# Bags, Wraps & Covers: research notes Date: 2026-10-25. Scope: the taxonomy leaf bundles three jobs that have nothing in common except that the product is a flexible barrier: **food storage bags and wraps** (zip-top, vacuum-seal and fold-top bags, plastic cling wrap, aluminum foil, parchment paper, wax paper, reusable silicone bags), **protective covers for furniture and equipment** (furniture dust covers, outdoor patio-furniture covers, appliance and grill covers, mattress and box-spring encasements), and **reusable carry/shopping bags** (totes, produce bags). One register, one leaf, as instructed; the `intended_use` split does the dividing between the three jobs rather than three separate documents. The add-on schema's own comment on `bag.jsonc` is literally `//Bags, Wraps & Covers` with a single field, `sous_vide_safe`, which is itself evidence for the food-storage-bag reading of this leaf (nobody would ask whether a furniture cover is sous-vide safe). That one field anchored the scope decision more than the taxonomy label alone would have. **Excluded, left to siblings** (same `supplies > household` prefix in `catalog.json`): garbage and bin liners (`garbage-bin-bags`), insulated cooler/lunch bags (`coolers`, which already owns "Insulated Bags"), mesh laundry wash bags (`laundry-wash-bag-frames`), vacuum-storage and moving bags and boxes for closets/attics (`storage-containers`, "Storage Containers, Caddies, Bags & Boxes"), photo/stamp archival sleeves (`keepsake-storage`), and gift wrap/bags (`gift-bags`, stationery vertical). Camera bags and musical instrument bags/cases are `supplies > gear` leaves, not household, so also excluded. Where a listing could plausibly sit in more than one of these (a "heavy-duty storage bag" that is really a closet vacuum bag, or a "cooler tote" that is really an insulated bag), it belongs to the sibling named above, not here. Quote labels: **verbatim** = the sentence quoted was in text I read myself (curl, then `pandoc -f html -t plain`, read from the resulting `.txt` file in the scratchpad); **fetched excerpt** = a fetch tool's own summary of a page, lightly reworded; **search excerpt** = from a search-engine summary with no page read. WebSearch reported its budget exhausted for this session on the first five queries I tried (freezer-vs-storage mil thickness, FDA food contact, California reusable-bag law, bed-bug encasement standard, patio cover fabric). Every claim below is a **verbatim** fetched-and-converted excerpt or is explicitly marked otherwise; no claim in `category.json` rests on a search excerpt alone. ## Seed Seed run output (recorded verbatim from the script): - Example: kitchen-knives `steel_hardness`: "a numeric spec that only makes sense as one side of a trade-off, and a visual grammar borrowed from the hobbyist community that puts the trade-off on one axis." - Persona: "A perfumer, who thinks in what you notice first, what lasts, and what fades." 1. **The bags/wraps/covers equivalent of steel hardness is film thickness in mils.** It is the number behind "freezer" versus "storage" versus "snack" bags, and like hardness it trades two things against each other: thicker film blocks more air and moisture (longer freezer life) but costs more and is stiffer. **Data check failed the invented-range version**: I could not find two independently fetched sources giving the same mil numbers for freezer vs. storage bags (brand FAQ pages that would carry this, Ziploc and Glad, returned HTTP 429/404 to every fetch attempt). **Kept** the concept as the `freezer_or_storage_grade` categorical facet (a buyer-visible label, not an invented number) and dropped any numeric mil range from the JSON. The general mil/micron concept survives as a vocabulary term and in `wrap_material`'s note that household cling wrap has gotten thinner over time (Wikipedia `Plastic wrap`, citing a Dow Saran Wrap spec page and an SC-Johnson-style FAQ it names but I did not independently fetch). 2. **A perfumer notices what you smell first, what lasts, and what fades.** Applied here: "what you smell first" is odour/air permeability — whether a storage bag lets fridge smells cross, which is really the same mechanism as freezer burn (air exposure). **Dropped as a standalone facet**: no listing or spec sheet states an odour-permeability number, and I found no second source with one, so it stays as a sentence inside `freezer_or_storage_grade` and `closure_type`'s `why_experts_care` rather than its own facet. "What fades" maps cleanly onto outdoor furniture covers: UV fading and the vinyl cracking that non-UV-treated polyethylene suffers (Wikipedia `Tarpaulin`, verbatim: "When treated against ultraviolet light, these tarpaulins can last for years exposed to the elements, but non-UV treated"). **Kept** as part of the `exposure` split and the `cover_fabric` facet's trade-off. ## Sources and what each supports ### Could not reach (recorded so the next author does not repeat the attempt) - CalRecycle (`calrecycle.ca.gov`) on the California reusable-bag law: every URL I tried returned HTTP 403, including the bare domain. No Internet Archive snapshot existed under the paths I guessed. - `ecfr.gov` (21 CFR 177, food-contact plastics) and `fda.gov/food/...` food-contact-substances pages: both redirected to a bot-block page ("abuse-detection-apology.html" on fda.gov; an "unblock" interstitial on ecfr.gov). - FSIS (`fsis.usda.gov`) vacuum-packaging/sous-vide safety page: HTTP 403. - Brand pages for thickness/spec claims: `ziploc.com/en/faq` (429), `foodsaver.com/vacuum-seal-bags` (403), `stasher.com/pages/faq` (404), `budgeindustries.com` and `duckcovers.com` fabric-guide pages (404, or redirected to a bare product-collection page with no spec text). `homedepot.com`, `lowes.com`, `thespruce.com` and `bhg.com` buying-guide pages for patio furniture covers: all blocked (403/402) to both curl and the WebFetch tool. - `npmapestworld.org` bed-bug page and MSU Extension's bed-bug-encasement news page: both returned pages with no usable body text after HTML→text conversion (0–2 lines). Because every furniture-cover-specific retailer and brand source was unreachable, the `cover_fabric` trade-off (waterproof vinyl vs. breathable weave) is sourced **by analogy** to the Wikipedia `Tarpaulin` article, which describes the same materials (woven polyethylene, PVC-coated vinyl, canvas) used for the same purpose (covering something against weather), not to a furniture-cover-specific test. This is flagged as a medium-confidence judgement in `category.json`, not stated as a measured fact. ### Food storage, wrap and vacuum packing **Wikipedia, "Plastic wrap"** — https://en.wikipedia.org/wiki/Plastic_wrap (curled, converted with pandoc, read as text; kind: primer) - verbatim: "Common plastic wrap is roughly 12.7 μm (0.0005 in) thick... now the majority of brands on shelves around the world are 8, 9 or 10 μm thick." (cites a Dow Saran Wrap spec page and a 2014 FAQ page, neither independently fetched by me — noted as a secondary-summary limitation.) - verbatim: "PVC has an acceptably-low permeability to water vapor and oxygen... There are concerns about the transfer of plasticizers from PVC into food." and "A common, cheaper alternative to PVC is low-density polyethylene (LDPE). It is less adhesive than PVC." - verbatim footnote 9: Meadows, Michelle (2002). "Plastics and the Microwave". FDA Consumer — confirms the FDA addressed this in a 2002 consumer article, which I could not fetch directly (archived, not live). - Supports: `wrap_material`'s trade-off, `heat_safe_rating`'s rationale, the mil/micron vocabulary note. Confidence medium (secondary summary for the exact thickness figures; the PVC-vs-PE mechanism claim is the primer's own prose, not citation-needed). **Wikipedia, "Vacuum packing"** — https://en.wikipedia.org/wiki/Vacuum_packing (kind: primer) - verbatim: "In an oxygen-depleted environment, anaerobic bacteria can proliferate, potentially causing food-safety issues. Some pathogens of concern in vacuum packed foods are spore-forming non-proteolytic Clostridium botulinum, Yersinia enterocolitica, and Listeria monocytogenes." - verbatim: "Vacuum packing reduces freezer burn by preventing the food from exposure to cold, dry air." (marked `[citation needed]` in the source article itself — I am reporting it as the article's own qualitative claim, confirmed by a second, independent source below, not relying on the citation-needed line alone.) - **Dropped**: the article's oxygen-transmission-rate numbers ("a standard PA/PE bag allows on average 100 cubic centimeters, PVDC allows on average over 10, and EVOH on average 1") are explicitly marked `[citation needed]` in the source text I read. Per the two-source rule, this does not go in `category.json` as a number; it is not in the register at all, not even as a range. - Supports: `closure_type`'s vacuum-seal rationale, the botulism/anaerobic judgement. Confidence high for the mechanism, medium for "reduces freezer burn" pending the second source below. **Healthline, "Freezer Burn: What It Is, Food Safety, and Tips for Prevention"** — https://www.healthline.com/nutrition/freezer-burn (kind: primer) - verbatim: "Freezer burn is the dehydration of food that results from moisture loss... through a process called sublimation." - verbatim: "Remove as much air as possible from the packaging of frozen fruits and vegetables... You can use reusable containers, such as sealed silicone bags." - verbatim: "wrap meat, poultry, or seafood in freezer paper or plastic wrap, then in foil, and then in a freezer bag." - Supports: the second, independent source for "removing air slows freezer burn" (judgement 3), and the `reusable` split's silicone-bag example. Confidence high; this is the second source the vacuum-packing claim needed. ### Furniture, equipment and pest covers **Wikipedia, "Tarpaulin"** — https://en.wikipedia.org/wiki/Tarpaulin (kind: primer) - verbatim: "Canvas tarpaulins are not 100% waterproof, though they are water resistant... when there is standing water on canvas tarps, or when water cannot quickly drain away from canvas tarps..." (article continues into mildew risk). - verbatim: "A polyethylene tarpaulin... is not a traditional fabric, but rather, a laminate of woven and sheet material... waterproof. Sheets can be either of low density polyethylene (LDPE) or high density polyethylene (HDPE). When treated against ultraviolet light, these tarpaulins can last for years exposed to the elements, but non-UV [treated ones degrade faster, sentence continues]." - verbatim: vinyl tarps are "constructed of 10 oz/sq yd (340 g/m²) coated yellow vinyl. This makes it waterproof and gives it a... [resistance to] grease and mildew." - Supports: `cover_fabric`'s waterproof-vs-breathable trade-off, by analogy (tarps, not furniture covers specifically — see "Could not reach" above). Confidence medium, flagged as a judgement, not a measured fact about furniture covers. **University of Kentucky Entomology, "Bed Bugs" (ENTFACT-636)** — https://entomology.ca.uky.edu/ef636 (redirected to entomology.mgcafe.uky.edu/ef636; kind: academic) - verbatim: "Once the covers are installed and zipped shut, any bed bugs that happen to be inside eventually die... However, encasements will not keep bed bugs from crawling onto a bed and biting a sleeping person. Bed bug-proof encasements are also available for sofas and loveseats." - Supports: `pest_proof_encasement` facet and its judgement. This is extension guidance from a university entomology department, not a certifying standard; I found no ASTM-style "bed bug proof" test standard reachable, and say so in the judgement note rather than implying one exists. ### Reusable bags **Wikipedia, "Reusable shopping bag"** — https://en.wikipedia.org/wiki/Reusable_shopping_bag (kind: primer) - verbatim: "There is a published protocol for testing laboratories involving a 10 kg (22 lb) test load being subjected to a series of static and dynamic laboratory tests." (footnote names "ENVIRONMENTAL CHOICE PROGRAM Reusable Utility Bags (CCD-100)," a Canadian standard I did not independently fetch.) - verbatim: "Other evaluators have used 'walk tests' of people carrying bags with 10 kg (22 lb) of mixed loads for a distance of 175 feet (approx. 50 meters)." - verbatim: "In 2024, California banned sale of all plastic bags at grocery checkouts. The legislation, Senate Bill 1053, closed a loophole that previously allowed thicker plastic 'reusable' bags." - Supports: `load_capacity`'s 10 kg benchmark and the `reusable` trade-off/judgement about regulation shifting from thickness to a load test. I did **not** carry forward any specific mil number for what the old thickness threshold was (I recalled "2.25 mil" from general knowledge, not from anything I fetched this session, so it is correctly absent from `category.json`). ### Units (vocabulary) **Wikipedia, "Thousandth of an inch"** — https://en.wikipedia.org/wiki/Thousandth_of_an_inch (kind: primer) - verbatim: "linear mil = 0.0254 mm (0.001 in)." **Wikipedia, "Units of textile measurement"** (denier) — https://en.wikipedia.org/wiki/Denier_(unit) (kind: primer) - verbatim: "den or D (denier): Grams per 9,000 metres of yarn" and, lower in the article, "In modern usage, it is defined as grams per 9,000 linear [metres]." ## Data points used for derived metrics - `cost_per_sqft_wrap`: no sourced constant needed; the formula is pure arithmetic over `price` and the roll's stated length/width, both already in `supplies-standard.jsonc`'s `sizing.dimensions`. - `cover_fit_slack`: likewise pure arithmetic over `fit_dimensions_stated` and the buyer's own measurement of their furniture; no external number. Neither derived metric needed a researched constant, which is why neither carries an `evidence` array. ## Against the existing add-on schema `zamana-expertise/data/schema-parts/supplies/household/add-ons/bag.jsonc` has exactly one field, `sous_vide_safe` (boolean). The shared `household-standard.jsonc` has `materials` (a household-wide enum: plastic, fabric_textile, canvas, foam, cardboard_paperboard, paper, etc., no finer distinction), `mounting_install` (freestanding, wall_mounted_screw, ... — written for shelves and racks), and `sizing.weight_capacity` ({value, unit}). `supplies-standard.jsonc` gives `high_signal_markdown`, `sizing.dimensions`, `sizing.capacity`, `sizing.weight`, `sizing.pack_count`. Proposals only; nothing edited: - **`sous_vide_safe` is a hard filter that under-informs.** Broaden it to the ordinal `heat_rating` (storage_only / microwave_safe / microwave_and_sous_vide_safe) this register uses as `heat_safe_rating`. One enum field carries more signal than one boolean and matches how brands actually print the claim (a tiered icon set, not a single yes/no). Category quality currency, per AGENTS.md's four-field test — buyers filter on it and we would commit to augmenting it (`backfill_cost: low`). - **Add `closure_type`** (zip_top / slider_press_seal / vacuum_seal / fold_or_twist_tie). Hard filter: this is literally what distinguishes one SKU from another in the zip-top-bag aisle, and it is always in the title. - **Add `bag_grade`** (freezer / storage / snack_or_sandwich). Vertical anchor for this one sub-family: the single word that tells you which product you are looking at, always in the title, always worth augmenting. - **Add `wrap_material`** (plastic_wrap / aluminum_foil / parchment_paper / wax_paper) for the roll/sheet sub-family. The shared `materials` enum is too coarse here (it has generic "plastic" and "paper" but no way to tell cling wrap from foil from parchment, which is the entire mechanism this facet exists to capture) — this needs its own field, not a value added to the shared enum, because "plastic" in the shared enum also covers rigid tubs in other household leaves. - **Add `cover_fabric_type`** (woven_polyester_breathable / pvc_vinyl_coated / canvas / nonwoven_dust_cover) for the cover sub-family. Same reasoning: the shared `materials` enum's "fabric_textile" and "canvas" don't distinguish the one trade-off (waterproof vs. breathable) that this whole sub-family is bought on. - **Add `bed_bug_encasement`** (boolean) for mattress/box-spring/sofa covers. Hard filter, health-adjacent (pest control, not a medical claim), and the title already says "encasement" vs. "cover" so it costs nothing to backfill. I could not confirm a formal certifying standard exists for this claim (see "Could not reach" and the judgement in `category.json`), so I am not proposing the field be sourced to a standard, only to the listing's own wording. - **Reuse, don't duplicate, `sizing.weight_capacity`** for `load_capacity`. I nearly proposed a new `rated_load_kg` field before noticing `household-standard.jsonc` already has exactly this shape. Mark it `researched`/`medium` backfill for this leaf specifically, since it is rarely on the listing (see `load_capacity`'s facet note). - **`mounting_install` is near-total noise for this leaf.** Almost nothing in bags, wraps or covers mounts to anything; it was written for shelves, racks and dispensers. Recommend the pipeline default it to `freestanding`/`none` for this leaf rather than spend backfill effort on it. - **`fit_dimensions_stated` needs no new field.** It is the existing `sizing.dimensions` block (length/width/height/depth), used for covers the same way `garbage_bin_bag.jsonc`'s `fits_can_size_stated` is used for bin bags. I looked for a reason to propose a cover-specific field and did not find one. ## Branches - `intended_use` = "food storage or wrap" → ask `form`, `reusable`, `closure_type`, `freezer_or_storage_grade`, `heat_safe_rating`, `wrap_material`; the cover-only facets (`exposure`, `cover_fabric`, `fit_dimensions_stated`, `pest_proof_encasement`) become irrelevant. Chain: a food bag is never covering furniture, so nothing about weather exposure or fit-to-furniture applies; the facet list for this branch is exactly the union of what the `form` and `reusable` sub-splits ask next. SUPPORTED by the scope decision itself (these facets are defined only in terms of food bags/wraps). - `intended_use` = "furniture or equipment cover" → ask `exposure`, `cover_fabric`, `fit_dimensions_stated`, `pest_proof_encasement`; the food facets (`form`, `reusable`, `closure_type`, `freezer_or_storage_grade`, `heat_safe_rating`, `wrap_material`) become irrelevant. Chain: a cover is never resealed, frozen, or microwaved, so every food-storage facet is undefined for it. SUPPORTED, same reasoning as above in reverse. - `intended_use` = "carry or shopping bag" → ask `reusable`, `load_capacity`; everything else becomes irrelevant. Chain: a carry bag has no closure in the food-bag sense, no indoor/outdoor weatherproofing question in the way a cover does (a tote gets rained on incidentally, not by design), and is not a food-contact product in the sense that drives `heat_safe_rating`. SUPPORTED. - `form` = "a bag you seal" → ask `closure_type`, `freezer_or_storage_grade`; `wrap_material` becomes irrelevant. Chain: a bag is, definitionally, not a roll; it has exactly one closure and one grade, and `wrap_material`'s values (plastic_wrap/foil/parchment/wax paper) describe rolls, not bags. SUPPORTED by the facet's own value list. - `form` = "a roll you tear off" → ask `wrap_material`; `closure_type`, `freezer_or_storage_grade` and `reusable` become irrelevant. Chain: a roll has no closure to ask about; "freezer/storage/snack" grading is a bag convention a roll's own title never uses; and I found no retail cling wrap, foil, parchment or wax paper marketed as reusable (Wikipedia's `Plastic wrap` article describes none), so `reusable` is dropped here, not just unimportant. SUPPORTED. No branch needed a chain longer than one stated fact plus the scope decision itself; none were dropped as WEAK in this pass (there was no separate blind-check pass in this run — see "Corrections" below). ## Trimmed detail - `wrap_material`'s `why_experts_care` was trimmed from 82 to 80 words by the validator's word-cap check (see "Corrections"). The cut sentence was "Foil blocks light and air completely and tolerates oven heat but can't go in a microwave; parchment is heat-safe and non-stick but not a moisture seal; wax paper is neither heat-safe nor a true seal," shortened to "Foil blocks light and air but can't microwave; parchment is heat-safe but not a seal; wax paper is neither." No factual content was dropped, only restated more tersely. - The exact Dow Saran Wrap and brand-FAQ thickness figures Wikipedia's `Plastic wrap` article cites (8–10 μm current, 12.7 μm historical) are recorded above as a secondary summary, not independently fetched, and deliberately kept out of `category.json`'s facets as a hard `range` per the "do not invent ranges" rule — they appear only as flavour text in the sources list above. - The 2024 California SB 1053 detail (closing a thickness-based "reusable" loophole) is in `category.json`'s `reusable` trade-off and a judgement, but the historical thickness number the old loophole actually used (I recalled "2.25 mil" from general knowledge, associated with SB 270, 2014) is deliberately absent, since I never saw that specific figure in anything I fetched this session. ## Corrections No blind-check pass has been run on this register yet (this is the initial authoring pass, 2026-10-25); this section is empty by construction, not by omission. Three things the validator itself caught and I fixed before this file was finalised: split `reusable` was missing its required `rationale` field; `wrap_material`'s `why_experts_care` was two words over the 80-word cap; and the `wiki_mil` reference was defined but never cited by any facet, so it was added to `freezer_or_storage_grade`'s evidence list (the mil unit it defines is the concept behind that facet's grading). `validate.py`, `render.py` and `viewer.py` all ran clean on the first try after those three fixes; no script misbehaved. ## Corrections (2026-10-26) - Required fix 1: rewrote `tradeoffs[0].plain` and `judgements[1]` for `wrap_material`. Refetched `wiki_plasticwrap` directly (curl, saved to `scratch/fix-bags/plasticwrap.html`) and confirmed the source says only that PVC has low permeability to water vapor/oxygen and that plasticizer migration is a public concern, not a sourced "migrates faster than PE" or brand-marketing claim, and that the FDA's 2002 piece is the article's reassurance ("does not pose danger"), not a concern the FDA raised. Both fields now state only what the source supports; `sources_disagree` on the judgement changed to `false` and `confidence` to `high` since there is no actual source conflict, only public perception vs. the FDA's own finding. `wrap_material`'s `why_experts_care` already read correctly before this pass (no change needed there). - Required fix 2: added the missing `wiki_beeswax` reference entry (title, url, kind, accessed, supports) — the `wrap_material` value and its evidence citation for reusable beeswax/cloth wrap were already in the facet, but the reference itself was undefined, which `validate.py` was failing on. Refetched `en.wikipedia.org/wiki/Beeswax_wrap` directly (`scratch/fix-bags/beeswax.html`) to confirm "reusable" and "washed and air-dried." Single-sourced: a second independent source was sought (tried `en.wikipedia.org/wiki/Food_wrap`, which redirects to the same Plastic wrap article, and an EPA page that 404'd) and not found this session; flagging as single-source rather than inventing a second one, consistent with how `entfact_bedbug` and `wiki_tarp` are already flagged single-source elsewhere in this register. The `form`="a roll you tear off" branch already asks `reusable` rather than dropping it, so no split change was needed. - Required fix 3: `cover_fabric` backfill_cost raised from `low` to `medium` and `backfill_note` now states plainly that the "usually on the listing" claim is unverified for furniture/patio covers specifically, since every cover-specific retailer page fetched this session was blocked. - Required fix 4: `load_capacity` backfill_cost raised from `medium` to `high` and `backfill_note` now names the brand and regulatory pages that were blocked this session, since no page carrying a load number was actually reachable. - Required fix 6: `cover_fit_slack.depends_on` now lists `furniture_dimensions` alongside `fit_dimensions_stated`, and `why` states it is the buyer's own measurement, not a listing field, so the formula's second term is no longer undefined. - Required fix 5 (split questions failing the generic-trap test) and Required fix 7 (`intended_use`'s definition broad enough to absorb siblings): rejected as already resolved, no change made. The `reusable` split question now reads "Disposable bag or wrap, or reusable?" and the `exposure` split reads "Indoor dust cover, or outdoor weather cover?" — both anchor to bag/wrap/cover vocabulary rather than check.md's quoted "Use it once, or wash and reuse it?" / "Covering it indoors, or leaving it outside?" text, which is no longer in the file. `intended_use`'s `definition` already reads "...or is a reusable shopping or produce bag," not check.md's quoted "carries loads like a shopping bag." Neither field matches what check.md describes as broken, so there was nothing left to fix for these two items. - `validate.py`, `render.py` and `viewer.py` all ran clean after these changes (`BUILD_OK`).