LL2-22 — Ch22 Nanotechnology — early lessons from early warnings#

Report: Late lessons from early warnings: science, precaution, innovation (EEA Report No 1/2013). This is the last chapter of Part C, “Emerging issues”; Part D starts on p. 561. Report pages: 530–560. The chapter text runs pp. 530–551, Table 22.2 is on p. 551, the references are on pp. 552–559 and p. 560 is blank. PDF pages: 532–562.

Reading record: I read the full text extract page by page, through to the final marker (PDF 562 / p. 560). I also checked the rendered PDF pages for: - the chapter summary box (p. 530); - Figure 22.1 (p. 532); - Box 22.1 (p. 544); - Table 22.1 (p. 548), which the extraction garbled badly; I re-transcribed it below from the rendered page; - Figure 22.2 (p. 549); - Table 22.2 (p. 551); - two suspect unit values (pp. 535, 536).

Author affiliations come from the report’s own contributor biographies (Annex 1, pp. 685–699). The chapter has no panels or commentaries.

Authors and standpoint#

Authors (p. 530): Steffen Foss Hansen, Andrew Maynard, Anders Baun, Joel A. Tickner and Diana M. Bowman.

Footnote 1 (p. 530) says the chapter “is based on and in parts identical to” Hansen, Maynard, Baun & Tickner (2008), ‘Late lessons from early warnings for nanotechnology’, Nature Nanotechnology 3: 444–447. Bowman is not an author of the 2008 article, so she joined the team for this version. My inference (not stated in the chapter): the regulatory section (22.5) probably carries much of her input, since it leans on her publications for the EU cosmetics, novel foods and REACH detail and for the idea of “regulatory inheritance” (Bowman & van Calster 2007; Bowman et al. 2010; Ludlow, Bowman & Kirk 2009; Chaudhry, Gergely & Bowman 2012; Stokes & Bowman 2012). I have not compared the chapter with the 2008 article.

Stated positions (from the report’s Annex 1, not from the chapter itself): - Hansen was a senior researcher at DTU Environment and NanoDTU Environment & Health. He worked on risk analysis, regulation and governance of nanotechnologies. He sat on the WHO nano expert advisory group and worked on the EC’s REACH Implementation Projects on nanomaterials (p. 691). The chapter cites his 2009 PhD thesis, titled Regulation and risk assessment of nanomaterials: Too little, too late? (p. 554). Hansen and Tickner also wrote Ch 2 of this report, on precaution and “false alarms”. - Maynard was the Charles and Rita Gelman Risk Science Professor and director of the University of Michigan Risk Science Center, and a member of the WEF Global Agenda Council on the Challenges of Emerging Technologies (p. 693). The chapter doesn’t say so, but the reference list includes a Project on Emerging Nanotechnologies (PEN, Woodrow Wilson Center) report by him (Maynard 2006). Background, not in the report: he was PEN’s chief science adviser, and PEN’s Consumer Products Inventory is the chapter’s main source on products (pp. 533–535, 539). He also co-authored toxicology papers the chapter cites: Poland et al. 2008, a key early warning, and Shvedova et al. 2005, which the chapter cites only as a source of the residual-metal hypothesis (p. 535). He co-authored the research agendas the chapter uses as its benchmark examples as well (Maynard et al. 2006; Oberdörster et al. 2005b; Tsuji et al. 2006). - Baun was professor of risk assessment of nanomaterials at DTU, working on environmental risk assessment and ecotoxicology. He was DTU project manager for the EU FP7 projects NanoImpactNet and ENRHES (pp. 685–686). - Tickner was an associate professor at UMass Lowell and a principal investigator at the Lowell Center for Sustainable Production. His training was in toxics policy, epidemiology, risk assessment and pollution prevention. The bio calls him a “noted authority on chemicals alternatives assessment”, and he had advised government agencies, NGOs and trade unions (p. 699). The alternatives-assessment framework in 22.6.5 is his own (Rossi, Tickner & Geiser 2006). - Bowman was an assistant professor in the Risk Science Center and the Department of Health Management and Policy at Michigan. Her background is in science and law; she is admitted as a barrister and solicitor in Victoria, Australia, and was a member of an Australian Government expert forum on enabling technologies (p. 687).

Evident stance: - The authors write as insiders to the nanotechnology environment, health and safety (EHS) research and governance community. They are not opponents of the technology. - They favour precaution and regulatory reform, but present precaution as compatible with commercial development, and even as something that enables it: - “precautionary yet socially and economically responsive strategies” (pp. 530, 551); - “act on what we know now, to enable industry to produce and market nanotechnology-enabled products that are as safe as possible” (p. 547). - The tone is measured on hazard evidence (“preliminary”, p. 536) and sharper on governance: - “patchy at best … sometimes complacent” (p. 542); - “Substantial time is being wasted” (p. 540).

Genre: The historical Late Lessons chapters look back at harms that actually happened. This one sits in Part C, “Emerging issues” (pp. 429–560), and looks forward. It audits a young technology against the 12 lessons of the 2001 volume. Its title, “early lessons from early warnings”, plays on the report’s title. That makes it a different kind of evidence from the historical chapters (see the Limitations section).

Self-citation: By my rough count, 38 distinct works among roughly 160 reference entries (around a quarter) are written or edited by at least one of the five authors. That count includes chapters in books the authors edited. Some of the key “early warning” studies (Poland et al. 2008; Aitken et al. 2009) and the research agendas the chapter uses as benchmarks have chapter authors among their co-authors.

Panels and other voices: There are no panels, and no response from industry, regulators or developers. - The only regulator voice is a short quote from Hermann Stamm of the EC Joint Research Centre on the nanomaterial definition (p. 540). - The main NGO voice is Richard Denison of Environmental Defense (p. 546). - The views of scientists developing the technology appear only as summarised by Powell (2007), and the authors then reject them (p. 545).

Note for readers of these notes: The project lead, Andrew Maynard, is the chapter’s second author. I have treated the chapter at arm’s length, as I would any other.

Section-by-section notes#

Chapter summary box (p. 530)#

Framing: Nanotechnology is “the latest in a long series of technologies heralded as ushering in a new era of technology-driven prosperity”. The expected benefits are: - economic development and jobs; - better materials at lower environmental cost; - new diagnostics and treatments.

The central question is: “have we learnt the lessons of past ‘wonder technologies’ or are we destined to repeat past mistakes?” (p. 530).

Scope: The chapter introduces and defines nanotechnology and nanomaterials, describes uses, summarises early warnings and regulatory responses, and draws lessons “notwithstanding nanotechnology’s immaturity” (p. 530).

Two headline messages: 1. Development has happened “in the absence of clear design rules for chemists and materials developers on how to integrate health, safety and environmental concerns into design”. “Green nanotechnology” shows promise. But it needs sustainability research “funded at levels significant enough to identify early warnings”, and regulatory incentives for safer materials (p. 530). 2. Political decision-makers “have yet to address many of the shortcomings” in legislation, R&D, risk assessment, management and governance. The result is “a developmental environment that hinders the adoption of precautionary yet socially and economically responsive strategies” (p. 530).

These two paragraphs come back almost word for word as the chapter’s closing text (pp. 550–551).

22.1 What is nanotechnology and what are nanomaterials? (pp. 531–533)#

Origins and definition: - The field has roots in physics, chemistry, biology, materials science and electronics (RS & RAE 2004). - Taniguchi first used the term in 1974, for precision engineering at the nanometre level. - The term has since been “framed and reframed by various actors”, and many definitions exist (p. 531). - The chapter adopts the US National Nanotechnology Initiative (NNI) definition: understanding and control of matter at about 1–100 nm “where unique phenomena enable novel applications” (NNI 2009) (p. 531). - The scientific rationale: the nanoscale sits between the quantum and classical (Newtonian) regimes. It therefore “holds the possibility of revealing and exploiting unique novel phenomena” (p. 531).

Two common criteria for engineered nanomaterials (p. 531): - a purposely engineered structure with at least one dimension of about 1–100 nm; - nanostructure that gives properties different from the bulk material.

The chapter claims that “in most materials or systems it can be determined whether they involve nanomaterials or not” (Hansen et al. 2007) (p. 531).

How nanomaterials are made (p. 531): - Top-down: ball milling, etching, sonication, laser ablation. - Bottom-up: sol-gel, chemical vapour deposition, plasma or flame spraying, supercritical fluids, spinning, self-assembly. - Top-down struggles to make ever-smaller structures accurately. Bottom-up struggles to make structures large enough and of good enough quality.

The range of materials (p. 531): C60 and C70, carbon nanotubes (CNTs), nanoscale liposomes, self-assembled monolayers, dendrimers and aerogels. Standards bodies moved from trying to define “nanotechnology” to defining kinds of nano-object (BSI 2007; ISO 2008).

Figure 22.1 (p. 532) reproduces a categorisation framework from Hansen et al. (2007), co-authored by Hansen and Baun. It sorts materials by where the nanostructure sits: - Bulk: Ia single-phase; Ib multi-phase. - Surface: IIa structured surface; IIb film; IIc structured film. - Particles: IIIa bound to a surface; IIIb suspended in liquids; IIIc suspended in solids; IIId airborne.

The ISO (2008) definition of a nanoparticle is given (three external dimensions of 1–100 nm). Footnote 3 notes that this definition is not universally accepted (p. 532).

Where concern sits: “Most health and environmental impact concerns have been raised over nanoparticles that fall into subcategories IIIc and IIId” (p. 533). Query: this is as printed. I would have expected IIIb (free particles in liquid) alongside IIId; worth checking against Hansen et al. 2007. The claimed benefit of the framework is that it breaks systems into parts, so exposure routes and the relevance of effect studies can be judged (p. 533).

Analytical note: The definitional choices matter later in the chapter. Definitions require novel properties, so nanomaterials become “novel by definition” (Box 22.1, p. 544). That automatically triggers one of the EEA’s warning signs.

22.2 Development of nanotechnology and nanomaterials (p. 533)#

• Growth: Rapid “when assessed by a number of metrics”, including government funding, publications and patents (Chen & Roco 2009; Youtie et al. 2008; Sylvester & Bowman 2011).
• Early drivers: the discoveries of fullerenes, quantum dots and CNTs (Iijima 1991), plus instruments that made nanostructures visible, such as the scanning tunnelling microscope and the atomic force microscope (p. 533).
• Policy turning point: The US NNI was set up in 2000 with large increases in R&D funding. Since then, “most developed and many emerging economies” have launched national initiatives (Roco 2011) (p. 533).
• Investment: Lux Research (2008) estimated global nano R&D in 2008 at about USD 18.2 billion: 8.4 bn from governments, 8.6 bn from companies and 1.2 bn from venture capital. The chapter flags this as “somewhat speculative” (p. 533).
• Diffusion: Government funding increased publications (Linkov et al. 2009). Government-funded science had “a crucial role” in knowledge creation and technology transfer, but there is a lag before knowledge turns into useful inventions (Igami & Okazaki 2007) (p. 533).

Note: The chapter treats state promotion as a background fact here. It returns as a problem under Lesson 10, where the bodies promoting the technology also oversee it (pp. 546–547).

22.3 Current production and application (pp. 533–534)#

Companies: About 2,000 nanotechnology-focused companies worldwide, of which about 1,100 are in the US and 670 are headquartered in the EU (Nanowerk 2010). They range from multinationals to university spin-offs. Sectors include energy, analysis, textiles, antimicrobial wound dressings, paints and coatings, fuel catalysts, lubricants, cosmetics and food packaging (p. 533).

The authors’ own promotional language: “one could argue that nanotechnology is entering in a new era” of more and more sophisticated products, and “Mundane products will soon, it would appear, be superseded by a range of innovative nanotechnology-based products” (p. 533). The wording is hedged (“one could argue”, “it would appear”), but here the chapter still echoes the kind of enthusiasm it criticises elsewhere (pp. 545–546). This is my observation, not the authors’.

The PEN Consumer Products Inventory: - Launched March 2006 with 212 products; 580 in 2007; 1,317 products from about 30 countries by March 2011. - 738 were health and fitness products. - Nanoscale silver was the primary material in many of them (p. 533).

What the inventory lacks: - It records no unit sales and no mass or volume of nanomaterial per product. - “Such information is only available if the producers themselves make it available, which occurs rarely.” - The public, “and even the relevant regulators themselves”, know little about production volumes, and the scattered data are of questionable accuracy (p. 534).

Production data, which the chapter calls “at best patchy” (p. 534): - Several hundred tonnes a year of carbon-based nanomaterials in 2001. - About 900 tonnes of nanotubes alone by 2003 (Kleiner & Hogan 2003). - Frontier Carbon Corp produced more than 40 tonnes a year of C60. - Cientifica estimated 2006 output of nanotubes and fibres at 65 tonnes, worth about EUR 144 million, and forecast more than EUR 3 billion by 2010 (over 60% annual growth). The chapter comments: “The veracity of these claims is still to be tested.” - Even less is known about quantum dots, nano-metals and nanostructured surfaces.

Analytical note: The production figures contradict each other (900 t of nanotubes in 2003 against 65 t of nanotubes and fibres in 2006), and the chapter does not comment on this. The contradiction makes the chapter’s own point: nobody, regulators included, could establish basic facts about scale. Consultancy forecasts are passed on with a caveat but not examined.

22.4 Signs of early warnings (pp. 534–536)#

Framing: Concerns have been raised “almost since the emergence of nanotechnology (Drexler, 1986)”. Analogies have been drawn with ambient ultrafine particles and asbestos (RS & RAE 2004; Seaton et al. 2009; Mullins 2010) (p. 534).

Background note (not from the chapter): Drexler’s 1986 concerns were mainly about molecular manufacturing and self-replicating machines. That is a different class of risk from the toxicity of nanomaterials, but the chapter treats the two as one lineage.

(a) The ultrafine-particle lineage, and titanium dioxide (TiO2) (p. 534)#

• Ultrafine air pollution: Ambient ultrafine particles (under 100 nm; for example from diesel exhaust or incinerators) have been linked to cardiovascular and lung illness and death “inversely correlated with size i.e. the smaller the particles, the more dangerous” (Oberdörster et al. 2005a; Pope & Dockery 2006) (p. 534).
• My assessment: this states the air-pollution evidence of the time more strongly than it deserves. Pope & Dockery is mainly about fine particle mass (PM2.5).
• Foundations of nanotoxicology: Ultrafine-particle research in the 1990s supplied the working hypotheses: hazard may depend on size, shape, crystal structure, surface area, surface chemistry and surface charge, rather than only on the properties normally used for chemicals (Oberdörster et al. 2007) (p. 534).
• 1990: Oberdörster et al. and Ferin et al. found that ultrafine TiO2 (30 nm) and aluminium oxide (20 nm) caused “a very striking inflammatory reaction” in rat lungs, compared with 250 nm and 500 nm particles (p. 534).
• 1992: Oberdörster et al. found that TiO2 crystal structure affects toxicity, and that surface area is a better dose measure than mass (rats). Donaldson et al. (2002) found the same for carbon black (p. 534).
• Contrary evidence: Warheit et al. (2006) and Sayes et al. (2007) found no link with surface area for nano TiO2, silica and other particles in rats (p. 534).
• Background: Warheit worked in industry (DuPont) at the time. The chapter doesn’t say so.
• Cancer evidence: One study (Heinrich et al. 1995) found a statistically significant increase in malignant lung tumours in rats after chronic inhalation of nano TiO2. On that basis NIOSH (2011) concluded that ultrafine TiO2 “should be considered a potential occupational carcinogen”. NIOSH judged it not a direct-acting carcinogen: it acts through a secondary genotoxicity mechanism linked to particle size and surface area, and surface area is the key exposure measure (p. 534).
• Omitted (background): the long-running debate about whether rat tumours under “lung overload” apply to humans, and IARC’s 2006 classification of TiO2 as possibly carcinogenic to humans (Group 2B).

(b) Carbon nanotubes (p. 535)#

• The asbestos analogy: CNTs look like asbestos fibres, which prompted concern (Huczko et al. 2001; Warheit 2009) (pp. 534–535).
• Lam et al. (2004): In mice, single-walled CNTs (SWCNTs) of several purities all caused dose-dependent granulomas and interstitial inflammation in the lungs (p. 535).
• Warheit et al. (2004): In rats, SWCNT soot caused lung granulomas, but not in a dose-dependent way. The absence of the usual lung biomarkers suggested “a potentially new mechanism” (p. 535).
• Poland et al. (2008) (Maynard is a co-author):
• Four multi-walled CNTs (MWCNTs) were injected into the abdominal cavity of mice, as a stand-in for the lining of the chest cavity.
• Long MWCNTs caused length-dependent inflammation, foreign-body giant cells and granulomas “qualitatively and quantitatively similar” to the response to long asbestos.
• Short MWCNTs caused no significant inflammation at 1 day and no giant cells at 7 days.
• Water-soluble components had no effect. That ruled out residual metals, which earlier in-vitro studies had suggested might be the cause (Shvedova et al. 2005; Kagan et al. 2006) (p. 535).
• Takagi et al. (2008) and Muller et al. (2009) support the length effect (p. 535).
• Unit query: the dose is printed as “50 mg” to “three mice” (confirmed on the rendered page). Poland et al. used 50 µg per mouse, so this is probably a unit error. Verify.
• Method limits (acknowledged): Most CNT studies used intratracheal or intraperitoneal instillation. These routes bypass the upper airway’s defences and deposit particles unevenly, so their relevance has long been questioned (Osier & Oberdörster 1997, misprinted as “Oiser”) (p. 535).
• Industry inhalation studies: Nose-only inhalation studies published by industry (BASF, Nanocyl and Bayer) “support previous findings” (Ellinger-Ziegelbauer & Pauluhn 2009; Ma-Hock et al. 2009; Pauluhn 2010).
• Ma-Hock’s 90-day study found concentration-dependent granulomatous inflammation in the lung and the lymph nodes that drain it.
• Inhalation also revealed inflammation in the nasal cavity, larynx and trachea, and alveolar lipoproteinosis. None of these had been seen with instillation (p. 535).

Analytical note — two points the chapter doesn’t draw out: 1. Here, toxicology produced by industry strengthened the warning rather than muddying it. That appears to contrast with several of the historical Late Lessons cases (a cross-chapter comparison to check at synthesis). 2. Pauluhn (2010) is the Bayer study. Its title, as given in the chapter’s own reference list (p. 557), states that the toxic effects “are determined by density of agglomerate structures, not fibrillar structures”. In other words, Bayer’s study questioned the asbestos-like fibre mechanism even while it confirmed the inflammation. The chapter cites it only as supporting evidence.

(c) Nanosilver and human health (pp. 535–536)#

• Prevalence: Nanosilver is “reportedly one of the most widely used nanomaterials in consumer products today” (PEN 2011). Uses include dietary supplements, personal care products, paints, textiles, paper, water and air purification, polymers, and antibacterial foils for washing machines, kitchenware and food storage (p. 535).
• Unknown scale: “The scale of use is currently unknown as there are no labelling requirements for nanoproducts, and the concentrations used are also unknown for most of the products on the market” (Boxall et al. 2008) (p. 535).
• Why it matters: Many uses involve direct human exposure (p. 535).
• Baseline toxicity: Silver metal’s toxicity is “generally considered to be relatively low”. Repeated high intake of colloidal silver causes argyria, an irreversible blue-grey skin discolouration that is “not life threatening” (p. 535).
• Distribution in the body: Silver from nanoparticles enters by mouth and by inhalation, and reaches the liver, olfactory bulb, lungs, skin, brain, kidneys and testes (Sung et al. 2008, 2009; Kim et al. 2008). Whether it travels as particles, ions or complexes is unclear (Mikkelsen et al. 2011) (pp. 535–536).
• Effects:
• Oral nanosilver caused inflammation and slight liver damage in mice (Cha et al. 2008; Kim et al. 2008).
• Prolonged inhalation caused lung inflammation and changes in lung function in rats (Sung et al. 2008) (p. 536).
• In-vitro studies point to reactive oxygen species driving inflammation and cell death. How relevant this is remains unclear (p. 536).
• State of evidence: Wijnhoven et al. (2009) concluded that there are too few well-controlled studies, and too little known about how nanosilver moves through the body, “to provide a proper foundation for human risk assessment” (p. 536).

(d) Nanosilver and the environment (p. 536)#

• Ecotoxicity: Silver is ecotoxic, but how toxic depends heavily on its chemical form.
• REACH predicted no-effect concentrations (PNECs) for silver (ECHA 2011):
• 0.04 µg/L in freshwater;
• 0.86 mg/L in seawater (as printed and confirmed on the rendered page; probably µg/L — check the ECHA dossier);
• 0.025 mg/L in sewage treatment plants (p. 536).
• Effect levels for silver nanoparticles:
• EC50s as low as 4 µg/L for freshwater algae, and far below 1 mg/L for crustaceans (Navarro et al. 2008; Griffitt et al. 2008).
• Nitrifying bacteria can be inhibited below 1 mg/L (Hu 2010), so wastewater treatment plants may be affected (p. 536).
• The text defines EC50 wrongly, as “the maximum concentration that induces a response halfway”.
• Exposure:
• Chemical form probably governs bioavailability for nanoparticles as it does for silver ions, but this is little studied.
• Silver nanoparticles in textiles come out in the wash (Benn et al. 2010).
• Modelled environmental concentrations are in the low ng/L range (Gottschalk et al. 2010) (p. 536).
• The key open question: “It remains uncertain whether silver nanoparticles are more toxic than their bulk counterpart or ionic silver, since the effects can in many cases be ascribed to the ionic form of silver (Ag+)”. Some studies found a more pronounced effect from nanosilver (e.g. Navarro et al. 2008), “but the data so far are not conclusive” (p. 536).
• Aitken et al. (2009) (EMERGNANO review for Defra; Maynard is a co-author) found:
• “indicative evidence of the harm of silver nanoparticles at low concentrations on aquatic invertebrates”, and said industry should reduce or eliminate environmental exposure;
• not enough evidence for a risk assessment;
• but “sufficient evidence to suggest that silver nanoparticles may be harmful to the environment and therefore the use of the precautionary principle should be considered” (p. 536).

Analytical note (my arithmetic; the chapter does none of this): - The lowest reported EC50 for nanosilver (4 µg/L, algae) is roughly three orders of magnitude above the modelled “low ng/L” environmental concentrations. - But EC50s are acute, 50%-effect levels, not safe levels. The better comparison is the REACH freshwater PNEC for silver that the chapter itself reports: 0.04 µg/L, or 40 ng/L. That is only about one order of magnitude above “low ng/L”, so the margin is much thinner than the EC50 comparison suggests. (The PNEC is for silver in general, not nano-specific.) - The evidence therefore supports “plausible hazard, uncertain and possibly narrow margin of safety”, not demonstrated harm. The case for precaution rests on persistence, wide use, lack of data and potential effects on a key piece of infrastructure (wastewater treatment), rather than on observed damage.

(e) The chapter’s own assessment of the warnings (p. 536)#

• “Although preliminary, these studies on the nanoforms of TiO2 and silver as well as carbon nanotubes are indicative of wider concerns that materials intentionally designed and engineered at the nanoscale to exhibit novel properties may also pose emergent risks” (p. 536).
• They “arguably trigger indicators of early warnings”, and have brought more attention and funding to EHS research (Hankin et al. 2011; Aitken et al. 2011; NAS 2012) (p. 536).

The wording here is careful: “preliminary”, “arguably”. Compare the stronger wording later in the regulatory discussion (p. 539; see Limitations).

22.5 Current (lack of nano-specific) regulation (pp. 537–542)#

Framing (p. 537): - Governments have funded some EHS research, but there has been “limited action from regulatory decision-makers towards changing existing technology-neutral regulation”. - The authors concede that “This is not surprising given the current state of scientific understanding”, and that the lag is not “unique to nanotechnologies”. Ludlow et al. (2009) note that a new technology typically goes through a perceived period of under-regulation, with specific rules only following R&D and commercialisation. - Existing frameworks are “not perfect”, many are outdated, and overhauls were needed even before nanotechnology arrived. Nanomaterials “highlight many of the deficiencies that have existed for some time” (p. 537).

This last point is fair-minded and important: nanotechnology acts as a stress test that exposes weaknesses already present in the regulatory system.

Voluntary reporting (p. 537): - The UK (DEFRA), the US (EPA) and Australia (NICNAS) ran voluntary reporting schemes. “Voluntary in nature, and somewhat onerous in operation, the schemes can be described as at best underwhelming.” - DEFRA received 13 submissions over the scheme’s two years. - The US scheme ended in 2009 with submissions from 31 organisations (DEFRA 2008; Hansen 2009; Maynard & Rejeski 2009). - “Given the lack of buy-in from stakeholders”, France and California moved to mandatory reporting (p. 537).

US law: the Toxic Substances Control Act (TSCA) (pp. 537–538): - Identity problem: Chemicals are identified by their Chemical Abstracts Service (CAS) number, which distinguishes molecular structure, not size. Nanosilver and bulk silver share a CAS number, so the TSCA Inventory can’t tell them apart. “This approach ignores evidence that size and shape often lead to nanomaterials behaving in substantially different ways from their bulk counterparts.” This gap for nanosilver has already caused considerable debate (p. 537). - Not unique to the US: Most jurisdictions classify chemicals as new or existing by CAS number. Most nanoscale substances count as “existing” (p. 537). - Significant new use rules (SNURs): Under TSCA section 5(a)(2), EPA can demand data on existing chemicals when their use changes significantly. It has used this for several nanomaterials, including single- and multi-walled CNTs. EPA has also proposed a broader SNUR requiring 90 days’ notice before anyone makes, imports or processes new nanoscale forms of listed substances (p. 537). - The burden-of-proof paradox: Even for a material with a new molecular structure, EPA “must show that it may pose an unreasonable risk of significant exposure” (wording as printed) before manufacturers can be required to test. “These are just the data the agency needs to determine whether the substance poses an unreasonable risk — a classic regulatory paradox” (p. 537). - EPA has proposed a data-collection rule covering production volume, manufacturing, exposure and release, and health and safety data. TSCA reform had not been enacted (p. 538). - EPA Inspector General report (December 2011) found that: - programme offices had no formal process for sharing the information they might collect; - EPA had no overall message for stakeholders; - regulating nanomaterials as chemicals ties EPA’s success to the limits of those statutes; - EPA’s management was “limited by lack of risk information and reliance on industry-submitted data”.

Its conclusion: without better internal processes and communication, “the Agency will not be able to assure that it is effectively managing nanomaterial risks” (p. 538).

EU law: REACH (pp. 538–539): - Design: Adopted in 2006 and in force progressively from 2007. Article 1 aims at a high level of protection. The precautionary principle is written into the text. Article 5 sets “no data, no market”. REACH applies to new and existing chemicals alike, which avoids some of TSCA’s problems (p. 538). - The same identity problem: REACH also relies on CAS identity. It is unresolved whether a nano form counts as the same substance as the bulk form or a different one. - If different, hazard data are required above 1 tonne a year. - If the same, whether the bulk data are adequate is “open to discussion”. - The only change so far has been to remove REACH’s exemption for carbon and graphite (CEC 2008a) (p. 538). - Industry split on CNTs: Companies formed two separate data-gathering groups. One treats CNTs as new substances; the other, including Arkema and Bayer, treats them as bulk graphite (Milmo 2009). The authors comment that whether nanomaterials are new “is not just a theoretical question, but a source of confusion among regulated parties”. Guidance was expected from the 2012 review (p. 538). - Test methods: Above 10 tonnes a year, a chemical safety assessment is required. But the Commission (CEC 2008a) and the Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR 2007) warned that standard test guidelines may not suit nanomaterials. So manufacturers can rely neither on the bulk material’s profile nor on existing guidelines (pp. 538–539). - History: REACH was due for review in 2012. There had been last-minute attempts to add nanomaterials at second reading in 2006 (Bowman & van Calster 2007). How to change REACH was “still up for debate” (p. 539).

The Cosmetics Regulation (p. 539): - The recast was not started because of nanomaterials, but debate centred on them (Bowman et al. 2010). - Adopted in 2009. It defines a nanomaterial as an insoluble or bio-persistent, intentionally manufactured material with an external dimension or internal structure of 1–100 nm (Art. 2(k)). - Products must label nano ingredients with “[nano]” (Art. 19(1)(g)), from 2012. There is no minimum threshold. - Producers must provide a safety assessment. The Commission must publish a catalogue of nanomaterials used, and their “reasonably foreseeable exposure conditions” (Art. 16(10)(a)). - Examples: TiO2, zinc oxide and lipid nanocapsules in sunscreens and moisturisers; fullerenes in a few face creams (p. 539).

Novel Foods, and the politics of the REACH review (pp. 539–540): - The cosmetics recast “could be interpreted as a successful political effort”. But the Novel Foods recast failed: the Parliament and Council could not agree changes that would have covered foods produced with new processes such as nanotechnology (p. 539). - The existing Novel Foods Regulation already requires pre-market approval and safety assessment of new foods. Had the revisions passed, nano-specific information “might have assisted” with current concerns (CEC 2008b; Chaudhry et al. 2012) (p. 539). - That failure “should act as a warning sign” for REACH, where “the stakes appear to be significantly higher”. “There is, we would argue, the potential for nanomaterials to be overlooked in the 2012 REACH revision discussion”, with attention going to the “myriad of other issues in play”: dossier quality, registration bureaucracy and the burden on small and medium enterprises (p. 539). - The Commission was playing down expectations, saying there would be “no fundamental overhaul” (EurActiv 2011). The authors respond: “such efforts to maintain the status quo are worrying given the rapidly increasing evidence of risks as well as the swift growth of production and commercialisation of nanomaterials and products” (pp. 539–540). - The next formal REACH revision relevant to nanomaterials was not scheduled before 2019. “Substantial time is being wasted and effective regulation of nanomaterials is being pushed even further into the future” (p. 540). - In 2009 the European Parliament’s Environment Committee adopted a report (Schlyter, printed “Schylter”) calling for “no data, no market” to apply to nanomaterials until safety assessments can be made. Its fate was unclear, but the authors think it added pressure on the Commission and Council (p. 540).

The European Commission’s definition (October 2011) (p. 540): - The text: A nanomaterial is a natural, incidental or manufactured material with particles (unbound, aggregated or agglomerated) where 50% or more of particles in the number size distribution have one or more external dimensions of 1–100 nm. The 50% threshold may be lowered to 1–50% “where warranted by concerns for the environment, health, safety or competitiveness”. - Note: competitiveness is named alongside environment, health and safety as a legitimate reason to move the regulatory boundary. The threshold can only be lowered (from 50% to 1–50%), which brings more materials into scope. The authors don’t comment on this. - The dispute: The definition differs “considerably” from the one in the Cosmetics Regulation, was “mooted in relation to the recast of the Novel Foods Regulation”, and has been controversial. - According to the chapter, Maynard (2011) argued that the Commission’s “one size fits all policy-based definition has the potential to sideline the science and may fail to capture what is important for addressing risk”. This is the chapter’s paraphrase of Maynard 2011, not a direct quote from it. ChemSec (2011), which the chapter places “within the scientific community” (it is in fact an NGO; background), said the definition ignores the key physical and chemical characteristics linked to risk. - Stamm (JRC) replied that such a definition “is urgently needed, especially for particulate nanomaterials. The aim should be to identify a general class of materials for attention — whether they are benign or hazardous”. - The authors’ position: “both camps have valid points”, but the definition must not become “a barrier to the effective regulation of nanomaterials” (p. 540). - Note: a chapter author’s own published view (Maynard 2011) is presented in the third person as one of the two camps.

Cross-cutting reviews and “regulatory inheritance” (pp. 540–541): - Government and independent reviews (Chaudhry et al. 2006; Ludlow et al. 2007; European Commission 2008) all found that nanomaterials are “captured” by existing regimes. But those regimes’ failure to tell nano from conventional products raises concerns (p. 540). - Four cross-cutting problems (p. 540): 1. regimes don’t distinguish new from known substances when deciding what triggers oversight; 2. volume and mass thresholds don’t fit nano production volumes; 3. there is little trust in conventional risk-assessment protocols; 4. thresholds and exposure limits set by existing methods are questionable. - Regulatory inheritance (Stokes & Bowman 2012): nanomaterials “have inherited the scope and features of the previous analogous regime” (pp. 540–541). - The US, Australia, China, India, the OECD and the EU all propose to treat nanomaterials mainly like their conventional counterparts. “In doing so, they have opted to retain the regulatory status quo despite the growing body of literature that suggests that some nanomaterials may cause harm”. The authors call this “not surprising” given knowledge gaps and “a general lack of express reliance on the precautionary principle in most jurisdictions” (p. 541). - Australia: From January 2011, NICNAS removed some low-volume and low-concentration exemptions for new industrial nanoscale chemicals. The authors call this “minor and incremental”, and typical of how countries tweak before attempting wholesale change (p. 541).

Why nanomaterials strain hazard identification (pp. 541–542): - Dose metric: Risk assessment rests on Paracelsus’s rule that the dose makes the poison, and assumes toxicity scales with mass. That may not hold. Surface area and surface chemistry may be better guides, which raises the question of how to measure relevant exposure in labs, workplaces and the environment. SCENIHR (2007): behaviour and effects “are not solely dependent on exposure in terms of the mass concentration” (p. 541). - Clumping: Nanoparticles agglomerate (loosely) and aggregate (tightly). - Clumping depends on concentration, with smaller clusters forming at lower concentrations. - If toxicity rises as cluster size falls, higher concentrations may not mean higher toxicity. - Clusters might break apart inside the body. - “the statement that lower exposure equals lower effects should be seriously scrutinised before it can be considered valid for engineered nanoparticles” (Baun & Hansen 2008; Baun et al. 2009) (p. 541). - Environmental criteria: Environmental hazard identification relies on degradability and bioaccumulation, and very few studies exist for nanomaterials. SCENIHR (2007) said the criteria for persistence, bioaccumulation and toxicity (PBT) “should be assessed for applicability to nanoparticles” (p. 541). - Case-by-case assessment: Experts often recommend assessing nanomaterials case by case (SCENIHR 2007, 2009; Stone et al. 2010). But past experience with ordinary chemicals shows this is “very time- and resource-intensive”, and “one has to wonder” whether it is the right approach (p. 541). “It has been claimed” that there are up to 50,000 possible combinations of SWCNT (Schmidt 2007), each potentially with different properties. The chapter concedes that not all of them are expected to be commercially relevant. But add fullerenes, quantum dots, and metal and metal-oxide particles, and the variety implies “a great complexity” for case-by-case assessment (p. 542).

Analytical note: This is a real structural bind. Regulators are told to assess case by case because general rules are unreliable, yet the sheer variety threatens to make case-by-case assessment impractical at scale. The chapter’s own wording is more cautious: “one has to wonder”, “great complexity”. The chapter’s way out comes later, as design criteria or grouping by “emergent risk, plausibility and impact” (p. 547), but it is only sketched.

22.6 Late lessons from early warnings for nanotechnology (pp. 542–547)#

Opening verdict: “stakeholders are doing some things right, but we are still in danger of repeating old, and potentially costly, mistakes” (p. 542). The 2008 Nature Nanotechnology article shares this lesson-by-lesson structure, so this section probably follows it most closely (not verified against the article).

22.6.1 Lessons 1–3: heed the “warnings” (pp. 542–543)#

• Ignorance and humility: The section opens with the 2001 report’s point that knowledge “will always be subject to some degree of ignorance”, so being “alert to — and humble about” its gaps is fundamental (p. 542).
• A distinctive claim: “Perhaps more than any preceding technology, the early development of nanotechnology has been characterised by discussions of potential risks and the need for regulatory reform” (Grieger et al. 2009; Fiedler & Reynolds 1994).
• Risk discussion was built into the NNI from the start.
• The Royal Society and Royal Academy of Engineering report (RS & RAE 2004) stressed uncertainties.
• Levi-Faur and Comhanester (2007): talk about the right regulatory framework is accompanying the technology’s development rather than following it (p. 542).
• Industry self-regulation: Codes of conduct, guidance and risk frameworks have emerged (Bowman & Hodge 2009; Meili & Widmer 2010). “Voluntary in nature, they sit within the shadow of formal regulatory obligations and do not seek to usurp legislative requirements” (p. 542).
• But action lags talk: Despite some moves such as early EHS funding, “coordinated action seems slow to emerge” (p. 542).
• The EEA warning signs: materials that are novel, bio-persistent, readily dispersed or bioaccumulative, and/or that lead to “irreversible action” (for example mesothelioma from inhaling asbestiform fibres). The authors call these “clearly relevant to many nanomaterials”:
• they have novel properties and can go into very diverse products;
• they may travel through the body in new ways, such as across the blood–brain or placental barriers;
• they “may be designed to be persistent”;
• their environmental fate can’t yet be predicted, and monitoring of dispersal isn’t expected soon (SCENIHR 2007);
• whether they bioaccumulate or cause irreversible harm is “largely unknown”, but the potential exists (p. 542).
• The response: The global response is “patchy at best, with governments being slow, and sometimes complacent”, about gathering data on production, use patterns and whether protective equipment works (pp. 542–543). But “efforts to date have been better than those seen in response to the emergence of earlier technologies, but they are still far from ideal” (p. 543).
• Research strategies: Maynard et al. (2006, Nature) called for, within stated timeframes:
• strategic, risk-focused research programmes within 12 months;
• instruments to assess exposure within 3–10 years;
• robust systems to evaluate life-cycle impacts within 5 years;
• validated toxicity-testing methods within 5–15 years;
• predictive models of health and environmental impact within 10 years (p. 543).

EU Seventh Framework Programme calls reflect some of this, and countries are developing integrated EHS research programmes, such as the NNI’s 2008 cross-agency strategy (p. 543). - Remaining gaps (p. 543): - epidemiology of exposed populations; - what happens to ingested nanomaterials; - fate, behaviour and (eco)toxicity across the life cycle; - interactions with natural organic matter, sediments and existing pollutants. - Blind spots: Strategies aimed at recognised uncertainties “should be relatively easy to develop”, because the questions are broadly agreed. The greater danger is missing areas entirely because “the right questions have not been identified”. The EEA’s examples: antimicrobial growth promoters in farm animals, MTBE and tributyltin (p. 543). “At present it is not clear whether the recognition of ignorance … is sufficient to avoid blind spots, or whether the novel properties of nanomaterials inherently will generate blind spots because of their novelty” (p. 543).

22.6.2 Lessons 4 and 11: reduce obstacles to action (p. 543)#

• Institutional ignorance: Useful research can still be ignored. The EEA cites regulators whose disciplines blinkered them, such as clinicians focused on acute effects when dealing with radiation and asbestos. Nanotechnology spans physics, chemistry, computer science, health, environmental science and law, so there is “a real danger of similar errors” (Karn et al. 2003) (p. 543).
• Centres without EHS: Many multidisciplinary nanoscience centres have been set up, “but only a few of these address health, environmental and social aspects”. Resources are needed for infrastructure that gets disciplines working together (Lynch 2006) (p. 543).
• The regulator’s worldview: In deciding how TSCA covers nanomaterials, the US EPA “appeared to be constrained by a world-view rooted in chemistry”: it held that a nanomaterial is legally new only if it has a unique molecular identity (US EPA 2007). “However, it is now clear that characteristics other than molecular identity — such as particle size and shape — can affect exposure and response” (SCENIHR 2007) (p. 543).

Box 22.1: The EEA’s warning signs applied to fullerenes and CNTs (p. 544)#

• The challenge: Responding to ignorance “seems almost impossible when it comes to a rapidly emerging technology”. The EEA’s warning signs “bear an uncanny resemblance to many of the concerns now being raised” about C60 and CNTs (p. 544).
• Novelty: No single taxonomy exists, and one may not be possible or desirable (RCEP 2008). But “one could argue that nanomaterials are novel by definition”, because definitions require novel applications or properties (p. 544).
• Persistence:
• Few degradability studies exist, but the structures are expected to persist.
• C60 and CNTs also form naturally in forest fires and eruptions, and are chemically identical to engineered ones.
• Geological evidence shows great resistance to breakdown: C60 in the 1.85-billion-year-old Sudbury impact breccias in Canada (Becker et al. 1994); C60 in a 70-million-year-old dinosaur eggshell from Xixia, China (Zhenxia et al. 1998); CNTs and fullerenes in 10,000-year-old ice-core meltwater (Murr et al. 2004) (p. 544).
• Dispersal:
• Depends on the environmental compartment.
• In the lab, hydrophobic C60 and CNTs aggregate fast and may settle out rather than spread in water.
• But changing the surface alters this: hydroxylated C60 is much more soluble.
• Interactions with humic substances and aquatic organisms are unclear (p. 544).
• Bioaccumulation: CNTs are non-biodegradable, insoluble in water and fat-soluble, which suggests potential to bioaccumulate. There is “a profound lack of studies” (RCEP 2008) (p. 544).
• Irreversibility: It is “almost impossible to say” whether harm could be irreversible, though some CNT effects might be (Poland et al. 2009; Smith et al. 2007, on rainbow trout). But “Widespread production and use of C60 and CNT will inevitably lead to the release of these materials into the environment, and hence an irreversible action, as they would be practically impossible to locate and recover after release” (Hansen et al. 2009) (p. 544).

Analytical note: 1. Using geological natural analogues as evidence of persistence is inventive and fairly convincing. 2. The way the Box applies the criteria is permissive. “Novel by definition”, plus “irreversible because unrecoverable once released”, would flag almost any widely dispersed new material. That is useful for deciding where to look first, but weak for telling materials apart. 3. The reference list includes Hansen et al. (2012), ‘Operationalization and application of “early warning signs” to screen nanomaterials for harmful properties’. It is directly relevant to the Box but never cited in the text. Also, neither Poland et al. 2009 nor Hansen et al. 2009 appears in the reference list.

22.6.3 Lessons 5 and 8: stay in the real world (pp. 544–545)#

• Ideal versus actual: The 2001 panel observed that technologies are often assumed to perform to specification, although real life “can be far from ideal”. This “echoes claims made of nanotechnology” (p. 544).
• The Weiss visit: In 2006, Rick Weiss of the Washington Post visited a nanomaterial company expecting a high-tech workplace. He found “men in grease-stained blue coats”, “story-tall spray-drying machines”, “noisy milling operations and workers with face masks covered by a pale dust” from emptying buckets of fresh powder (Weiss 2006) (p. 544).
• The containment assumption: Nanotechnology is often assumed to use small quantities in sealed processes. The past shows that persistent substances used in closed settings or locked in solid materials, such as PCBs, “will eventually end up in the environment” (p. 545).
• Developers’ mental model: “there is evidence that the R&D community is entrenched in the philosophy that basic research will ultimately solve real-world problems through a one-way process of knowledge diffusion, and that they do not need to worry about EHS issues” (p. 545).
• Powell (2007) found that many scientists developing nanotechnologies don’t think it poses new or substantial risks, and think risk concerns rest on “invalid science”.
• The authors: “This is a mistake in our view”, pointing to the 2001 report and “the sorry tale of asbestos” (p. 545).
• Reference gap: Powell 2007 is missing from the reference list. It is probably Powell, M.C. (2007), Health, Risk & Society 9(2); verify.
• Who should inform policy: Applied researchers and the EHS community. Following the EEA, workers and users can bring information to regulatory appraisal, “although such knowledge of course needs as much critical appraisal as specialist knowledge” (p. 545).
• Lay insight: “it can be argued that non-experts have little to contribute” to such a complex field. But non-specialists closely involved with a technology can bring unique insight into what matters and what will work (Gavelin et al. 2007, the UK Nanotechnology Engagement Group) (p. 545).

Note: The evidence for the “one-way diffusion” claim is thin. It rests on a single survey (Powell 2007), which is missing from the reference list. The Weiss newspaper account supports a different point, the gap between assumed and actual working conditions. The claim is plausible, not shown.

22.6.4 Lessons 6 and 9: consider wider issues (p. 545)#

• Two opposite dangers, both acknowledged: that “speculation on risks overshadows real benefits”, or that “an unbalanced promotion of possible benefits will prevent potential risks from being critically scrutinised”. Nanotechnology “is in such a position” (Maynard et al. 2011) (p. 545).
• The claimed pros: economic benefits, better materials, lower resource use and new medical treatments. The cons “mainly revolve around worker health, consumer exposure and environmental impacts” (p. 545).
• Testing the claims: It is hard to judge either side’s claims over the short and long term, but working out more likely scenarios is vital, “As we emerge from the first flush of nano-enthusiasm and begin the hard work of translating good ideas into viable products” (p. 545).
• Trust: “If proclaimed pros do not materialise in the foreseeable future despite heavy public investments, or if projected cons are not investigated, but later prove to be significant, decision-making processes will be undermined, and public trust may be compromised” (p. 545).
• BSE: The public’s “surprised revulsion” that cattle were fed offal. The 2001 panel speculated that paying earlier attention to wider social values might have limited the BSE problem. “The extent to which societal interests and values can prevent real risks with emerging technologies is debatable”, but values shape what is accepted or rejected (p. 545).
• Citizens as stakeholders: Because applications are so widespread, “citizens around the world are as much stakeholders in the technology as the governments, industries and scientists promoting it”. Yet “so far the deliberate engagement of citizens and the public in risk-related decisions on nanotechnology has been very limited” (p. 545).

22.6.5 Lesson 7: evaluate alternative solutions (pp. 545–546)#

• The lesson in one line: “don’t become so enamoured by a new technology, that you are blinded to alternative solutions” (p. 545).
• The pattern: Proponents tend to justify heavy investment “by promoting its application to every conceivable problem”, so alternatives get too little scrutiny and the best solution isn’t always chosen (p. 545).
• Applied to nano: “the number of nanoscale solutions looking for a problem continues to grow. And with international nano-fever running high, everyone wants to be at the forefront of the nanotechnology revolution”. The goal should be the best solution to a given problem, “rather than to squeeze a solution out of the latest technology”. “while nanotechnology could be used, it may be questionable whether it should” (p. 546).
• Alternatives assessment (Rossi, Tickner & Geiser 2006) has six steps (p. 546): 1. identify targets for action; 2. characterise and prioritise end uses; 3. identify alternatives; 4. evaluate and compare them; 5. select the preferred option; 6. review it.
• Scope: Look at the service or function needed, not just at swapping one chemical for a safer one. For nanomaterials, the alternatives need not be other nanomaterials; they could be process or administrative changes that reduce the need for the material at all (p. 546).
• Timing: Alternatives thinking can be applied at the design stage, “which did not occur in the case of nanotechnology” (p. 546).

Note: No example is given of a better alternative that nanotechnology displaced; the lesson is argued by analogy. Table 22.1 shows that none of the reports or regulations reviewed took any notice of this lesson.

22.6.6 Lesson 10: maintain regulatory independence (pp. 546–547)#

• The EEA’s finding: Interested parties often influence regulators unduly. One factor in the slow UK response to BSE was that the regulator was responsible “first to the industry and only second to consumers” (p. 546).
• Applied to nano: “In many countries, the organisations responsible for overseeing the development of nanotechnologies through R&D are the very ones that address health and environmental issues” (p. 546).
• Denison’s testimony: Richard Denison of Environmental Defense (spelled “Dennison” in the text) told the US House Science and Technology Committee in 2007 that housing both promotion and risk mitigation within the NNI had created a conflict of interest, and that this “conflict of interest is both slowing and compromising efforts by NNI and its member agencies and departments to effectively address nanotechnology’s implications” (p. 546).
• The National Academy of Sciences (NAS, 2012): The dual roles may “impede implementation and evaluation of the EHS risk research”, so “a clear separation of management and budgetary authority and accountability is needed” between promoting applications and assessing health and environmental implications (p. 546).
• The authors: An integrated approach to risks and benefits is critical. But “when the promoters of nanotechnology, whether government or industry, have a strong influence over oversight, independent regulatory decision-making becomes compromised. Perhaps more insidiously, research and development decisions end up being influenced by what will ultimately promote the technology, rather than what will protect producers, users and the environment” (pp. 546–547).

Note: The evidence is testimony and an expert committee’s stated concern. The chapter doesn’t show that any particular EHS decision was distorted. A side observation (mine): Denison’s organisation co-developed the voluntary Nano Risk Framework with DuPont in the same year (p. 549). That is an NGO–industry partnership, not a regulatory arrangement, so it doesn’t bear directly on the regulatory-independence lesson. But it shows that the NGO voice quoted here was not simply an outside adversary of industry.

22.6.7 Lesson 12: avoid paralysis by analysis (p. 547)#

• The EEA’s point: Experts have often argued early that “we know enough to take protective action”. Good policy balances information against action, keeps the end point (preventing harm) in view, and builds in reviews to correct course (p. 547).
• The nano record: “Twenty years have elapsed since first indications of nanomaterial harm were published (Ferin et al., 1990; Oberdorster et al., 1990)”, and the literature has grown. “Yet many governments still call for more information as a substitute for action; there are indications that understanding and managing the risks of engineered nanomaterials are being paralysed by analysis” (p. 547).
• Act now: “we need to act on what we know now, to enable industry to produce and market nanotechnology-enabled products that are as safe as possible”. Engineered nanomaterials are already on the market, some with poorly understood risks, and “may therefore be ineffectively regulated”. Using current knowledge “will help prevent basic mistakes being made” while better knowledge is developed (p. 547).
• Design criteria: Criteria could sort nanomaterials into higher and lower concern by intrinsic properties, use or exposure. Maynard et al. (2011) proposed principles of “emergent risk, plausibility and impact” for identifying materials of high concern (p. 547).
• Timing and lock-in: Considering health and safety at the design stage, including safer production methods and alternatives, “will facilitate decisions as economic interests are not fully entrenched at that point” (p. 547).

Note: The chapter doesn’t say which concrete actions “what we know now” justifies, beyond data requirements, design criteria and the recommendations surveyed in 22.8. It proposes no exposure limits, restrictions or product withdrawals.

22.7 So have we learnt the lessons? (pp. 547–548)#

Framing: Some of the 12 lessons don’t apply directly to all emerging technologies, but many are relevant. “Yet the picture is not as bleak as it might be” (p. 547).

Table 22.1 (p. 548): “Late lessons learned, as indicated in 10 EU Member State nanomaterials reports”. Each document is scored against the 2001 lessons: - + mentioned in passing; - ++ substantially discussed or analysed; - +++ strategy suggested or implemented; - blank: no notice taken.

Transcription from the rendered page. Report columns: RS&RAE 04 | DG Sanco 04 | Chaudhry 06 | “IRGB” 06 [presumably IRGC] | SCENIHR 07 | RCEP 08 | CEC 08 | Stone 09 | RIVM 09 | Aitken 09 | SCENIHR 09. Regulation columns: Cosmetics | Biocides | Food Additives | RoHS & WEEE.

The rows follow the printed order, which puts the lay-knowledge row before the alternatives row. (Audit: every cell re-checked against the rendered page; the transcription is correct.)

What the text says the table covers (p. 547): “10 main EU Member State national and multilateral scientific reports”, plus, in its “second half”, “the main EU regulatory actions taken over the course of the last decade”.

The authors’ reading (p. 547): Policy-makers and stakeholders “seem to have learnt at least some of the lessons”. They are: - asking critical questions early about health and environmental fate and effects; - building collaborations across disciplines, departments and national borders; - beginning to target research; - engaging stakeholders; - asking whether existing oversight is fit for purpose.

The regulations (pp. 547–548): - The Cosmetics and Biocides regulations acknowledge high uncertainty. They require industry to submit data on physical and chemical characteristics, exposure and toxicology, which provides “some elements of a strategy” for long-term monitoring and for reducing blind spots. - Putting “the burden of providing health and safety data” on industry “helps to overcome the problem of paralysis by analysis since companies are in theory not able to market their products without proper data”. - Independence is partly ensured because the scientific committees that evaluate the data (the Scientific Committee on Consumer Safety, SCCS, and the European Food Safety Authority, EFSA) are separate from the agencies promoting nanomaterials in those products. However, “these agencies have recently come under attack for not being independent from industry interests” (Muilerman & Tweedale 2011, PAN Europe) (p. 548).

Verdict: “the question here seems not to be whether we have learnt the lessons, but whether we are applying them effectively enough to prevent nanotechnology becoming yet another future case study on how not to introduce a new technology” (p. 548). The diagnosis: - we have become “distracted by the way that nanotechnology is being overseen by the very government organisations that promote it”; - research strategies aren’t producing clear answers; - disciplinary and institutional barriers persist; - stakeholders aren’t engaged, or not early enough; - this is partly “bureaucratic inertia”, while remarks such as “risk research jeopardises innovation or regulation is bad for business, only cloud the waters when clarity of thought and action are needed” (p. 548).

Close: To realise the benefits “without leaving a legacy of harm”, “perhaps it is time to go back to the classroom and re-learn these late lessons from early warnings” (p. 548).

My reading of Table 22.1 (not the authors’): 1. There is no “+++” anywhere in the table. Taken literally, the scoring says that none of the 11 reports or 4 regulations suggested or put in place a strategy for any lesson. That is bleaker than the text’s “not as bleak as it might be”. But treat this with care: section 22.8 itself describes several of the same bodies as recommending strategies, among them RS & RAE 2004 (six key recommendations), RCEP 2008 and IRGC. And the text credits the Cosmetics and Biocides regulations with “some elements of a strategy” (p. 547). So the empty “+++” column reflects either a strict, unexplained coding threshold or an inconsistency between table and text. It should not be read as evidence that no one proposed a strategy. 2. Uptake is uneven. The three “science” lessons (ignorance, monitoring, blind spots) are noted almost everywhere. The governance lessons are thin: - alternatives: the row is completely blank; - regulatory independence: one passing mention (RS & RAE) plus the Cosmetics Regulation; - lay knowledge: three reports, all from 2004–2006; - interdisciplinary and institutional obstacles: reports only, no regulation. 3. The later reports cover fewer lessons, but genre may explain this. The broadest coverage is in the earliest document (RS & RAE 2004, a wide-ranging policy report). The 2009 documents (Stone, RIVM, Aitken, SCENIHR 2009) score mostly on the science lessons. Those are largely scientific EHS reviews and risk-assessment opinions, so the pattern may reflect what kind of document was sampled rather than a real decline in attention to governance. The table alone can’t separate the two. 4. Method problems: - The title says “10 … Member State … reports”, but there are 11 report columns. The text’s fuller wording, “national and multilateral”, covers the EU-level documents (DG Sanco, SCENIHR ×2, CEC). IRGC (printed “IRGB”) is an international body, and RS & RAE and RCEP are UK bodies. - No account is given of how documents were chosen, who scored them, on what criteria, or with what cross-check. - The scores measure attention in documents, not action or outcomes. 5. Table–text inconsistencies: - The text says regulatory independence is “to some extent” ensured in both the Cosmetics and the Biocides regulations (pp. 547–548), but the table scores only Cosmetics (++) on that row; Biocides is blank. - The table column is “Stone et al. (2009)”; the text and reference list give Stone et al. 2010 (ENRHES).

22.8 Precautionary strategies for nanomaterials (pp. 549–551)#

The lag (Linkov et al. 2009; Hansen is a co-author): - There is a substantial time lag between products appearing, EHS data being produced, and regulators using those data. The cause is agencies’ limited resources and the time needed to adapt risk-assessment methods (p. 549). - Figure 22.2 is a schematic, not plotted data. It shows three rising curves: emerging nanoproducts (earliest and steepest), generated EHS data (later), and EHS data analysed by regulators (latest), with a horizontal “Gap” between them. The figure contains typos (“EHA”, “anylysed”) (p. 549).

How long the lag will be: “The precise extent of the time lag is unclear, but there is historical evidence indicating that it will not be less than two decades” (p. 549). The evidence offered: - Lawless (1977) analysed 45 episodes of public alarm about technologies (reproduction and genetics, food and medicine, the environment). Institutions grapple with a problem while papers on its effects pile up. The average delay was “one or two decades”. - The 2001 volume found gaps of decades, sometimes over a century, between the first report of harm and effective regulation. - These cases “may not reflect all emerging technologies”, but “they do represent plausible worst-case scenarios” (p. 549).

Two problems: - Lawless (1977) is not in the reference list. - Neither source supports a minimum of two decades. Lawless’s average is “one or two decades”, and the 2001 volume says “decades”. The chapter also calls these cases “plausible worst-case scenarios”, which sits awkwardly with presenting two decades as a floor (“will not be less than”).

Short product lifetimes: Because specific nano products will probably be short-lived as the technology improves, Linkov et al. (2009) argue that regulation should adapt to the field’s changing nature. “The question however is how this should be done” (p. 549).

RCEP (2008): Existing regulatory approaches “cannot be relied on to detect and manage problems before a novel technology such as nanomaterials has become ubiquitous”. The authors call this “rather bleak and discouraging” (p. 549).

Three families of precautionary strategy: 1. Adapt legislation (Chaudhry et al. 2006; Fuhr et al. 2007; Franco et al. 2007; Ludlow et al. 2007; Breggin et al. 2009): clarify key terms and definitions; make sure legislation’s scope and objectives are relevant; set thresholds that fit nanomaterials; assess risk before or after environmental release (p. 549). 2. Adapt risk assessment and management. The main example is the Nano Risk Framework developed jointly by the NGO Environmental Defense and DuPont (2007). - It is applied repeatedly through basic R&D, prototyping, pilot testing, test marketing, and whenever new information arrives. - Its six steps, as printed: “develop the nanomaterial and its intended uses” (the framework itself says “describe”; background); develop hazard and exposure profiles over the full life cycle; evaluate the information to assess the probability of risks; evaluate risk-management options and recommend a course of action; decide “alongside key stakeholders” whether to continue R&D and production; update and re-run the evaluation regularly and share information with stakeholders (pp. 549–550). 3. Broader governance. - The Royal Commission on Environmental Pollution (RCEP, cited as “2009”; probably the 2008 report) calls for “flexible and resilient forms of adaptive management”, modification and extension of existing regulation “as a matter of urgency”, and an early warning system with robust monitoring informed by “the full range of perspectives on innovation” (p. 550). - The International Risk Governance Council (IRGC, 2006 and 2007; printed “ICRG”) offers a framework of pre-assessment, risk appraisal, and judgement of tolerability and acceptability, tied together by risk management and communication. Its nano recommendations fall into five groups: improve the knowledge base; strengthen risk-management structures; promote stakeholder communication and participation; ensure social benefits and acceptance; foster collaboration between stakeholders and nations (p. 550). - The German Advisory Council on the Environment (SRU 2011) calls for more risk research, social dialogue, “a single piece of nano-specific legislation based on the precautionary principle”, a labelling scheme and product register, and reform of chemical, product and environmental law (p. 550).

These echo the RS & RAE (2004) recommendations (p. 550): - health, safety and environmental research as an integral part of innovation and product design; - avoiding the release of manufactured nanoparticles and nanotubes into the environment “as far as possible”; - regulators to review whether existing rules are adequate and to horizon-scan; - considering ethics and social implications in the training of research students and staff; - government-funded public dialogue; - a multi-stakeholder group to spot emerging issues at the earliest stage.

Design: - “one must not forget the critical role of design in ensuring that technology development occurs in parallel with technology assessment”. - Nano development has gone ahead without design rules for integrating health, safety and environmental concerns. “This is not surprising given that most chemists and materials designers are not trained to recognise these issues.” - “Green nanotechnology” shows promise (pp. 550–551). - Preventive design needs “a cultural transition” (pp. 550–551): - chemists and materials developers educated in health, safety and environment; - these concerns treated as quality criteria “equal to economic and performance considerations”; - sustainability research funded at levels high enough to detect early warnings; - regulatory incentives for safer materials.

Conclusion: The common thread across all these recommendations is that “many of them are not or have yet to be successfully implemented by political decision-makers”. The result is a development environment that hinders “precautionary yet socially and economically responsive strategies”, which, if unresolved, could hamper responsible development (p. 551).

Table 22.2: Early warnings and actions (p. 551)#

The table lists: 1974 Taniguchi; 1981 STM; 1985 AFM; 1985 fullerenes; 1986 Drexler; 1991 CNTs; 1992 surface-area finding; 2000 NNI; 2003 RS & RAE; 2004 SWCNT granulomas; 2006 PEN inventory (212 products); 2006 DEFRA voluntary scheme; 2007 ED–DuPont framework; 2007 US EPA voluntary scheme; 2008 MWCNTs asbestos-like; 2009 US scheme ends with 31 submissions; 2009 Cosmetics Regulation; 2009 EP Environment Committee “no data, no market”; 2011 NICNAS change; 2011 NIOSH TiO2; 2011 EC definition.

Observations: - Missing warnings: The table leaves out the 1990 “first indications” and the 1995 rat tumour study, although the text relies on both (pp. 534, 547). - Dating: It dates the RS & RAE report to 2003; the text cites it as 2004. - Missing events: The Novel Foods failure, the EPA Inspector General report, the US SNURs for CNTs and NAS 2012 are absent. - Smaller discrepancies with the text: - “31 submissions” in the table against “31 organisations” in the text (p. 537). - The EC definition is called a “proposal”, but the text quotes the adopted October 2011 Recommendation (p. 540). - STM and AFM are dated 1981 and 1985, but the text cites Binning et al. 1982 and 1986. - What it shows: A decade and a half of mostly information-gathering and voluntary measures after the key hazard findings. The only EU legislation in the table that expressly differentiates nanomaterials is the 2009 Cosmetics Regulation. The text also credits the Biocides Regulation with nano data requirements (p. 547), but the table omits it. The binding non-EU measures in the chapter are incremental: Australia’s 2011 NICNAS changes (in the table) and US SNURs for some CNTs (text only, p. 537).

References (pp. 552–559): citation integrity#

Cited in the text but missing from the reference list: - Powell 2007 (p. 545) and Lawless 1977 (p. 549); - Navarro et al. 2008 (p. 536); - Takagi et al. 2008 and Muller et al. 2009 (p. 535); - Hansen et al. 2009 and Poland et al. 2009 (Box 22.1, p. 544); - Stone et al. 2011 (p. 542); - RCEP 2009 (p. 550); - Bowman et al. 2010 (p. 539); - DEFRA 2006a/b, Weiss 2005, NICNAS 2008 and Fuhr et al. 2007 (pp. 537, 549); - Ludlow et al. 2007 (pp. 540, 549), a key source for the cross-cutting regulatory problems; only Ludlow et al. 2009 is listed; - Biswas & Wu 2005 and BSI 2007b (p. 531); Sung et al. 2009 (p. 535); Davies 2009 (p. 538); - Igami & Okazaki 2007 (p. 533): the list has Igami & Saka 2007 instead; - Kroto et al. is cited as 1986 (p. 533) but listed as 1985.

Listed but never cited in the text: - Crocetti & Miller 2011 (Friends of the Earth, ‘Nano-Silver: policy failure puts public health at risk’); - Hansen & Tickner 2007 (on voluntary measures); - Hansen et al. 2012 (on operationalising the early warning signs); - Denison 2007b (post-hearing answers); - Thayer 2007; Cientifica 2007; Technology Transfer Center 2007; Roco & Renn 2006; Roco & Bainbridge 2001. - Sayes et al. 2004 and ISO 2008 each appear twice.

Why it matters: These look like traces of layered drafting. Two of the missing references carry real weight. Powell is the only evidence for developers’ mental models, and Lawless is the only evidence for the “two decades” lag.

Case timeline#

This is a prospective case, so the timeline records warnings and responses. There are no realised harms.

Lag between warning and action: - TiO2: First credible warnings in 1990/1992 (inflammation; dose metric) and 1995 (tumours). The main “action” in the chapter is NIOSH’s 2011 designation, some 16–21 years later. That designation is guidance, not binding regulation (background). Nano-TiO2 in cosmetics also falls under the 2009 Cosmetics Regulation (p. 539). - CNTs: Concern from 2004, sharpening in 2008. The responses noted are: - US SNURs for some single- and multi-walled CNTs (undated in the chapter; p. 537); - the removal of REACH’s exemption for carbon and graphite (CEC 2008a; p. 538); - RS & RAE’s earlier call to avoid releasing nanotubes (2004; p. 550).

But registrants split on whether CNTs are new substances at all (p. 538), and the EU had no nano-specific REACH provisions by 2012. - Binding EU nano-specific law: The 2009 Cosmetics Regulation. Its adoption came about 19 years after the 1990 warning and 5 years after RS & RAE 2004; labelling was due to take effect in 2012, per the chapter (p. 539). - The authors’ own summary: twenty years after the first indications, governments still call for more information “as a substitute for action” (p. 547).

Harms and costs: No human harm is documented in the chapter. The evidence is of hazard in animals and in environmental tests. The main exposure concerns are: - workers handling powders (p. 544); - consumers directly exposed to nanosilver and to cosmetics (pp. 535, 539); - aquatic life and wastewater treatment (p. 536).

No cost figures, for harm or for regulation, are given. The benefits are claimed but not assessed.

What was known when: - By 1992, that the mass-based dose measure may fail for ultrafine particles. - By 2004, that the leading scientific academies judged the risks uncertain enough to recommend avoiding environmental release. - By 2008, that some CNTs behave like asbestos in a sensitive model. - By 2009, that nanosilver is widespread in consumer products, with unknown volumes and plausible ecotoxicity. - Throughout, that regulators could not identify nano forms under chemical-identity systems, and did not know production volumes.

The authors’ own lessons and conclusions#

A. Conclusions the authors derive from evidence (descriptive or diagnostic)#

1. Early warnings exist for TiO2, CNTs and nanosilver. They are “preliminary” but “indicative of wider concerns” about emergent risks from engineered novelty (p. 536).
2. Existing technology-neutral regulation, keyed to chemical identity and tonnage, formally captures nanomaterials but fails to distinguish them. Nanomaterials have “inherited” older regimes, and most governments have kept the status quo (pp. 537–541).
3. Voluntary data-gathering has failed (“at best underwhelming”), and regulators lack basic production and use data (pp. 534, 537).
4. Nanomaterials challenge core assumptions of hazard identification: the mass-based dose measure, monotonic dose–response, the PBT criteria, and case-by-case feasibility (pp. 541–542).
5. Nanotechnology is unusual in that risk discussion accompanied early development, and efforts have been “better than” for earlier technologies. But coordinated action has been slow and “patchy” (pp. 542–543).
6. Lessons are recognised in expert documents (Table 22.1) but not applied effectively. The obstacles are promoter–overseer conflation, research that yields no clear answers, disciplinary and institutional barriers, late or weak public engagement, and bureaucratic inertia (pp. 547–548).
7. There is a structural lag, which the authors put at probably two decades or more, between products and regulators’ use of EHS data (p. 549).
8. Recommendations from expert bodies (RS & RAE, RCEP, IRGC, SRU) have largely not been implemented by political decision-makers (p. 551).

B. Recommendations and advocacy#

• Act now on existing knowledge rather than waiting for more; use design criteria to prioritise materials of concern (p. 547).
• Separate the promotion of nanotechnology from EHS research management and budgets. This is implicit rather than a stated recommendation: the authors quote NAS 2012 and Denison approvingly and conclude that promoter influence over oversight compromises independence (pp. 546–548).
• Revise REACH now for nanomaterials rather than wait until 2019: “immediate revisions are needed to address the most obvious and short-term limitations of the current legislative framework” (p. 540).
• Fund interdisciplinary infrastructure: “It is critical to set aside resources to create an infrastructure that gets people working together across disciplines” (p. 543).
• Close critical research gaps: epidemiology of exposed populations, ingested nanomaterials, life-cycle fate and (eco)toxicity, and interactions with environmental matrices (p. 543).
• Consider alternatives by function, ideally at the design stage (“could vs should”) (p. 546).
• Design: Integrate health, safety and environment at the design stage; build up green nanotechnology; educate chemists; treat EHS as a quality criterion equal to performance and cost; fund sustainability research; create regulatory incentives for safer materials (pp. 547, 550–551).
• Engage the public and use workers’ and users’ knowledge, critically appraised (p. 545).
• Adapt regulation to fast product cycles (reported from Linkov; p. 549). The authors report, without explicitly endorsing, the SRU call for a single nano-specific precautionary law and the RCEP call for adaptive management (p. 550).
• Rhetorical: “go back to the classroom and re-learn these late lessons” (p. 548).

Mechanisms and dynamics#

1. Promise, hype and the investment race. The chapter places nanotechnology in a line of “wonder technologies” (p. 530). It describes: - national initiatives multiplying after the NNI (p. 533); - an estimated USD 18.2 billion of R&D in 2008 alone, a figure the chapter calls “somewhat speculative” (p. 533); - “international nano-fever” in which “everyone wants to be at the forefront” (p. 546); - “nanoscale solutions looking for a problem” (p. 546).

The mechanism it proposes is that heavy investment creates pressure to apply the technology “to every conceivable problem”. That crowds out the assessment of alternatives (pp. 545–546), and unbalanced promotion keeps risks from being scrutinised (p. 545). It also warns of a trust trap: public trust suffers if promised benefits fail to appear, or if neglected risks later emerge (p. 545). The chapter is not immune to the enthusiasm itself (p. 533).

2. Invisibility: who knows what. - Governance depends on data held by producers, who “rarely” disclose it (p. 534). - There is no labelling, so consumer exposure can’t be estimated (p. 535). - Market and production figures come from consultancies and contradict each other (p. 534). - Voluntary reporting attracted few participants (13 submissions and 31 organisations; p. 537). - The EPA’s management relied on industry-submitted data (p. 538).

The result is a regulator that can’t see the scale, distribution or forms of what it regulates. My inference: missing exposure data then becomes a ready reason to wait for more information. That is the “paralysis by analysis” the authors criticise when they urge action on what is already known (p. 547).

3. Legacy categories and regulatory inheritance. - Chemical-control regimes key identity to molecular structure (CAS numbers) and trigger scrutiny by tonnage. - Nano forms of existing substances inherit the bulk material’s status. Under TSCA they are “existing” chemicals; under REACH their status is ambiguous; the thresholds don’t fit nano volumes (pp. 537–541). - “Regulatory inheritance” (Stokes & Bowman 2012; p. 541) names the dynamic: new things arrive inside old legal boxes, whose scope and assumptions travel with them. - This is reinforced by a disciplinary worldview. The EPA’s view was “rooted in chemistry” (p. 543).

4. Burden and standard of proof. - The TSCA paradox: the regulator must show potential risk before it can require the data needed to show risk (p. 537). - REACH reverses the burden (“no data, no market”), but the reversal is weakened when the regulated object is ambiguous: firms can register CNTs as graphite (p. 538). - The Cosmetics and Biocides regulations put the data burden on industry. The authors say this “helps to overcome” paralysis by analysis, since companies are “in theory” unable to market without data (p. 547).

The chapter’s implicit model: whoever carries the burden of producing data determines whether uncertainty leads to delay or to data.

5. Paradigm strain: when the toolkit doesn’t fit. Novel properties undermine assumptions built into assessment methods: - mass-based dose measures (p. 541); - “lower exposure equals lower effects”, given concentration-dependent clumping (p. 541); - PBT criteria (p. 541); - routes of exposure in tests, since instillation differs from inhalation (p. 535).

The chapter raises the possibility that novelty “inherently will generate blind spots” (p. 543). Recognising uncertainty isn’t enough if the instruments used to look for harm are miscalibrated for the thing being examined.

6. Complexity and combinatorial variety. - There may be up to 50,000 SWCNT variants (p. 542). - Properties change with surface coating (p. 544). - Interactions with environmental matrices and co-pollutants matter (p. 543).

Variety outstrips case-by-case assessment (pp. 541–542). The proposed answer is to screen by “emergent risk, plausibility and impact” (p. 547), but this remains a proposal.

7. Pace and time lags. - Products appear first, EHS data later, and regulators use the data later still (Fig. 22.2; p. 549). - Institutions “grapple” for one or two decades while the literature grows (Lawless; p. 549). - Individual products may be short-lived (p. 549), so the object of regulation keeps moving while the regulatory cycle turns slowly: REACH reviews in 2012 and then 2019 (p. 540). - The authors frame this as systematic, not accidental: following Ludlow et al. (2009), a new technology’s emergence is “likely to be perceived as a period of under-regulation”, with specific rules following R&D and commercialisation (p. 537).

8. Institutional design: promoters as overseers. - The same bodies fund and promote development and also address health and environment (pp. 546–547). - The claimed mechanism is subtle: the effect that operates “Perhaps more insidiously” is on research agendas. R&D decisions are steered by “what will ultimately promote the technology” rather than by what protects people (p. 547). - The chapter’s evidence is testimony and the NAS’s concern. It also notes that the ostensibly independent EU committees faced accusations of industry influence (p. 548).

9. How the actors thought (mental models): - Developers and R&D scientists: believe in a linear model in which basic research solves real problems through one-way diffusion, and see risk concerns as based on “invalid science” (Powell 2007; p. 545). Most chemists and materials designers are “not trained to recognise these issues” (p. 550). The authors call this view “a mistake” and call for “a cultural transition” in how chemists and materials developers are educated (pp. 545, 550–551). - Proponents and governments: see nanotechnology as a strategic race (“everyone wants to be at the forefront”) and a potential solution to cancer, energy and clean water (pp. 545–546). - Unnamed critics of caution: the chapter attributes “risk research jeopardises innovation” and “regulation is bad for business” to “comments from some quarters”, without saying whose (p. 548). - Regulators: hold chemistry-based identity models (p. 543) and incrementalist instincts: tweak before overhauling (NICNAS, p. 541); “no fundamental overhaul” (Commission, p. 539); keep the status quo (p. 541). Their default is to call for more information (p. 547). Some regulatory scientists took a pragmatic line: define a class for attention “whether they are benign or hazardous” (Stamm; p. 540). - Industry: mixed. There are voluntary codes (p. 542), a joint NGO–industry framework (p. 549), and inhalation studies that confirmed hazards (p. 535). But few firms responded to voluntary reporting (p. 537), registrants split on classification (p. 538), and industry science offered alternative mechanistic interpretations. That last point is my inference from the Pauluhn (2010) title in the reference list (p. 557); the chapter doesn’t discuss it. - NGOs: watchdogs (Denison’s testimony, p. 546; ChemSec, p. 540; PAN Europe, p. 548) and also partners (Environmental Defense with DuPont, p. 549). - Publics: absent from risk decisions (p. 545). The BSE analogy suggests public values can reshape acceptability suddenly (p. 545).

10. The politics of reform windows. - Legislative reform competes with other agendas: in the REACH review, dossier quality, red tape and small-firm burden (p. 539). - Failure in one venue (Novel Foods) predicts difficulty in higher-stakes ones (p. 539). - Fixed review cycles set the tempo (2012, then 2019; p. 540). - Both nano-relevant recasts were started for other reasons. In the cosmetics recast, nano provisions were added; in the Novel Foods recast, the same route failed (p. 539). So a reform started for other reasons can carry nano provisions, but it doesn’t guarantee them.

11. Definitions as governance battlegrounds. - Definitions decide what gets attention. - Competing definitions (NNI, ISO, Cosmetics, the Commission’s) produce inconsistency (pp. 531–532, 539–540). - A size-based policy definition may miss hazard-relevant properties (Maynard; ChemSec). Without some boundary, nothing is triggered (Stamm) (p. 540). - The Commission’s definition lists competitiveness, alongside environment, health and safety, as a reason to lower the 50% threshold (p. 540). Lowering the threshold widens coverage. My inference: economic considerations are written into the very act of deciding what counts as the regulated object. The authors don’t comment on this.

12. Voluntarism and “the shadow of regulation”. Voluntary codes and frameworks multiplied where hard law lagged (pp. 542, 549). They “sit within the shadow of formal regulatory obligations” (p. 542). Where voluntarism was tested by a measurable output, disclosure, it largely failed (p. 537).

13. Lock-in and the design window. The chapter’s timing argument: - intervention is easiest at the design stage, because “economic interests are not fully entrenched at that point” (p. 547); - thinking about alternatives at the design stage “did not occur” for nanotechnology (p. 546); - design rules for integrating EHS were missing (pp. 530, 550).

14. Idealised versus real-world use. Assumptions of small quantities and sealed processes contrast with dusty powder-handling workplaces (pp. 544–545), and with the PCB history of “closed” uses leaking into the environment (p. 545).

15. Irreversibility through dispersal. For persistent materials, release is itself irreversible, because the material can’t be recovered (p. 544). Natural analogues show persistence over geological time (p. 544).

16. Distribution of risks and benefits. - Benefits go to firms, investors and states in a competitive race (p. 546), and to consumers through products (p. 533). - Risks fall on workers (p. 544), consumers with direct exposure (p. 535), aquatic ecosystems and wastewater infrastructure (p. 536), and potentially future generations (“legacy of harm”, p. 548).

The chapter says little about distribution beyond these lists, and quantifies nothing.

17. Anticipatory talk as a substitute for action. The chapter’s most distinctive dynamic. Nanotechnology had unusually early and extensive risk discussion (p. 542), yet action was slow (pp. 542–543, 547–548), and “calls for more information” worked as “a substitute for action” (p. 547). Table 22.1 shows the texture: lessons about knowledge and uncertainty were widely noted, while lessons about power, alternatives and participation were rarely noted. No document scores “+++” (strategy). But section 22.8 shows that several of these bodies did propose strategies, so the gap the chapter identifies is between recommendation and implementation by “political decision-makers” (p. 551), not an absence of proposals.

Framing and language to note: - “wonder technologies” (p. 530); - “heralded” (p. 530); - “first flush of nano-enthusiasm” (p. 545); - “nano-fever” (p. 546); - “solutions looking for a problem” (p. 546); - “uncanny resemblance” (p. 544); - “the sorry tale of asbestos” (p. 545); - “a classic regulatory paradox” (p. 537); - “at best underwhelming” (p. 537); - “paralysed by analysis” (p. 547); - “legacy of harm” and “go back to the classroom” (p. 548).

Unattributed remarks “from some quarters” are quoted and rejected as clouding the waters: “risk research jeopardises innovation”, “regulation is bad for business” (p. 548). The authors’ own positive frame is “responsible development” and “precautionary yet socially and economically responsive” (pp. 530, 551). This presents precaution as a way to secure benefits, not as a brake.

Transferable insights (technology-neutral)#

1. Legacy identity categories can make new variants legally invisible. When a regime identifies regulated objects by a legacy key and triggers scrutiny by volume, new forms whose behaviour depends on other properties inherit the old status and escape specific assessment. Evidence: CAS-based identity under TSCA and REACH; nanosilver versus silver; tonnage thresholds; “regulatory inheritance” (pp. 537–541, 543). Strength: strong for the descriptive claim, since these are documented features of the regimes, confirmed by several independent reviews (p. 540). The harm consequence is inferred, not shown.

2. Voluntary disclosure schemes produce little data, which leaves regulators blind to scale and use. Evidence: 13 submissions in two years (DEFRA) and 31 organisations (EPA); producers “rarely” publish volumes; jurisdictions moved to mandatory schemes (pp. 534, 537). Strength: strong for these cases (quantified and uncontested); moderate as a generalisation (a small number of schemes). Caveat: the chapter also calls the schemes “somewhat onerous in operation” (p. 537), so poor design may explain the low response as much as voluntariness does. The evidence doesn’t separate the two.

3. If the regulator must prove risk before it can compel data, the data may never come. Evidence: the TSCA paradox (p. 537); the EPA Inspector General on reliance on industry data (p. 538); the contrast with “no data, no market” (p. 538) and the burden on industry in the cosmetics and biocides rules (p. 547). Strength: moderate–strong. The structural logic is clear and supported by an official audit, but the chapter shows no comparison of outcomes.

4. Reversing the burden of proof only works if the regulated object is clearly defined. Ambiguity lets regulated parties choose the classification that suits them. Evidence: CNT registrants split between “new substance” and “graphite” (p. 538); the same-or-different problem under REACH (p. 538). Strength: suggestive. One telling example; motives are not established.

5. Products and variants multiply faster than safety evidence, and regulators take up that evidence later still. Evidence: Linkov’s gap (Fig. 22.2); Lawless’s one to two decades; the 2001 volume’s decades; “twenty years” since first indications (pp. 547, 549). Strength: moderate. The direction is well supported, but the figure is schematic, Lawless is unreferenced, and “not less than two decades” overstates the source.

6. Novelty can defeat the assessment toolkit itself, so acknowledging uncertainty doesn’t prevent blind spots. Dose measures, assumptions of monotonic dose–response, persistence criteria and test routes may all be miscalibrated for the new thing. Evidence: p. 541; the instillation-versus-inhalation differences (p. 535); “inherently will generate blind spots” (p. 543). Strength: moderate. Supported by expert scientific committees (SCENIHR) and some data, but contrary results exist (p. 534).

7. Combinatorial variety strains case-by-case assessment. Screening or grouping by property, use and exposure is needed. Evidence: a claimed 50,000 possible SWCNT variants, not all commercially relevant; the resource intensity of case-by-case work, which the chapter questions (“one has to wonder”) rather than rejects (pp. 541–542); proposed screening principles (p. 547). Strength: moderate for the problem; asserted for the proposed solution.

8. Talking early about risks is not the same as acting early. Anticipatory discussion can coexist with, and even stand in for, slow action. Calls for “more information” become a way of not acting. Evidence: unusually early risk discourse (p. 542) versus slow, patchy action (pp. 542–543, 547–548). In Table 22.1, knowledge lessons are widely noted and governance lessons rarely. Expert recommendations existed but were “not or have yet to be successfully implemented by political decision-makers” (p. 551). Strength: moderate. The authors’ judgement is backed by the regulatory chronology, but Table 22.1’s method is undocumented and its “+++” coding is inconsistent with the text (see 22.7).

9. When the same body promotes a technology and oversees its risks, risk research tends to be subordinated, and, the authors suggest, “perhaps more insidiously” through agenda-setting. Evidence: the NNI’s dual role; Denison’s testimony; the NAS 2012 call to separate management and budgets; the BSE precedent (pp. 546–548). Strength: moderate. Authoritative and plausible, but no specific distorted decision is shown.

10. Idealised assumptions about controlled use diverge from real production, and persistent materials escape containment. Evidence: the Weiss workplace description; the PCB analogy (pp. 544–545). Strength: suggestive. One journalistic observation plus a historical analogy.

11. Developer communities’ mental models can resist integrating safety. Examples are belief in one-way knowledge diffusion, the view that risk concerns are “invalid science”, and a lack of training. Evidence: Powell 2007; the authors’ claim about training (pp. 545, 550). Strength: suggestive. A single survey, missing from the references, and assertion.

12. Pushing a technology’s solutions crowds out comparison by function. The right question is the best way to meet a need, not how to use the new capability (“could versus should”). Evidence: pp. 545–546; the alternatives row of Table 22.1 is completely blank (p. 548). Strength: suggestive. Argued by analogy; no case of a better alternative displaced by nanotechnology is shown. The blank row does show institutional neglect.

13. The easiest time to change course is at design, before economic interests become entrenched (“economic interests are not fully entrenched at that point”, p. 547). Evidence: pp. 547, 550–551; no design rules existed (p. 530). Strength: asserted within this chapter. It rests on logic and the earlier volume, not on nano evidence.

14. Definitions are governance instruments and are contested. A policy bright line can miss what matters, yet some boundary is needed to trigger attention. Economic considerations can be written into where the boundary sits: the EC definition lets competitiveness, as well as environment, health and safety, justify lowering the threshold. That last point is my inference; the authors don’t make it. Evidence: the 2011 EC definition dispute; the competitiveness clause; inconsistent definitions across instruments (pp. 539–540). Strength: moderate. A documented controversy, with both sides presented.

15. Reform windows are scarce and crowded. A failure in one instrument signals difficulty in others, and fixed review cycles can push effective rules years out. Evidence: the Novel Foods failure; the REACH 2012 review being played down; the 2019 horizon (pp. 539–540). Strength: moderate. Documented events, with a predictive element to check.

16. Dispersal alone can make a decision irreversible. Persistent materials released at scale can’t be recovered. Evidence: Box 22.1 (p. 544). Strength: moderate. The persistence evidence from natural analogues is strong for C60 and CNTs, but the criterion applies so broadly that it discriminates poorly.

17. Both unmet promises and neglected risks erode trust. Heavy public investment raises the stakes of each. Evidence: p. 545, drawing on the BSE lesson. Strength: asserted. No evidence is offered within the chapter.

18. Industry-generated science can supply early warnings, not only contest them, even while offering different interpretations of mechanism. Evidence: BASF and Bayer inhalation studies confirmed inflammation (p. 535); the title of Pauluhn (2010) argues against the fibre mechanism (reference list, p. 557). Strength: suggestive. This is my inference; the authors don’t draw it.

19. A new technology acts as a stress test that reveals weaknesses already present in the regulatory system. Evidence: nanomaterials “highlight many of the deficiencies that have existed for some time” (p. 537); the cross-cutting review findings (p. 540). Strength: moderate. Reasonable, and supported by multiple reviews.

Limitations, contestation and bias check#

1. Genre: a prospective audit, not a case of realised harm. - Unlike the historical chapters, there is no outcome against which to test the lessons. The 2001 lessons serve as a normative checklist, and any delay tends to read as “failure”, whether or not harm follows. - The chapter can’t tell apart three possibilities: - a warning that will prove real; - one that will be averted by action; - a false positive. - Hansen and Tickner discuss false positives in general in Ch 2 of the report, but not here. The chapter doesn’t consider what an over-reaction would have cost.

2. Hazard versus risk, and escalating language. - The evidence section is careful: “preliminary”, “arguably”, “not conclusive” (p. 536). - The regulatory section escalates to “rapidly increasing evidence of risks” (p. 539), and the lessons section to “act on what we know now” (p. 547). - Most of the evidence concerns hazard, from animal studies with high doses and non-physiological routes (acknowledged, p. 535). Exposure is largely unknown (pp. 534–535). The tumour evidence is a single rat study (p. 534). - Contrary findings (Warheit 2006; Sayes 2007; the ionic explanation for silver) are presented (pp. 534, 536) but not weighed. - Background: the rat lung-overload debate over TiO2 is not mentioned. - The claim that ultrafine particles are more dangerous the smaller they are (p. 534) is stated too strongly for the epidemiology of the time.

3. Treating “nanotechnology” as a single thing. - The chapter’s own Figure 22.1 shows great heterogeneity, and the evidence covers three materials. - Yet the lessons and verdicts are applied to “nanotechnology” as a whole. - The chapter partly guards against this (“some nanomaterials”), but the Box 22.1 criteria (novel by definition; irreversible because unrecoverable) would flag almost any new dispersive material.

4. Missing voices. - There are no panels and no response from industry, regulators or developers. - Developers’ views are summarised, not heard, and then dismissed (“This is a mistake in our view”, p. 545). - Benefits are listed but not assessed. The costs of regulation and the effects on innovation and small firms appear only as items competing for REACH’s attention (p. 539). - The chapter acknowledges once that risk speculation might overshadow real benefits (p. 545), and says that weighing likely pros and cons is “vital”, but it doesn’t develop that side of the argument.

5. The authors’ proximity to the field. - About a quarter of the references are the authors’ own work. - Chapter authors co-produced: - some of the warning evidence (Poland 2008 and Aitken 2009, both with Maynard; the ENRHES review, Stone et al. 2010, with Baun and Hansen); - the research agenda used as the benchmark example (Maynard et al. 2006); - the categorisation framework (Hansen et al. 2007); - the alternatives framework (Rossi, Tickner & Geiser 2006); - the lag argument (Linkov et al. 2009, with Hansen). - Background: the product data come mainly from PEN’s inventory, and Maynard was PEN’s chief science adviser. - One author’s view is presented in the third person as one side of the definition dispute (Maynard 2011, p. 540). - None of this is illegitimate; these were central figures in the field. But the chapter is partly the field’s leaders assessing their own field, with little external challenge.

6. Internal inconsistencies and errors (for accuracy when citing): - Probable unit errors: “50 mg” in Poland et al. (probably 50 µg); seawater PNEC “0.86 mg/L” (probably µg/L). Both confirmed as printed (pp. 535, 536). - EC50 is defined wrongly (p. 536). - Table 22.1: - “10 reports” but 11 columns; - “IRGB”/”ICRG” for IRGC; - no “+++” despite the legend, although 22.8 describes strategic recommendations from several of the same bodies; - Biocides is scored blank on regulatory independence, though the text credits it (pp. 547–548). - Table 22.2 dates RS & RAE to 2003, against 2004 in the text, and omits the 1990 and 1995 warnings the text relies on (p. 551). - The “two decades” claim: presented as a floor, but its sources give “one or two decades” (Lawless) and “decades” (EEA 2001), and the chapter itself calls them “plausible worst-case scenarios” (p. 549). - Production figures conflict (900 t in 2003 versus 65 t in 2006; p. 534). - Missing references for Powell and Lawless, among others. - Query: the “IIIc and IIId” categorisation (p. 533). - Temporal layering: - text written around 2008–2011 is presented unchanged in a 2013 report: “will come into effect in 2012” (p. 539); the REACH 2012 review discussed in the future tense (pp. 538–540); - the Maynard et al. (2006) research milestones are reported (p. 543), but the chapter doesn’t check whether the ones that had fallen due by 2012–13 were met. The 12-month and 5-year targets, for example, had passed; - background: the Commission published its Second Regulatory Review on Nanomaterials in October 2012, before the report appeared (May 2013), and the chapter doesn’t mention it.

7. Omissions (from background knowledge, not verified here): - In the US, antimicrobial nanosilver products fall mainly under pesticide law (FIFRA), not only TSCA. The chapter discusses nanosilver only under TSCA (p. 537). - IARC’s 2006 classification of TiO2 as Group 2B is not mentioned.

8. Credit where due: the chapter is more balanced than it might have been. - It presents contrary toxicology (p. 534) and the method limits of instillation studies (p. 535). - It concedes that the regulatory lag is “not surprising” and not “unique to nanotechnologies” (p. 537). - It judges efforts “better than” for earlier technologies (p. 543). - It grants that “both camps have valid points” on the definition (p. 540). - It insists that lay and worker knowledge “needs as much critical appraisal as specialist knowledge” (p. 545). - It acknowledges that risk speculation can overshadow benefits (p. 545). - It frames precaution as enabling safe products (p. 547). - In its evidence section, it ties the asbestos-like findings specifically to long MWCNTs; short ones caused no significant inflammation (p. 535). The rhetorical use elsewhere (“the sorry tale of asbestos”, p. 545) is broader.

9. Foresight bias. The warnings were chosen by analogy with known disasters (asbestos, ultrafine particles). That is appropriate for horizon-scanning, but it selects for resemblance and may miss problems that don’t resemble past cases. That is exactly the “blind spot” risk the chapter itself raises (p. 543).

10. Later-evidence pointers (background knowledge; unverified here; for the hindsight strand): - REACH: - The Commission’s Second Regulatory Review on Nanomaterials (October 2012) kept REACH as the framework and proposed changes to its annexes. - Nano-specific REACH annex amendments were adopted as Commission Regulation (EU) 2018/1881 and applied from 1 January 2020. This partly bears out the chapter’s warning that nanomaterials might be “overlooked” in 2012, and its sense that effective rules would come late. - The EC definition was revised in 2022. - EU product law: - Nano labelling in cosmetics applied from July 2013. - The Novel Foods Regulation (EU) 2015/2283 covers engineered nanomaterials. - TiO2 as a food additive (E171) was banned from 2022, after EFSA’s 2021 opinion. - The EU classified TiO2 as a suspected carcinogen by inhalation (powder forms, 2019/2020). The EU General Court annulled that classification in 2022, and reportedly the Court of Justice upheld the annulment in 2025 (verify). - US: - TSCA reform (the Lautenberg Act, 2016). - A one-time TSCA section 8(a) reporting rule for nanoscale materials (2017). - CNTs: - NIOSH recommended exposure limit of 1 µg/m³ for CNTs and nanofibres (2013). - IARC (2014) classified one MWCNT type (MWCNT-7) as Group 2B, and other CNTs as Group 3. - National registers: Mandatory nanomaterial registers in France (from 2013), Belgium and Denmark. - Human harm: As far as I know, no clear population-level human harm from engineered nanomaterials had been documented by the mid-2020s. But epidemiology is limited, and the latency of fibre-type diseases is long. Whether nanotechnology became “another case study” remains open and needs careful checking.

Notable quotes#

1. “have we learnt the lessons of past ‘wonder technologies’ or are we destined to repeat past mistakes?” (p. 530)
2. “Such information is only available if the producers themselves make it available, which occurs rarely.” (p. 534)
3. “These are just the data the agency needs to determine whether the substance poses an unreasonable risk — a classic regulatory paradox.” (p. 537)
4. “Voluntary in nature, and somewhat onerous in operation, the schemes can be described as at best underwhelming.” (p. 537)
5. “Perhaps more than any preceding technology, the early development of nanotechnology has been characterised by discussions of potential risks and the need for regulatory reform” (p. 542)
6. “the R&D community is entrenched in the philosophy that basic research will ultimately solve real-world problems through a one-way process of knowledge diffusion” (p. 545)
7. “while nanotechnology could be used, it may be questionable whether it should.” (p. 546)
8. “Perhaps more insidiously, research and development decisions end up being influenced by what will ultimately promote the technology, rather than what will protect producers, users and the environment.” (p. 547)
9. “Yet many governments still call for more information as a substitute for action” (p. 547)
10. “the question here seems not to be whether we have learnt the lessons, but whether we are applying them effectively enough” (p. 548)

Open questions#

1. Did nanotechnology become “yet another future case study on how not to introduce a new technology” (p. 548)? Or does the absence (so far) of documented population-level harm count as a false positive, a success of anticipatory governance, or simply latency? How could these be told apart?
2. Was the prediction of a lag of at least two decades (p. 549) borne out? Measure from 1990/1992, or from 2004 (RS & RAE), to binding nano-specific requirements: REACH annexes applying in 2020, the TSCA 8(a) rule in 2017.
3. Were the Maynard et al. (2006) research milestones met? These were exposure instruments in 3–10 years, validated toxicity methods in 5–15 years, and predictive models in 10 years (p. 543). This is a clean, checkable test of the chapter’s benchmark.
4. Did the NNI ever separate promotion from EHS management and budgets, as NAS (2012) urged? If not, can any effect on the direction of EHS research be detected?
5. How were the CNT and TiO2 hazard questions resolved? Did the asbestos-like findings for long, rigid MWCNTs hold up in inhalation carcinogenicity studies? How did the TiO2 classification saga unfold, and what does the court challenge say about how precautionary classifications are contested?
6. Did the nanosilver question (particle versus ion) resolve, and did the regulatory route (biocides and pesticides law) turn out to matter more than chemicals law?
7. Did “green nanotechnology” and safe-by-design become embedded in practice (EHS as a quality criterion “equal to economic and performance considerations”; p. 551), or remain aspirational?
8. How was Table 22.1 coded, and by whom? Is there an underlying dataset? A re-coding of later documents would test whether governance lessons stayed neglected.
9. What changed between the 2008 Nature Nanotechnology article and this 2013 chapter? A comparison would separate the original argument from later updates. The article is the source for most of section 22.6.
10. Could Andrew retrieve the unreferenced sources? Powell (2007) (likely Health, Risk & Society 9(2)) and Lawless (1977, Technology and Social Shock) underpin two claims.
11. Is the Box 22.1 screen, as operationalised in the uncited Hansen et al. (2012), discriminating enough to be useful? Or does “novel by definition” flag everything?
12. What costs did the precautionary proposals (a single nano law, labelling, registers) turn out to impose, and on whom? The chapter doesn’t assess this.

Audit log#

Independent audit against the full text extract (PDF 532–562) and the rendered PDF. Tables 22.1 and 22.2, Figures 22.1 and 22.2, both unit values, the Annex 1 bios, the table of contents (Part C, pp. 429–560) and the reference list were all re-checked. Changes to the notes:

• Authors: Bowman’s role in section 22.5, previously stated as fact, is now flagged as inference, with the supporting citations listed.
• Maynard bio: Shvedova et al. 2005 is no longer called an “early warning” (the chapter cites it only for the residual-metal hypothesis). Only Maynard 2006 is a PEN publication. Added background on his PEN role and the chapter’s reliance on the PEN inventory.
• Genre: “deliberately flips” changed to “plays on”; the chapter placed in Part C, “Emerging issues” (checked against the table of contents).
• Self-citation: now 38 distinct works among roughly 160 entries, including chapters in edited volumes. “Several key” changed to “some”; “judged against” changed to “used as benchmarks”.
• 22.2: “by every measure” corrected to the source’s “a number of metrics”; added the Linkov citation and the “crucial role” nuance.
• 22.3: the “promotional language” remark now notes the authors’ hedges and is marked as my observation.
• 22.4(d): added that some studies found a more pronounced nano effect (Navarro 2008), as the source says.
• 22.4(d) analytical note: corrected the margin-of-safety claim. Against the chapter’s own freshwater PNEC (0.04 µg/L), modelled ng/L levels are about one order of magnitude below, not three.
• 22.5 Novel Foods: added the existing regulation’s requirements and what the revisions might have done; replaced “fights over” with the source’s “myriad of other issues”.
• 22.5: added that the fate of the Schlyter report was unclear, and that the authors think it added pressure.
• 22.5 EC definition: noted that the threshold can only be lowered; added “differs considerably” and the Novel Foods link; marked the Maynard 2011 wording as the chapter’s paraphrase; noted that ChemSec is an NGO (background).
• 22.5 case-by-case: restored the “it has been claimed” hedge on 50,000 SWCNT variants and the “not all commercially relevant” concession; softened “unworkable”.
• 22.6: the claim that this section follows the 2008 article most closely is now flagged as unverified.
• 22.6.1: “irreversible harm” changed back to the EEA’s term, “irreversible action”.
• 22.6.3 note: corrected. The Weiss anecdote supports the real-world-conditions point, not the one-way-diffusion claim, which rests on Powell alone.
• 22.6.6 note: rewrote the Denison/DuPont remark. An NGO–industry voluntary partnership is not a regulatory-independence case.
• 22.7: confirmed the Table 22.1 transcription cell by cell against the rendered page; added the text’s “national and multilateral” description.
• 22.7 reading 1: added a caveat. The empty “+++” column conflicts with the strategic recommendations described in 22.8, so it is not evidence that no one proposed a strategy.
• 22.7 reading 3: “coverage shrank over time” now carries a genre caveat (the 2009 documents are scientific reviews).
• 22.7 reading 4: corrected the “not Member State reports” critique in light of the text’s “national and multilateral”.
• 22.7 reading 5 (new): Biocides is scored blank on regulatory independence although the text credits it; Stone 2009 (table) versus 2010 (text).
• 22.7 regulations: “burden of proof” corrected to the source’s “burden of providing health and safety data”.
• 22.8 Lawless: broadened the critique to cover EEA 2001’s “decades” and the tension with “plausible worst-case scenarios”.
• 22.8 Nano Risk Framework: restored the printed wording of the six steps (“develop …”, flagged).
• Table 22.2: corrected “the only binding nano-specific rule is the Cosmetics Regulation”; added the Biocides, NICNAS and SNUR points and three minor table–text discrepancies.
• References: added more cited-but-missing items (Ludlow 2007, Biswas & Wu 2005, BSI 2007b, Sung 2009, Davies 2009, NICNAS 2008, the Igami & Okazaki versus Saka mismatch, the Kroto date) and more listed-but-uncited items (Denison 2007b, Technology Transfer Center 2007), plus duplicate entries.
• Timeline: the 1990 row’s dose and route marked as background; “First binding nano-specific EU rule” softened; the dates of the Novel Foods failure and the Biocides Regulation marked as background.
• Lag analysis: CNT responses expanded (REACH carbon/graphite exemption, RS & RAE); NIOSH’s non-binding status marked as background; the cosmetics timing clarified against the chapter’s “2012”.
• Recommendations (B): “endorsing” separation changed to “implicit”; the REACH-revision scope now quoted accurately; added the interdisciplinary-infrastructure and research-gap recommendations (p. 543).
• Mechanism 1: “USD 18 billion a year” corrected to a 2008 estimate that the chapter calls “somewhat speculative”.
• Mechanism 2: “almost no one” changed to “few”; replaced an inference (“can’t act on exposure you can’t measure”) that contradicted the authors’ “act on what we know now”.
• Mechanism 4: “the cure” corrected to the source’s “helps to overcome … in theory”.
• Mechanism 7: the Ludlow claim restated in the source’s wording (“likely to be perceived as a period of under-regulation”).
• Mechanism 9: softened “core cultural obstacle”; moved the anti-caution quotes from “proponents and governments” to the unnamed “some quarters” the source gives; labelled the Pauluhn inference as mine.
• Mechanism 10: corrected an over-generalisation from the cosmetics recast (the Novel Foods recast is the counterexample).
• Mechanism 11: added the threshold-lowering nuance and labelled the inference as mine.
• Mechanism 17: corrected “no document proposed a strategy”.
• Insight 2: added the source’s own alternative explanation for low uptake (the schemes were “onerous”).
• Insight 7: “unworkable” changed to “strains”, with the “claimed” hedge.
• Insight 8: removed “no strategy appears anywhere”; tied the insight to the recommendation–implementation gap (p. 551).
• Insight 9: “above all” changed to the source’s “perhaps more insidiously”.
• Insight 13: “cheapest” changed to “easiest”, with the source quote.
• Insight 14: added the threshold nuance and labelled the inference as mine.
• Limitations 4: “doesn’t take that seriously” changed to “doesn’t develop”.
• Limitations 5: corrected the co-production list (Shvedova removed; ENRHES/Stone 2010 added); added the PEN background.
• Limitations 6: added the Table 22.1 inconsistencies and restated the two-decades point. Replaced the mis-flagged “within 12 months” layering example (it reports a 2006 call) with the point that the elapsed milestones were never assessed.
• Limitations 8: removed “rigid”, which is not in the source.
• Framing language: “Opponents’ rhetoric” changed to unattributed remarks “from some quarters”.
• Analytical note on industry toxicology: the cross-chapter comparison is flagged to check at synthesis.
• Verified without change: all ten notable quotes are verbatim; the Annex 1 bios and page references are accurate; both probable unit errors are confirmed as printed; the Figure 22.1 and 22.2 descriptions are accurate; no contemporary-technology content was found.
• Digest: corresponding fixes, namely removing “the report’s clearest prospective case”, correcting the Cosmetics “one binding rule” claim, adding the PNEC margin, adding the “+++” caveat, correcting the advocacy wording, softening the insight-table wording and correcting the caveats (Stamm as a counter-voice, Ludlow 2007, the two-decades nuance).