age (p. 206). Growers saw DBCP as “essential” and the concerns as “too small” (p. 211). Strength: suggestive. The letter is vivid but a single artefact; the claim about growers is largely asserted.

11. Confident safety claims made without studies, accepted by regulators who had voiced doubts, are a recurring failure pattern. “Authoritative assertion but without evidence” (p. 211); the USDA “accepted these reassurances even though no studies had been performed” (p. 205). Strength: strong for this case. The chapter explicitly links it to other chapters.

12. Waiting for “adverse effects” in people rather than acting on “effects” in test systems builds in a harm-first trigger. Lesson 5 (p. 211) and the 1961–1977 gap. Strength: strong as a description of what happened. The prescription (act on biological effects) is a normative choice whose costs the chapter does not examine.

13. Interim or emergency powers let regulators act while longer proceedings run, including under mixed evidence. Examples are EPA’s FIFRA suspension authority and OSHA’s emergency temporary standard (pp. 206–207). Strength: moderate. Well documented, but the chapter’s chronology on the “mixed” sprayer evidence is compressed, and it does not evaluate these powers.

14. After-the-fact accountability is slow, uneven and itself contested. Litigation both surfaces hidden information and can be used to suppress it. Settlements came about 20 years later. One 1997 settlement averaged USD 1,500 per worker, and the 1992 Costa Rica settlement about USD 20,000 each. Disclosure came through litigation. A SLAPP was used against a documentary (pp. 208–211). Strength: moderate. The figures are documented. The chapter omits contested aspects of the litigation (see Limitations), so it gives only a partial picture of how the litigation played out.

15. Persistence turns a time-limited decision into a long-lasting liability. Early persistence warnings were available at the start and not acted on. The contamination remains decades after use ended (pp. 210, 212). Strength: strong for continued presence. Moderate for health significance at current levels: the exceedances are mainly against health guidelines, and EWG is an advocacy source.

16. Test methods can harden into a standard of proof that decides what counts as evidence. Disputes over high-dose animal testing (p. 207) and the discounting of animal data (p. 210) show the terms of evidence being contested in the proponent’s favour. The authors’ counter-claim, that high doses predict low-dose hazards “in very many cases”, is asserted without citation. Strength: moderate for the dispute pattern. Asserted for the authors’ general methodological claim (though DBCP’s own reproductive data support it in this instance).

Limitations, contestation and bias check#

Where the chapter is strong. - The core occupational story is well documented from primary sources: Whorton et al. 1977 in the Lancet; the NIOSH HHE; the OSHA Federal Register notices; Potashnik et al.; Torkelson et al. 1961. The human effect is large (9/13 azoospermic; 50% of 106; 78% of 23), replicated across three plants in two countries, and shows dose–response. Causation for production workers is not seriously in doubt. Hindsight bias is limited on the central claim, because the industry’s own consultant and the 1961 authors recommended impermeable clothing and instrumental rather than odour-based control, which were not adopted. (The sub-1 ppm limit they recommended was adopted as a recommendation but proved inadequate.)

Where it is advocacy or thin. - Insider authorship. One author ran the agency whose response is praised (pp. 204, 206). The chapter discloses her role but does not reflect on it, and it examines OSHA’s post-1977 conduct less critically than the pre-1970 regulators. - Source base partly adversarial. Key historical documents (the Lykken 1958 document, the Panamanian medical data) are cited via Misko et al. (1993), whose provenance is not stated. Others come from “reports collected by lawyers” (p. 209), a law-firm press release (Beasley Allen 2007, in the references), and an advocacy group’s analysis (EWG, p. 210). The chapter’s own lesson that information emerged via litigation (p. 211) explains why this is so. But it means the account of internal decision-making and of harm abroad rests partly on plaintiff-side material. - Scale of harm abroad. “Tens of thousands” is hedged with “allegedly” once (p. 203). Later the harm is treated more firmly: the Amvac settlement went to “the now-sterile workers” (p. 210), and the firms are said to have sold DBCP abroad “without ensuring that worker health would be adequately protected” (p. 210). (The heading “spreading sterility?” (p. 207), by contrast, keeps a question mark.) The best field evidence cited is 25 of 28 Panamanian workers with sperm damage (p. 209), with no comparison group or sampling method described. The early sprayer studies in California and Israel were “mixed” (p. 207). User exposures (field application) differ from production-plant exposures. The chapter does not address this gap in evidence between manufacturing and field use. - Omission on litigation integrity (background knowledge; verify). Box 9.2 calls the plaintiffs’ lawyer Juan Dominguez “controversial” (p. 208) without explaining why. From background knowledge, not verified in this pass: in 2009 a Los Angeles Superior Court judge dismissed two DBCP suits against Dole (Mejia and Rivera) after finding that claims had been fabricated and witnesses coached through recruiting in Nicaragua. Dole’s defamation suit against the filmmakers related to how Bananas! presented the earlier Tellez verdict in light of those findings. US federal courts also declined to enforce Nicaraguan DBCP judgments under Nicaragua’s Special Law 364 on due-process grounds (e.g. Osorio v. Dole, S.D. Fla. 2009). If confirmed, these facts do not affect the documented harm to US and Israeli production workers or the SLAPP ruling. They do bear on how far the compensation narrative and the USD 489 million figure (p. 210) can be taken at face value, and on how fairly Box 9.2 represents Dole’s position. The box’s framing is one-sided even on its own terms: Dole’s reasons are reduced to a clause. - Normative counterfactuals. “The toxicological data available in 1961 … was sufficient to have required specific health warnings, personal protective equipment and medical surveillance” (p. 205) is plausible, since the industry’s own consultant recommended part of it (protective clothing). The chapter reports no contemporary recommendation of health warnings or medical surveillance, and the verdict is judged against later norms without discussing what was standard practice in 1961 or what the statutes then required. - Generalised methodological claim. High doses have been “reliable predictors … in very many cases” (p. 207) is asserted without citation. It is a live dispute in toxicology (background, not from the chapter): critics argue high-dose effects can come from mechanisms, such as cytotoxicity and regenerative proliferation, that don’t operate at low doses. The chapter acknowledges the industry view only to dismiss it. - Extension to endocrine disruptors. The chapter does not claim DBCP is itself an endocrine disrupter. It makes two claims of different strength. The summary calls DBCP “one of many examples supporting the growing concerns about increasing rates of reproductive and developmental disease, and about the endocrine disrupting chemicals that seem to be playing a role” (p. 203), which is an evidential claim. The conclusion says its lessons are “very relevant” to those concerns (p. 211), which is closer to an analogy. My assessment (background): DBCP is primarily a direct testicular toxicant. The elevated FSH the chapter reports is a consequence of testicular damage (p. 207), not evidence of hormone-mimicking disruption. A single high-exposure occupational cluster says little about population-level trends. The link is strongest as an analogy about how warnings get handled, weaker as evidence about endocrine disruption. The chapter’s hedge (“seem to be playing a role”) acknowledges this only lightly. - “First clear example” (p. 203). This is a strong priority claim with no support given. Background, to verify: sperm abnormalities were reported among chlordecone (Kepone) production workers in Hopewell, Virginia in 1975–76, before the DBCP discovery. Lead’s effects on male reproduction were described much earlier. The claim may hold only in a narrow sense (e.g. first clearly recognised occupational sterility cluster from a synthetic pesticide). - No benefits side. The chapter does not discuss DBCP’s agronomic value, the economic stakes for producing countries, the substitutes adopted, or whether any substitute caused its own problems. There is no cost-benefit analysis, so it cannot show that earlier action would have been net-beneficial. The case for that is intuitively strong given the severity of harm, but the chapter does not argue it. - Chronology and fact inconsistencies. Internal to the chapter: registration in 1961 vs 1964; first production in 1955 vs 1956; the section heading “1961–1975” vs 1958 data; the final OSHA standard on 11 vs 17 March 1978; the 1979–80 sprayer studies cited as informing a 1977 decision; the November 1979 EPA action called a permanent prohibition in the text but a “Suspension order and notice of intent to cancel” in the references; the Lykken document called an internal Shell memorandum in the text but a letter to an outside laboratory in the references; “no medical surveillance” (p. 205) vs a 1961 plant medical examination (Table 9.1); Table 9.1’s “South America”. Against background knowledge (to verify): FDA vs USDA as registering authority. None changes the core story. Together they suggest the chapter was not tightly edited and should be cross-checked against primary sources before any specific date is quoted. - Unused evidence. The sex-ratio and family-outcome studies (Potashnik et al. 1984; Goldsmith et al. 1984) are in the references but not discussed. The Clark and Snedeker (2005) cancer evaluation is cited only for trade names and uses (p. 204), not in the cancer section. Their findings might strengthen or complicate the account. - Case selection and hindsight. DBCP is a “clean” case: one agent, a specific endpoint, a huge effect, cheap detection. That makes it a persuasive illustration of precautionary failure. It may overstate how easy it is to act on early warnings in cases where the signal is weaker, the endpoint is diffuse, or the evidence is genuinely mixed. The pro-precaution framing is justified here, but the case cannot on its own tell us how precaution performs when warnings turn out to be false positives.

Fairness in the other direction. The chapter could have been harsher. It reports rather than editorialises the Peach Council letter (“This argument found little favour with OSHA”, p. 206). It concedes dermal uncertainty in the human dose (p. 206), reports “mixed” sprayer evidence (p. 207), notes the non-significance at 5 ppm (p. 205), and records that the growers contest the facts (p. 210). It also credits the company scientists with recommending protective limits (p. 205), which complicates a simple story of industry villainy.

Dissent within panels. Not applicable: there are no panels. Box 9.2 is aligned with the chapter authors.

Notable quotes#

1. “If anyone wants to use a male birth control drug, I think we have identified one, but it is not very pleasant to use.” Charles Hine, c. 1960, via Goldsmith 1997 (p. 204)
2. “We understand that Dow Chemical Company have similar data and are very upset by the effects noted on the testes.” Shell document described in the text as an internal memorandum, Lykken 1958 (p. 204)
3. “However, the Shell representative considered the Hine recommendations to be impractical.” (p. 205)
4. “USDA accepted these reassurances even though no studies had been performed to indicate that this approach was safe.” (p. 205)
5. “The toxicological data available in 1961 … was sufficient to have required specific health warnings, personal protective equipment and medical surveillance. None was provided.” (p. 205)
6. “While involuntary sterility caused by a manufactured chemical may be bad, it is not necessarily so.” US National Peach Council, 1977 (p. 206)
7. “Moreover, the labels on the pesticides exported were in English.” (p. 207)
8. “These settlements came 20 years after each of these firms knew about the potential reproductive health risks to DBCP pesticide spray workers” (p. 210)
9. “No action was taken to avoid the earlier biological ‘effects’ in animals until they had become ‘adverse effects’ in people.” (p. 211)
10. “The original evidence for DBCP effects on human male sterility came from the lay and local knowledge of the workers and their wives.” (p. 211)

Open questions#

1. Registration history. Was DBCP registered in 1961 or 1964, and by which agency (FDA tolerance vs USDA FIFRA registration)? Primary source: USDA/EPA registration records; Thrupp (1991).
2. EPA sequence 1977–1985. What exactly did EPA do in September 1977 (a suspension of distribution? conditional? crop-specific?) and in November 1979? Primary source: the Federal Register notices, including the one the chapter cites as “Federal Register 65135, 9 November 1979” (volume 44, by year), and the September 1977 notice, which the chapter does not cite.
3. The internal decision record. Can the Lykken 1958 document (an internal memo per the text, a letter to M.R. Zavon at the Kettering Laboratory per the references) and the FDA–Shell–Dow discussions (uncited in the chapter) be read in their original form, not via Misko et al. (1993)? What is Misko et al. (1993)? It appears to be a litigation compilation, but this is unverified.
4. Historical exposures. Were the 1977 Lathrop and Magnolia air levels representative of past exposures? This affects the “below 1 ppm” inference (p. 206).
5. Scale of harm among plantation workers. Is there any controlled epidemiology of field applicators, beyond litigation-derived medical evaluations? How did the “mixed” US and Israeli sprayer studies turn out on later follow-up?
6. Litigation outcomes after 2013. What happened to the pending suits of “tens of thousands of banana workers” (p. 210)? Were the Nicaraguan judgments ever enforced? How did the 2009 fraud findings (if confirmed) affect later cases? What about later cases in Delaware and elsewhere, and in the Philippines and Côte d’Ivoire?
7. International rules. Did DBCP ever enter international prior-informed-consent or phase-out regimes (e.g. the Rotterdam Convention’s Annex III)? If not, why not? This bears directly on lesson 9 (p. 211).
8. Drinking water after 2010. What do later California and US monitoring data show about DBCP levels, treatment costs, and cost recovery from manufacturers (e.g. utility suits)?
9. Carcinogenicity. Has any human cancer evidence emerged since IARC 1999? Has the classification changed?
10. “First clear example”. Does the chlordecone (Kepone) episode of 1975–76, or any other, predate DBCP as a recognised case of chemically induced male reproductive harm in workers?
11. Long-term reproductive outcomes. What do the sex-ratio and family-outcome papers show? What did longer follow-ups of the Israeli and US cohorts find?
12. Substitutes. What nematicides replaced DBCP, and did any cause comparable problems (e.g. ethylene dibromide, which appears in Kloos 1996)? What did the ban cost growers, if anything?
13. Why the 1961 regulators deferred. What statutory tools and capacities did the FDA and USDA have in 1958–61, and what would a demand for surveillance have required legally?
14. The EDC link. Does DBCP’s mechanism (direct testicular toxicity) justify its use as an exemplar for endocrine-disruptor policy, or only as an exemplar of how warnings are handled?

Audit log#

Independent audit against the text extract (all pages) and PDF pp. 211–214 (report pp. 209–212), re-rendered to check the photo, the bold phrase in communication lesson 2 (confirmed bold), Table 9.1 and the garbled well-trend sentence (garbled in the original too).

1. Header: the p. 209 photograph fills the lower half of the page; it is not full-page.
2. Authors: marked “head of OSHA” as background, and “ran the rulemaking” as inference from her post; noted that the chapter discloses Bingham’s role but does not reflect on it.
3. Discipline: NIOSH 1977 is a primary source, so the claim that the chapter relies on “secondary sources” was corrected.
4. Discipline: added that the reference list describes Lykken 1958 as a letter to M.R. Zavon at the Kettering Laboratory, not an internal Shell memo; updated the 9.2 bullet, the timeline, the quotes list, Limitations and Open questions to match.
5. Discipline: added the list of key passages that carry no in-text citation (FDA–Shell–Dow talks, the cancer bioassays, the Dow–Standard Fruit contract, all 9.8 settlement figures, the 2010 well data).
6. Summary box: softened the 1961 vs 1964 “inconsistency” to “apparent”, since 1964 may refer to a separate fumigant approval.
7. Summary box: corrected “only place the chapter hedges” the scale of harm abroad; added the “reports collected by lawyers” and “unclear and contested” hedges.
8. 9.1: the gendered-sequence point was presented as fact (“which delayed the men’s testing”); relabelled it as my inference, with no evidence of delay (and likewise in Mechanisms 1).
9. Box 9.1: the “5 ppm non-significance is the crux” claim was the note-writer’s link, not the chapter’s; softened it and added that Hine’s own data as reported showed shrinkage at 5 ppm (p. 204), contradicting Shell’s “no adverse effects” claim.
10. 9.3: noted that the FDA–Shell–Dow paragraph is uncited and describes discussions with the regulator, not an internal decision.
11. 9.3: noted that the 1 ppm figure was “recommended”, not stated to be enforceable (also in the timeline, the translation failures and the digest).
12. 9.3: credited the chapter’s own statement that the odour threshold was “too high to ensure reliable protection”; added the omitted “Virtually no attempts were made” sentence; flagged the tension between “no medical surveillance” and Table 9.1’s 1961 examination.
13. 9.3 and Limitations: corrected “the consultant recommended most of it” (warnings, PPE, surveillance). The chapter reports only protective clothing and the <1 ppm limit.
14. 9.4: added that OCAW’s petition sought testing for sterility and cancer; replaced “the chapter concludes” with the chapter’s own “suggested” on the Magnolia data.
15. 9.4 and Mechanisms 8: added the omitted finding that some Israeli workers’ testicular function improved, and changed “irreversibility” to “partial irreversibility”.
16. 9.6: softened the sprayer-study “chronology problem” (the data may have predated publication); added that the chapter’s own 1979 EPA reference is titled “Suspension order and notice of intent to cancel”, which conflicts with “permanent”.
17. 9.7: the Standard Fruit indemnity deal settled a threatened breach-of-contract claim; continued supply is implied, not stated (fixed here, in Mechanisms 4 and 6, in insight 9 and in the digest, where “enforced its supply contract” was an overstatement).
18. 9.7: “There were no controls” became “no comparison group or selection method is described” (also in Limitations and the digest).
19. Box 9.2: completed the SLAPP quote and added the cross-complaint; noted that the box’s web sources are partly primary legal documents hosted on the film’s site.
20. 9.8: added the ~USD 20,000 per-worker arithmetic for the 1992 settlement, the omitted “without ensuring that worker health would be adequately protected” sentence, and the absence of in-text citations; corrected “tens of thousands of suits” to “tens of thousands of banana workers” (also in the timeline and Open questions).
21. 9.9: noted that the 2010 well-inventory figures are uncited and that the trend sentence is garbled in the PDF itself.
22. 9.11: added the conclusion’s “in wildlife, workers and consumers” and its citations.
23. Table 9.1: corrected “entries only in the table”, since 1961 soil persistence is also in the body (p. 210).
24. References: Clark and Snedeker 2005 is cited in the text (p. 204, for trade names), so I moved it out of the “not cited” list; added OSHA 1978 as listed but uncited; fixed Unused evidence to match.
25. Timeline and Lags: plantation use is documented “until at least 1985” and “about 1986”, so “use abroad ends” became “at least”; lags became “at least ~8–9 / ~27–28 years” (also in insight 8 and the digest).
26. Lags: removed the unchecked cross-report claim “one of the fastest responses in either report”; added a caveat that the 1999 date is a public health goal document, so the lag to first drinking-water regulation may be shorter.
27. Communication lesson 7 and insight 7: removed the incorrect objection that “specialist journals” overstates the case (Torkelson 1961 was in a specialist toxicology journal). Lesson 7 is now rated “derived” and insight 7 “moderate to strong”.
28. Mechanisms 2(5): “weaker EPA action for US users” was unsupported (EPA suspended distribution in 1977 and prohibited use in 1979); I rewrote it so that the manufacturing-to-user gap applies mainly before 1977 and to exports.
29. Mechanisms 3: marked the reading of non-significance as absence of effect as my inference, and added the Dow/Hazleton non-significance parallel; qualified “affected at 0.04–0.43 ppm” as plant-wide air ranges, not individual doses (also in insight 2 and the digest).
30. Mechanisms 4 and insight 1: removed the unsupported “commercial representative” and “internal override”. The chapter says only “the Shell representative”, in discussions with the FDA. Moved “instrumental monitoring” to the 1961 paper rather than Hine’s registration report, and downgraded insight 1’s override dynamic to “suggestive” (it rests on one uncited paragraph).
31. Mechanisms 5: corrected “no federal occupational health agency” before 1970 to “neither NIOSH nor OSHA existed” (background); added the industry criticism at the 1977 OSHA hearing as recorded pushback besides the Peach Council letter.
32. Mechanisms 6: corrected the harm sequence (plantation use began in 1969, before the 1977 US discovery); marked the benefit distribution and the unrealised-promise link as my inference; attributed “no PPE / no information” to lawyers’ reports.
33. Mechanisms 7: “compensation small” became “uneven”; the chapter does not characterise the sums, which range from ~USD 1,500 to ~USD 20,000 each. Qualified the Swedish-pressure point: it got a lawsuit withdrawn, not worker protection (also in insight 14 and the digest).
34. Mechanisms 10: widened the list of hedges (“suggested”, “may have contributed”, “unclear and contested”).
35. Insight 3: downgraded from moderate to suggestive, as the note-writer’s own inference (also in the digest).
36. Insight 8: reworded “move” to “persist elsewhere”, since export use predated the US action; the US ban failed to stop that use rather than displacing it.
37. Limitations: the “treated more firmly” example cited a question-marked heading; replaced it with “now-sterile workers” and “without ensuring … adequately protected” (p. 210). Corrected “US sprayer studies” to “California and Israel”.
38. Limitations, EDC extension: separated the summary’s evidential claim (“supporting”) from the conclusion’s analogy (“very relevant”); noted that the chapter never calls DBCP an endocrine disrupter; marked the mechanistic critique as my background assessment (also in the digest).
39. Limitations, chronology list: split internal inconsistencies from background ones (FDA vs USDA) and added the three new internal inconsistencies (EPA 1979 title, Lykken description, surveillance vs Table 9.1).
40. Limitations: marked the high-dose critique as background; narrowed “hindsight bias limited” to the protections actually recommended and not adopted (protective clothing, instrumental control).
41. Digest: changes as above; also added three omitted author lessons (surveillance, specialist journals, persistence), and marked the Kepone point as unverified background.