Sources for the standardization-by-layer chart — physics and nutrition

• 26th CGPM (2018), Resolution 1 — the revised SI, effective 20 May 2019, defined by seven fixed constants (Δν_Cs, c, h, e, k, N_A, K_cd)readEffective date and the seven constants read off the resolution page on 2026-08-22.
• BIPM, The International System of Units (SI), 9th edition (2019, updated 2026)read

The layer-0 ceiling, stated precisely. Three facts get conflated here and they take three different levers. (1) The molecule is stable — lycopene has one structure, one mass, one ChEBI identifier; at the analyte level the measurand is as good as physics's, which is why the deficiency diseases closed. (2) A specific tomato is measurable — AOAC methods plus NIST food-matrix SRMs already give a value with an uncertainty on a named sample; that is layer 1 and it works. (3) 'Tomato' is not a measurand, it is a category — the real object is a distribution over cultivar, soil, ripeness, season, storage and cooking. Publishing one number against the name is the mass of a rock: no physicist would put that in CODATA. So the variation is layer 0 and irreducible, but the damage is layer 3 (reference data reports a point where a distribution belongs) and partly layer 2 (the identifier 'tomato' is too coarse; FoodOn has cultivar-level terms, composition tables rarely use them).

• Novotny, Gebauer & Baer, Discrepancy between the Atwater factor predicted and empirically measured energy values of almonds, Am J Clin Nutr 96, 296–301 (2012) — 32% overestimationread10.3945/ajcn.112.035782Abstract read via Europe PMC 2026-08-31; the paper states the figure in its own words: "The Atwater factors, when applied to almonds, resulted in a 32% overestimation of their measured energy content." The label matches the source exactly.
• Baer, Gebauer & Novotny, Walnuts consumed by healthy adults provide less available energy than predicted by the Atwater factors, J Nutr 146, 9–13 (2016) — measured energy 21% BELOW the Atwater predictionread10.3945/jn.115.217372Abstract read via Europe PMC 2026-08-31: "The ME of the walnuts was 21% less than that predicted by the Atwater factors." WORDING CORRECTED: the label previously read "21% overestimation", which flips the denominator — the paper measures the shortfall against the PREDICTION, not the overshoot against the measured value. The almond paper genuinely says "32% overestimation"; this one does not, and the two must not be described in the same terms just because they sit side by side.
• Baer & Novotny, Metabolizable energy from cashew nuts is less than that predicted by Atwater factors, Nutrients 11, 33 (2018)crossref10.3390/nu11010033
• Harnack LJ. Dietary assessment tools and food databases are needed to support research on the health effects of ultra-processed foods and food additives. J Acad Nutr Diet, online 2026-08-29 (sole author; byline: Director, Nutrition Coordinating Center)read10.1016/j.jand.2026.156840L0 claim it sources: for additives the measurand itself is capped by labeling law — amounts need not be declared (21 CFR 101.4), so 'frequency of ingestion of foods containing a specific substance may be estimated but not the amount of the substance ingested' (pre-proof L154-160). Read in full 2026-09-01 from the CC BY-NC-ND journal pre-proof; line numbers are pre-proof — re-verify against the Version of Record.

• JCGM 100:2008 — Guide to the expression of uncertainty in measurement (GUM); Amd.1:2026 on nonlinearity in measurement modelsreadCurrent edition confirmed from the BIPM JCGM publications index on 2026-08-22.
• ISO/IEC 17025:2017 — General requirements for the competence of testing and calibration laboratoriesreadiso.org refuses automated HEAD requests (403); title, year and standard number confirmed by search on 2026-08-22.
• CIPM MRA (1999) — mutual recognition of national measurement standards and calibration certificates200
• ILAC — international laboratory accreditation cooperation200

• AOAC INTERNATIONAL, Official Methods of Analysis, 22nd edition (2023) — 3,000+ validated methods; the eight coexisting dietary-fibre methodsunverifiedaoac.org refuses automated requests (403). Edition and method numbers are as recorded in the Stack Audit and are NOT independently verified here.
• Codex Alimentarius dietary fibre definition — agreed at CCNFSDU 30 (Nov 2008), adopted by the Commission at its 32nd session (2009); permits multiple methods for one analyteunverifiedPortal reachable; the specific session dates are from the Stack Audit and are not independently verified here.
• USDA FoodData Central Foundation Foods — the one data type that records analytical approach, sample count, location and collection date200
• Harnack LJ. Dietary assessment tools and food databases are needed to support research on the health effects of ultra-processed foods and food additives. J Acad Nutr Diet, online 2026-08-29 (sole author; byline: Director, Nutrition Coordinating Center)read10.1016/j.jand.2026.156840L1 claim it sources: the field's instruments are additive-blind — the NCC Food and Nutrient Database and NDSR include non-nutritive sweeteners, sugar alcohols and caffeine 'but no other food additives' (L44-46); the toolmaker grades the instruments: records/24-hour recalls 'well suited', FFQs 'generally poorly suited' (L164-181). Read in full 2026-09-01 from the CC BY-NC-ND journal pre-proof; line numbers are pre-proof — re-verify against the Version of Record.

• Particle Data Group, Monte Carlo particle numbering scheme (Review of Particle Physics, 2024)200
• Navas et al. (Particle Data Group), Review of Particle Physics, Phys. Rev. D 110, 030001 (2024)crossref10.1103/PhysRevD.110.030001
• Gazetteer of Planetary Nomenclature — the IAU-authorized naming register200Cited in place of iau.org, which returns 404 to automated requests.
• IUPAC nomenclature — identity for chemical species200

• GS1 Global Trade Item Number (GTIN) — solves packaged-food identity; nothing equivalent exists for 'a raw apple'unverifiedgs1.org refuses automated HEAD requests (403).
• Jarvis et al., Periodic Table of Food Initiative, Nature Food 5, 189–193 (2024) — a curated list of 1,650 foods, published because no canonical list existedcrossref10.1038/s43016-024-00941-y
• EFSA, The food classification and description system FoodEx2 (revision 2), EFSA Supporting Publication 2015:EN-804crossref10.2903/sp.efsa.2015.EN-804
• Harnack LJ. Dietary assessment tools and food databases are needed to support research on the health effects of ultra-processed foods and food additives. J Acad Nutr Diet, online 2026-08-29 (sole author; byline: Director, Nutrition Coordinating Center)read10.1016/j.jand.2026.156840L2 claim it sources: ingredient identity fails without curation — FD&C Blue No. 1 is listed as patent blue AE, brilliant blue FCF, D&C blue no. 1, or C.I. food blue no. 1, and 'without data curation it is not possible to know with confidence which foods contain an additive of interest' (L54-60). Read in full 2026-09-01 from the CC BY-NC-ND journal pre-proof; line numbers are pre-proof — re-verify against the Version of Record.

• Tiesinga, Mohr, Newell & Taylor, CODATA recommended values of the fundamental physical constants: 2018, Rev. Mod. Phys. 93, 025010 (2021)crossref10.1103/RevModPhys.93.025010
• Mohr, Newell, Taylor & Tiesinga, CODATA recommended values of the fundamental physical constants: 2022, Rev. Mod. Phys. 97, 025002 (2025)crossref10.1103/RevModPhys.97.025002
• NIST Standard Reference Data200
• NIST CODATA Fundamental Physical Constants portal200
• ISO 80000-1:2022 — Quantities and units, Part 1: General (2nd edition, Dec 2022)readiso.org refuses automated HEAD requests (403); edition and year confirmed by search on 2026-08-22.
• PhySH — Physics Subject Headings (APS, unveiled January 2016), which replaced PACS; PACS was last maintained in 2010read

The removable half of the tomato problem. Layer 0's variation cannot be engineered away, but the downstream harm can: carry mean ± SD, n, sampling year, cultivar or origin where known, and the database version. That is the physics move — you do not eliminate the uncertainty, you declare it and carry it. USDA FoodData Central's Foundation Foods already publishes that shape (n, SD, min/max, year acquired) for a small subset: the right format, thin coverage, which is a maturity-high / adoption-low row by construction. The 20–45% cross-database variance on this layer and the Davis-style 1950-to-1999 drift are the same finding at two timescales — different countries' tomatoes really do differ, and so did 1950's.

• Wood, Barber, Scruggs et al., Value assignment of SRM 1849b infant/adult nutritional formula, NIST SP 260-233r1 (2024)crossref10.6028/NIST.SP.260-233r1CORRECTION: the Stack Audit attributes this to 'Phillips et al., 2024'. Crossref returns Wood, Barber, Scruggs et al.
• USDA FoodData Central — the databases where the traceability chain is absent200
• FAO/INFOODS — guidelines on units and component identifiers (tagnames)200
• Davis, Epp & Riordan, Changes in USDA food composition data for 43 garden crops, 1950 to 1999, J Am Coll Nutr 23, 669–682 (2004)crossref10.1080/07315724.2004.10719409CITE THE PAPER, NOT THE POPULAR VERSION. Verified against PubMed (PMID 15637215) on 2026-08-22. Of 13 nutrients, 6 showed statistically reliable declines (protein, Ca, P, Fe, riboflavin, ascorbic acid), median declines 6%–38%; the other 7 showed no reliable change, individual R-values are 'usually not distinguishable from 1 with current data', and about 28% of R-values EXCEED 1. The authors attribute any real decline to cultivar change and yield/nutrient trade-offs. The 1950 record carries no method, cultivar or sampling metadata, which is why real change cannot be separated from measurement change — the strongest argument in the corpus for making database-plus-version a required field (rows N05, R01).
• Dooley, Griffiths, Gosal et al., FoodOn: a harmonized food ontology, npj Science of Food 2, 23 (2018) — 24,000+ food-product termscrossref10.1038/s41538-018-0032-6The Stack Audit writes 'Dooley, Hsiao et al.'; Crossref returns Dooley, Griffiths, Gosal as the first three authors.
• Andrés-Hernández et al., Compositional Dietary Nutrition Ontology (CDNO), Crop Science 60 (2020) — 1,000+ component termscrossref10.1002/csc2.20092
• Dooley, Andrés-Hernández, Bordea et al., OBO Foundry food ontology interconnectivity, Semantic Web 15 (2024)crossref10.3233/SW-233458CORRECTION: the Stack Audit attributes this to 'Griffiths et al.'. Crossref returns Dooley, Andrés-Hernández, Bordea.
• FoodOn project site200
• CDNO in the OBO Foundry200
• Davis, Epp & Riordan, Changes in USDA food composition data for 43 garden crops, 1950 to 1999, J Am Coll Nutr 23, 669–682 (2004)crossref10.1080/07315724.2004.10719409CITE THE PAPER, NOT THE POPULAR VERSION. Verified against PubMed (PMID 15637215) on 2026-08-22. Of 13 nutrients, 6 showed statistically reliable declines (protein, Ca, P, Fe, riboflavin, ascorbic acid), median declines 6%–38%; the other 7 showed no reliable change, individual R-values are 'usually not distinguishable from 1 with current data', and about 28% of R-values EXCEED 1. The authors attribute any real decline to cultivar change and yield/nutrient trade-offs. The 1950 record carries no method, cultivar or sampling metadata, which is why real change cannot be separated from measurement change — the strongest argument in the corpus for making database-plus-version a required field (rows N05, R01).
• Brinkley, Gallo-Franco, Vázquez-Manjarrez et al., The state of food composition databases: data attributes and FAIR data harmonization in the era of digital innovation, Frontiers in Nutrition 12, 1552367 (2025)read10.3389/fnut.2025.1552367First author confirmed as Sarah Brinkley against Crossref on 2026-08-31; an earlier source-check on this project reported a conflicting first author for this DOI and was wrong. FULL TEXT READ 2026-08-31 (17 pp). Figures verified verbatim: "an integrative review of 35 data attributes across 101 FCDBs from 110 countries"; "only 17 databases (17%) satisfied all 13 FAIR criteria"; "every database included in this study met the findable criteria"; aggregated Accessibility, Interoperability and Reusability of 30, 69 and 43% respectively. THIS READ CORRECTED THE LAYER: the rationale had claimed FCDBs are findable far more often than interoperable, but interoperability at 69% is the second-strongest of the four, not a weak point. Accessibility (30%) and reusability (43%) are the weak ones, and the paper's own conclusion names "machine-readability and data reusability" as the gap. LOCATION OF THE 17/101 FIGURE, for citation: it is stated verbatim in Brinkley's Results, in the FAIR-principles subsection, as "Notably, only 17 databases (17%) satisfied all 13 FAIR criteria evaluated in this integrative review", immediately followed by "All of these FCDBs are accessible through web interfaces and are associated with High-Income countries (Table 1)." It is NOT in the abstract, which is why a reviewer working from secondary sources could not confirm it. It is the authors' own count, not a recount by this project, and the 17 databases are identified in Brinkley's Table 1.
• Hakel-Smith & Lewis, A standardized nutrition care process and language are essential components of a conceptual model to guide and document nutrition care and patient outcomes, J Am Diet Assoc 104, 1878-1884 (2004)crossref10.1016/j.jada.2004.10.015The concession source. Clinical dietetics DOES have a standardized process and controlled language (the Nutrition Care Process and NCPT), and it is the strongest counter-example to a flat reading of this chart. It is recorded here because the L3 score must survive it: the NCP standardises the clinical encounter record, not the identity or composition of the food, so it raises L3 without touching L0 or L2. A reviewer will raise this; the paper should raise it first.
• EN 16104:2012 — Food data: structure and interchange format (CEN/TC 387, published 3 November 2012)readThe CEN catalogue entry returns HTTP 500 to automated requests; standard number, title, committee and publication date confirmed on 2026-08-31 against EuroFIR, which developed it. The normative text is a paid CEN document and has NOT been read — the L3 rationale cites its existence and reach, never its clauses. This is the register's one money-gated item; pricing it is tracked as V1/V2.
• LanguaL — the international framework for food description (faceted thesaurus, 14 facets)unverifiedlangual.org returns HTTP 455 to automated requests. Named in the L2 and L3 rationales and in Table 1's v1.3 basis line for Identity, so it is recorded here rather than left as an unbacked name in the chart. NOT independently verified in this pass.
• Harnack LJ. Dietary assessment tools and food databases are needed to support research on the health effects of ultra-processed foods and food additives. J Acad Nutr Diet, online 2026-08-29 (sole author; byline: Director, Nutrition Coordinating Center)read10.1016/j.jand.2026.156840L3 claim it sources: the reference-data layer is missing — the only comprehensive US food-substance inventory (FDA Substances Added to Food: ~4,000 substances, 43 technical-effect categories) lacks chemical-composition classes and self-affirmed GRAS entirely (L125-153), and ingredient databases (Mintel GNPD, Nutritionix, USDA Branded Foods) are unparsed and unlinked to research assessment tools (L47-64). Read in full 2026-09-01 from the CC BY-NC-ND journal pre-proof; line numbers are pre-proof — re-verify against the Version of Record.

• arXiv (1991) — the universal preprint record200
• INSPIRE-HEP — the literature and data index for high-energy physics200
• Maguire, Heinrich & Watt, HEPData: a repository for high energy physics data, J. Phys. Conf. Ser. 898, 102006 (2017)crossref10.1088/1742-6596/898/10/102006
• HEPData — the numbers behind published HEP plots200
• CERN Open Data Portal200
• FITS Standard 4.0 — approved by the IAU FITS Working Group 22 July 2016, released 2018read
• Pence, Chiappetti, Page, Shaw et al., Definition of the Flexible Image Transport System (FITS), version 3.0, A&A 524, A42 (2010)crossref10.1051/0004-6361/201015362This DOI is FITS version 3.0, not 4.0 — do not relabel it.

• W3C PROV-O — the provenance ontology (W3C Recommendation, 30 April 2013)200
• Chhetri, Halchenko, Jarecka et al., A survey of provenance, assertion and evidence ontologies, Companion Proc. ACM Web Conference 2025 — 23 ontologies reviewed, none nutrition-specificcrossref10.1145/3701716.3715483
• Groth, Gibson & Velterop, The anatomy of a nanopublication, Information Services and Use 30, 51–56 (2010)crossref10.3233/ISU-2010-0613
• Wilkinson et al., The FAIR Guiding Principles, Scientific Data 3, 160018 (2016)crossref10.1038/sdata.2016.18
• EuroFIR — the quality-index framework, the nearest nutrition-native provenance practice200

• EQUATOR Network reporting-guideline library — searched for a physics equivalent of CONSORT/STROBE; none is registered200WEAKEST CELL ON THE PHYSICS SIDE. This is an absence claim resting on a negative search of one registry. Absence from EQUATOR is not proof no such guideline exists.
• Lyons, Discovering the significance of 5 sigma, arXiv:1310.1284 (2013) — the discovery convention that substitutes for a checklist200
• Klein & Roodman, Blind analysis in nuclear and particle physics, Annu. Rev. Nucl. Part. Sci. 55, 141–163 (2005)crossref10.1146/annurev.nucl.55.090704.151521

• Lachat, Hawwash, Ocké et al., STROBE-nut, PLOS Medicine 13, e1002036 (2016) — 24 nutrition-specific additions to STROBEcrossref10.1371/journal.pmed.1002036
• STROBE-nut project site200
• Brazma, Hingamp, Quackenbush et al., MIAME, Nature Genetics 29, 365–371 (2001) — the counterexample: Nature gated publication on it from December 2002crossref10.1038/ng1201-365
• EQUATOR Network — where STROBE-nut is curated, and endorsed without being enforced200

• CODATA least-squares adjustment — the methodology that forces every measurement into one consistent set (Rev. Mod. Phys. 93, 025010 and 97, 025002)crossref10.1103/RevModPhys.97.025002
• Particle Data Group scale factor (S-factor) — the published rule for averaging inputs that disagree, Phys. Rev. D 110, 030001 (2024)crossref10.1103/PhysRevD.110.030001
• Gross & Vitells, Trial factors for the look elsewhere effect in high energy physics, Eur. Phys. J. C 70, 525–530 (2010)crossref10.1140/epjc/s10052-010-1470-8

• Atkins et al., Grading quality of evidence and strength of recommendations, BMJ 328, 1490 (2004) — the foundational GRADE papercrossref10.1136/bmj.328.7454.1490
• Werner, Binder, Toews, Schünemann, Meerpohl & Schwingshackl, Use of GRADE in evidence syntheses published in high-impact-factor nutrition journals, J Clin Epidemiol 135, 54–69 (2021)crossref10.1016/j.jclinepi.2021.02.010THE SOURCE OF THE 5.9% KPI. Verified against the PubMed abstract (PMID 33588024) on 2026-08-22: of 800 SRs, 55 used GRADE and 47 (5.9%) rated outcome-specific certainty; of 465 ratings, very-low 28.8% and low 41.0%. CORRECTION: the hub printed 68.8% low/very-low; the paper's figures sum to 69.8%.
• Schwingshackl et al., NutriGrade, Advances in Nutrition 7, 994–1004 (2016)crossref10.3945/an.116.013052
• Johnston et al., NutriRECS red-meat guideline, Ann Intern Med 171, 756–764 (2019) — GRADE applied to nutrition, and the dispute that followedcrossref10.7326/M19-1621
• GRADE Working Group200
• Bailey & Stover, Precision Nutrition: The Hype Is Exceeding the Science and Evidentiary Standards Needed to Inform Public Health Recommendations for Prevention of Chronic Disease, Annu Rev Nutr 43, 385-407 (2023)crossref10.1146/annurev-nutr-061021-025153Argues from inside the field that the evidentiary standards needed to move from finding to public-health recommendation are not met by current precision-nutrition evidence. Sources the L6 claim that the grading step is asked to carry weight the underlying record cannot support.

• 'Regime of validity' is a disciplinary norm, not a document — there is no citable standard that requires a physics claim to state its domain of validityunverifiedWEAKEST-SOURCED PHYSICS CELL, AND THE ONE THE WHOLE COMPARISON LEANS ON. The claim that physics states scope as part of the claim is an observation about practice. The two entries below are the nearest formal instruments, and neither governs research claims.
• ISO/IEC 17025:2017 scope of accreditation — the formal instrument that states exactly what a laboratory is accredited to measureread
• ASME Boiler and Pressure Vessel Code — applied translation with explicit safety factors200

• GRADE indirectness domain — captures population/intervention mismatch but resolves it by downgrading certainty, collapsing INAPPLICABLE into LOW CONFIDENCE200
• Loudon et al., The PRECIS-2 tool, BMJ 350, h2147 (2015) — describes where results should generalize, without encoding a machine-checkable boundarycrossref10.1136/bmj.h2147
• Glasgow, Vogt & Boles, RE-AIM, Am J Public Health 89, 1322–1327 (1999)crossref10.2105/AJPH.89.9.1322
• Transportability and generalizability theory (Pearl & Bareinboim; Dahabreh, Hernán, Stuart, Robertson, Steingrimsson) — the most rigorous 'does this apply here' apparatus in existence, and essentially unused in nutritionunverifiedNamed in the Stack Audit with journal/volume but no DOI; not independently verified here.
• López-Moreno, Quesada-Fernández & López-Gil, The Appearance of Validity in Nutrition Science: Why Weak Inferences Persist, Advances in Nutrition (August 2026), PMID 42364696crossref10.1016/j.advnut.2026.100693Independent peer-reviewed arrival at an adjacent conclusion, from inside the field, in the same month as v1.4. Categorises eight recurrent inferential fallacies — including an exposure contrast fallacy and a replacement fallacy — and closes by calling for alignment between estimands, exposure definitions, comparators and causal interpretation. Cited for CONVERGENCE, at a different level: it argues inference distorts even when accepted methods are correctly applied, where this framework argues the infrastructure to specify the contrast is absent. The two compose; neither derives from the other.

• Metre Convention, signed 20 May 1875 (17 original signatories); BIPM lists 68 Member States and 35 Associate States and Economies as of 2026-08-22read
• OIML — international legal metrology, the commerce tier200
• ISO/IEC 17025:2017 — accreditation is claimable and auditedread
• ILAC arrangement — mutual recognition of accreditation200
• No statute governs the conduct of physics research — statutes exist only at the units, commerce and engineering tierunverifiedArgument from absence, stated as such in the working paper. The claim is that no law dictates how a physicist runs or reports an experiment; OIML legal metrology and national weights-and-measures law govern trade and engineering only.

• EuroFIR quality index — a quality-scoring rubric, not a claimable public conformance badge200
• W3C WCAG 2.2 — the contrast: a claimable A/AA/AAA tier, which nutrition data has no equivalent of200
• CoreTrustSeal — a claimable tier for repositories200
• OBO Foundry principles — a conformance regime for ontologies200
• DeVito, Bacon & Goldacre, Compliance with legal requirement to report clinical trial results on ClinicalTrials.gov, The Lancet 395, 361–369 (2020) — 40.9% of due trials reported on time under FDAAA 801crossref10.1016/S0140-6736(19)33220-9The governance evidence: a statute without a hard gate underperforms. Cited in the working paper §5.
• Nutrition Labeling and Education Act of 1990 (NLEA) — nutrition's actual labelling statute, and the enforcement point it does have200Shows nutrition accepts standardised formats where a clear point of enforcement exists (the package label).
• Penders, Wolters, Feskens et al., Capable and credible? Challenging nutrition science, Eur J Nutr 56, 2009-2012 (2017)crossref10.1007/s00394-017-1507-yA field-internal challenge to nutrition science's own credibility and capability, written by nutrition researchers and science-studies scholars together. Sources the L8 claim that the incentive and governance critique is not external to the field but published within it, and has been for a decade without moving the layer.

• JCGM 200:2012 (VIM, 3rd ed.) §2.41 'metrological traceability' — a documented unbroken chain of calibrations to a stated reference, each link contributing to the measurement uncertaintyreadTHE CELL THAT LICENSES SCORING A COMPARATOR AT LAYER 9 AT ALL. Physics has a defined term here; nutrition has an aspiration.
• JCGM publications index — confirms JCGM 200:2012 is the current VIM editionread
• CIPM MRA — how traceability is realized across national metrology institutes200
• ISO/IEC 17025:2017 — the accreditation that operationalizes the chainread

• No external standard covers the CHAIN end to end — value → identity → vocabulary → provenance resolving in one record, from food measurement through the study to the guideline. That absence is the findingunverifiedNOT AN EXTERNAL SOURCE. This score is the audit's own provisional estimate of how often the full chain resolves in one record; nothing outside this project measures it. Scope corrected 2026-08-31: the earlier wording said 'there is no external standard for this layer', which overreached. Compiler-level capture of derivation IS standardised, on three continents (see the three sources below). What is unstandardised is carry-forward beyond the compiler.
• Westenbrink, Presser, Roe & Ireland, Documentation of aggregated/compiled values in food composition databases; EuroFIR default to improve harmonization, J Food Compos Anal 102, 103968 (2021)crossref10.1016/j.jfca.2021.103968EuroFIR's Acquisition Type, Value Type and Method Type descriptors record how a value was obtained, including a coded case for a value borrowed from another food. Cited in the L9 rationale since v1.4 but absent from this register until 2026-08-31.
• Westenbrink, Kadvan, Roe & Koroušić Seljak, Evaluation of harmonized EuroFIR documentation for macronutrient values in 26 European food composition databases, J Food Compos Anal 83, 103257 (2019)crossref10.1016/j.jfca.2019.03.006Establishes the reach of the EuroFIR documentation standard — 26 European databases — which is what makes the capture claim European-wide rather than a single-compiler practice.
• USDA FoodData Central, Download Field Descriptions (October 2020) — the food_nutrient_derivation tablereadRead off the PDF on 2026-08-31. USDA ships a dedicated food_nutrient_derivation table: 'Procedure indicating how a food nutrient value was obtained', with 'Code used for the derivation (e.g. A means analytical)' and a source code where '4 means calculated or imputed'. FNDDS carries derivation codes at both the food-nutrient and the ingredient-nutrient level. This is the US half of the capture claim, and it is why 'imputation status is rarely recorded' was wrong for the United States too, not only for Europe.