# Notebooks, notepads and sticky notes: research notes

Date: 2026-10-25. Scope: blank, ruled, dotted and grid paper sold bound or stacked for handwriting: bound notebooks (sewn, stapled, spiral or wire, glued, ring or disc-bound, including composition books, memo books, and bound refill inserts that work on their own, such as stitched cover refills), tear-off pads (legal, memo, steno, letter pads, and paper desk notepads) and sticky notes. Left to siblings: dated planners, diaries and dated desk pads (`planners`), loose filler, punched refill sheets for ring and disc systems, printer paper, flip-chart and easel pads (including Post-it Super Sticky easel and wall pads), and art paper sold by weight and texture, including sketchbooks for dry media (`paper`; `stationery-boards` also sends flip-chart paper there), index cards (`cards`), page flags and tabs (`labels`), binders and loose rings (`binders`), leather, felt and cork desk pads and blotters (`desk-mats`), pen refills (`ink-cartridges`). Notebook covers sold empty: no sibling register claims them and no facet here applies, so they are flagged for the taxonomy owner rather than assigned. Ring and disc notebooks sold complete with paper stay here; empty binders go to `binders`. (Scope revised after the blind check; see Corrections after the blind check.)

Quote labels: **verbatim** = the sentence in quotation marks is in text I read myself (curl plus tag stripping, or a Shopify product JSON `body_html`); **fetched excerpt** = the WebFetch tool summarised the page; **search excerpt** = the claim came from a search-engine summary only. A search-excerpt-only number does not go in category.json.

Research constraint: the session's WebSearch budget ran out after three searches in this run (the budget is shared with other agents in the session). The rest of the research used curl on known URLs, Shopify `products.json` feeds from two specialist pen-and-paper shops, and Internet Archive snapshots. JetPens (Cloudflare), Goulet Pens (404 on the guessed blog URL), post-it.com and 3M (Access Denied), Staples (403) and ISO (403) could not be read directly.

## Seed

Seed output, verbatim:

```
=== Example entry (from an unrelated category) ===
category: olive-oil
what this shows: a facet that experts rank first, that almost no listing carries, and that is expensive to backfill. The honest answer is to list it, mark it researched, and limit backfill to the top of the catalog. Also shows a derived metric hanging off a researched facet.
[harvest_date example entry and months_since_harvest derived metric]

=== Persona ===
An allergist, who thinks in exposure, dose and what is hidden in the ingredients list

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

1. **The notebooks equivalent of harvest date is ink tolerance: whether the paper takes a fountain pen (or gel) without feathering or bleeding.** Pen and paper reviewers rank it first (The Pen Addict, The Well-Appointed Desk, Scribbright, Inkophile all lead on it), almost no listing states it as a field, and the honest backfill is reviewer writing tests for the top N. Like harvest date versus best-before, the number listings do carry (gsm) is a loose proxy: Tomoe River at 52 gsm resists bleed better than Moleskine at 70. **Kept** as split 2 `ink_tolerance` and its costly companion `show_through`.
2. **The allergist notices what is hidden in the paper.** The paper mill and its coating are rarely named; "acid-free" is printed as if it were rare when most commercial paper is alkaline already (Wikipedia, acid-free paper); and mills change the paper without changing the label (Tomoegawa's 2020 change to 52 gsm Tomoe River, reported by The Well-Appointed Desk; Moleskine commenters comparing 2015 and recent books). **Kept**: `permanence` with "acid-free" read as weak, and the batch-change judgement. **Dropped**: BPA or other residues in recycled paper (no source ties it to notebooks); scented notebooks (no data); sticky-note residue on surfaces (only 3M's own no-residue claim, via Wikipedia). Exposure in the literal sense: the only regulator text found treats paper and CMYK inks as within US children's lead limits (16 CFR 1500.91), so it became a low-stakes judgement, not a facet.

## Sources and what each supports

### Regulator

**16 CFR 1500.91, via Cornell LII** https://www.law.cornell.edu/cfr/text/16/1500.91
- verbatim: "(d) The following materials do not exceed the lead content limits under section 101(a) of the CPSIA provided that these materials have neither been treated or adulterated ... (5) Paper and similar materials made from wood or other cellulosic fiber, including, but not limited to, paperboard, linerboard and medium, and coatings on such paper which become part of the substrate. (6) CMYK process printing inks (excluding spot colors ...)".
- Supports: no safety standard is a buying factor for notebooks; paper and process inks are pre-cleared for children's products. Confidence medium (primary text; no second source read on how it applies to notebooks specifically). No number from it is in the JSON.

### How paper and ink interact (primers)

**Wikipedia, Sizing** https://en.wikipedia.org/wiki/Sizing
- verbatim: "Sizing is used during paper manufacture to reduce the paper's tendency when dry to absorb liquid, with the goal of allowing inks and paints to remain on the surface of the paper and to dry there, rather than be absorbed into the paper." "There are two types of sizing: internal sizing ... and surface sizing". "hard sized papers have the highest water resistance, such as coated fine papers".
- Supports: the mechanism. Confidence high.

**Wikipedia, Fountain pen** https://en.wikipedia.org/wiki/Fountain_pen
- verbatim: "Inks intended for use with fountain pens are water-based." "Contemporary fountain pen inks are almost exclusively dye-based".

**Wikipedia, Ballpoint pen** https://en.wikipedia.org/wiki/Ballpoint_pen
- verbatim: "Ballpoint pen ink is normally a paste containing around 25 to 40 percent dye. The dyes are suspended in a mixture of solvents and fatty acids." "... to create a smooth paste that dries quick[ly]". "Bíró's innovation successfully coupled viscous ink with a ball-and-socket mechanism".

**Wikipedia, Gel pen** https://en.wikipedia.org/wiki/Gel_pen
- verbatim: "A gel pen uses ink in which pigment is suspended in a water-based gel."
- The three pen pages together support the ink tiers in split 2 and the "ballpoint makes dry time irrelevant" branch. Confidence high.

### Specialist reviewers and communities

**The Pen Addict, Tomoe River Paper Review (2013)** https://www.penaddict.com/blog/2013/6/3/tomoe-river-paper-review
- verbatim: "It is thin, transparent, and crinkles with ease". "No matter the size of the nib - from 1.5 mm stub nibs down to my Japanese EF nibs - they all glide across the page with no feedback at all. On top of that, there is no bleed or feathering." "With the Rhodia-level smoothness also comes Rhodia-length dry time. Lefties beware. You also will not be able to use the back of the page either."
- Supports: coated thin paper stops bleed but shows through; smooth paper dries slowly. Confidence high.

**JetPens, Tomoe River 52 gsm Loose Leaf Paper (Internet Archive snapshot 2025-11-26; live page behind Cloudflare)** https://web.archive.org/web/20251126231635/https://www.jetpens.com/Tomoe-River-52-gsm-Loose-Leaf-Paper-A4-Blank-White-100-Sheets/pd/11514
- verbatim: "At 52 gsm, the thin and delicate loose leaf pages are designed to reduce bulk. Despite its thinness, the paper is incredibly reliable. The sheets are resistant to bleedthrough and feathering, can handle a variety of inks, and are acid-free." Staff note: "The ink doesn't feather or bleed through to underlying pages, but it does show through heavily on the reverse side of the sheet you are using." Customer review fragment: "proof on regular paper, but on the Tomoe it smears."
- Supports: 52 gsm, bleed resistance, heavy show-through. Confidence high. The smear fragment is not used any more: the blind checker read the full review, and it is about a Zebra brush pen lifting under Copic markers, not dry time (see Corrections after the blind check).

**The Well-Appointed Desk, Jesi answers your questions about Tomoe River (2021)** https://www.wellappointeddesk.com/2021/06/jesi-answers-all-your-questions-about-tomoe-river/
- verbatim: "Tomoe River 52 gsm paper is highly prized by many fountain pen users. The paper is incredibly thin and light, yet fountain pen ink doesn't bleed through the page (well, it can if you really try). The coating on the paper allows characteristics of fountain pen ink to show up". "In 2020, rumors started up that the manufacturer of Tomoe River paper, Tomoegawa, had changed the manufacturing process of their 52 gsm offering without making the change public. Although the company denied this at first, Tomoegawa did release information confirming the changes." Quoting Sakae TP: no longer manufacturing "due to the discontinuation of the base paper product".
- Supports: second source for 52 gsm and bleed resistance; the batch-change judgement. Confidence high.

**The Well-Appointed Desk, Reconsidering Moleskine (2016), with reader comments** https://www.wellappointeddesk.com/2016/01/reconsidering-moleskine/
- verbatim (author): "From the reverse of the pages, you can see some ghosting and show through". (commenters): "the bleed-through makes the reverse sides of the pages unusable. However, pencils work fine, ballpoints work fine, gel pens work fine." "Things have changed. They are now unusable with a fountain pen. To check, I pulled out an old one (from 2015) and compared it with a recently purchased one."
- Supports: Moleskine judgement; batch change. Confidence medium (comments). Re-read 2026-10-25 (WebFetch): the author's ghosting sentence reads in full "From the reverse of the pages, you can see some ghosting and show through on the top row which is the planner and the Cahier", so it is about the planner and Cahier, not the Classic.

**Inkophile, Moleskine Paper Meets Ink (2013)** https://inkophile.wordpress.com/2013/04/02/moleskine-paper-meets-ink-and-it-aint-pretty/
- verbatim: "In fact any writing instrument, save a chunky Sharpie or a fountain pen, will work just fine. Bleed-through might be an issue since the paper is very thin".

**Scribbright, Moleskine Classic Hardcover Notebook Review** https://scribbright.com/moleskine-classic-hardcover-notebook-review/
- verbatim: "Paper Weight 70 gsm / 47 lb, acid-free, ivory". "240 ivory-tinted pages at 70 gsm, acid-free and FSC-certified." "The 70 gsm paper handles most ballpoint and gel pens without issue. Fountain pen users will see ghosting on the reverse side and occasional bleed-through with wetter nibs". "240 pages in the hardcover large version, which translates to 120 sheets."
- Supports: Moleskine 70 gsm; pages versus sheets; acid-free claim on a mass premium book. Confidence high.

**Little Coffee Fox, Moleskine Classic Notebook Review** https://littlecoffeefox.com/moleskine-classic-notebook-review/
- verbatim: "The paper in the Moleskine classic notebook is only 70 gsm, as compared to the Leuchtturm1917 journal's 80 gsm." "Of all the journals and notebooks that I see used in the bullet journal community, the Moleskine is the one that I see in the wild the most."
- Supports: second source for 70 and 80 gsm. Confidence high.

### Brand spec pages and brand guides

**Dingbats, Best Notebook That Doesn't Bleed Through in 2026 (brand buying guide)** https://www.dingbats-notebooks.com/blogs/the-dingbats-blog/best-notebook-that-doesnt-bleed-through-in-2026
- verbatim: "GSM (grams per square meter) is the primary predictor of bleed-through." "Moleskine (70gsm) sits at this threshold". "Leuchtturm (80gsm) shows occasional bleed-through with wet inks." "Rhodia (90gsm) and similar papers rarely bleed with fountain pens." "Japanese papers like Tomoe River (68gsm) use specialized coatings that prevent bleed-through despite thin paper, but they trade this for heavy show-through (ghosting)". Own books: "Paper : 100gsm cream coated, acid-free / Pages : 192 (96 sheets)"; "Paper : 160gsm matte natural white, acid-free / Pages : 128 (64 sheets)"; "The trade-off: fewer pages (128 vs 192)".
- Re-read 2026-10-25 (WebFetch excerpt): pen table row "Gel pen | 80gsm+ | Leuchtturm, Dingbats*, Rhodia"; "Gel pens deposit minimal ink; fountain pens and markers deposit significantly more."; "Moleskine (70gsm) will bleed with markers."; "Moleskine (70gsm) remains the worst choice for anyone who uses wet inks." The blind checker read verbatim: "For ballpoint users, Moleskine is fine. For anyone who uses fountain pens, gel pens, or markers, it's a poor choice".
- Supports: the disagreement on gsm (it calls gsm primary, then concedes coating); the disagreement on gel with Scribbright and Inkophile; the marker mapping (markers grouped with fountain pens as heavy ink); third source for 70/80; second for Rhodia 90; the heavier-paper-fewer-pages trade-off (its own products). Confidence medium: a seller ranking its own books first. Its bleed bands by gsm are not used in the JSON.

**Rhodia, View All Collections** https://rhodiapads.com/collections/pads
- verbatim: "micro-perforated sheets for ease and 80 g exceptionally smooth extra white paper"; "enjoy premium 90 g ivory paper"; "Superior 80g Rhodia paper for those who prefer spiral binding."
- Supports: Rhodia 80 and 90 g. Confidence high.

**Leuchtturm1917 US, Hardcover Notebook (Shopify product JSON)** https://www.leuchtturm1917.us/products/hardcover-notebook (read as `.json`)
- verbatim: "Six formats, from Pocket (A6) to Master (A4+)"; "80–100 gsm FSC-certified paper"; "Four ruling options"; "Numbered pages and two page markers". Options: "Ruled, Squared, Plain, Dotted".

**Dromgoole's, Leuchtturm1917 A5 Medium Hardcover Dotted Notebook (Shopify product JSON; brand copy)** https://www.dromgooles.com/products/leuchtturm1917-a5-medium-hardcover-dotted-notebook-black
- verbatim: "251 numbered pages using 80 g/m² ink-proof paper. Sized at 145 x 210 mm". "Dotted - Discreet dots, set at a distance of 5mm". "Ruled - A clear layout with 6mm line spacing." "Squared - 5x5mm squares". "Thread-bound book opens flat".

**Dromgoole's catalogue (Shopify products.json)** https://www.dromgooles.com/products.json
- verbatim (Apica CD Semi B5): "stitched binding to lay flat as you write. Paper inside is 6.5 mm ruled ... 68 pages 7 x 10"". (Midori Soft Color A5): "56 pages", "Glue binding for clean and easy page removal." (Midori Yuru Log B6): "5mm dot grid 80 pages".

**Urban Kit Supply catalogue (Shopify products.json)** https://www.urbankitsupply.com/products.json
- verbatim (Midori MD A5): "exposed spine and cheesecloth binding ... allow the notebook to open perfectly flat"; "a subtle "tooth" that provides just the right amount of feedback to the pen. It is specifically engineered to resist bleeding and feathering". (Field Notes 1943): "bound with heavy-duty silver staples". (Maruman Mnemosyne N165): "Pages tear perfectly because of the micro-perforation and the ring binding system."

### Construction, sizes and formats (primers)

**Wikipedia, Notebook (the Legal_pad URL resolves here)** https://en.wikipedia.org/wiki/Notebook#Legal_pad
- verbatim: "Binding methods can affect whether a notebook can lie flat when open and whether the pages are likely to remain attached." "Other bound notebooks are available that use glue to hold the pages together; this process is "padding."" "Principal types of binding are padding, perfect, spiral, comb, sewn, clasp, disc, and pressure".

**Wikipedia, Bookbinding** https://en.wikipedia.org/wiki/Bookbinding
- verbatim: sewing through the fold "has been called the "gold standard" for binding ... through-the-fold books have wide margins and can open completely flat. Pages cannot fall out unless they are ripped." Mass-market glued paperbacks "are likely to fall apart or lose sheets after much handling or several years."

**Wikipedia, Coil binding** https://en.wikipedia.org/wiki/Coil_binding
- verbatim: "Documents bound with helical coil (usually called spiral coil) can open flat on a desk or table and offer 360 degree rotation for easy note taking."

**Wikipedia, Exercise book** https://en.wikipedia.org/wiki/Exercise_book
- verbatim: thin school notebooks "are stapled through the fold, which, aided by their thinness, allows to open them completely flat". "'Thick' notebooks must be either ruled with 6 mm, 7 mm, 8 mm, 9 mm spacing". UK sizes "A5 (148×210 mm), A4 (210×297 mm)".

**Wikipedia, Ruled paper** https://en.wikipedia.org/wiki/Ruled_paper
- verbatim: "Medium ruled (or college ruled) paper has 9⁄32 in (7.1 mm) spacing"; "Narrow ruled paper has 1⁄4 in ... 6.4 mm ... used by those with smaller handwriting"; "Wide ruled (or legal ruled) paper has 11⁄32 in (8.7 mm) spacing ... commonly used by American children in grade school, as well as by those with larger handwriting." "Grid paper for maths is commonly a 5 x 5 mm square grid."
- Supports: the handwriting-to-spacing mapping. The US inch values are single-source here (the second source I tried, Staples, returned 403; search budget was gone), so they stay in these notes for the backfill converter and are not in the JSON.

**Wikipedia, Recto and verso** https://en.wikipedia.org/wiki/Recto_and_verso
- verbatim: "In double-sided printing, each leaf has two pages – front and back."

**PaperSizes.org** https://www.papersizes.org/a-paper-sizes.htm and **Wikipedia, Paper size** https://en.wikipedia.org/wiki/Paper_size
- verbatim: "A5 148 x 210 mm"; "A6 105 x 148 mm" (papersizes.org); "148 × 210" row and "The Japanese standard JIS P 0138 defines two main series of paper sizes" (Wikipedia). JIS B5 182 × 257 and ISO B5 176 × 250 appear only on Wikipedia pages, so B5 dimensions are not in the JSON.

**Wikipedia, Acid-free paper** https://en.wikipedia.org/wiki/Acid-free_paper
- verbatim: "Much commercially produced paper is acid-free but this is largely the result of a shift from kaolin clay to precipitated calcium carbonate (PCC) as the main filler material". "Manufacturers of acid-free paper can indicate the compliance of their product with the test requirements of the ISO 9706 or ANSI Z39.48-1992 standards using a circled infinity symbol". "For paper to last at least 100 years it must have an alkaline reserve of 2 percent or more." (single source; not in JSON).

### Sticky notes

**Wikipedia, Post-it note** https://en.wikipedia.org/wiki/Post-it_note
- verbatim: "a small piece of paper with a re-adherable strip of glue on its back, made for temporarily attaching notes". "In 2003, the company introduced Post-it Brand Super Sticky Notes, with a stronger glue that adheres better to vertical and non-smooth surfaces." "In 2018, 3M launched Post-it Extreme Notes, which are more durable and water-resistant and which stick to wood and other materials in industrial environments." "The typical Post-it is 3 inches by 3 inches". Sizes "1 3⁄8 x 1 7⁄8, 2 x 2 ... 11 x 11". "the typical pad containing 100 sheets ... 30, 45, 50, 70, 75, 90, 100, and 400 sheets per pad."

**Target, search results for Post-it Super Sticky notes** https://www.target.com/s?searchTerm=post-it+super+sticky+notes
- fetched excerpt: titles "Post-it Super Sticky Notes 3"x 3" Energy Boost Collection: 6 Multicolor Pads, 70 Sheets Each"; "Post-it 3pk 3" x 3" Super Sticky Notes 45 Sheets/Pad"; listings "highlight "2X sticking power""; "30 sheets per pad (Big Notes) to 90 sheets per pad"; sizes 1.88 x 1.88 to 6 x 6 in.
- Supports: Super Sticky pads at 45/70/90; size and sheet count in titles. The "2X" figure is a fetched excerpt from one retailer of 3M's own claim, so it is not in the JSON.
- Blind checker (Target product feed, verbatim): "Super Sticky Big Notes Blue 30 Sheets"; "Canary Yellow, 4 Pads per pack, 50 sheets per pad"; totals "180 Sheets", "210 Sheets", "135 Sheets", cubes "400 Sheets"; ambiguous "12 Pads/Pack: Multicolored, 90 Sheets"; "Post-it Super Sticky Mini Easel Pad, 15 x 18 Inches".
- Re-read 2026-10-25, https://www.target.com/s?searchTerm=post-it+notes (WebFetch excerpt): "Post-it Notes, 3 in x 3 in, Canary Yellow" "50 sheets per pad"; "Post-it Notes 4pk 3\" x 3\" Beachside Café" "50 Sheet/Pad"; "Post-it 5pk 3\" x 3\" Pop-Up Notes" "100 Sheets/Pad"; Super Sticky at "45 Sheets/Pad" and "70 Sheets/Pad"; totals "Post-it Recycled Super Sticky Notes 4\"x6\"" "180 Sheets", "4\"x4\"" "210 Sheets", "3pk Super Sticky Summer Joy" "135 Sheets"; "up&up ... Cube" "400 Sheets". So standard Post-it pads come at 50 as often as 100, and Super Sticky pads are not reliably smaller than standard ones.

**Target, search results for Five Star college ruled notebooks** https://www.target.com/s?searchTerm=five+star+college+ruled+notebook
- fetched excerpt: "Five Star 8.5"x11" 1 Subject College Ruled Notebook ... 100 Pages"; "Five Star 3 Subjects 150pg"; "Five Star 80ct College Ruled Reinforced Filler Paper ... 80 Sheets"; "most highlight ink bleed resistance".
- Supports: titles carry ruling, size, a count and its unit word. Low confidence on whether Five Star's "100 Pages" is the tool's paraphrase of "100 Sheets".
- Resolved after the blind check. The checker read Target's product feed verbatim: title "1 Subject College Ruled Notebook ... 100 Pages", bullet "Number of Pages: 100", copy "100 double-sided, college ruled sheets"; the 3-subject "150pg" and 5-subject "200 Pages" books are 150 and 200 sheets. My re-read 2026-10-25 (WebFetch excerpt of the same search) confirms the titles verbatim: "Five Star 8.5\"x11\" 1 Subject College Ruled Notebook Sky Blue 100 Pages", "Five Star 3 Subjects 150pg College Ruled Spiral Notebook (Colors May Vary): Perforated, Acid-Free, 3 Hole Punched, Flexible Cover", "Five Star 80pg 2 Subject College Ruled Notebook 9.5\"x6\": Spiral, Wire-Bound, Acid-Free, Perforated, Water-Resistant, Pocket", "Five Star 80pg College Rule Composition Notebook 9.75\"x7.5\": Soft Cover, Acid-Free, Sewn Binding, Adult Stationery", "Five Star 200pg 5 Subjects Wide Ruled Spiral Notebook (Colors May Vary): Multi Subject, Perforated Pages, Flexible Cover". The excerpt did not carry the bullets, so the "double-sided sheets" line rests on the checker's read. The "100 Pages" label is a Target listing labelling sheets as pages, not a tool paraphrase. Also supports: Perforated, Acid-Free, Spiral, Wire-Bound and Sewn Binding as title words.

### Could not reach
- JetPens blog guides, Goulet Pens blog, post-it.com and 3M product pages, Staples, ISO 9706 page, NISO Z39.48 page, Moleskine product page (404). Fountain Pen Network archived thread timed out. Reddit not attempted (blocked in earlier runs).

## Data points used for derived metrics and ranges

| Product | gsm | Pages (sheets) | Size | Ruling spacing | Binding | Source |
|---|---|---|---|---|---|---|
| Moleskine Classic large | 70 | 240 (120) | n/a | n/a | lay-flat hardcover | Scribbright; LCF; Dingbats |
| Leuchtturm1917 Medium A5 | 80 | 251 numbered | 145 x 210 mm | 6 mm ruled, 5 mm dots and squares | thread-bound | Dromgoole's; LCF; Dingbats; Leuchtturm US says 80 to 100 across the range |
| Rhodia pads / Webnotebook | 80 / 90 | n/a | n/a | n/a | stapled pad / spiral / book | Rhodia; Dingbats (90) |
| Tomoe River loose leaf | 52 | 100 sheets | A4 | blank | loose | JetPens archive; WAD |
| Dingbats Wildlife | 100 | 192 (96) | n/a | n/a | n/a | Dingbats |
| Dingbats Pro | 160 | 128 (64) | n/a | n/a | n/a | Dingbats |
| Apica CD semi B5 | not stated | 68 | 7 x 10 in | 6.5 mm | stitched | Dromgoole's |
| Midori Soft Color A5 | not stated | 56 | A5 | dot grid | glued | Dromgoole's |
| Post-it standard | n/a | 100 sheets per pad | 3 x 3 in | n/a | adhesive | Wikipedia |
| Post-it Super Sticky | n/a | 45, 70, 90 per pad | 3 x 3 in | n/a | adhesive | Target (fetched excerpt); Wikipedia lists these counts |

Listing-field scan (rough, one run, regex over description text): 266 notebook, pad and sticky-note listings from the two Shopify catalogues above (titles matching notebook, notepad, memo, sticky, pad or journal; covers, planners, diaries and refills excluded). Share of listings whose title or description mentions each field: ruling 72%, size 68%, binding words 51%, page or sheet count 37%, spacing in mm 32%, fountain pen 16%, bleed or ghosting words 11%, perforation 6%, gsm 4%, acid 3%. These shops sell mostly Japanese and European brands whose structured specs sit in images or variant names, so the gsm figure is likely an undercount. Used only to set `source` (gsm, ink tolerance, binding, spacing marked researched) and not quoted in the JSON.

`price_per_page`: pages = sheets × 2 (Recto and verso; Scribbright's 240 = 120; Dingbats' 192 = 96).
`price_per_note`: pack_count × sheets_per_pad, both in Target titles.

## Against the existing add-on schema

Add-on read: `supplies/stationery/add-ons/notebook.jsonc` (`ruling` enum [lined_college, lined_wide, lined_narrow, dot_grid, grid_graph, blank, mixed, not_applicable, unspecified], `paper_gsm` "bleed-through currency", `sizing.page_count`), composed with `stationery-standard.jsonc` (empty) and `supplies-standard.jsonc` (`high_signal_markdown`, `sizing.size_stated`, `sizing.pack_count`, `sizing.pieces_included_text`, `sizing.dimensions`, `sizing.capacity`, `sizing.weight`, `sizing.measurements_text`). Kinds follow `zamana-expertise/AGENTS.md` rule 4. Proposals only; nothing in zamana-expertise was edited.

| Facet in this doc | In add-on today | Proposed kind | Note |
|---|---|---|---|
| format | no | hard filter (enum: bound_notebook, tear_off_pad, sticky_notes) | The leaf bundles three products; without this field sticky notes and legal pads share every filter with journals. Cheap from title words. |
| ink_tolerance | no; `paper_gsm` is commented as the "bleed-through currency" | category quality currency (ordinal enum) | Proposal: add `ink_tolerance` [ballpoint_pencil, gel_rollerball, fountain_pen, unspecified] and stop treating gsm as the bleed currency. Tomoe River 52 gsm against Moleskine 70 gsm is the counterexample. Willing to augment from mill name and brand claims; reviewer tests for the top N. |
| ruling | yes, `ruling` | hard filter (already) | Keep. The lined_college / lined_wide / lined_narrow values fold spacing into ruling for US goods; European and Japanese books state mm instead (6, 6.5 mm), which none of the three values names. Proposal: keep `ruling` coarse (lined, dot_grid, grid_graph, blank, mixed) and move spacing to a separate `line_spacing_mm` number outside sizing. |
| line_spacing_mm | partly, inside `ruling` values | high_signal_markdown, or a number if the ruling enum is coarsened | A number only premium brands state; US names convert (wide 8.7, college 7.1, narrow 6.4 mm, Wikipedia). |
| binding | no | hard filter (enum: sewn, stapled, spiral_wire, glued, ring_disc) | Buyers filter on lies-flat and tear-out. Spiral and rings are visible in photos (rule 1), but sewn versus glued is not, so the field earns its place. |
| paper_gsm | yes, `paper_gsm` | category quality currency, secondary | Keep as a number, but as an annotation next to ink tolerance, not as the bleed proxy its comment implies. |
| page_count | yes, `sizing.page_count` | sizing (already) | Proposal: document the unit as pages and convert sheets × 2 at extraction; listings mix "pages", "sheets" and "ct". |
| page_size | partly, `sizing.size_stated` and `sizing.dimensions` | hard filter (enum A6_or_smaller, A5, B5, letter_A4, other) inside sizing | Size varies across variants of one listing (Leuchtturm sells six formats), so it belongs in sizing. |
| show_through | no | high_signal_markdown | Review-only fact; a sentence, not a column. |
| paper_smoothness | no | high_signal_markdown | Brand copy words ("silky smooth", "tooth"). |
| permanence | no; rule 5 cuts certifications | high_signal_markdown | "Acid-free" is near-universal; only an ISO 9706 or Z39.48 claim is worth a line. |
| adhesive_strength | no | hard filter (enum standard, super_sticky, extreme) for sticky notes | Title product-line word. |
| note_size | partly, `sizing.dimensions` | sizing (already, via dimensions) | 3 x 3 in default; no new field. |
| sheets_per_pad | partly; `sizing.pack_count` holds pads, nothing holds sheets per pad | sizing (number) | Proposal: add `sizing.sheets_per_pad`; needed for cost per note. |
| perforated | no | hard filter (boolean) | Added after the blind check. Title word on mass-market books ("Perforated" at Target); brand copy on Rhodia and Maruman. |

Add-on fields this register considers noise for this leaf: `sizing.capacity` and `sizing.weight` from supplies-standard (no buyer compares notebooks by stated weight or capacity; weight is a photo-and-gsm question), and `ruling: not_applicable` should be reserved for sticky notes only, where most are blank anyway.

## Branches

The `when` labels are values of each split facet. Split 2's question offers pens; the mapping rule (fountain pen or markers → fountain pen safe; gel or rollerball → gel and rollerball safe or better; ballpoint or pencil → any tier) is in `ink_tolerance.why_experts_care`. Branches on split 2 fire on the tier the buyer's answer maps to; since a ballpoint or pencil answer maps to "any tier", no branch is keyed to the lowest tier (revised after the blind check).

**Split 1, format**
- when `sticky notes`: then_ask adhesive_strength, note_size, sheets_per_pad.
  - "Sticky notes are held by a strip of low-tack adhesive on each sheet, not a binding, and are sold by note size and sheets per pad" (format, why experts care), and "Standard notes are made to peel off paper without residue but fall off walls and screens" (adhesive_strength). sheets_per_pad is not novice-answerable and has no question, so on this leg it is shown, not asked.
- when `sticky notes`: makes_irrelevant binding, page_count, page_size, permanence.
  - binding: same sentence, "not a binding" (Wikipedia: "a re-adherable strip of glue on its back"). Sticky titles at Target say "Glue Binding"; the binding hint says to ignore it.
  - page_count and page_size: replaced by sheets_per_pad and note_size on this branch (same sentence: "sold by note size and sheets per pad"). page_size is split 4, so on this leg split 4 is skipped.
  - permanence: chain of two facts. Sticky notes are "made for temporarily attaching notes" (Wikipedia, verbatim), and permanence "matters for journals meant to last" (permanence, why experts care). A temporary note has no use for archival paper.
- when `tear-off pad` (added after the blind check): makes_irrelevant binding.
  - "A pad is glued or wired at one edge so every sheet tears off, whatever holds it" (format, why experts care); Wikipedia, Notebook: glued stacks are "padding". Steno pads are wire-bound but sold as pads (format backfill note), and their sheets still come off, so the binding question would ask a pad buyer about a choice that does not change use.
- when `bound notebook`: then_ask binding, page_count.
  - "A notebook keeps pages, so binding and page count matter" (format, why experts care).
  - page_size removed from this leg after the blind check: pads need size as much as books, so it is not a change of path. page_size is now split 4 for every form except sticky notes.

**Split 2, ink_tolerance**
- when `fountain pen safe`: then_ask show_through, paper_smoothness.
  - show_through: "Thin coated paper can stop bleed yet still look like tracing paper from behind" (show_through, why experts care; Pen Addict, JetPens).
  - paper_smoothness: "Very smooth paper keeps wet ink on the surface, which shows off colour but dries slowly" (paper_smoothness; Pen Addict "Rhodia-length dry time. Lefties beware.").
  - paper_gsm removed from this leg after the blind check: nothing says gsm becomes more useful once paper is already fountain pen safe, and judgement 1 says coating dominates for fountain pens.
- Removed after the blind check: `ballpoint and pencil only` making paper_smoothness irrelevant. The facet covers surface feel as well as dry time, and pencil writers choose tooth (blind checker: Midori MD copy, "the precise scratch of a pencil"); the `when` value was also the paper's lowest tier while a ballpoint answer maps to any tier, so the leg's firing condition was ambiguous.
- Dropped before delivery: `ballpoint and pencil only` making show_through irrelevant (show-through is a property of the sheet).
- No branch for `gel and rollerball safe`: nothing in the register says a gel writer's next question differs from the default.

**Split 3, ruling**
- when `lined`: then_ask line_spacing_mm. "Spacing should match handwriting" (line_spacing_mm; Wikipedia, Ruled paper).
  - `dot grid` and `grid` removed from this leg after the blind check: every cited dot and square pitch is 5 mm (Leuchtturm, Midori Yuru Log, maths grid paper), and the handwriting question maps only to line spacing.
- when `blank`: makes_irrelevant line_spacing_mm. Blank pages have no lines (ruling definition: "lines, a grid of dots, squares, nothing, or a mix").
- No branch for `dot grid`, `grid` or `mixed`.

**Split 4, page_size** (replaced binding after the blind check): no branches. Its answer filters on itself and changes nothing downstream. Binding stays a filter with its own question, asked on the bound-notebook leg of split 1.

Renderer note: see Corrections item 4.

## Numbers in category.json and the sources behind each

| Number | Sources |
|---|---|
| Moleskine 70 gsm | Scribbright, Little Coffee Fox, Dingbats (all verbatim) |
| Leuchtturm 80 gsm | Dromgoole's (brand copy), Little Coffee Fox, Dingbats; Leuchtturm US range 80 to 100 |
| Rhodia 80 or 90 g | Rhodia (both); Dingbats (90) |
| Tomoe River 52 gsm | JetPens archive, The Well-Appointed Desk |
| 70 to 90 gsm typical premium | the four rows above |
| Moleskine 240 pages = 120 sheets; one sheet two pages | Scribbright; Wikipedia Recto and verso; Dingbats (192 = 96, 128 = 64) |
| Leuchtturm 251 pages, 145 x 210 mm, 6 mm lines, 5 mm dots | Dromgoole's brand copy (one listing; brand-specific facts, attributed by name and labelled "one retailer's brand copy" in the JSON, not generalised). 5 mm dots also on Midori Yuru Log (Dromgoole's) |
| Apica 6.5 mm | Dromgoole's (one listing); removed from category.json after the blind check |
| Dingbats 160 gsm 128 pages vs 192 | Dingbats (its own products; labelled in the JSON as the seller's figures, with the coating confound) |
| A5 148 x 210 mm | PaperSizes.org, Wikipedia Paper size, Wikipedia Exercise book |
| 3 x 3 in standard note | Wikipedia Post-it; Target titles |
| 100 sheets typical; 30 to 400 per pad | Wikipedia Post-it (100 typical; 30, 45, 50, 70, 75, 90, 100, 400) |
| Standard Post-it at 50 or 100; Super Sticky at 30 to 90 | Target (two search reads plus the checker's feed read), with each count also in Wikipedia's list; attributed to Target in the JSON |
| Five Star "100 Pages" = 100 sheets | Target feed (checker, verbatim) and Target titles (my WebFetch re-read); Wikipedia Recto and verso for one sheet = two pages |
| Note sizes from under 2 in | Wikipedia (1 3/8 x 1 7/8, 2 x 2); Target (1.88 x 1.88) |

Removed during drafting for lack of a second source: Super Sticky "2X sticking power" (Target excerpt only); the 2003 and 2018 launch years (Wikipedia only; kept here); 11 x 11 in maximum note size; US ruling inch values; the 2 percent alkaline reserve; JIS B5 dimensions; Dingbats' gsm bleed bands.

## Dropped

- Pop-up (Z-fold) sticky notes as a dispenser-fit facet. Real (Wikipedia lists Pop-Up Notes; Target lists a dispenser), but no source gives which pads fit which dispensers.
- Numbered pages and an index page. On Leuchtturm and Apica copy, popular with bullet journalers by reputation, but I read no source on why it matters; left as high-signal markdown material.
- Perforation as its own facet; folded into the `pages come out` answer of binding. Reversed after the blind check: binding values cannot carry perforation, so `perforated` is now its own facet.
- Left-handedness as a split. One reviewer sentence ("Lefties beware") ties it to dry time; spiral-on-the-left complaints were not sourced. Kept inside paper_smoothness.
- Recycled content and FSC. Leuchtturm and Moleskine state FSC; no source ties it to performance; rule 5 cuts certifications.

## Trimmed detail

- **mechanism.plain** (154 words → under 150): "by capillary action" (Wikipedia, Sizing: "curbing the paper fibers' tendency to absorb liquids by capillary action"); "instead of soaking in".
- **adhesive_strength.why_experts_care**: launch years 2003 (Super Sticky) and 2018 (Extreme), Wikipedia only.
- **note_size.why_experts_care**: "from 1 3/8 x 1 7/8 in up to 11 by 11 in" replaced by "under 2 in to large sheets for walls" (the 11 x 11 bound is single-source).
- **ink_tolerance**: Dingbats' bleed bands (under 70 high risk, 70 to 80 moderate, 80 to 90 lower, 90 to 100 minimal, 100+ effectively bleed-proof; alcohol markers and watercolour need 120+) are a seller's rubric, so they stay here.
- **page_size**: Leuchtturm sells Pocket (A6), Medium (A5), Master Classic and Slim (A4+), Composition (B5) and Paperback (B6+).
- **ruling split rationale**: "Dot grid is the journaling default; lined is the school and office default" removed as unsourced.

## Corrections

1. `vertical`: the brief requires `"supplies"`, but the skill's schema.json enum is formulations, machines, fashion, furnishing, other. The delivered category.json carries `"supplies"` as briefed (matching the sibling runs `cookware-sets` and `correction-products`), so validate.py reports exactly one problem: `schema  vertical: 'supplies' is not one of ['formulations', 'machines', 'fashion', 'furnishing', 'other']`. A copy with `"vertical": "other"` validates clean ("14 facets, 4 splits, 30 references"). render.py and viewer.py accept "supplies" and ran clean. Scripts and schema were not modified; the fix belongs in the skill's schema enum.
2. Before delivery: the ballpoint branch no longer makes show_through irrelevant (see Branches); split 2 rationale no longer claims fountain pen writers lose "most mass-market" notebooks (unsourced); the ruling facet no longer says journaling buyers "favour" dot grid (unsourced); page_count typical range now attributes each figure to its brand.
3. STYLE checklist run on the JSON prose: first paragraph says what decides the choice; every vocabulary term is used after a gloss in the body (gsm in the summary, feathering and ghosting in the mechanism); each facet answers what, why, values, source in that order; generic test passed on all four splits (each question names category-specific values). No blind check has been run yet.
4. Viewer: `tests/viewer-browser.py` (Playwright, pyenv 3.11, offline) passed 6 of 6 on a temporary copy (1380, 768 and 390 px, light and dark). In the vertical-tree view the `bound notebook` leg shows "ask Pages, Page size" without Binding, because Binding is itself split 4 and the renderer folds it into the sequence; and, as the `correction-products` run recorded, the tree chains every leg into the next split in order, so the `sticky notes` leg still flows on to "4 · Binding" although the branch marks binding irrelevant. The timeline view and the step list in category.md show the branches as written. Recorded only; scripts out of scope.

## Corrections after the blind check (2026-10-25)

Source for each item: the blind check report `check.md` (fix numbers in brackets). Re-reads this session used WebFetch only; WebSearch was out of budget, curl to Target was refused, and the JetPens archive and WAD Tomoe River page did not load (archive unreachable, 403), so those items rest on the checker's verbatim reads.

1. [1] page_count: the backfill note no longer trusts the unit word; it checks the description for a sheet count, uses sheets x 2, and keeps title-only "pages" on mass-market school notebooks low confidence. Backfill cost raised from low to medium. why_experts_care now says some titles call sheets pages, citing Five Star at Target.
2. [1] price_per_page: why now warns that the metric is off by 2 where a title's "pages" are sheets and should be shown only for a checked count; target_fivestar added as evidence.
3. [1] judgement 3: claim and note now say some titles call sheets pages (Five Star "100 Pages" for 100 double-sided sheets). target_fivestar supports line and title rewritten; "(fetched excerpt)" dropped from the title, method kept in these notes.
4. [2] judgement 2: `sources_disagree` set to true and confidence lowered to medium; claim now says reviewers split on gel; note names Scribbright, Inkophile and the WAD commenter against Dingbats ("Gel pen 80gsm+"). "often bleeds" changed to "sometimes bleeds" (Scribbright "occasional").
5. [2] ink_tolerance: why now says reviewers split on gel; the backfill default "untested 70 gsm stock to gel safe at most" replaced with "never infer a tier from gsm alone"; dingbats_bleed and wad_moleskine added as evidence. Split 2 rationale "bleeds" changed to "sometimes bleeds".
6. [3] binding question rewritten from "Lies flat for good, folds back, or pages come out?" to "Bound like a book, folds back on itself, or rearrangeable pages?". The answers are now exclusive by construction (spine, coil, rings): bound like a book is sewn, stapled or glued; folds back is spiral or wire; rearrangeable is ring or disc. Tear-out moved to a new facet (item 7).
7. [3] New facet `perforated` (boolean, listing, low): Target titles carry "Perforated" (Five Star, re-read), Rhodia "micro-perforated sheets", Maruman "micro-perforation" (Urban Kit). A perforated spiral notebook is no longer excluded by the binding answer. Vocabulary term "Perforation" added. Add-on table row added.
8. [3] binding backfill: the "record glued" default for unknown hardcovers removed; such hardcovers stay unknown and still match "bound like a book". Hint now lists the Target title words (Spiral, Wire-Bound, Sewn Binding) and says to ignore "Glue Binding" on sticky-note titles. Glued pads and glued books are now separate: pads are handled by the new tear-off pad branch (item 13), books by "cheap glued books can shed sheets".
9. [4] Swapped split 4: page_size (narrows 4, cost 1) replaces binding (3, 2), following the register's own scores. page_size roles now include split; binding roles are filter only. New rationale and skip_if_known on the page_size split. Summary updated to name page size.
10. [5] sheets_per_pad: backfill cost none → low, with a per-pad versus pack-total disambiguation step. why and range restated: Wikipedia lists 30 to 400 and calls 100 typical; at Target, standard Post-it pads come at 50 or 100 and Super Sticky at 30 to 90 (attributed).
11. [5] price_per_note formula now uses the pack total when the title states one; why rewritten.
12. [5] Trade-off "adhesive_strength vs sheets_per_pad" removed, and its sentence removed from adhesive_strength: with standard pads at 50 and Super Sticky at 70 or 90 on Target, stronger adhesive does not reliably mean fewer sheets.
13. [13] format: new branch `when: tear-off pad` makes binding irrelevant (trace in Branches). Hint now states keyword order: drop easel and wall pads, then sticky keywords (sticky notes, Post-it, repositionable) win over "pad" and "notepad". why now says pads are "glued or wired" so every sheet tears off.
14. [6] paper_smoothness: the JetPens smear evidence removed from why, from the facet's evidence and from the jetpens_tomoe supports line (it concerns a brush pen lifting under Copic markers). why now ends "Dry time matters for wet inks; ballpoint paste dries quickly"; wiki_ballpoint added as evidence. Trade-off 3 cites penaddict_tomoe and wad_tomoe (sheen) instead of jetpens_tomoe.
15. [6, 7, 8] Split 2 branch `ballpoint and pencil only` → makes_irrelevant paper_smoothness removed (pencil writers choose tooth; the when value was ambiguous against "any tier"). Branches intro now says split 2 branches fire on the tier the answer maps to.
16. [7] Markers and highlighters: ink_tolerance why maps markers to fountain pen safe, labelled our call (Dingbats: "fountain pens and markers deposit significantly more"). The question is unchanged.
17. [7] Sketchbooks for dry media moved to `paper` in the scope line (paper owns art paper sold by weight and texture); blank notebooks stay here.
18. [7] ink_tolerance backfill now says to filter only where a value exists and show unknowns rather than drop them.
19. [8] Fountain-pen leg no longer asks paper_gsm; bound-notebook leg no longer asks page_size; ruling leg asks line_spacing_mm for `lined` only. line_spacing_mm why now says dots and squares sit at 5 mm on the books read, so spacing is asked for lined pages.
20. [9] line_spacing_mm backfill now labels the US wide, college and narrow values as Wikipedia-only and the converted values as approximate, and leaves Law Ruled spacing unknown. ruling backfill now maps Quad Ruled and Quadrille to grid and Law Ruled and Legal Ruled to lined; wiki_ruled_paper and target_fivestar added to ruling evidence. No second source for the inch values was found (search budget gone), so they stay out of the JSON.
21. [10] permanence: "most commercial paper is now alkaline" changed to "much commercial paper is acid-free" (Wikipedia: "Much commercially produced paper is acid-free"); same change in judgement 4's note and the wiki_acid_free supports line. Backfill cost medium → low, hint names "Acid-Free" in Five Star titles; target_fivestar added as evidence.
22. [11] Single-source figures labelled in the JSON: Dingbats 160 vs 100 gsm page counts ("by the seller's own figures, though the two papers also differ in coating"); Leuchtturm 6 mm, 251 pages and 145 by 210 mm ("one retailer's brand copy"); Super Sticky counts attributed to Target. Apica 6.5 mm removed from the JSON, since no second source exists and Leuchtturm already illustrates mm ruling.
23. [4, Part A] paper_gsm typical range: the synthesised "70 to 90 gsm" band replaced by the sourced points (Moleskine 70, Leuchtturm 80 with its 80 to 100 range, Rhodia 80 or 90, Dingbats 100 to 160, Tomoe River 52), so no range is invented.
24. [12] Scope: the schema has no scope field, so the out-of-scope list sits in format.backfill_note: easel and wall pads (flip-chart paper, `paper`, as `stationery-boards` also assigns it), leather and felt desk pads (`desk-mats`), dated desk pads (`planners`), loose filler and punched refill sheets (`paper`). note_size no longer says sizes run "to large sheets for walls"; it says Super Sticky easel and wall pads sit outside. Paper desk notepads stay here, as `desk-mats` expects.
25. [12] Scope line in these notes: notebook covers no longer sent to `ink-cartridges` (whose scope is pen refills only); no sibling claims empty covers, so they are flagged for the taxonomy owner rather than assigned. Bound refill inserts that work on their own stay here (matching `planners`, which treats a stitched refill as usable alone); loose punched refill sheets go to `paper`.
26. [14] dromgooles_catalog: supports now credit the 5 mm dots to Midori Yuru Log and drop the Midori Soft Color line, whose product is titled a "Ring Notebook"; title updated. The binding facet no longer rests on that glue example.
27. [14] wad_moleskine: supports now say the author's ghosting sentence concerns the Moleskine planner and Cahier (re-read, verbatim above); show_through why says the same and keeps Scribbright for the Classic.
28. [14] Tomoe River mill map: ink_tolerance backfill says Tomoegawa stopped making it, so later paper under the name needs its own test; judgement 5's note and a wad_tomoe supports line add the commenters' report that Sanzen took production over. That report rests on the checker's read; the WAD page returned 403 to me.
29. [15] `vertical` changed from "supplies" to "other", so validate.py passes. This supersedes Correction 1; the schema still lacks `supplies`, and most sibling supplies registers already use `other`.
30. Rejected or partly applied: none of the 15 fixes was rejected outright. [7] The checker offered a new answer for markers; I kept the question at three answers and stated the mapping instead, because a fourth pen answer would push the question past a quick menu and no source defines a marker tier. [12] "Put a scope line in category.json": there is no top-level field for it (`additionalProperties: false`), so it went into format.backfill_note. Part D's mass-market bleed-claim rule ("Perforated sheets fight ink bleed") was applied as "do not read generic copy as a tier".
31. Build: validate.py, render.py and viewer.py rerun after these changes (results below).