# Cutting boards, cooling racks and prep surfaces: research notes

Date: 2026-10-25. Scope: food-prep cutting boards and flexible cutting mats (hardwood, bamboo, plastic, soft rubber or soft polyethylene, paper composite, glass or stone), wire cooling and roasting racks sold on their own, and flat prep slabs (marble, granite or glass pastry slabs, wooden pastry boards); not knives (sibling `knives`), sheet pans or silicone baking and pastry mats (sibling `baking-sheets`, "Mats & Liners"), pizza stones and peels (sibling `baking-pans`), or sink drain boards (`dish-racks`, household). Taxonomy: `supplies > kitchen > boards`.

Overlaps left to siblings: a rack sold as a set with a sheet pan belongs to `baking-sheets` when the pan is the lead product; this register only uses the pan's size as the fit target for a standalone rack. Silicone pastry mats with printed rings are "Mats & Liners" and stay with `baking-sheets`. Serving-only platters (slate cheese boards) are borderline with `bottle-wine-serving-tools`; boards sold for cutting and serving stay here.

Quote labels: **verbatim** = the sentence in quotation marks was in raw text I read myself (curl plus tag stripping, pdftotext, or the PubMed E-utilities plain-text abstract); **fetched excerpt** = paraphrased from raw text I read, because the passage is long; **search excerpt** = the claim came only from a search-engine summary. No search-excerpt-only number is in category.json.

Research constraints this run: Serious Eats refuses Anthropic's agent at the search API (HTTP 400, "not accessible to our user agent"), so I did not fetch it by other means. Wirecutter (nytimes.com) returned 403 to curl. NSF's own pages returned 403 and have no Wayback snapshot. The session's shared web-search budget (200 calls across concurrent agents) ran out partway through; the remaining sources were reached by direct fetches of known URLs, sitemaps and Wikipedia's citations.

## Seed

Seed output, verbatim:

```
=== Example entry (from an unrelated category) ===
category: bedsheets
what this shows: the expert split is often not the number everyone advertises. Thread count is the generic trap here. Weave is what changes how the sheet feels and sleeps.

<example>
{
  "id": "weave",
  "label": "Weave",
  "definition": "The pattern the threads are crossed in, which decides how the sheet feels and how it handles heat.",
  "why_experts_care": "Percale is a simple one-over-one weave that feels crisp and cool and gets softer with washing. Sateen puts more thread on the surface so it feels silky and warmer and shows wear sooner. Thread count, the number most listings lead with, matters far less than which of these two you pick, and above about 400 it is mostly marketing.",
  "kind": "categorical",
  "values": ["percale", "sateen", "linen", "flannel", "jersey"],
  "roles": ["split", "filter"],
  "novice_answerable": true,
  "cold_start_question": "Do you sleep hot, or do you like a sheet that feels smooth and a bit warm?",
  "source": "listing",
  "listing_field_hint": "Usually in the title or the first bullet. Sometimes only in the fabric care tab.",
  "backfill_cost": "low",
  "backfill_note": "When missing, the product photo and the description words 'crisp' or 'silky' usually settle it.",
  "visual": { "chartable": false },
  "priority": { "narrows_set": 4, "cost_to_answer": 1 },
  "evidence": ["wirecutter_sheets", "textile_primer"]
}
</example>

=== Persona ===
A cycling mechanic, who thinks in compatibility, standards and the part that has to match

Ask yourself two things and write the answers under 'Seed' in notes.md:
  1. What is the kitchen-boards equivalent of the example above?
  2. What would this persona notice first about kitchen-boards that a typical buyer would not?
Keep what survives the data check. Drop the rest.
```

1. **Equivalent of weave versus thread count.** The words listings lead with on boards are "antibacterial", "knife-safe", "eco" and the size. The thing that changes how a board behaves is what the surface is made of and, for wood, which way the grain faces. That decides how fast a knife dulls and how deep the scars get, which is what decides hygiene in the long run (Cliver: knife-scarred plastic is the hard case). Survived: `board_material` as a split, `grain_construction` and the derived `knife_friendliness` scale. Survived as a judgement: Epicurean labels its paper-composite boards "Knife Safe" while TechGearLab found composite dulled knives fastest. Dropped: an "antibacterial claim" facet; every material's marketing claims it, and the studies are about surface wear and drying, not the claim.
2. **What the cycling mechanic notices first: the part that has to match.**
   - The rack has to match the pan. A half-sheet rack is about 11.8 × 16.9 in so that it sits inside an 18 × 13 in half-sheet pan (Kitchenatics verbatim; Wikipedia and TechGearLab for the pan). Survived as `sheet_pan_fit` and the `pan_size_from_rack` derived metric.
   - The board has to match the knife. Hard, thin Japanese edges are what Knifewear and Hasegawa sell soft boards for. Survived as the `knife_friendliness` scale and its visual grammar; no split, because the register cannot ask the knife's hardness without a lecture.
   - The board has to fit the dishwasher and the cabinet. Tom's Guide: the 21 in OXO "could be too big to fit in the lower-rack snugly"; IKEA's 18 × 20.8 in Lämplig is "too big to fit under standard-height cabinets if you stand it up". Survived as a note inside `board_size`.
   - The oil has to match. Wikipedia says plant oils go rancid and most linseed and tung oils are not food grade; only one source, so kept here and in the vocabulary entry for board oil in plain words without the rancid claim.
   - Dropped: replacement feet fitting only one series (Epicurean's "Non-Slip Board Corner Replacement Feet ... exclusively for the Non-Slip Series boards"). True, trivial.

## Sources and what each supports

### Regulators and standards

**FDA Food Code 2022, Chapter 4** (copy hosted by Champaign-Urbana Public Health District). https://www.c-uphd.org/documents/eh/2022-FDA-Food-Code-Chapter-4-Equipment-Utensils-and-Linens.pdf (fda.gov download links 404'd). pdftotext. Confidence high.
- **verbatim** 4-101.17: "(A) Except as specified in ¶ (B), (C), and (D) of this section, wood and wood wicker may not be used as a FOOD-CONTACT SURFACE. (B) Hard maple or an equivalently hard, close-grained wood may be used for: (1) Cutting boards; cutting blocks; bakers' tables; and UTENSILS such as rolling pins, doughnut dowels, salad bowls, and chopsticks".
- **verbatim** 4-501.12 Cutting Surfaces: "Surfaces such as cutting blocks and boards that are subject to scratching and scoring shall be resurfaced if they can no longer be effectively cleaned and SANITIZED, or discarded if they are not capable of being resurfaced."
- **verbatim** 4-101.11: food-contact materials shall be "(B) Durable, CORROSION-RESISTANT, and nonabsorbent ... (E) Resistant to pitting, chipping, crazing, scratching, scoring, distortion, and decomposition."
- Supports: the governing-standard judgement; why hard close-grained species are the norm; resurface-or-discard rule.

**USDA FSIS, "Cutting Boards"**. https://www.fsis.usda.gov/food-safety/safe-food-handling-and-preparation/food-safety-basics/cutting-boards (live page 403; read via Internet Archive snapshot `web.archive.org/web/2025/...`). Confidence high.
- **verbatim** "Consumers may choose either wood or a nonporous surface cutting board such as plastic, marble, glass, or pyroceramic. Nonporous surfaces are easier to clean than wood."
- **verbatim** "consider using one cutting board for fresh produce and bread and a separate one for raw meat, poultry, and seafood."
- **verbatim** "Nonporous acrylic, plastic, or glass boards and solid wood boards can be washed in a dishwasher (laminated boards may crack and split)."
- **verbatim** "Both wooden and plastic cutting boards can be sanitized with a solution of 1 tablespoon of unscented, liquid chlorine bleach per gallon of water."
- **verbatim** "Bamboo cutting boards are harder and less porous than hardwoods. Bamboo absorbs very little moisture and resists scarring from knives".
- **verbatim** "Once cutting boards become excessively worn or develop hard-to-clean grooves, they should be discarded."
- Supports: separate raw-meat board; the dishwasher disagreement; bamboo hardness; replace when grooved.

### Academic and extension

**Ak, Cliver, Kaspar 1994a, "Cutting Boards of Plastic and Wood Contaminated Experimentally with Bacteria"**, J Food Prot 57(1):16-22, PMID 31113021. PubMed abstract via E-utilities. Confidence high.
- **verbatim** "New and used Plastic (four polymers plus hard rubber) and wood (nine hardwoods) cutting boards"; "Bacteria inoculated onto Plastic blocks were readily recovered for minutes to hours and would multiply if held overnight. Recoveries from wooden blocks were generally less than those from plastic blocks"; "Clean wood blocks usually absorbed the inoculum completely within 3-10 min"; "If ≥106 CFU were applied, bacteria might be recovered from wood after 12 h ... but numbers were reduced by at least 98%, and often more than 99.9%. Mineral oil treatment of the wood surface had little effect"; "These results do not support the often-heard assertion that Plastic cutting boards are more sanitary than wood."
- Note: the popular "99.9 percent die within 3 minutes" version (search excerpts) overstates the abstract. The 3 to 10 minutes is absorption; the 98 to 99.9 percent reduction is after 12 hours at high inoculum. The register says only what the abstract says.

**Ak, Cliver, Kaspar 1994b, "Decontamination of Plastic and Wooden Cutting Boards for Kitchen Use"**, J Food Prot 57(1):23-30, PMID 31113026. Confidence high.
- **verbatim** "New plastic cutting surfaces were relatively easy to clean and were microbiologically neutral, but plastic boards with extensive knife scars were difficult to clean manually, especially if they had deposits of chicken fat on them."

**Cliver 2006, "Cutting boards in Salmonella cross-contamination"**, J AOAC Int 89(2):538-42, PMID 16640304. Confidence high.
- **verbatim** "In general, wood is said to dull knives less than plastic, and plastic is seen as less porous than wood."; "High-density polyethylene, which is most used in commercial applications, has been shown to delaminate in response to knife scarring."; "The moisture is drawn in by capillary action until there is no more free fluid on the surface"; "2 epidemiological studies seem to show that cutting board cleaning habits have little influence on the incidence of sporadic salmonellosis. Further, one of these studies indicated that use of plastic cutting boards in home kitchens is hazardous, whereas use of wooden cutting boards is not."

**Yadav et al. 2023, "Cutting Boards: An Overlooked Source of Microplastics in Human Food?"**, Environ Sci Technol 57(22):8225-8235, PMID 37220346. PubMed abstract (pubs.acs.org returned 403). Confidence medium (one lab, their own assumptions).
- **verbatim** "The mass and number of microplastics released from polypropylene chopping boards were greater than polyethylene by 5-60% and 14-71%, respectively."; "Based on our assumptions, we estimated a per-person annual exposure of 7.4-50.7 g of microplastics from a polyethylene chopping board and 49.5 g of microplastics from a polypropylene chopping board."; "The preliminary toxicity study of the polyethylene microplastics did not show adverse effects on the viability of mouse fibroblast cells for 72 h."

**TODAY, "Are plastic cutting boards safe to use? We asked a chemical engineer"**. https://www.today.com/shop/are-plastic-cutting-boards-safe-rcna196304. curl. Second source for the order of magnitude. Confidence medium.
- **verbatim** "it was estimated that one person could be exposed to up to around 50 grams of microplastics annually from a polyethylene chopping board or a polypropylene chopping board."

**NC State News, "Fast Facts About Cutting Boards and Food Safety in Your Kitchen"** (Ben Chapman, food safety researcher), 2014. https://news.ncsu.edu/2014/09/cutting-boards-food-safety/. curl. Confidence medium-high.
- **verbatim** "Hardwoods, like maple, are fine-grained, and the capillary action of those grains pulls down fluid, trapping the bacteria – which are killed off as the board dries after cleaning"; "Soft woods, like cypress, are less likely to dull the edge of your knife, but also pose a greater food safety risk"; "Chapman recommends using plastic cutting boards for meat and wood cutting boards for fruit, vegetables, or any ready-to-eat foods"; "But wood cutting boards would quickly be ruined by a dishwasher".
- Also: chlorine bleach for plastic, quaternary ammonium for wood, because chlorine binds to organic material (kept in Trimmed detail).

**University of Maine Cooperative Extension, "Cutting Board Safety"** (2022). https://extension.umaine.edu/food-health/2022/10/27/cutting-board-safety/. curl. Confidence medium.
- **verbatim** "it might be helpful to use color coded cutting boards for specific food products."; "When choosing a cutting board, lean more towards nonporous surfaces"; "Nonporous and solid wood boards can also be cleaned in the dishwasher".
- Sides with FSIS on the dishwasher and leans nonporous, against Cliver; logged under judgements.

**USDA Forest Products Laboratory, Wood Handbook (FPL-GTR-190), Chapter 5, Kretschmann 2010**. https://research.fs.usda.gov/download/treesearch/37427.pdf. pdftotext. Confidence high.
- **verbatim** table 5-3b (inch-pound), side hardness at 12% moisture: Cherry, black 950; Maple, sugar 1,450; Walnut, Black 1,010 (lbf). "side hardness is hardness measured when load is perpendicular to grain."

### Primers and wood references

**The Wood Database** species pages (Eric Meier). curl. Confidence high for numbers.
- Hard maple https://www.wood-database.com/hard-maple/ **verbatim** "Janka Hardness: 1,450 lbf (6,450 N)"; common uses include "cutting boards, butcher blocks".
- Black walnut https://www.wood-database.com/black-walnut/ **verbatim** "Janka Hardness: 1,010 lbf (4,490 N)".
- Black cherry https://www.wood-database.com/black-cherry/ **verbatim** "950 lbf (4,230 N)".
- Teak https://www.wood-database.com/teak/ **verbatim** "Janka Hardness: 1,070 lbf (4,740 N)"; "teak contains a high level of silica (up to 1.4%) which has a pronounced blunting effect on cutting edges." (The context is woodworking tools; the step to kitchen knives is Tom's Guide's, below.) Teak's 1,070 is single-source and is not in the JSON.

**Wikipedia, "Cutting board"**. https://en.wikipedia.org/wiki/Cutting_board. curl. Confidence low to medium (much is "citation needed"); used only as a second source.
- **verbatim** "Hardwoods with tightly grained wood and small pores are best"; "Red oak, though a hardwood, has large pores which retain dirt even after washing."; "glass cutting boards can damage knives because of the high hardness of the material. Cutting on glass tends to dent, roll or even chip knife edges"; "There are also cutting boards made of glass, steel, or marble ... but tend to damage blades due to their hardness."; "Wooden boards should never be placed in the dishwasher"; "To prevent cracking, wood cutting boards should be treated monthly under normal usage."; "HDPE will be warped by the hot water"; "bamboo is commonly thought to be naturally antimicrobial (although studies show otherwise [Xi et al. 2013])"; the colour code (blue raw seafood, red raw meat, green salads and fruit, yellow cooked meat, brown root vegetables, white dairy and bakery).

**Wikipedia, "Sheet pan"**. https://en.wikipedia.org/wiki/Sheet_pan. curl. Confidence high for sizes (cites a trade source).
- **verbatim** "A half sheet pan is 18 in × 13 in (460 mm × 330 mm); quarter sheets are 9 in × 13 in"; "Another popular size for home baking is 10 in × 15 in ... jelly roll"; "quarter sheets will not necessarily nest inside a half sheet side by side".

**Center for Environmental Health, "What's in Wood Cutting Boards?"** https://ceh.org/yourhealth/whats-in-wood-cutting-boards/ (live 403; read via Wayback 2025). Confidence low (advocacy group; no measurements).
- **verbatim** "If you choose a cutting board made from multiple glued pieces ... make sure the glue used is free from formaldehyde and melamine."; "When FDA-approved food contact substances like glue resins/polymers are completely cured (totally dried), they are considered food safe".

### Specialist reviewers

**TechGearLab, "The Best Cutting Boards"** (Ian Nicholson). https://www.techgearlab.com/topics/kitchen/best-cutting-board. curl. Tested 11 boards; a dedicated knife per board for a week, then re-sharpened. Confidence high for relative findings.
- **verbatim** "Models with end-grain hard construction are by far the best, as the knife blade can slip in between the fibers"; "Bamboo and wood are generally similar but tougher on knives than end-grain hardwoods."; "There was no question that composite boards were the hardest of all materials and the ones that dulled our knives the quickest"; "The least knife-friendly options are composite models like the Epicurean"; plastic "will dull knives faster ... than hardwood and bamboo options"; "We found the Imeko TPU and the HOMWE to be the most knife-friendly" (among plastics); "At 1.5 inches thick and roughly 10 pounds" (Sonder) and "this 1.5-inch thick model (weighing 10 pounds)" (Boos); "hardwood boards require good care, such as oiling them once a month"; "a downside of all wood models is that they should not be cleaned in a dishwasher"; "thicker than many other plastic options, which helps it reduce warping, even after repeated cycles in the dishwasher" (Gorilla Grip); Farberware "will generally start to warp after repeated washes"; "most from 8 x 10" to 18 x 24""; "We often encourage people to get a bigger board than they think they need"; rubberised feet, weight or tacky material are the three ways boards stay put.

**TechGearLab, "The Best Cooling Rack"** (Aaron Rice). https://www.techgearlab.com/topics/kitchen/best-cooling-rack. curl. Tested 8 racks. Confidence high for relative findings.
- **verbatim** "This rack is double the height of the other single racks we tested, standing at 1 inch high and offering superior airflow." (Kitchenatics); "the manufacturer's claim that this rack can support up to 45 pounds is true — we tested it"; "oven safe up to 575℉"; "A half sheet ... with a standard measurement of 18 x 13 inches ... a half-sheet rack is designed to fit inside a half-sheet pan."; "This comes down to two main design points: elevation height and size and spacing of the lattice frame."; "the straight rack wires are spaced ⅝-inch apart, and cookies easily slide through" (Wilton Excelle); "The majority of the racks we tested are 100% food-grade stainless steel and are oven-safe up to 575℉. Only the three-tiered racks ... have a non-stick coating and should never be put into the oven."; "It is tough to beat the price-per-square-inch of the HIWARE Half Sheet 2-Pack"; the Ultra Cuisine was "terribly bowed" so cookies slid off; the Wilton 3-Tier collapsible gives "480 square inches of cooling space from only a 160 square inch footprint".
- Single-source and left out of the JSON: "standard ⅜-inch gridded rack"; "the standard height is ½-inch".

**Tom's Guide, "Best cutting boards: tested and rated"**. https://www.tomsguide.com/best-picks/best-cutting-boards. curl. Ten boards. Confidence medium.
- **verbatim** "teak's high silica content may accelerate the dulling of your knives over time."; Teakhaus "smallest version of this board measures 16x12x1.5 inches and weighs 10 pounds, and the largest comes in at 24x18x1.5 inches and 15 pounds"; Boos "at 13 pounds the board is extremely heavy"; OXO "may be too big for smaller countertops at 21 inches long" and "could be too big to fit in the lower-rack snugly"; Lämplig "too big to fit under standard-height cabinets if you stand it up"; "If you do place a wooden cutting board in a dishwasher the high heat and steam will cause it to warp"; "Another bonus to having colour-coded boards is you can assign different boards for veggies and raw meat."; chef Michael Handal (ICE): "maple, cherry, and walnut are all popular choices"; board wax "a combination of food-safe beeswax and mineral oil".

### Enthusiast community

**Knifewear, "The Ultimate Cutting Board Comparison Guide"** (Nathan, Knifewear; a Japanese-knife retailer whose blog is the knife-enthusiast reference). https://knifewear.com/blogs/articles/the-ultimate-cutting-board-comparison-guide-home-professional-use. curl. Confidence medium (seller, but specific).
- **verbatim** "Countertops, glass, granite, and even bamboo are very hard, so they quickly dull your knives."; end grain: "your edge contacts the board, it's cutting between the fibres of the wood, rather than against them"; "the moisture in the wood will cause it to swell slightly, and those gaps will slowly fill with time"; "end-grain doesn't warp horizontally like regular boards"; side grain "Remember, wood is always better than a harder material"; end grain boards are "bulky and heavy" with "a higher price point"; Hasegawa "The outer layer is quite soft and easier on your knife edges than any other board on this list" and "They're also our only option that's dishwasher safe!"; plastic "can gouge and harbour bacteria easily".

**Knifewear, Hasegawa FSR Pro-Soft product page**. https://knifewear.com/en-us/products/hasegawa-soft-cutting-board-with-wood-core. curl. Confidence medium.
- **verbatim** "soft surface that gives way to your knife's edge ... while the rigid wooden core keeps the board sturdy and prevents warping"; "The super-soft surface can scratch up more easily than wooden boards."; "Dishwasher & bleach safe."

**MADE in JAPAN.JP, "Asahi Cookin' Cut"**. https://made-in-japan.jp/en/parkerasahi/. curl (Parker Asahi's own PDF would not download). Confidence medium-low (secondary site summarising the maker's PDF).
- **verbatim** "Parker Asahi was the first in Japan to develop a synthetic-rubber chopping board, and the product has been sold since 1965."; "avoid dishwashers when cautioned by the seller due to deformation risk"; sizes "240×210×13mm" to "1000×330×20mm"; heat guidance "PRO: 'up to 130°C for 30 minutes' and HOME: 'up to 100°C.'" (single source; not in JSON).

### Spec sheets (fields brands list)

**John Boos, R-Board series**. https://www.johnboos.com/products/maple-cutting-boards-1-1-2-thick-r-board-series. curl. Confidence high for what the page says.
- **verbatim** "Made of solid North American Maple, these NSF Certified John Boos maple cutting boards are 1-1/2” thickness ... Each maple end grain cutting board includes recessed finger grips and is reversible"; and on the same page "Sustainably sourced and renewable, John Boos maple R-Board cutting boards feature robust edge-grain construction with a 1-1/2” thickness". The brand's own page calls one product both end grain and edge grain.
- Fields: species ("Northern Hard Rock Maple (NSF Certified)"), thickness, size, reversible, grips.

**John Boos, edge-grain collection**. https://www.johnboos.com/collections/edge-grain-cutting-boards. curl.
- Thickness lines on the page: 1", 1-1/4", 1-1/2", 1-3/4", 2-1/4". **verbatim** Prestige: "one side includes the juice groove for cooked meats and the other side is flat for fruits and vegetables"; "The Northern Hard Rock Maple has the ideal hardness and is resilient durable, long lasting, sanitary and knife-friendly."

**Epicurean, Kitchen Series**. https://www.epicurean.com/products/kitchen-series (and products.json). curl.
- Feature labels on page: "Dishwasher Safe", "Heat Resistant", "Knife Safe", "NSF Certified", "Paper Composite"; variants "Natural / 8" × 6" / 1/4"" up to "17.5" × 13" / 1/4"". Board Butter: "A proprietary blend of mineral oil and beeswax".
- Not confirmed from raw text: the "350 °F" heat figure from search summaries. Left out.

**Kitchenatics, half-sheet stainless rack**. https://kitchenatics.com/shop/premium-quality-cooling-roasting-stainless-steel-rack/half-size-cooking-rack-commercial-quality/. curl.
- **verbatim** "COMMERCIAL GRADE 304 (18/8) STAINLESS STEEL ... DISHWASHER-SAFE"; "OVEN-SAFE to 575˚F."; "PERFECT FIT FOR POPULAR BRANDS of large 13″ x 18″ half sheet pan ... Rack Dimensions: 11.81″ x 16.85″"; "with 1″-inch high legs"; quarter rack "8.5″ x 12″ x 1”".

**Wilton** (BigCommerce product pages, curl).
- Gold Non-Stick Cooling Rack 16 × 10 in, https://wilton.com/perfect-results-premium-non-stick-cooling-rack-16-x-10-inch/191003003/ **verbatim** "Nonstick cooling grids are not oven-safe"; "The grid-like surface won't let small baked goods fall through the cracks".
- Chrome-Plated Cooling Grid 10 × 16 in, https://wilton.com/chrome-plated-cooling-grid-10-x-16-inch/191004716/ **verbatim** "Chrome-plated tightly woven grid keeps small items from falling through"; "hand wash in warm, soapy water; rinse and dry thoroughly".
- 3-Tier Collapsible, https://wilton.com/3-tier-collapsible-cooling-rack/191003104/ **verbatim** "Not Oven Safe. Handwash for best results."

### Could not reach or did not use

Serious Eats (agent blocked by the site; not circumvented). Wirecutter (403). America's Test Kitchen (paywall; not attempted). NSF/ANSI 2 pages (403, no snapshot). Parker Asahi PDF (connection refused). Nordic Ware product pages (Incapsula block; the sitemap shows an "oven-safe nonstick grid" product, which suggests some coated racks are oven-rated, but I could not read the rating, so the register only says what Wilton and TechGearLab say about the coated racks they cover). Amazon listings were not attempted; earlier registers record that they block fetches.

## Data points used for derived metrics

| Rack or pan | Size | Source |
|---|---|---|
| Half-sheet pan (outer) | 18 × 13 in | Wikipedia Sheet pan; TechGearLab racks |
| Quarter-sheet pan | 9 × 13 in | Wikipedia; TechGearLab |
| Jelly-roll pan | 10 × 15 in | Wikipedia |
| Kitchenatics half-sheet rack | 11.81 × 16.85 in, 1 in legs | Kitchenatics page; TechGearLab (1 in) |
| Kitchenatics quarter-sheet rack | 8.5 × 12 in | Kitchenatics page |
| Wilton chrome and nonstick grids | 10 × 16 in | Wilton pages (not a standard pan size) |

`pan_size_from_rack`: the smallest standard pan whose length and width both exceed the rack's. Both Kitchenatics racks sit about 1 in inside their pans on each dimension (13 − 11.8, 18 − 16.85; 9 − 8.5, 13 − 12), so "exceeds" is the rule, not a fixed margin; I did not find a sourced margin. A 10 × 16 in grid exceeds a jelly-roll pan's 15 in length and fits a half sheet.

`price_per_sq_in`: TechGearLab's own comparison for racks ("price-per-square-inch") and the square-inch areas it reports (half-sheet HIWARE 195 sq in, Wilton tiered 480 sq in on a 160 sq in footprint). Applied to boards with length × width.

## Against the existing add-on schema

Read: `supplies/kitchen/add-ons/board.jsonc` (`grain_construction` enum, `sizing.length`, `sizing.width`, `sizing.thickness` each with in/cm), composed with `kitchen-standard.jsonc` (`materials[]` with material and role, `max_heat_temp`, `dishwasher_safe`) and `supplies-standard.jsonc` (`high_signal_markdown`, `sizing.dimensions`, `sizing.weight`, `sizing.pieces_included_text`, `sizing.pack_count`). Proposals only; nothing in zamana-expertise was edited.

| Facet in this register | In schema today | Proposed kind (AGENTS.md) | Note |
|---|---|---|---|
| product_job | no | hard filter | Proposal: `board_type` enum [cutting_board, flexible_mat, cooling_rack, prep_slab]. The leaf bundles three products that share almost no other field; every downstream field depends on it. |
| board_material | yes, `materials[]` | vertical anchor | Works today with roles: Hasegawa = plastic(exterior_surface) + wood(core). Gaps: `plastic` does not separate polyethylene from polypropylene (see plastic_polymer); "paper composite" lands in `composite`, which is fine. |
| grain_construction | yes | quality currency (wood only) | Keep. Borderline under rule 1 because the checkerboard is visible in photos, but listings state it and the photo cannot settle edge versus face grain. Boos's own page calls one board both end and edge grain, so extraction needs a tie-break (photo wins when text contradicts itself). |
| wood_species | no; `materials` has only `wood` | quality currency (wood only) | Proposal: `wood_species` enum [maple, walnut, cherry, teak, acacia, olive, beech, other, unspecified]. Titles carry it; it drives hardness and the teak-silica note. |
| knife_friendliness | no | derived at query time | From board_material plus grain_construction. Not stored. |
| board_size | partly: add-on `sizing.length`/`width` and standard `sizing.dimensions` | sizing | The add-on's length, width and thickness duplicate `supplies-standard.sizing.dimensions` (length, width, height). Rule 5 cuts leaf duplicates of standard fields: keep one set. |
| thickness | yes, `sizing.thickness` (duplicate of dimensions.height) | sizing | Keep in the standard block only. Bought by number for wood boards (1 to 2¼ in at Boos). |
| juice_groove | no | visual / high_signal_markdown | A photo shows it; rule 1 keeps it out of the schema. Register keeps it as an annotation. |
| dishwasher_safe | yes, kitchen standard | hard filter | Keep. The "top_rack_only" value is real for some plastics. FSIS and brands disagree for solid wood; extract the brand's statement. |
| nsf_certified | no; rule 5 cuts certifications | high_signal_markdown | Boos and Epicurean pages state it. Belongs in the markdown line. |
| plastic_polymer | no | quality currency (plastic only), or a `materials` enum extension | Proposal: add `polyethylene`, `polypropylene`, `tpu` to the kitchen `materials` enum. Yadav et al. found PP shed 5 to 60 percent more microplastic mass than PE. |
| colour_coded_set | partly, `pack_count` | high_signal_markdown | Count is in sizing; the colour scheme is a photo fact. |
| rack_material | partly: `materials` has stainless_steel and ptfe_nonstick (coating) | vertical anchor | Gap: no chrome value. Proposal: add `chrome_plating` to `materials` for role `coating`. |
| sheet_pan_fit | no | hard filter | Proposal: `fits_pan_size` enum [quarter_sheet, jelly_roll, half_sheet, two_thirds_sheet, full_sheet, not_pan_sized]. Titles say "fits half sheet pan". Derivable from rack dimensions when absent. |
| rack_grid | no | visual | A photo shows grid versus parallel wires. |
| max_heat_temp | yes, kitchen standard | hard filter (racks) | Keep. Stainless racks state 575 °F; coated racks state "not oven-safe", which should land as an explicit null-with-flag, not as unspecified. |
| leg_height | no | high_signal_markdown | Eyeballable in relative terms; the number is in bullets. Not worth a column. |
| rack_format | partly, `pieces_included_text` | high_signal_markdown or visual | Tiered and collapsible racks are obvious in photos. |

Schema fields this register treats as noise for this leaf: the add-on's `sizing.length/width/thickness` (duplicates of the standard dimensions block); `supplies-standard.sizing.capacity` (no board or rack is bought by capacity); `grain_construction: not_applicable` should be implied by a non-wood material rather than extracted.

## Branches

The `when` values are entries of each split facet's values list. Question-to-value rules: split 1 offers its three values verbatim. Split 2 offers four answers; "wood" maps to `hardwood`, the other three map to the value of the same name (`bamboo`, `plastic`, `soft rubber or soft polyethylene`); `paper composite` and `glass or stone` are reachable as filters but not offered. Split 3 offers its three values verbatim.

**Split 1, product_job ("Cutting on it, cooling bakes, or rolling dough?")**
- when `cooling rack`: then_ask rack_material, sheet_pan_fit, rack_format, rack_grid; makes_irrelevant board_material, grain_construction, wood_species, knife_friendliness, juice_groove, plastic_polymer, colour_coded_set.
  - then_ask: TechGearLab racks "two main design points: elevation height and size and spacing of the lattice frame" (rack_grid, leg_height via rack_material's stainless branch); "a half-sheet rack is designed to fit inside a half-sheet pan" (sheet_pan_fit); the tiered racks are a separate class it tested for counter space (rack_format); material decides oven use (rack_material, Wilton verbatim).
  - makes_irrelevant: each of these facets is defined on a cutting surface (their definitions in the register). A wire rack has no cutting surface. Trivial.
- when `cutting board`: then_ask board_material, board_size, dishwasher_safe; makes_irrelevant rack_material, sheet_pan_fit, rack_format, rack_grid, leg_height.
  - board_material: mechanism, "surface hardness trades knife wear against scar depth". board_size: TechGearLab "get a bigger board than you think". dishwasher_safe: the dishwasher is the cleaning question the sources argue over (judgement 2).
  - makes_irrelevant: rack facets describe wire racks only. Trivial.
- when `prep slab`: makes_irrelevant rack_material, sheet_pan_fit, rack_format, rack_grid, leg_height. Trivial. No then_ask: I found no sourced spec that a pastry slab buyer compares on, so the branch only prunes.

**Split 2, board_material ("Wood, bamboo, plastic, or soft rubber?")**
- when `hardwood`: then_ask grain_construction, wood_species; makes_irrelevant plastic_polymer.
  - grain_construction: TechGearLab "end-grain ... by far the best" for knives; Knifewear on end grain self-healing and not warping. wood_species: Food Code "hard maple or an equivalently hard, close-grained wood"; Wood Handbook and Wood Database hardness; teak silica (Tom's Guide, Wood Database).
  - plastic_polymer irrelevant: it is defined for plastic boards. Trivial.
  - Dropped leg: "dishwasher_safe irrelevant on hardwood". FSIS says solid wood can go in a dishwasher, so the question is contested (judgement 2), not settled. Kept asked.
- when `bamboo`: then_ask knife_friendliness; makes_irrelevant wood_species, plastic_polymer.
  - knife_friendliness: Knifewear "even bamboo are very hard, so they quickly dull your knives"; TechGearLab bamboo tougher on knives than end-grain hardwood; FSIS "harder ... than hardwoods". The buyer who chose bamboo for price or weight should see the knife cost.
  - wood_species irrelevant: bamboo is the species. Trivial.
- when `plastic`: then_ask plastic_polymer, thickness, colour_coded_set; makes_irrelevant grain_construction, wood_species.
  - plastic_polymer: Yadav et al. PP versus PE (5 to 60 percent more mass).
  - thickness: chain of two facts. (1) Plastic boards go through the dishwasher (FSIS; NCSU). (2) Hot water warps thin HDPE and thicker plastic resists it (TechGearLab Gorilla Grip and Farberware; Wikipedia). So thickness predicts warping on this branch.
  - colour_coded_set: chain of two facts. (1) Sources recommend a separate board for raw meat (FSIS) and plastic for meat (NCSU Chapman). (2) The colour-coded sets the sources describe are plastic (Tom's Guide Joseph Joseph Folio; UMaine). WEAK-acceptable: colour-coded wood sets may exist; none was in my sources.
  - makes_irrelevant: grain and species are wood facets. Trivial.
- when `soft rubber or soft polyethylene`: then_ask dishwasher_safe; makes_irrelevant grain_construction, wood_species.
  - dishwasher_safe: the two named products differ. Hasegawa is "Dishwasher & bleach safe" (Knifewear product page; Knifewear guide "our only option that's dishwasher safe"); Asahi rubber boards carry "deformation risk" in dishwashers (MADE in JAPAN.JP). So on this branch the answer splits the set.
  - makes_irrelevant: grain and species are wood facets. Hasegawa's wood core is not a cutting surface. Trivial.

**Split 3, rack_material ("Stainless (oven-safe), chrome-plated, or nonstick (not for the oven)?")**
- when `stainless steel`: then_ask sheet_pan_fit, max_heat_temp, leg_height.
  - sheet_pan_fit: oven use is rack-in-pan ("the pairing of a rack and a baking sheet", TechGearLab; Kitchenatics "PERFECT FIT ... half sheet pan"). max_heat_temp: "oven safe up to 575℉" (TechGearLab; Kitchenatics). leg_height: TechGearLab's 1 in rack "offering superior airflow" for "roasting and air frying".
- when `chrome-plated steel`: then_ask dishwasher_safe.
  - Wilton chrome grid: "hand wash ... rinse and dry thoroughly"; Kitchenatics stainless: "DISHWASHER-SAFE". The material changes the care answer.
- when `nonstick-coated steel`: makes_irrelevant max_heat_temp, leg_height.
  - max_heat_temp: Wilton "Nonstick cooling grids are not oven-safe"; TechGearLab "should never be put into the oven". No oven, no oven temperature.
  - leg_height: chain. (1) TechGearLab values height for airflow when "cooking and cooling" and singles out roasting and air-frying. (2) A nonstick rack is cooling-only. WEAK: height still helps cooling (TechGearLab says height also cools faster). Dropped this leg; only max_heat_temp is made irrelevant.

## Numbers in category.json and how many sources stand behind each

| Number | Sources |
|---|---|
| Janka / side hardness: maple 1,450, walnut 1,010, cherry 950 lbf | Wood Handbook table 5-3b; Wood Database pages |
| PP sheds 5 to 60 percent more mass than PE; 7.4 to 50.7 g/yr PE, 49.5 g/yr PP | Yadav et al. abstract (primary); TODAY "up to around 50 grams" for the order of magnitude |
| Wood absorbs inoculum in 3 to 10 min; 98 to 99.9 percent reduction at 12 h | Ak et al. 1994a abstract (primary, single study); Cliver 2006 for the capillary mechanism |
| 1.5 in wood boards weigh about 10 lb (Boos 13 lb) | TechGearLab (Sonder, Boos 10 lb); Tom's Guide (Teakhaus 16×12×1.5 at 10 lb; Boos 13 lb) |
| Boos thickness 1 to 2¼ in; Epicurean ¼ in | Boos collection and R-Board pages; Epicurean page. Brand facts, not generalised |
| Board sizes about 8 × 10 to 18 × 24 in | TechGearLab; Tom's Guide (Teakhaus 24 × 18); Epicurean 8 × 6 to 17.5 × 13 |
| Half sheet 18 × 13 in; quarter 9 × 13; jelly roll 10 × 15 | Wikipedia Sheet pan; TechGearLab (half and quarter) |
| Half-sheet rack 11.8 × 16.9 in | Kitchenatics (single brand; given as an example) |
| Stainless racks oven-safe to 575 °F | Kitchenatics; TechGearLab (several stainless racks) |
| 1 in leg height on the tallest rack tested | TechGearLab; Kitchenatics |
| Bleach 1 tbsp per gallon | FSIS; NCSU |
| Oil wood about monthly | TechGearLab; Wikipedia |

## Dropped

- Non-slip feet as a facet. TechGearLab tests stability and Tom's Guide faults Epicurean for lacking feet, but feet are a photo fact and the stability test ranks weight as much as feet. Mentioned in thickness.
- Glue type (formaldehyde- or melamine-free) as a facet. One advocacy source (CEH, repeated by Wikipedia); listings rarely state the glue. Kept as a low-confidence judgement.
- Board oil type. One source on rancidity. Vocabulary entry only.
- Marble keeping pastry cool. I could not find a fetchable source; the register makes no such claim.
- Microwave disinfection of wood (Park and Cliver 1996). Real and primary, but no buyer chooses a board by it.
- Asahi heat limits (130 °C PRO, 100 °C HOME). Single secondary source.
- Epicurean "heat resistant to 350 °F". Search excerpt only.
- Rack load rating (Kitchenatics 45 lb, TechGearLab confirmed). Only one brand states one; a note in rack_material would overstate how often listings carry it.

## Trimmed detail

- `board_material`: NCSU (Chapman) recommends chlorine bleach for plastic and quaternary ammonium for wood "because chlorine binds very easily to organic materials, like the wood"; FSIS gives bleach for both. Not in the JSON; sanitiser choice is a care detail, not a buying decision.
- `knife_friendliness`: TechGearLab rates bamboo and edge-grain hardwood as "generally similar", where Knifewear puts bamboo with glass and granite. The scale places bamboo one step harder than edge-grain wood, following FSIS ("harder ... than hardwoods").
- `grain_construction`: face grain (the wide face of a plank up) is rare in cutting boards and is the add-on's third value; no source discussed it, so the register lists it without a claim.
- `wood_species`: red oak is the standard counter-example: hard but open-pored, "retain dirt even after washing" (Wikipedia). Soft woods such as cypress are kinder to edges but split into grooves (NCSU, Chapman).
- `sheet_pan_fit`: Checkered Chef sells quarter racks slightly undersize so two fill a half sheet and one fits some toaster ovens (TechGearLab).
- Summary: "Racks are wire; the material decides whether one can go in the oven" shortened.

## Script problems

- The skill's `schema.json` `vertical` enum is `formulations, machines, fashion, furnishing, other`. The brief requires `"vertical": "supplies"`, which the validator rejects as a schema error. I set `supplies` as instructed; with that one value changed to `other` on a scratch copy the register validates clean (see Corrections). Render and viewer were run directly. Reported, not fixed, since the skill's files are out of scope.

## Where "on listing" comes from

Retail marketplaces were not fetched. Facets marked `listing` rest on titles and fields seen in fetched brand and retailer pages, plus marketplace titles that appeared as search-result titles (for example "End Grain Maple Cutting Board, 20" x 15" x 1.5", John Boos" on Amazon; "Teakhaus Edge Grain Reversible Cutting Board 18 x 14"; a Royal Craft bamboo title "with Deep Juice Groove"; a Walmart "Half Sheet Pan Cooling Rack"; Kitchenatics' own title "fits Half Sheet Baking Pan"; Wilton titles "3-Tier Collapsible", "Chrome-Plated Cooling Grid, 10 x 16-Inch"). Those are product names, so material, grain, size, groove, pan fit and format are in the title text a scraper reads. Facets whose value sits in bullets or brand spec rows rather than titles (`max_heat_temp`, `leg_height`, `nsf_certified`, `plastic_polymer`) are marked `researched` with low or medium backfill.

## Corrections

**STYLE.md checklist pass (2026-10-25), before any blind check.**

1. `vertical` is `supplies` per the brief; the validator rejects it (see Script problems). A scratch copy with `other` validates clean: "18 facets, 3 splits, 30 references".
2. Split 1 rationale said the answer "removes two thirds of the facets", a number I had not measured. Now "drops most of the other facets".
3. `knife_friendliness`: "This is the expert's main reason to pay more for a board" was an opinion with no source. Replaced with TechGearLab's own line that knife wear is "nearly as important as hygienic factors". "Owners of hard Japanese knives filter on this first" read as question ordering and was unsourced; now says who it matters most to, sourced to Knifewear and Hasegawa.
4. `board_size`: "A small board spills food and makes you work in a corner" was unsourced; cut, leaving TechGearLab's "bigger than you think".
5. `juice_groove`: "a smaller flat area on that side" was unsourced; replaced with Boos's groove-one-side design and Tom's Guide's reversibility complaint about the Lämplig.
6. `sheet_pan_fit`: the line "two quarter racks do not always fill a half pan" misapplied Wikipedia, which says that of quarter pans, not racks. Replaced with the quarter and jelly-roll pan sizes.
7. `nsf_certified`: "where the US Food Code rules that drive it do not apply" claimed a link between NSF and the Food Code that no source states. Now says the Food Code's material rules apply to food businesses.
8. `max_heat_temp` moved from `listing` to `researched` (low): the rating is in brand bullets and care lines; marketplace coverage unconfirmed.
9. Paper composite: "pressed with resin" came only from search summaries. Family and vocabulary now say "made from paper fibre, sold under names such as Richlite" (Tom's Guide verbatim "RichLite paper composite"; Epicurean "Paper Composite").
10. Vocabulary: Janka now uses The Wood Database's definition ("a .444" (11.28 mm) diameter steel ball halfway into the wood"); microplastics no longer uses the unsourced 5 mm cut-off, and says most particles in the study were under 0.1 mm (abstract: "dominated by <100 μm"); "white" boards, "slightly harder" polypropylene, "cheaper" chrome, "uncommon" face grain and silica "in their cells" were unsourced and cut.
11. Trade-off 2 said dishwasher-safe boards are all plastic or composite; Hasegawa's soft board is dishwasher-safe and gentle (Knifewear), so it is now named as the exception.

Viewer: `tests/viewer-browser.py` on a temporary copy passed 6 of 6 (1380, 768, 390 px; light and dark; offline). The 1380 light first screen was read: tree root and two branches visible, the third (cooling rack) reached by horizontal scroll inside the diagram; map chips legible.
12. Dishwasher and wood: FSIS's own sentence ends "(laminated boards may crack and split)", and most wood boards are glued from several pieces (CEH: "more commonly they contain pieces that are glued together"; Wikipedia: glued boards "are more widely available"). The judgement now says even FSIS limits dishwasher use to one-piece solid wood, and `dishwasher_safe` says the same. The disagreement is narrower than first written. Revalidated (scratch copy with `other`), re-rendered, viewer check 6 of 6 again.

**Corrections (2026-10-26), applying the blind check (check.md).**

1. Deleted the "nonstick = not oven-safe" rule. Dropped "(not for the oven)" from the `rack_material` split question and cold-start question, deleted the nonstick-coated-steel branch's `makes_irrelevant: max_heat_temp` leg, rewrote `rack_material` and `max_heat_temp`'s `why_experts_care` and mechanism.plain to say coatings vary (Wilton not oven-safe; Nordic Ware's nonstick grids rated to 400 °F, two Wayback-fetched Nordic Ware product pages, new refs `nordicware_large_grid` and `nordicware_half_sheet_grid`), and changed the `max_heat_temp` backfill note to record "not oven-safe" only when the maker states it.
2. Fixed `knife_friendliness`: plastic now sits below bamboo and edge-grain hardwood in the derived tier rule (no sourced soft-plastic exception), matching TechGearLab and Cliver. Removed the bamboo → `knife_friendliness` then_ask leg, since it is derived with no question.
3. Narrowed scope instead of inventing slab facets: removed `prep slab` from `product_job` (value, branch, family, definitions). The slab branch had no facets of its own and left every board facet live with no way to answer "marble" or "granite", so it is out of scope here until it gets its own research rather than riding on board facets. This also removes "rolling dough" from the cold-start question, which was pulling in silicone pastry mats that `baking-sheets` owns.
4. Restated `pan_size_from_rack` on a sourced floor, not the outer pan size: each rim sacrifices 0.5 to 0.75 in of flat surface per edge (Wikipedia, now also in its `supports`), which is consistent with Kitchenatics' rack-to-pan gap. Added two-thirds (21 × 15 in) and full sheet (26 × 18 in) to `sheet_pan_fit`'s values and to the formula, with new vocabulary entries for both.
5. `rack_grid`: hint now says the photo settles cross-hatched versus parallel wires, since the title word "grid" is also used for Wilton's parallel-wire tiered rack (TechGearLab: "not gridded"). Added `wilton_tiered_rack` to its evidence.
6. Reworked the `rack_material` branches: chrome-plated now asks `max_heat_temp` (the real open question) instead of inferring `dishwasher_safe` from one brand pair; removed the duplicate `sheet_pan_fit` leg under stainless steel, since `product_job`'s cooling-rack branch already asks it.
7. `thickness` and the grain_construction trade-off now say weight follows area as well as thickness (Tom's Guide: ~10 lb at 16 × 12 in, ~15 lb at 24 × 18 in for 1.5 in boards), instead of stating one weight for the thickness alone.
8. `nsf_certified`: restored the Food Code link via 4-205.10 ("ANSI-accredited certification... is deemed to comply with Parts 4-1 and 4-2"), now cited directly from the already-fetched Food Code PDF text and added to `fda_food_code_2022`'s supports. Dropped "the US body" (any ANSI-accredited program qualifies) and added "where adopted" to the Food Code judgement's note. Did not name the specific standard behind "NSF Certified" for boards; nsf.org is still 403 live and Wayback was unavailable (temporarily offline) when checked.
9. Trade-off 1: moved TPU out of the "takes visible cuts" group; TechGearLab found TPU mats resisted cuts, which matches `plastic_polymer`'s own claim.
10. `plastic_polymer`: replaced "tens of grams a year... from either" with "about 7 to 50 g a year by the authors' own assumptions, with overlapping ranges for the two plastics", removing the implied PP > PE per-person ordering the abstract does not support.
11. Wording/attribution: `board_size` now says Tom's Guide "said... could be too big", not that it was; the Hasegawa `board_material` hint now quotes "plastic coating over a wood core" instead of an unsupported "polyethylene"; the `usda_fsis_boards` reference title now gives the actual snapshot date (2026-01-03) instead of "2025"; the unsourced "gold or dark finish usually means a coating" claim was dropped from `rack_material`'s backfill note (folded into fix 1's rewrite).
12. Demoted the plastic → `thickness` and plastic → `colour_coded_set` legs from then_ask to nothing (both facets already carry the `annotation` role), since the per-board-type chains behind them were single-product and weak.

Rebuilt and revalidated after all of the above: `validate.py` clean (18 facets, 3 splits, 32 references), `render.py` and `viewer.py` ran clean.