Prose-style & consumer-language sweep — 2026-06-04 (batch 4, ingredients)
Counts: P1 0 · P2 16 · P3 7. Report-only; no pages were edited.
Pages scanned (22): cinnamon.md, clementines-tangerines.md, coconut.md, coffee.md, collard-greens.md, corn-flakes.md, corn-grits.md, corn-oil.md, corn.md, cornbread.md, cream-of-wheat.md, crisped-rice-cereal.md, cucumber.md, custard.md, dill-pickles.md, dried-herbs.md, egg-noodles.md, eggs.md, fermented-beverage-bases.md, fish-containing-baby-foods.md, fish-fingers.md, fish.md
Stubs skipped (2): cow-milk-infant-formula.md (provisional_scaffold + literature_scope: thin), cumin.md (provisional_scaffold + literature_scope: thin). cocoa.md skipped (covered in batch 1).
Clean pages (no findings): collard-greens.md, corn-flakes.md, corn-grits.md, cream-of-wheat.md, crisped-rice-cereal.md, cucumber.md, dried-herbs.md, fish-containing-baby-foods.md.
Next run resumes at wiki/ingredients/flavored-carbonated-bottled-waters.md.
Dominant pattern this batch
Pages whose corpus data show all-zero or near-zero TDS values (eggs, custard, egg-noodles, coconut) describe the ingredient with unanchored magnitude qualifiers (“negligible,” “low-to-negligible,” “trace,” “low-risk”) in opening prose that precedes the data tables. The fix in every case is the same: move the quantitative anchor (TDS zeros, ppb range, or transfer-factor percentage) forward to the same sentence, or remove the qualifier and let the data section carry the conclusion. Coconut is the most severe case: the contamination profile contains all data gaps, yet the prose asserts “generally low,” “trace levels,” and “well below dietary-significance thresholds.”
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? |
|---|---|---|---|---|---|
| clementines-tangerines.md | 95 | accumulate heavy metals at very low concentrations in the edible flesh / Cadmium uptake from soil into citrus fruit is consistently low in the literature; the FSA/Fera FS102048 survey…found very low metal concentrations | ”Very low” and “consistently low” are unanchored magnitude qualifiers; the FSA data are cited but Table 6 parsing is explicitly pending. The contamination profile table shows Pb typical 10.2–19.7 ppb — not obviously “very low.” The same unanchored “very low” phrase recurs at lines 128, 132, 138, and 154. | Defer these qualifiers until Table 6 values are parsed; then anchor to measured ppb ranges (e.g., “Pb 10–20 ppb in the FSA/Fera survey, placing clementines below leafy vegetables and root crops”). If the flesh is genuinely lower than the table suggests, note the disparity explicitly. | ⚠ yes — the “very low” claim across 5 instances could anchor HMTc derivative-risk decisions for products using citrus |
| clementines-tangerines.md | 132 | low-risk derivatives given the flesh-only basis of most juice production | ”low-risk” characterizes consumer risk for clementine juice without a ppb anchor; the flesh ppb values are not parsed | Rephrase to reference the expected Pb/Cd range once Table 6 is parsed, or qualify as “expected lower-risk given flesh-only basis” with an explicit note that quantitative data are pending | ⚠ yes — this risk characterization applies to a juice form consumed by children |
| coconut.md | 130 | accumulates heavy metals at generally low concentrations relative to other tropical commodities | ”Generally low” characterizes contamination magnitude; the contamination profile table shows all data gaps — no corpus data support the claim | Replace with mechanism only (“coastal sandy soils host lower Cd than the volcanic-tropical soils of cocoa and coffee growing regions”) and note that no corpus occurrence data are currently available | ⚠ yes — “generally low” vs all-data-gap profile is a direct inconsistency |
| coconut.md | 132 | HMTc panel concerns for coconut are low across the board. Lead, cadmium, mercury, and arsenic in coconut meat are typically at trace levels. Aluminum in coconut water is documented but at concentrations well below dietary-significance thresholds. | Three unanchored magnitude qualifiers in a single paragraph; the contamination profile immediately above shows all analytes as data gaps — no measured ppb values are in the corpus to support “trace levels” or “well below dietary-significance thresholds” | Replace with a transparent data-gap acknowledgment: “No occurrence data have been integrated into this corpus for coconut meat, oil, or water. The claims below reflect published literature review that has not yet been structured into the contamination profile; values are pending.” Remove the risk-magnitude language until data are loaded. | ⚠ yes — making substantive risk claims when the profile is all data-gaps is the most significant inconsistency in this batch |
| custard.md | 95 | Its heavy metal risk profile is low for most analytes / egg metal content is low | ”Low” risk profile for custard and “low” egg metal content are unanchored in these sentences; transfer factors below 1% are cited in the same paragraph as a mechanism, but no ppb values appear until the contamination table | Anchor inline: “custard’s heavy metal profile is dominated by the dairy and egg fractions; Pb in the contamination profile ranges 0–14.8 ppb (FSA survey), Cd 0–4.3 ppb, with most analytes at or near zero” — and link the egg claim to the eggs page TDS data | ⚠ yes — risk profile claim in the opening orientation paragraph; editors should verify FSA values before confirming “low” |
| custard.md | 97 | even if any single component carries trace metals, the dilution effect of mixing multiple low-risk ingredients reduces the per-gram metal burden | ”Trace metals” and “low-risk ingredients” are unanchored magnitude terms for individual components; the dilution calculation is not performed | Rephrase to name the components and their approximate contribution: “milk, eggs, sugar, and starch each carry metal loads at or near detection limits in TDS surveys; the blended product inherits a weighted average of these low baselines” | no — this is about the aggregate calculation logic, less claim-sensitive than the per-analyte risk statement |
| custard.md | 132 | Sugar and starch thickeners, if used, add negligible metals. | ”Negligible” is an unanchored magnitude qualifier for sugar and starch contributions; no ppb or cross-reference to a sugar or starch page | Replace with “Sugar and starch thickeners carry metals below detection in TDS surveys (see white-sugar when available)” or cite a supporting reference | no — the relative statement (sugar/starch < dairy/egg) is mechanistically reasonable; the trigger word is “negligible” |
| custard.md | 160 | custard is a low-risk matrix under normal production conditions | ”Low-risk matrix” characterizes consumer risk; FSA data (Pb p95 51.1 ppb, Sn p95 245.4 ppb) are in the contamination table — the Sn value at 245.4 ppb p95 is notable and arguably inconsistent with an unqualified “low-risk” label | Rephrase to be analyte-specific: “custard carries low Cd and Pb under normal commercial production; Sn from tinplate packaging (p95 245.4 ppb in FSA data) warrants attention for products in tinned packaging” | ⚠ yes — the Sn p95 at 245.4 ppb is material to the risk characterization; “low-risk” should not subsume it |
| corn.md | 165 | corn oil is therefore a low-risk derivative for the analytes tracked in this wiki | ”Low-risk derivative” characterizes consumer risk for corn oil; the mechanism (metals are hydrophilic, do not partition into oil) is given, but no ppb anchor for corn oil appears in this page | Add a parenthetical reference to the corn-oil page: “corn oil is a lower-risk derivative for HMTc analytes (see corn-oil for occurrence data)” — without asserting “low-risk” without data in this page’s prose | no — the mechanism is sound; the risk label without ppb is the trigger |
| corn-oil.md | 127 | picks up modest additional metals from packaging-migration over shelf life | ”Modest” characterizes the magnitude of packaging-derived metal pickup for tin and aluminium vs glass, without a ppb figure for the packaging-contact increment | Add a quantitative comparison or note the absence of data: “picks up additional metals from packaging contact (no specific migration data in the current corpus for corn oil in tin or aluminium vs glass)“ | no — the relative claim (PET/tin/Al > glass) is plausible; “modest” is the unanchored qualifier |
| cornbread.md | 161 | Milk or buttermilk, when used, is a low-risk ingredient for heavy metals / Eggs contribute negligibly to the metal burden of the finished product | Two unanchored magnitude qualifiers for dairy and egg contributions; no cross-reference ppb values anchor these in this sentence | Anchor to the relevant pages: “Milk (see whole-milk) and eggs (TDS: all analytes zero, fda2022-tds-elements-fy2018-fy2020) contribute minimal metal loads to the finished product” | no — both claims are defensible given TDS data; the unanchored “low-risk” and “negligibly” are the triggers |
| eggs.md | 95 | Eggs are a low-to-negligible heavy metal risk food under normal commercial production conditions / mercury exposure is minimal | ”Low-to-negligible heavy metal risk food” characterizes consumer risk before any ppb data appears; “mercury exposure is minimal” under commercial conditions is likewise unanchored | The TDS data (all zero across 27 composites) at line 97 would anchor “low-to-negligible” if cited here. Rephrase line 95: “Eggs carry metals at or below detection in FDA TDS surveys (see contamination profile below); cadmium and lead transfer rates from feed to egg are below 1% by weight.” | no — the TDS data at line 97 does support the claim; the issue is the opening sentence states the conclusion before showing the data |
| egg-noodles.md | 99 | the egg fraction adding negligible metals under normal commercial production conditions | ”Negligible” for the egg fraction’s metal contribution; the TDS hard-boiled egg data (all zero at line 103) would anchor this but appears after the qualifier | Anchor inline: “the egg fraction adding metals at or below detection (FDA TDS: all analytes zero for hard-boiled eggs, fda2022-tds-elements-fy2018-fy2020)“ | no — TDS support is available and cited in the same paragraph; the proximity of the anchor is the issue |
| fish-fingers.md | 95 | The frying oil contributes negligible additional metal load / the breadcrumb coating…adds a small cadmium contribution | Two unanchored qualifiers: “negligible” for frying oil and “small” for breadcrumb cadmium; no ppb range is given for either contribution | For frying oil, rephrase to note absence of corpus data: “frying oil contributes little additional metal given the lipophilic-partition argument, though no fish-fry-oil-specific migration data are in the corpus.” For breadcrumb, anchor to wheat Cd range (typically 20–100 ppb grain Cd per wheat when available) | ⚠ yes — the “small” cadmium claim for the breadcrumb fraction is material in products marketed to children |
| fish.md | 291 | Cadmium in fin-fish is generally low because cadmium is not significantly biomagnified through the aquatic food web / Lead in fish is similarly low | ”Generally low” for Cd and Pb in fin-fish are synthesis-level magnitude claims; the contamination table shows Cd and Pb occurrence data exist (n=15 for some analytes) but no typical ppb is given for Cd in this prose paragraph | Anchor to the contamination profile typical values: “cadmium in fin-fish typically falls in the single-digit to low tens of ppb range (per contamination profile above), below bivalve mollusc levels” | ⚠ yes — this is a synthesis claim about all fin-fish; editors should verify the typical ppb in the table supports “generally low” before confirming |
| dill-pickles.md | 99 | the metal burden of the underlying cucumber (generally low, as discussed on [[ingredients/cucumber]]) | ”Generally low” for cucumber is a cross-page reference without any ppb anchor in dill-pickles.md itself; readers see the qualifier without the data | Rephrase to pull the relevant number across: “the metal burden of the underlying cucumber (Pb 0 ppb typical, Cd below detection in TDS data per cucumber)“ | no — the cross-reference supplies the data; “generally low” without the number is the trigger |
P3 — em dashes & inline bold
| Page | Line | Offending text | Why | Suggested fix | Claim-adjacent? |
|---|---|---|---|---|---|
| cinnamon.md | 94 | Cinnamon — the dried inner bark of Cinnamomum trees, sold as Ceylon (Cinnamomum verum) or cassia (Cinnamomum aromaticum / cassia / loureiroi / burmannii, the dominant US-market form) — is the spice second only to turmeric | Em-dash pair around an extended appositive in the opening sentence of running prose | Replace with parentheses: “Cinnamon (the dried inner bark of Cinnamomum trees, sold as Ceylon… or cassia…, the dominant US-market form) is the spice second only to turmeric” | no |
| dried-herbs.md | 94 | Dried herbs — the dehydrated leaves, flowers, and aerial plant parts of culinary and medicinal aromatic plants (oregano, basil, thyme, parsley, sage, rosemary, mint, chamomile, lemon balm, dill, marjoram, savory) — share the drying-concentration heavy-metals pattern | Em-dash pair around an extended appositive in the opening sentence of running prose | Replace with parentheses: “Dried herbs (the dehydrated leaves, flowers, and aerial plant parts of culinary and medicinal aromatic plants …) share the drying-concentration heavy-metals pattern” | no |
| fermented-beverage-bases.md | 125 | (from source tea and from container migration — the documented case of Pb poisoning from kombucha fermented in ceramic containers per the Munilla-García 2023 case established the ceramic-container Pb-migration pathway as a serious concern) | Em dash inside a parenthetical connects a pathway label to an elaborating clause | Replace with a comma: “(from source tea and from container migration, a pathway established by the Munilla-García 2023 ceramic-kombucha Pb poisoning case)“ | no |
| fermented-beverage-bases.md | 131 | (glass vs stainless-steel vs ceramic — ceramic carries the documented Pb-migration risk) | Em dash inside a parenthetical list attaches a risk gloss to one item in the list | Replace with nested parentheses or a comma: “(glass vs stainless-steel vs ceramic, the last of which carries the documented Pb-migration risk)“ | no |
| fermented-beverage-bases.md | 147 | include fermentation-vessel material specification — this is the dominant operative lever | Em dash connecting a subject noun phrase to an explanatory predicate clause inside a lever paragraph | Replace with a colon or period: “include fermentation-vessel material specification; this is the dominant operative lever” | no |
| coffee.md | 159–167 | **Wet vs dry processing** … **Hulling** … **Roasting** … **Grinding** … **Brewing** | Five bold paragraph-opening labels for processing steps in the “Processing effects” section; each names a discrete processing step but the bold serves as structural emphasis rather than a definition label. Same borderline pattern flagged in camellia-sinensis.md (batch 3, line 160). The paragraphs are not in a bullet list. | If intended as a catalog: convert to a bullet list with explicit bold definition labels. If intended as prose: remove the bold; the section heading already establishes the topic, so structure is carried without bold. Applies to all five paragraphs at lines 159, 161, 163, 165, 167. | no |
| coconut.md | 144–154 | **Husking** … **Cracking and meat extraction** … **Drying to copra** … **Coconut oil extraction** … **Drying to coconut flour and coconut chips** … **Coconut water extraction and packaging** | Six bold paragraph-opening labels for processing steps in the “Processing effects” section; same borderline pattern as coffee.md above. None are in a bullet list. | Same fix as coffee.md: either convert the section to a bullet list with bold definition labels, or remove the bold and let the “Processing effects” heading carry the structure. Applies to lines 144, 146, 148, 150, 152, 154. | no |