Skip to content

Prose-style & consumer-language sweep — 2026-06-09 (batch 9, ingredients)

Counts: P1 0 · P2 14 · P3 21. Report-only; no pages were edited.

Pages scanned (25): saltine-crackers.md, seafood.md, semi-skimmed-milk.md, sesame-oil.md, shark.md, shellfish.md, skim-milk.md, soft-drink-bases.md, soy-based-infant-formula.md, soy-protein-isolate.md, soy.md, soybean-oil.md, spices.md, sports-drink-bases.md, squash.md, strawberries.md, sunflower-oil.md, sunflower-seeds.md, tea.md, teething-and-snacks.md, tinned-fish.md, tomato-paste.md, tomato-soup.md, tomato.md, tortilla-chips.md

Stubs/previously-covered skipped (6): salt.md (provisional_scaffold), seaweed.md (covered batch 1), soy-milk.md (provisional_scaffold), spinach.md (covered batch 1), sweet-potato.md (covered batch 1), swordfish.md (provisional_scaffold).

Clean pages (no findings): sunflower-seeds.md, tinned-fish.md, tomato-soup.md, tomato.md.

Next run resumes at wiki/ingredients/turkey.md.


Dominant pattern this batch

Oil-family “modest” cascade. soybean-oil.md and sunflower-oil.md each use “modest” three to five times across different sections without ever anchoring to a ppb value — exactly the same pattern flagged for rapeseed-oil.md in batch 7. This is now clearly a template-level issue for the oil-ingredient pages. A single editing pass across all oil pages should replace “modest baseline Pb-and-Cd,” “modest rates,” “modest additional metals from packaging,” and “modest net change” with the actual corpus ppb values or a note that no quantified data are available.

Dairy family repeats the batch 6/7 pattern. semi-skimmed-milk.md and skim-milk.md each declare dairy metal contributions “negligible” and products “low-risk” without per-ppb anchors — repeating the pattern in fromage-frais (batch 5) and reduced-fat-milk (batch 8). The same fix applies: replace “negligible” with “at or below detection limits (TDS data)” and link the specific profile data.

seafood.md:312–318 new bold variant. The seafood hub page uses bold paragraph-opening labels for analyte categories (Methylmercury, Cadmium, Inorganic arsenic, Total arsenic) in a synthesis/orientation section. This is closer to a legitimate catalog (each labels a discrete analyte with a cross-reference) but the paragraphs are prose, not a bullet list. Flag and let the editor decide.

soy.md:107–115 new bold variant. The soy page uses bold labels for derivative types (Soy oil, Soy protein isolate, Tofu, Soy milk, Fermented soy) in a derivatives section. Same structural assessment: discrete catalog in prose paragraphs, closer to legitimate than the generic lever labels, but not in a bullet list.


P1 — consumer safe/dangerous without anchor

PageLineOffending textWhySuggested fixClaim-adjacent?
(none found)

P2 — bullets-in-prose & other qualifiers

PageLineOffending textWhySuggested fixClaim-adjacent?
saltine-crackers.md99Salt (NaCl)...contributes negligible metals / Vegetable shortening or oil...is similarly low-risk after industrial refining”Negligible” for food-grade salt metals and “low-risk” for refined fat without ppb; both are mechanistically defensible but unanchoredReplace “negligible metals” with “metals below detection limits in food-grade salt”; replace “low-risk” with “metals at or below detection limits per TDS data for refined oils”no — mechanism is sound; the trigger words are “negligible” and “low-risk”
semi-skimmed-milk.md95Dairy is a low-risk matrix for heavy metals / is generally low (for dairy metals in commercial herds)“Low-risk matrix” and “generally low” without ppb; the supporting TDS data exist for fluid milk but are not cited in this sentenceAnchor: “Dairy is a low-metal matrix; Pb in fluid milk is typically below [X ppb] and Cd below [Y ppb] per TDS data (see reduced-fat-milk profile)”⚠ yes — this is the orientation paragraph for a high-traffic dairy-category page
semi-skimmed-milk.md132Its metal contribution in all these applications is negligible given the near-zero baseline / the resulting dried product remains low-risk”Negligible” for the milk-fraction contribution and “low-risk” for dried semi-skimmed milk without ppbRephrase: “Its metal contribution is at or below detection limits given the near-zero fluid-milk baseline; the dried product concentrates metals ~10× from the near-zero baseline but remains well below regulatory limits”no — near-zero baseline is accurate; “negligible” and “low-risk” are the triggers
skim-milk.md155the resulting powder remains low-risk”Low-risk” for skim milk powder after spray-drying concentration without ppbRephrase: “the resulting powder, though concentrating metals ~10×, remains well below the relevant regulatory limits for dairy given the near-zero fluid baseline”no — mechanistic claim; anchor to regulatory limits makes it concrete
skim-milk.md159The near-zero metal content of skim milk means its contribution...is negligible”Negligible” for the skim milk contribution to formulated products without ppbRephrase: “at or below detection limits means the skim milk fraction does not drive the metal profile of any downstream formulated product”no — TDS data (all ND) cited earlier in page; “negligible” is the trigger
soy.md56Pb is generally low except where contaminated soil is involved”Generally low” for soy Pb without ppb; the contamination profile table carries Pb dataAnchor: “Pb is typically below [X ppb] in commercial non-contaminated-soil soybean supply (profile below); elevated in produce from documented contaminated-soil regions”no — same pattern as legumes.md and non-soy-protein.md from prior batches
soybean-oil.md94modest baseline Pb-and-Cd from upstream seed agronomy”Modest baseline” without ppb; same oil-family pattern flagged in rapeseed-oil.md (batch 7) and sunflower-oil.md (this batch)Anchor from the profile table: “baseline Pb of [X ppb] and Cd of [Y ppb] from upstream seed agronomy (see contamination profile below)“no — profile carries the data; “modest” is the trigger
sunflower-oil.md94the seed itself carries a modest soil-derived metal load / Sunflower oil is consistently among the lowest-metal-load edible oils”Modest” and “lowest-metal-load” without ppb; the corpus data existAnchor: “the seed carries Pb typically [X ppb] and Cd [Y ppb] per corpus data; sunflower oil sits in the lower range of the edible-oil metals distribution (profile below)“no — both claims are supportable but unanchored
sunflower-oil.md98the seed accumulates Pb and Cd from soil...at modest rates relative to leafy vegetables”Modest rates” for Pb/Cd uptake without quantification; the comparison to leafy vegetables is relative without numbers for eitherRephrase: “the seed accumulates Pb and Cd at lower rates than leafy vegetables (Pb [X ppb] vs leafy-vegetable [A–B ppb])” — insert corpus valuesno — comparative mechanism claim; numbers would complete it
sunflower-oil.md123/127/153the net change is modest (refining); picks up modest additional metals from packaging-migration (shelf life); contribute modest additional metals (tin/Al packaging)Three “modest” qualifiers in different sections without ppb; same cascade as rapeseed-oil.md and soybean-oil.mdBatch-fix across all oil pages: replace each “modest” with either a corpus ppb value or “no corpus-specific quantification is available for this commodity; see the Charfi 2026 olive-oil packaging-comparison data for order-of-magnitude context”no — oil-family template issue; each instance lacks the same ppb anchor
tea.md123post-fermentation aging for pu-erh tea can produce modest additional Al concentration”Modest additional Al concentration” for pu-erh aging without ppb; the sentence acknowledges “this is not quantified in the loaded corpus”Rephrase: “post-fermentation aging for pu-erh tea can produce additional Al concentration (magnitude not quantified in the current corpus; flagged for future characterisation when pu-erh-specific data are ingested)“no — the absence-of-data acknowledgement is already there; “modest” adds an unanchored magnitude claim
teething-and-snacks.md125even at modest per-serving metal levels”Modest per-serving metal levels” for teething-and-snacks products without a ppb or per-serving dose valueRephrase: “even at single-digit-ppb per-serving metal levels (Pb, Cd at or below 10 ppb in most commercially available products)” — or anchor to relevant test data⚠ yes — infant/toddler context; “modest” understates the per-body-weight-adjusted exposure argument that is the reason the sentence exists
strawberries.md155the modest Cd values documented in the TDS data, the contribution is small relative to higher-risk ingredients”Modest Cd values” and “small” for the strawberry Cd contribution at typical inclusion levels; TDS median (8.5 ppb) IS cited in the prior paragraph. Borderline — the anchor is proximate.Minor: replace “the modest Cd values” with “the documented TDS Cd (median 8.5 ppb)” — no information loss, removes the unanchored qualifierno — the TDS data is cited in the same paragraph; this is a borderline but clean fix
tortilla-chips.md99The salt added during frying and seasoning contributes negligible metals”Negligible” for flavoring-salt metal contribution without ppb; mechanism is that food-grade salt is low in metalsRephrase: “The salt added during frying and seasoning contributes metals below detection limits at typical inclusion levels, assuming food-grade salt is used”no — same pattern as saltine-crackers.md salt finding

P3 — em dashes & inline bold

PageLineOffending textWhySuggested fixClaim-adjacent?
sesame-oil.md73sesame plants are documented Cd accumulators — sesame seeds carry Cd at concentrations multiples higher than most other oilseedsEm dash connecting the category label to its elaborating evidence clauseReplace with a semicolon: “sesame plants are documented Cd accumulators; sesame seeds carry Cd at concentrations…“no
sesame-oil.md91–101**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
seafood.md312–318**Methylmercury**..., **Cadmium**..., **Inorganic arsenic**..., **Total arsenic**...Four bold paragraph-opening labels for analyte categories in a synthesis/orientation section. Each names a discrete analyte with a cross-reference; closer to a legitimate catalog than the generic lever labels, but in prose paragraphs rather than a bullet list.Convert to a bullet list with bold definition labels (legitimate catalog per the task guidance) or remove bold and use the analyte as a plain-text paragraph opener.no
seafood.md346–356**Sourcing levers** (...)... through **Agronomic and formulation levers**...Bold lever-label paragraph openers in the Mitigation section; systemic pattern.Same fix as above.no
shellfish.md233bivalves filter large volumes of water — an oyster filters 5-50 liters per day — and concentrate dissolved and particle-bound metalsEm-dash pair around an illustrative parenthetical; same pattern as molluscs.md:114 (batch 7).Replace with parentheses: “bivalves filter large volumes of water (an oyster filters 5–50 liters per day) and concentrate dissolved and particle-bound metals”no
shellfish.md261–267**Sourcing levers** (...)... through **Packaging and storage levers** (...)...Four bold lever-label paragraph openers; systemic pattern.Same fix as above.no
soft-drink-bases.md129sweetener choice (HFCS vs cane sugar vs non-nutritive sweetener — different baseline impurity profiles)Em dash inside a parenthetical, connecting a list item to a characterising clauseReplace with a comma: “(HFCS vs cane sugar vs non-nutritive sweetener, each with different baseline impurity profiles)“no
soft-drink-bases.md141–151**Sourcing levers** (...)... through **Packaging and storage levers** (...)...Six bold lever-label paragraph openers; systemic pattern.Same fix as above.no
soy-based-infant-formula.md148–158**Sourcing levers** (...)... through **Packaging and storage levers** (...)...Six bold lever-label paragraph openers; systemic pattern.Same fix as above.no
soy.md107–115**Soy oil**..., **Soy protein isolate**..., **Tofu**..., **Soy milk**..., **Fermented soy**...Five bold paragraph-opening labels for derivative types in the “Processing effects / Ingredient-derivative risk” section. Each names a discrete product type. Closer to a legitimate catalog, but these are prose paragraphs, not a bullet list.Convert to a bullet list with bold definition labels (legitimate catalog), or remove bold and let the derivative name stand as plain text.no
soy.md129–139**Sourcing levers** (...)... through **Packaging and storage levers** (...)...Six bold lever-label paragraph openers; systemic pattern.Same fix as above.no
soy-protein-isolate.md131(a documented agronomic phenomenon — soy yields are Al-stress-sensitive in acidic soils because of root-zone Al toxicity, and the surviving plant tissue carries elevated Al)Em dash inside a parenthetical, connecting the label to its elaborating mechanism clauseReplace with a comma: “(a documented agronomic phenomenon: soy yields are Al-stress-sensitive in acidic soils because of root-zone Al toxicity…)“no
soy-protein-isolate.md139US Midwest soybean production, Brazilian soybean production, Argentine soybean production, Chinese domestic soybean production, European soybean production — each carrying different Al/Cd profilesEm dash connecting an enumerated list to a characterising clauseReplace with parentheses or a comma at list end: ”(…, European soybean production, each carrying different Al/Cd profiles reflecting regional soil characteristics)“no
soy-protein-isolate.md151–161**Sourcing levers** (...)... through **Packaging and storage levers** (...)...Six bold lever-label paragraph openers; systemic pattern.Same fix as above.no
spices.md94Spices — the broad seasoning category covering ground and whole dried plant parts (turmeric, paprika, cinnamon, black pepper, cumin, coriander, chili, cardamom, cloves, and dozens more) — sit at the top of the food-system heavy-metals risk distributionEm-dash pair around an extended appositive in the opening sentence of running proseReplace with parentheses: “Spices (the broad seasoning category covering ground and whole dried plant parts: turmeric, paprika, cinnamon, black pepper, cumin, coriander, chili, cardamom, cloves, and dozens more) sit at the top of the food-system heavy-metals risk distribution”no
sports-drink-bases.md139–149**Sourcing levers** (...)... through **Packaging and storage levers** (...)...Six bold lever-label paragraph openers; systemic pattern.Same fix as above.no
squash.md127summer squash, winter squash, pumpkin — broadly similar baselines but pumpkin in particular accumulates moderate CdEm dash connecting a cultivar list to a qualifying clause in running proseReplace with parentheses: “(summer squash, winter squash, pumpkin; broadly similar baselines, but pumpkin in particular accumulates moderate Cd)“no
squash.md139–149**Sourcing levers** (...)... through **Packaging and storage levers** (...)...Six bold lever-label paragraph openers; systemic pattern.Same fix as above.no
teething-and-snacks.md141–151**Sourcing levers** (...)... through **Packaging and storage levers** (...)...Six bold lever-label paragraph openers; systemic pattern.Same fix as above.no
shark.md93–99**Sourcing levers** (...)..., **Consumption-pattern levers**..., **Testing and QC levers** (...)..., **Processing, formulation, agronomic, and packaging levers**...Four bold lever-label paragraph openers (three levers combined into one paragraph for the last entry); systemic pattern.Same fix as above.no
tomato-paste.md95–105**Sourcing levers** (...)... through **Packaging and storage levers** (...)...Six bold lever-label paragraph openers; systemic pattern.Same fix as above.no