Methodology

This site makes claims about objects. This page explains on what basis, and where the limits of what we know actually fall.

The legend rubric

We do not guess at what counts as legendary. A model qualifies for the catalogue if it meets at least two of seven criteria, and every criterion marked as met cites a source that can be opened.

1. Longevity — at least 25 years of uninterrupted production in substantially unchanged construction.
2. Design recognition — presence in a design museum collection, or an award with an entry in the organiser's database.
3. Technical first — a solution that went on to become an industry standard.
4. Cultural footprint — a documented role outside stationery.
5. Scale — sales counted in the hundreds of millions.
6. Enthusiast canon — sustained presence in community recommendations.
7. Practice influence — a parts market, a modification culture, or institutional adoption.

The seventh criterion was added after the entire starting list had been researched. The reason was a specific model: the Zebra F-701 failed despite an authentic, documented modification culture, because the first six criteria measure only institutional recognition, scale and editorial attention. We assessed the criterion retroactively across every model, so that new entries would not receive a privilege the older ones never had.

The score is shown as a list of criteria with citations, not as a bare number. It is meant to be auditable: each criterion shows exactly what satisfies it, how strong the evidence is, and what it applies to.

The rule that changed the most

Evidence must concern this model and this writing class.

An award given to another variant is not the model's award. Sales of a whole series are not the model's sales. A museum object that is a fountain pen or a pencil is not recognition for a ballpoint.

We still display such evidence — in a separate box, marked as applying to the product line or the brand. It does not count toward the score. Without that distinction the catalogue would be a collection of other people's credit attached to objects that have none.

Examples from this site:

• MoMA's permanent collection holds the Fisher Bullet #400, not the AG7 — or so the manufacturer claims. The space evidence belongs to the AG7. Two different models, two different claims.
• The 1978 iF award "for the Lamy 2000" went to a desk set, not to the fountain pen or the ballpoint.
• The 2020 Good Design award for the Jetstream went to the Edge variant of 2019, not the base model of 2006.
• The billion EnerGels is a figure for the whole brand, not for the push-button barrel.

What we will not do with numbers

An annual figure, or a production rate multiplied by years, is our inference, not a source. The scale criterion requires a cumulative figure stated by a source.

The consequence is sometimes counterintuitive. We know that a million BIC 4 Couleurs leave the Montévrain factory each week. We may not multiply that by fifty-five years and present it as sales. The scale criterion stays unmet despite the product's obvious ubiquity.

The same rule removed the Pilot G-2 from the catalogue, despite it being the best-selling gel pen in the United States.

Confidence in dates

The confidence field on dates is mandatory and rendered. "1969 (approximate date)" is better than false precision.

We also separate design from manufacture. Many "Swiss" or "German" pens are designed in one country and assembled in another. For the Caran d'Ache 849 the Swiss Made marking covers both — and we explain what it means legally. For the Bic Cristal country of design and country of manufacture have parted entirely.

What is absent, and why

Prices. In any form — no amounts, no ranges, no "cheap" or "expensive". Prices go stale in days and would turn the site into a maintenance obligation impossible to keep. Instead we give a market tier: entry, mid, premium, collector. It is descriptive and stable over time.

Images without a licence. Every image requires a licence field and a source URL; the absence of either breaks the build. This is not a declaration but a schema — confirmed by a test. Attribution is rendered under the image, always visible, never in metadata.

We do not generate AI images posing as real products. We do not use manufacturers' logos: a brand name in text is permissible nominative use, a logo is a trademark.

Accounts, comments and star ratings. Not in this version of the site.

Models that are not here

We rejected more entries from the starting list than we accepted. Every rejection has a recorded justification and is reversible: an accession record, a manufacturer statement with a cumulative figure, or a second independent outlet restores a model to the catalogue.

A few cases worth naming:

• Tombow ZOOM 707 — the three design awards the assessment rested on have no records in the organisers' databases. Two supposedly independent reviews turned out to be one text by the same author under two addresses.
• rOtring 600 — the original German generation ran for about eleven years. The 600 line's canon belongs to the mechanical pencil, and reviews of the ballpoint concern the 2018 Japanese reissue, a different construction.
• Cross Classic Century — the presidential role belongs to the Townsend and the Century II according to Fortune. The claim about the Classic Century comes solely from the manufacturer's blog.
• Zenith 7 — the only candidate from Central Europe. Production was interrupted by an investor and the trademark passed to another company; today's pens are made under licence. Regional representation is not a reason to lower the evidentiary standard, so the catalogue remains without an entry from that region — and that is a finding, not an oversight.

One model came back: the Zebra F-701, after the seventh criterion was added.

Entries without a score

Some objects matter historically rather than by rubric. The Sakura Ball Sign 280 meets one criterion of seven, but it is the first commercially released gel pen — a catalogue holding three descendants without the original would tell the story from the middle.

Such entries are marked as artifacts and carry no numeric score. The criteria are assessed and shown exactly as elsewhere; they are simply not summed. Publishing the inventor of a category with a score of 1/7 would rank it below any pen with two reviews, and inflating the score would be bending the rubric.

Where to buy

Every retailer is checked by hand: we open the page and confirm that this model is on offer. Each link carries the date it was last checked, because honesty about the freshness of the data is the whole value here.

We do not list links to individual offers — auctions, classifieds, single listings. Such a link does not answer "where can I buy this", only "who was selling one unit on the day we checked".

Once a week an automated job checks every link. It distinguishes a server error from an absent resource: a 5xx response and an anti-bot barrier go into the inconclusive category rather than the dead one. A report that raises false alarms stops being read.

What we do not know

This list is part of the methodology, not a footnote to it.

Some museum collections and award databases block automated access. Where a record could not be opened, we say so plainly and do not treat inaccessibility as evidence either way. For the Lamy cp1 we give no Cooper Hewitt accession numbers because the collection site has been retired — an access limitation, not an absence of objects.

A manufacturer's claim without confirmation in an organiser's database remains a claim. Fisher's assertion that the Bullet is in MoMA has no support in the museum's public catalogue; we record it as unconfirmed, not as false.

Dimensions the manufacturer does not publish come from specialist retailers, and we say so on the model's page. Where data is missing, the field stays empty and the page shows "no data" — never a rounded number.

Who we did not admit

A threshold is a declaration until you can see who failed it. Candidates researched and rejected therefore have a page of their own with the rubric score, the reason for rejection and — for every entry — what would reopen the case. The content schema enforces this: an entry below the threshold cannot be saved without a way back.

The reserve is larger than the catalogue, and that is as it should be. Reversibility is not a rhetorical promise either: the Zebra F-701 was rejected and returned once the seventh criterion was added — the reserve shows what that earlier verdict looked like.