Prose-style & consumer-language sweep — 2026-06-09 (batch 10, ingredients — FINAL)
Counts: P1 0 · P2 16 · P3 12. Report-only; no pages were edited.
Pages scanned (24): turkey.md, turmeric.md, vegetable-juice.md, vegetable-oils.md, vegetable-purees.md, vegetables.md, vinegar.md, vitamin-mineral-premix.md, walnuts.md, water.md, watermelon.md, wheat-cereal-biscuits.md, wheat.md, white-bread.md, white-fish.md, white-sugar.md, whole-milk.md, whole-wheat-bread.md, wholemeal-bread.md, wild-mushrooms.md, wine-vinegar.md, winter-squash.md, yam.md, yogurt.md
Stubs/previously-covered skipped (1): vegetables-mixed.md (provisional_scaffold).
Clean pages (no findings): wholemeal-bread.md.
wiki/ingredients/ sweep complete. All substantive ingredient pages have now been covered across batches 2–10 (plus the batch 1 pages covered in the initial sweep). metals/ was swept in batch 1. Remaining directories: wiki/products/, wiki/regulations/, wiki/testing/, wiki/synthesis/, and the top-level narrative pages.
Dominant pattern this batch
Dairy-family qualifier density. whole-milk.md and yogurt.md each pack 3–5 unanchored magnitude qualifiers into their opening prose sections (“lowest-risk food categories,” “very low transfer factor,” “low-risk characterisation,” “very low even after concentration”), continuing the pattern seen in skim-milk, reduced-fat-milk, and semi-skimmed-milk. The dairy pages share a structural problem: the TDS all-zeros result IS in the contamination profile table below, but the opening section asserts the conclusion without citing it. The fix is the same across all dairy pages: pull the TDS citation forward into the opening paragraph or let the table carry the conclusion.
white-sugar.md packs three P2 qualifiers across three different sections (“lowest-risk food matrices,” “negligible metal content,” “low-risk matrix”) — all without a ppb anchor in those specific sentences, even though the TDS all-zeros result is cited in the same section. This page also exemplifies the fix: the TDS data are right there; the qualifiers should reference them.
vegetable-oils.md “modest” cascade repeats the rapeseed-soybean-sunflower pattern from batches 7–9: three “modest” qualifiers in different sections characterising baseline metal load, packaging pickup, and processing effects. This is now a confirmed cross-oil-page template issue — the editing pass should fix them all at once.
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? |
|---|---|---|---|---|---|
| turkey.md | 151 | concentrate non-volatile metals by a modest factor / risk classification of turkey as a low-accumulator matrix / contribute negligible additional metal load | ”Modest factor,” “low-accumulator matrix,” and “negligible” without ppb anchors in this paragraph; TDS data exist for the page but are not cited here | Anchor: “concentrate metals by a ~3–4× factor relative to raw muscle (TDS data for roasted turkey below); turkey processed products contribute metals proportional to their formula inclusion level” | no — three qualifiers in one paragraph; the TDS profile resolves all three |
| walnuts.md | 98 | Lead in walnuts is generally low because the woody tree tissue acts as a barrier | ”Generally low” for walnut Pb without ppb anchor; the contamination profile table is immediately below | Anchor inline: “Lead in walnuts is typically [X ppb] per corpus data (profile below), substantially lower than Ni given the woody-tissue barrier to Pb translocation” | no — same pattern as hazelnuts.md from batch 5 |
| watermelon.md | 99 | Watermelon is therefore among the lower-risk food matrices on the wiki for most heavy metals | ”Lower-risk food matrices” risk label without ppb; TDS data (all ND except occasional Ni, tAs, Cd at upper tail) IS cited in the next sentence, partially anchoring the claim | Minor: rephrase as “Watermelon flesh carries metals predominantly at or below detection limits per FDA TDS data (see contamination profile below), placing it at the lower end of the vegetable/fruit distribution” | no — the TDS data are immediately proximate; the risk label without the data attached is the trigger |
| white-fish.md | 95 | white fish (generally safer) / Lead and cadmium concentrations in white fish muscle are generally low | ”Generally safer” and “generally low” for white fish vs predatory species without ppb for either category | Anchor: “white fish species (Pb typically below [X ppb], Cd below [Y ppb] in muscle per corpus data) carry lower MeHg than apex predatory fish”; link to the contamination profile | ⚠ yes — “generally safer” is the most direct risk framing in the batch; used in a consumer-advisory context |
| white-sugar.md | 99 | White granulated sugar is among the lowest-risk food matrices on the wiki for heavy metals / The negligible metal content of white sugar | ”Lowest-risk” and “negligible” without ppb; TDS all-zeros result IS cited in the same paragraph | Replace with: “White granulated sugar carries metals at or below FDA TDS reporting limits across all seven analytes (n=3 composites, all ND)” — anchor to the data, no magnitude qualifier needed | no — TDS cited in-paragraph; “lowest-risk” and “negligible” substitute for the data-sentence |
| white-sugar.md | 159 | this is negligible (for starch in icing sugar) | “Negligible” for starch Cd contribution in icing sugar without ppb | Rephrase: “this contribution is expected to be small given the low starch-fraction weight (typically 2–5%) and the modest Cd load of commercial starch; no corpus data for icing sugar specifically are available” | no — mechanistic; “negligible” without data is the trigger |
| white-sugar.md | 189 | refined sugar is a low-risk matrix | ”Low-risk matrix” in the regulatory context section without ppb anchor | Rephrase: “refined sugar carries metals at or below regulatory detection thresholds; the Codex Pb limit (500 ppb) and commercial TDS results (all ND) create a wide compliance margin” | no — the regulatory threshold is cited in the same sentence; “low-risk” as a label is the trigger |
| whole-milk.md | 95 | whole milk is one of the lowest-risk food categories in the diet for most heavy metals / low-risk characterisation of fluid dairy from non-contaminated herds / The transfer factor from feed to milk for Cd and Pb is very low | Three unanchored magnitude qualifiers in the opening paragraph; TDS all-zeros result IS cited in the same paragraph | Anchor from the TDS result: “The FDA TDS FY2018-FY2020 records all seven analytes at zero across 27 composites, placing whole milk at the analytical floor for commercial dairy; transfer factors for Pb (<1% feed to milk) and Cd (<0.1%) are among the lowest of any food commodity” | ⚠ yes — opening orientation paragraph for the dairy category’s most-referenced page |
| whole-milk.md | 151 | the resulting whole milk powder concentrations remain very low even after this concentration step | ”Very low” for whole milk powder Pb/Cd after ~7–8× concentration; starting concentrations are TDS-zero but the sentence doesn’t name the resulting powder ppb range | Rephrase: “the resulting whole milk powder remains at or below detection limits even after 7–8× concentration from the near-zero baseline” | no — the mechanism is “near-zero baseline”; “very low” is the trigger |
| whole-milk.md | 155 | cheese metal concentrations remain very low given the near-zero starting point of fluid milk | ”Very low” for cheese without ppb; the near-zero baseline argument is made but no ppb for the concentrated cheese product | Rephrase: “cheese metal concentrations remain at or below detection limits in commercial monitoring data despite the 10× milk-to-cheese concentration factor, given the near-zero fluid-milk baseline” | no — mechanism is sound; “very low” without the monitoring-data anchor is the trigger |
| winter-squash.md | 99 | Cd concentrations in the edible flesh that are generally well below those observed in leafy greens or root vegetables | ”Generally well below” for squash Cd vs leafy greens, without ppb for squash in this sentence; TDS data cited in the next sentence | Minor: pull the TDS anchor forward: “Cd in winter squash flesh is typically below [X ppb] per TDS data (see contamination profile), well below leafy-vegetable concentrations” | no — comparative claim; anchoring squash to a ppb makes the comparison concrete |
| winter-squash.md | 165 | Given the generally low baseline contamination of winter squash, sourcing levers are of lower priority | ”Generally low baseline contamination” and “lower priority” without ppb or relative-risk anchor in this sentence | Rephrase: “Given Pb and Cd at or below detection limits in TDS data (profile above), sourcing intervention is lower priority for winter squash than for leafy greens or root vegetables” | no — the TDS profile above carries the data; “generally low” is the trigger |
| yogurt.md | 95 | yogurt is one of the lowest-risk food categories in the diet for Pb, Cd, and other heavy metals / bioavailability of heavy metals...is well established to be low | ”Lowest-risk” without ppb; “well established to be low” characterises the transfer factor without numbers; FDA CTZ action level and FSA data are cited later in the paragraph | Anchor: “Yogurt carries metals at or below detection limits in commercial monitoring (FSA/Fera survey; FDA CTZ establishes a 10 ppb Pb action level for infant yogurt products); the transfer factor from feed to milk is <3% for Pb and <0.1% for Cd” | ⚠ yes — orientation paragraph; “lowest-risk” sets expectations for QA practitioners reading the page |
| yogurt.md | 162 | confirming the regulatory expectation that dairy is a low-risk matrix for these analytes | ”Low-risk matrix” as a standing summary without ppb anchor in this sentence; the action level (10 ppb) is cited in the same sentence | Minor: rephrase as “confirming that yogurt operates within the 10 ppb Pb action level at the industry-standard level” — remove the “low-risk matrix” framing | no — the 10 ppb reference IS in the same sentence; “low-risk” is the trigger |
| vegetable-oils.md | 98 | contributing modest Pb-and-Cd dietary exposure relative to other food groups | ”Modest” for vegetable-oil dietary exposure contribution without a specific ppb or intake figure | Rephrase: “contributing Pb and Cd dietary exposure at [X µg/day] relative to other food groups (per Mehri 2024 probabilistic risk assessment)” — insert actual assessment values | no — the Mehri 2024 source is cited; “modest” could be replaced by the quantified contribution |
| vegetable-oils.md | 127/153 | pick up modest additional metals from packaging-migration over shelf life / Tin and aluminium packaging contribute modest additional metals | ”Modest” for packaging-migration increment without ppb; the same oil-page cascade flagged in rapeseed, soybean, and sunflower-oil batches. Now confirmed across five oil pages. | Batch fix across all oil pages: replace “modest additional metals” with either the corpus ppb migration values or “no specific corpus quantification is available; see the Charfi 2026 olive-oil packaging data for order-of-magnitude context” | no — template-level issue; affects all commodity-oil pages |
P3 — em dashes & inline bold
| Page | Line | Offending text | Why | Suggested fix | Claim-adjacent? |
|---|---|---|---|---|---|
| turmeric.md | 94 | the US-side case literature — Angelon-Gaetz 2018 North Carolina home-investigation cohort (n=386...), Lech 2020 ground turmeric US survey (n=127...), Huff 2025 Lancaster PA (n=116...) — links US-market adulterated turmeric to elevated childhood blood-lead levels | Em-dash pair around a parenthetical list of three citation-references in running prose | Replace with a comma or restructure: “US-side case literature (Angelon-Gaetz 2018 NC, n=386; Lech 2020 US survey, n=127; Huff 2025 Lancaster PA, n=116) links US-market adulterated turmeric to elevated childhood blood-lead levels” | no |
| turmeric.md | 98 | documented Pb concentrations in finished turmeric reaching 50,000-100,000 ppb (50-100 mg/kg) — two to three orders of magnitude above non-adulterated product | Em dash connecting a quantitative data point to its comparative characterization | Replace with a semicolon or comma: “reaching 50,000–100,000 ppb (50–100 mg/kg), two to three orders of magnitude above non-adulterated product” | no |
| vegetable-juice.md | 139–149 | **Sourcing levers** (...)... through **Packaging and storage levers** (...)... | Six bold lever-label paragraph openers; systemic pattern. | Convert to bulleted list with bold labels or use ### Lever type sub-headings. | no |
| vegetable-oils.md | 94 | Vegetable oils — the broad category of liquid edible oils pressed or solvent-extracted from seeds (sunflower, soybean, rapeseed/canola, corn, peanut, sesame, safflower), fruit (olive, palm), or nuts — share a common heavy-metals profile shaped by upstream seed-source agronomy | Em-dash pair around an extended appositive in the opening sentence of running prose | Replace with parentheses: “Vegetable oils (the broad category of liquid edible oils pressed or solvent-extracted from seeds, fruit, and nuts) share a common heavy-metals profile…“ | no |
| vegetable-purees.md | 125–127 | **[[products/root-vegetable-purees]]** — carrot, sweet potato... / **[[products/non-root-vegetable-purees]]** — spinach, kale... | Two bold wikilink paragraph-opening labels for product-route categories in a routing section; each names a discrete product-category with a brief description. Closer to a catalog but in prose paragraphs, not a bullet list. | Convert to bullet list: - **[[products/root-vegetable-purees]]** — carrot, sweet potato… (already uses em dashes as connectors; converting to a bullet list makes these legitimate catalog items) | no |
| vegetable-purees.md | 147–157 | **Sourcing levers** (...)... through **Packaging and storage levers** (...)... | Six bold lever-label paragraph openers; systemic pattern. | Same fix as above. | no |
| vegetables.md | 68–78 | **Sourcing levers** (...)... through **Packaging and storage levers** (...)... | Six bold lever-label paragraph openers; systemic pattern. | Same fix as above. | no |
| vinegar.md | 94 | Vinegar — the acidified aqueous product of acetic-acid fermentation of a sugar or alcohol substrate, sold as white distilled vinegar, apple cider vinegar, balsamic vinegar, wine vinegar, malt vinegar, and rice vinegar — accumulates heavy metals through two distinct pathways | Em-dash pair around an extended appositive in the opening sentence | Replace with parentheses: “Vinegar (the acidified aqueous product of acetic-acid fermentation of a sugar or alcohol substrate, sold as white distilled vinegar, apple cider vinegar, balsamic vinegar, wine vinegar, malt vinegar, and rice vinegar) accumulates heavy metals through two distinct pathways” | no |
| vitamin-mineral-premix.md | 154–164 | **Sourcing levers** (...)... through **Packaging and storage levers** (...)... | Six bold lever-label paragraph openers; systemic pattern. | Same fix as above. | no |
| wine-vinegar.md | 73 | red wine vinegar, white wine vinegar, sherry vinegar, champagne vinegar, rice-wine vinegar — each carrying different source-base profiles | Em dash connecting an enumerated list to a characterising clause | Replace with a comma: “(red wine vinegar, white wine vinegar, sherry vinegar, champagne vinegar, rice-wine vinegar, each carrying different source-base profiles)“ | no |
| wine-vinegar.md | 89–99 | **Sourcing levers** (...)... through **Packaging and storage levers** (...)... | Six bold lever-label paragraph openers; systemic pattern. Note: the Formulation-levers entry at line 95 also contains an em dash inside the lever paragraph (“cider vinegar, distilled white vinegar, rice vinegar — each with different source-base metal profiles”) — a double violation on that line. | Fix both the bold-label pattern (convert to bulleted list) and the em dash on line 95 (replace with a comma). | no |
ingredients/ sweep complete — aggregate patterns
Across all 10 batches covering wiki/ingredients/ (batches 2–10 plus batch 1 predecessor), the dominant findings were:
-
Unanchored magnitude qualifiers (P2): “generally low,” “very low,” “low-risk,” “modest,” and “negligible” appear throughout opening and derivative-risk sections without the contamination-profile ppb values that would anchor them. The fix is consistently the same: pull the data table anchor into the prose sentence, or remove the qualifier and let the table carry the conclusion.
-
Bold mitigation-lever-label pattern (P3): virtually every ingredient page uses
**Lever type** (wikilink)...as bold paragraph openers for the mitigation section. This is a template-level issue requiring one fix applied to the template rather than to individual pages. Converting each mitigation block to a bullet list with bold definition labels makes them legitimate catalogs. -
Em-dash openers (P3): most ingredient pages open their “Why this commodity accumulates heavy metals” section with
Ingredient — the [appositive description] — does X. Replace with parentheses in all cases. -
Oil-family “modest” cascade (P2): rapeseed-oil, soybean-oil, sunflower-oil, vegetable-oils, and multiple others all use “modest” for baseline load, packaging-migration, and processing-effects sections without ever anchoring to a corpus ppb. This is a cross-page template fix.
-
Dairy-family “low-risk” cluster (P2): fromage-frais, semi-skimmed-milk, skim-milk, reduced-fat-milk, whole-milk, yogurt, and related pages all declare dairy “low-risk,” “negligible,” or “very low” without citing the TDS all-zeros result in the same sentence. The TDS data are present elsewhere on the page; the fix is to cite them inline.