Prose-style & consumer-language sweep — 2026-06-09 (batch 6, ingredients)
Counts: P1 0 · P2 14 · P3 19. Report-only; no pages were edited.
Pages scanned (25): honey.md, iceberg-lettuce.md, infant-cereal-ingredients.md, infant-formula-concentrated-liquid.md, infant-formula-ingredients.md, infant-formula-powder.md, infant-formula-rtf-liquid.md, ketchup.md, lamb-chop.md, leafy-greens.md, leafy-vegetables.md, legumes.md, lentils.md, lettuce.md, lima-beans.md, macaroni-and-cheese.md, mango-juice.md, mango.md, margarine.md, matcha-powder.md, mayonnaise.md, meat-and-poultry-purees.md, meat-and-poultry.md, meat.md, melon.md
Stubs skipped (7): index.md (index page, not content), infant-formula.md (provisional_scaffold), infant-mixed-grain-cereal.md (provisional_scaffold), infant-oat-cereal.md (provisional_scaffold), infant-rice-cereal.md (provisional_scaffold), jujube-fruit.md (provisional_scaffold), marlin.md (provisional_scaffold).
Clean pages (no findings): honey.md, ketchup.md, lentils.md, lettuce.md, lima-beans.md.
Next run resumes at wiki/ingredients/milk-and-dairy.md.
Dominant pattern this batch
Two patterns again dominate.
First, category-level risk characterisations (“lower-risk food categories,” “lower-risk fruit categories,” “generally low-risk”) continue to appear in the opening “Why this commodity accumulates heavy metals” section on pages whose contamination profile tables contain the data that would anchor those claims. The fix is consistent: anchor to the typical ppb range from the profile table, or replace the characterisation with the mechanistic reason (e.g., “Cd does not partition efficiently from muscle to kidney tissue” rather than “lower-risk”).
Second, the bold-lever-label template pattern continues at high density. infant-formula-powder.md additionally uses bold paragraph-opening labels in its “Ranges by source, region, and variety” and “Processing effects” sections (a new variant of the pattern distinct from the mitigation-lever labels). legumes.md uses the same bold-opening-label format for processing-step descriptions in a way that mirrors the coffee/coconut processing-steps pattern flagged in batch 4. The infant-formula pages add the densest single-page instance of this batch: infant-formula-powder.md has bold labels in three distinct sections totalling 13+ labelled paragraphs.
Importantly: infant-formula-ingredients.md:52–55 and matcha-powder.md:146–149 use bold definition labels inside bullet/numbered lists respectively — these are legitimate catalogs and are NOT flagged.
The margarine and mayonnaise pages present a recurring structure: opening prose asserts “negligible heavy metal concentrations” with the FDA TDS all-below-detection citation in the same sentence, then the sourcing section states “negligible metal content” without any data citation at all. The first instance is borderline (citation is present); the second is a clear flag.
P1 — consumer safe/dangerous without anchor
| Page | Line | Offending text | Why | Suggested fix | Claim-adjacent? |
|---|---|---|---|---|---|
| (none found) | — | — | — | — | — |
P2 — bullets-in-prose & other qualifiers
| Page | Line | Offending text | Why | Suggested fix | Claim-adjacent? |
|---|---|---|---|---|---|
| lamb-chop.md | 165 | Domestic US lamb from regions without mining or industrial Cd soil loading is also generally low-risk | ”Generally low-risk” risk characterisation for lamb from clean-pasture regions without a dose, population, or ppb anchor | Rephrase with mechanism: “Domestic US lamb from clean-pasture regions carries Cd in muscle at concentrations similar to Australian/NZ equivalents; check corpus values once Cd muscle data are populated” | ⚠ yes — sourcing guidance for manufacturers; anchoring prevents over-generalisation to contaminated-pasture US regions |
| legumes.md | 92 | Pb (generally low except in contaminated soils) | ”Generally low” for Pb in legumes without a ppb anchor; legume-corpus Pb values are not cited in this sentence | Anchor: “Pb (typically below 50 ppb in commercial legumes from non-contaminated soils; elevated in produce from soils with documented Pb loading per beans and peanuts)“ | no — the exception is named; “generally low” is the trigger without a ppb floor |
| mango.md | 94 | The heavy-metals profile of mango sits at the lower end of the fruit-category distribution...the relatively low soil-uptake transfer factor for Cd-and-Pb keeps fresh-flesh concentrations modest | ”Lower end of the fruit-category distribution” and “modest” concentrations without ppb; the contamination profile table is immediately below but not cited in this sentence | Anchor inline: “Mango Pb is typically below 50 ppb and Cd below 20 ppb in the corpus data (see contamination profile below), placing it at the lower end of fruit-category distributions” | ⚠ yes — opening characterisation read before the data table; “modest” without a number could mislead on category rank |
| margarine.md | 99 | margarine carries negligible heavy metal concentrations | ”Negligible” characterises all metals without a ppb number in this sentence; FDA TDS (all below reporting limit, n=3) IS cited in the same sentence but “below reporting limit” is not a ppb value. Borderline. | Rephrase: “margarine carries metals at or below FDA TDS reporting limits across all seven analytes measured (n=3)” — remove “negligible” and let the reporting-limit result speak | no — TDS citation is present; the trigger is “negligible” substituting for a ppb anchor |
| margarine.md | 165 | Standard food-grade refined vegetable oils already satisfy the condition for negligible metal content | ”Negligible metal content” in the sourcing guidance without any data citation in this sentence; the closest data (TDS below-detection) is in the contamination profile section | Rephrase: “Standard food-grade refined vegetable oils already carry metals at or below detection limits per TDS data; no additional sourcing intervention is required for metal reduction” | no — mechanistic context is correct; “negligible” without citation is the trigger |
| mayonnaise.md | 99 | produces a finished product with negligible heavy metal concentrations across all analytes of concern | ”Negligible” for all metals in the opening prose; FDA TDS (all below reporting limit, n=3) IS cited in the same sentence but is not a ppb figure. Same borderline case as margarine:99. | Rephrase: “produces a finished product with metals at or below FDA TDS reporting limits across all seven measured analytes (n=3)“ | no — TDS cited in-sentence; “negligible” is the trigger |
| mayonnaise.md | 99 | Vinegar (acetic acid solution) contributes negligible metals | ”Negligible” for vinegar metal contribution without a ppb for vinegar metals | Rephrase: “Vinegar (acetic acid solution) contributes little additional metal given its dilute aqueous composition; no acetic-acid-specific occurrence data are in the current corpus” | no — mechanistic claim; the trigger is “negligible” |
| mayonnaise.md | 165 | Standard food-grade refined vegetable oils and food-grade egg products already satisfy the conditions for negligible metal content | ”Negligible metal content” without data citation in this sentence; same issue as margarine:165 | Rephrase: “Standard food-grade refined vegetable oils and food-grade egg products already carry metals at or below detection limits (TDS data); no additional sourcing intervention is required” | no — same pattern as margarine:165 |
| meat.md | 95 | Commercial muscle meat from mammals and poultry is among the lower-risk food categories for heavy metal accumulation | ”Lower-risk food categories” risk characterisation without dose, population, or ppb anchor; the mechanism is given but no typical ppb for muscle meat | Anchor: “Commercial muscle meat typically carries Pb below [X ppb] and Cd below [Y ppb] in TDS data (see contamination profile), placing it at the lower end of the food-system distribution” — insert corpus values | ⚠ yes — opening characterisation; the “lower-risk” label should be data-anchored given the infant-formula and meat-puree context this page serves |
| meat.md | 95 | methylmercury (MeHg) is negligible in terrestrial muscle compared with seafood | ”Negligible” for terrestrial meat MeHg vs seafood, comparative without ppb for either; mechanism (different bioaccumulation) is given | Anchor comparatively: “MeHg in terrestrial muscle is typically below method detection limits (FDA TDS: all meat composites ND for tHg), in contrast to seafood where MeHg can reach hundreds of ppb” | no — mechanistic claim; “negligible” without ppb floor is the trigger |
| meat.md | 126 | metal concentrations in muscle are low enough that geographic soil and water differences produce modest absolute changes against an already low baseline | ”Low enough,” “modest absolute changes,” and “already low baseline” together characterise muscle metal levels without any ppb anchor in this sentence | Rephrase: “muscle Pb and Cd are low enough in absolute terms (typically single to tens of ppb) that geographic variation produces changes of similar absolute magnitude to measurement uncertainty” — insert ppb range from profile | no — contextual claim about within-category variance; ppb anchoring would make the variance point more precise |
| melon.md | 95 | Melon (cucurbit fruits...) is among the lower-risk fruit categories for heavy metal accumulation / melon flesh typically contains concentrations of Pb, Cd, and most other regulated metals that are low relative to root vegetables or leafy greens | ”Lower-risk fruit categories” risk label and “low relative to root vegetables” without ppb; the contamination profile (Pb 0–4.8 ppb typical, Cd 0–10 ppb typical) is immediately below but not cited here | Anchor: “Melon Pb (typical 0–4.8 ppb) and Cd (0–10 ppb per FSA data, profile below) are below leafy-vegetable levels by an order of magnitude” | no — the table directly below has the values; anchor moves them into the claim |
| melon.md | 124 | Concentrations across survey samples in this and comparable European monitoring data are generally well below regulatory thresholds | ”Generally well below regulatory thresholds” without naming the thresholds or the survey concentrations; the EU ML for melon (Pb 100 ppb, Cd 50 ppb) is not cited | Rephrase: “Concentrations across FSA survey samples are well below the EU ML for fresh fruit (Pb 100 ppb, Cd 50 ppb); melon-specific corpus data show Pb 0–4.8 ppb typical and Cd 0–10 ppb typical” | no — thresholds are nameable; adding them converts an unanchored qualifier to a comparative claim |
| mayonnaise.md | 151 | mechanistic basis for expecting very low metal content | ”Very low” summarises expected metal levels; TDS below-detection data IS cited in the same sentence. Borderline. | Minor: replace “very low” with “at or below detection limits” — no information loss, removes the unanchored qualifier | no — borderline; TDS citation is present |
P3 — em dashes & inline bold
| Page | Line | Offending text | Why | Suggested fix | Claim-adjacent? |
|---|---|---|---|---|---|
| iceberg-lettuce.md | 155 | but for Cd — the dominant concern based on TDS data — washing provides minimal benefit | Em-dash pair around an appositive clause inside running prose | Replace with parentheses: “but for Cd (the dominant concern based on TDS data) washing provides minimal benefit” | no |
| infant-cereal-ingredients.md | 84–94 | **Sourcing levers** (...)... through **Packaging and storage levers** (...)... | Six bold lever-label paragraph openers in the Mitigation section; systemic template-level pattern. | Convert to bulleted list with bold labels, or remove bold and use ### Lever type sub-headings. Applies to all six paragraphs. | no |
| infant-formula-concentrated-liquid.md | 143–149 | **Sourcing levers** (...)... / **Agronomic, Processing, Formulation levers**... / **Testing and QC levers** (...)... / **Packaging and storage levers** (...)... | Four bold lever-label paragraph openers (three levers combined into one); systemic pattern. | Same fix as above. | no |
| infant-formula-ingredients.md | 100–110 | **Sourcing levers** (...)... through **Packaging and storage levers** (...)... | Six bold lever-label paragraph openers in the Mitigation section; systemic pattern. Note: the bullet list at lines 52–55 with bold labels is a legitimate catalog and is NOT flagged. | Same fix as above. | no |
| infant-formula-powder.md | 153–169 | **Base ingredient class**..., **Source-market variation**..., **Historical generation**..., **Spray-drying**..., **Reconstitution**..., **Fortification and additive contributions**..., **Packaging migration**... | Seven bold paragraph-opening labels in the “Ranges by source, region, and variety” and “Processing effects” sections. Each names a discrete variance axis or processing step; the surrounding prose carries the analysis, making this closer to a catalog than flattened analysis. However, these are prose paragraphs, not a bullet list. | Convert to a bullet list with bold definition labels (which would make them a legitimate catalog per the task guidance), or remove bold and let the section heading structure carry the organisation. Applies to all seven paragraphs. | no |
| infant-formula-powder.md | 181–191 | **Sourcing levers** (...)... through **Packaging and storage levers** (...)... | Six bold lever-label paragraph openers in the Mitigation section; systemic pattern. | Same fix as the general lever-label pattern. | no |
| infant-formula-rtf-liquid.md | 154–160 | **Sourcing levers** (...)... / **Agronomic, Processing, Formulation levers**... / **Testing and QC levers** (...)... / **Packaging and storage levers** (...)... | Four bold lever-label paragraph openers (three levers combined); systemic pattern. | Same fix as above. | no |
| leafy-greens.md | 177 | removing internalised metals — they do remove surface contamination at roughly the same rate as plain water | Em dash connecting two contrasting but related clauses in running prose (same sentence structure as leafy-vegetables.md:123) | Replace with a semicolon: “removing internalised metals; they do remove surface contamination at roughly the same rate as plain water” | no |
| leafy-greens.md | 191–201 | **Sourcing levers**... through **Packaging and storage levers**... | Six bold lever-label paragraph openers in the Mitigation section; systemic pattern. | Same fix as above. | no |
| leafy-vegetables.md | 94 | Leafy vegetables — spinach, lettuce, kale, chard, amaranth, cabbage, mustard greens, and dozens of regionally specific leafy greens consumed as salad, cooked greens, or supplement powder — sit at the top of the vegetable-category heavy-metals risk profile | Em-dash pair around an extended list-appositive in the opening sentence of running prose | Replace with parentheses: “Leafy vegetables (spinach, lettuce, kale, chard, amaranth, cabbage, mustard greens, and dozens more) sit at the top of the vegetable-category heavy-metals risk profile” | no |
| leafy-vegetables.md | 123 | vinegar rinses are not meaningfully more effective than water at removing internalised metals — they do remove surface contamination at roughly the same rate as plain water | Em dash connecting two contrasting clauses in running prose | Replace with a semicolon: “vinegar rinses are not meaningfully more effective than water at removing internalised metals; they do remove surface contamination at roughly the same rate as plain water” | no |
| legumes.md | 104–112 | **Soaking and discarding soak water**..., **Cooking**..., **Fermentation**..., **Milling and dehulling**..., **Protein extraction**... | Five bold paragraph-opening labels for discrete processing steps in the “Processing effects” section; same pattern as coffee/coconut processing steps flagged in batch 4. These are prose paragraphs, not a bullet list. | Convert to a bullet list with bold definition labels, or remove bold and let the “Processing effects” heading carry the section structure. Applies to all five paragraphs. | no |
| legumes.md | 124–134 | **Sourcing levers** (...)... through **Packaging and storage levers** (...)... | Six bold lever-label paragraph openers in the Mitigation section; systemic pattern. | Same fix as the general lever-label pattern. | no |
| mango-juice.md | 87–97 | **Sourcing levers** (...)... through **Packaging and storage levers** (...)... | Six bold lever-label paragraph openers in the Mitigation section; systemic pattern. | Same fix as above. | no |
| matcha-powder.md | 136 | The matcha grades — ceremonial, premium ceremonial, classic, culinary, ingredient — differ in Al content roughly inversely to grade tier | Em-dash pair around an enumerated list of grade names in running prose | Replace with parentheses: “The matcha grades (ceremonial, premium ceremonial, classic, culinary, ingredient) differ in Al content roughly inversely to grade tier” | no |
| matcha-powder.md | 151 | The consumer preparation step — whisking matcha powder into hot water and drinking the suspension — delivers 100 percent of the powder's metal content to the consumer | Em-dash pair around an appositive clause describing the preparation step | Replace with parentheses: “The consumer preparation step (whisking matcha powder into hot water and drinking the suspension) delivers 100 percent of the powder’s metal content to the consumer” | no |
| matcha-powder.md | 161–171 | **Sourcing levers** (...)... through **Packaging and storage levers** (...)... | Six bold lever-label paragraph openers in the Mitigation section; systemic pattern. Note: the numbered list at lines 146–149 with bold definition labels is a legitimate catalog and is NOT flagged. | Same fix as the general lever-label pattern. | no |
| meat-and-poultry-purees.md | 152–162 | **Sourcing levers** (...)... through **Packaging and storage levers** (...)... | Six bold lever-label paragraph openers in the Mitigation section; systemic pattern. | Same fix as above. | no |
| meat-and-poultry.md | 109–119 | **Sourcing levers** (...)... through **Packaging and storage levers** (...)... | Six bold lever-label paragraph openers in the Mitigation section; systemic pattern. | Same fix as above. | no |